memory representation in the medial prefrontal cortex...the meaning of the stimulus...

1

Upload: others

Post on 27-Sep-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

Memory representation in the medial prefrontal cortex

Cortical communication, and the development of a prefrontal ensemble code for associative memory

by

Mark Morrissey

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Psychology

University of Toronto

© Copyright by Mark Morrissey 2016

Page 2: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

ii

Memory representation in the medial prefrontal cortex

Cortical communication, and the development of a prefrontal ensemble code for associative memory

Mark Morrissey

Doctor of Philosophy

Department of Psychology

University of Toronto

2016

Abstract

Despite knowledge of the critical importance of the medial prefrontal cortex (mPFC) in both the

consolidation and long-term retrieval of event memory in the brain, it is unclear which aspects of

a memory are represented within this region and how they change with learning and memory

consolidation. Likewise, how the mPFC interacts with other facets of the memory network to

shape these representations is also uncertain. In this thesis I employed electrophysiological

techniques to examine individual prelimbic mPFC (PrL) neurons selectivity for various task

features in an associative memory paradigm in rats across learning, consolidation and over-

training. Further, I recorded oscillatory activity from the rhinal cortices and ventral hippocampus

to examine how these regions may modulate mPFC memory representations across time. I reveal

similar proportions at all time-points of both neurons highly selective for specific task features,

and neurons exhibiting more generalized activity. Looking at population level activity across

time however revealed a shift from more discrete encoding during learning to more generalized

encoding post-consolidation. Further, PrL neurons whose firing patterns were modulated by

rhinal and ventral hippocampal theta oscillations uncover a potential source of this change in

encoding across time. Specifically, I show that discrete populations of PrL neurons display

phase-locked spiking to theta oscillations in the lateral entorhinal cortex, the perirhinal cortex,

and ventral hippocampus. Whereas the proportions of these different populations were not

observed to change across time, the information contained within their activity did. PrL neurons

Page 3: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

iii

phase-locked to rhinal theta oscillations were highly selective to task features early on in the

acquisition of the memory and became more generalized with learning and consolidation.

Neurons phase-locked to ventral hippocampal theta on the other hand became more selective for

the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing

theories of the mPFC role in long-term memory including schema development and indexing of

cortical memory activity. Taken together, these results demonstrate memory encoding patterns

within the prelimbic mPFC and highlight, through communication with several important nodes

of the memory network, potential mechanisms through which the encoding of different memory

features evolve across learning.

Page 4: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

iv

Acknowledgments

I would like to express my deepest appreciation to my supervisor, Dr. Kaori Takehara-Nishiuchi,

who gave me the opportunity to embark on a career in science, has provided numerous and

generous opportunities for my continual growth and has supported me immensely throughout my

graduate studies. Thank you for being such a wonderful mentor.

I would also like to thank my loving wife for all of her patience, support and encouragement, and

for pushing me to chase my dreams.

Page 5: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

v

Table of Contents

Acknowledgments.......................................................................................................................... iv

Table of Contents .............................................................................................................................v

List of Tables ................................................................................................................................. ix

List of Figures ..................................................................................................................................x

Chapter 1 ..........................................................................................................................................1

Introduction .................................................................................................................................1

1.1 Theories on network organization underlying episodic memory ........................................2

1.1.1 Hippocampal index theory .......................................................................................4

1.1.2 Theories of memory consolidation ..........................................................................8

1.2 Role of the medial prefrontal cortex in consolidated memory ..........................................11

1.2.1 Sub-regions of the medial prefrontal cortex ..........................................................12

1.2.2 Medial prefrontal cortex and consolidated memory ..............................................14

1.2.3 Activity of single prefrontal neurons .....................................................................19

1.2.4 Networks of long-term memory .............................................................................21

1.3 Dissertation Objectives ......................................................................................................27

1.3.1 Questions................................................................................................................27

1.3.2 In-vivo electrophysiology ......................................................................................28

1.3.3 Trace eyeblink conditioning ..................................................................................30

1.3.4 Specific aims ..........................................................................................................35

Chapter 2 ........................................................................................................................................37

Materials and Methods ..............................................................................................................37

2.1 Animals ..............................................................................................................................37

2.2 Electrode Construction.......................................................................................................37

Page 6: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

vi

2.2.1 Electrodes for local field potential recording .........................................................37

2.2.2 Electrodes for single unit recording .......................................................................37

2.3 Surgical Procedures ...........................................................................................................38

2.3.1 Local field potential electrode implantation ..........................................................38

2.3.2 Microdrive-array and tetrode implantation ............................................................39

2.4 Adjustment of tetrode locations .........................................................................................39

2.5 Data Acquisition ................................................................................................................39

2.5.1 Trace eyeblink conditioning ..................................................................................40

2.5.2 Context Selectivity .................................................................................................41

2.5.3 EMG, LFP, and Unit recording..............................................................................41

2.5.4 Schedule .................................................................................................................42

2.6 Data Analysis .....................................................................................................................42

2.6.1 Behaviour analysis .................................................................................................42

2.6.2 Spike sorting ..........................................................................................................44

2.6.3 Firing rate analysis .................................................................................................44

2.6.4 Mutual information ................................................................................................46

2.6.5 Population firing rate matrix correlation................................................................47

2.6.6 Support vector machine learning ...........................................................................48

2.6.7 Phase-locked firings ...............................................................................................50

2.7 Statistical Analyses ............................................................................................................51

2.8 Histology ............................................................................................................................52

Chapter 3 ........................................................................................................................................53

Prelimbic Representations of Associative Memory Throughout Learning and

Consolidation ............................................................................................................................53

3.1 Introduction ........................................................................................................................53

3.2 Results ................................................................................................................................55

Page 7: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

vii

3.2.1 Acquisition and expression of the conditioned response .......................................55

3.2.2 Single unit activity patterns ...................................................................................57

3.2.3 Changes in selectivity across time .........................................................................61

3.2.4 Population activity patterns....................................................................................65

3.2.5 Decoding Prelimbic population activity ................................................................69

3.2.6 Contextual encoding of a consolidated associative memory .................................74

3.3 Discussion ..........................................................................................................................79

3.3.1 Summary ................................................................................................................79

3.3.2 Single Neuron Selectivity ......................................................................................80

3.3.3 Population level encoding ......................................................................................80

3.3.4 Implications of the varying degrees and changes in selectivity observed .............81

Chapter 4 ........................................................................................................................................87

Prelimbic-cortical communication shaping changes in the selectivity of the prelimbic

ensemble code with consolidation ............................................................................................87

4.1 Introduction ........................................................................................................................87

4.2 Results ................................................................................................................................90

4.2.1 Prelimbic neuron phase-locking to the rhinal cortices and hippocampus ..............90

4.2.2 Prelimbic phase-locking is not feature specific .....................................................94

4.2.3 Regional phase-locking does not change across the learning stages .....................96

4.2.4 Decoding of region-specific phase-locking populations across time ....................97

4.3 Discussion ........................................................................................................................103

4.3.1 Summary ..............................................................................................................103

4.3.2 Functionally defined cell types in the prelimbic cortex .......................................104

4.3.3 Implications of the role of each afferent input during memory acquisition and

retrieval ................................................................................................................108

4.3.4 Differences in selectivity between functional cell-types .....................................110

Discussion ...............................................................................................................................113

Page 8: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

viii

5.1 Summary ..........................................................................................................................113

5.2 Conclusions ......................................................................................................................115

5.2.1 Role of medial prefrontal cortex in memory........................................................115

5.2.2 Future directions ..................................................................................................120

References ....................................................................................................................................123

Page 9: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

ix

List of Tables

Table 1. Correlations between population firing rates across conditions ..................................... 68

Page 10: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

x

List of Figures

Figure 1. Overview of hippocampal-cortical patterns of connectivity. .......................................... 8

Figure 3. Connectivity and Organization of a Circuit that may support Memory Consolidation

and Retrieval. ................................................................................................................................ 26

Figure 4. Experimental Procedure and Trace Eyeblink Conditioning. ......................................... 41

Figure 5. Acquisition and Expression of the Conditioned Eyeblink Response. ........................... 57

Figure 6. Histology and Single Unit Isolation. ............................................................................. 58

Figure 7. Single Neuron Selectivity for Task Features. ................................................................ 60

Figure 8. Behaviourally Defined Learning Stages. ....................................................................... 62

Figure 9. Selectivity Across Learning Stages. .............................................................................. 63

Figure 10. Mutual Information defined Selectivity. ..................................................................... 65

Figure 11. Prelimbic Neuron Population Matrices. ...................................................................... 67

Figure 12. Matrix Correlation Differences Between Relational and Identity Features. ............... 69

Figure 13. Support Vector Machine Classification Accuracy. ..................................................... 70

Figure 14. SVM Classification Accuracy changes for Stimulus Features Across Time. ............. 72

Figure 15. Context Manipulation Paradigm and Behavioural Results. ........................................ 75

Figure 16. Single Neuron and Population Selectivity for Contextual Features. ........................... 76

Figure 17. SVM decoding of Contextual Features. ...................................................................... 78

Figure 18. Histology and examples of prefrontal locking and the proportion of locked cells. .... 93

Figure 19. Prelimbic selectivity of phase-locking to specific regions and type of information

encoded by these cells. .................................................................................................................. 95

Page 11: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

xi

Figure 20. Phase-locking across learning stages........................................................................... 97

Figure 21. SVM Classification Accuracy for Stimulus Features Across Time as a Function of

Functional Cell Types. ................................................................................................................ 102

Page 12: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

1

Chapter 1

Introduction

The memory system of the brain consists of a network of brain regions whose complexity

parallels the complexity of the underlying memory itself. Simple forms of reflexive memory may

only require a few circuits processing inputs and directing outputs outside of conscious

awareness (Christian & Thompson, 2003; Martinez, 2014; McCormick & Thompson, 1984).

Compare this to a memory of a previous event combining a vast amount of largely separate

sensory inputs, together containing information about the place and context in which something

occurred, detailed representations of features and stimuli, and the timing and order of events in

which they happened (Eichenbaum, Sauvage, Fortin, Komorowski, & Lipton, 2012;

Eichenbaum, Yonelinas, & Ranganath, 2007; Knierim, Neunuebel, & Deshmukh, 2013; Squire,

1992). Untangling the networks involved in the formation, storage and retrieval of this complex

form of memory, referred to as the episodic memory system, has been a difficult challenge for

the study of the brain. It requires an understanding of how the brain processes information,

untangling the complex circuitry involved in combining this information together, and finally

decoding the vast computations taking place within the brain, enabling the phenomenon of

retrieval or ‘remembering’. Several brain regions have been identified as crucial aspects of this

network and the goal of my dissertation was to deepen our understanding of the functioning of

one such region. Whereas the hippocampus, since its discovery as a fundamental node in the

episodic memory network, has been extensively studied for its role in memory formation and

retrieval, a region of the frontal cortex, the medial prefrontal cortex and its various sub-regions,

has relatively only recently become the focus of comprehensive investigation (Frankland &

Bontempi, 2005; Insel & Takehara-Nishiuchi, 2013).

Page 13: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

2

One of several prominent theories of the neuronal networks of memory formation and storage posits

that initial plasticity in the hippocampus rapidly forms the association among elements of an event,

and through bidirectional pathways between the hippocampus and neocortex, the memory gradually

becomes reorganized for long-term storage somewhere within the neocortex, independent of the

hippocampus (Alvarez & Squire, 1994; McClelland, McNaughton, & O’Reilly, 1995). Growing

evidence suggests that an area of the frontal cortex, the medial prefrontal cortex (mPFC), may play a

pivotal role in the neocortical network that supports stable event memories (Frankland & Bontempi,

2005; Insel & Takehara-Nishiuchi, 2013).

The main objective of my project was to characterize the aspects of an event memory that are

represented by the mPFC network and to investigate interactions between the mPFC and other

cortical structures of the memory network. Specifically I attempted to characterize the specific

aspects of an event that are encoded by prelimbic mPFC neurons, how this encoding may change

across time, from before learning the association to after the consolidation of the memory, and if

these representations are shaped by interactions with the rhinal cortices and hippocampus.

1.1 Theories on network organization underlying episodic memory

The memory systems of the brain can be divided by the type of information they contain, the neural

circuits governing their representation, and whether they can be explicitly recalled (‘declared’) or if

their recall lies outside of conscious awareness (Tulving, 1987). A common example is learning to

ride a bike, the ability of which is a form of procedural or skill learning likely involving the basal

ganglia circuitry, motor cortex and cerebellum (Bédard & Sanes, 2014; Charlesworth, Warren, &

Brainard, 2012; Hikosaka, Nakamura, Sakai, & Nakahara, 2002; Wiestler & Diedrichsen, 2013). The

retrieval of this memory, as demonstrated by producing the action, is completely independent from

the memory of the episode in which the learning took place, where you were, who was with you,

Page 14: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

3

what happened. These latter memories are classified as part of the explicit, or declarative memory

system, comprising episodic or event memories and rely on the functioning of the medial temporal

lobes. We know these two types of memory depend on independent systems because patients with

damage to the medial temporal lobes can still learn and retrieve procedural or skill memories despite

the inability to recall ever having learned them (Milner, 1959; Milner, 2005). Such patients are

shown to have a selective impairment in the episodic memory system (Milner, 1959; Squire, 1992;

Tulving, 1987) and studies of these types of patients were the first to inspire the notion that the

medial temporal lobes may serve a function specific to the storage and retrieval of episodic or

declarative memories (Scoville & Milner, 1957). Research using animal models has since made

significant contributions to our understanding of how these structures accomplish this.

Rodents do not have a medial temporal lobe, however the same structures exist in similar anatomical

positions with similar anatomical connectivity and function (Amaral & Witter, 1995; Manns &

Eichenbaum, 2006; Squire, 1992). Referred to more generally as the hippocampal system, in rodents

(and in humans and primates) the main structures consist of the hippocampus (dentate gyrus, CA1,

CA2, CA3, and subiculum), the entorhinal cortex, postrhinal cortex (parahippocampal in primates

and humans) and perirhinal cortex (Amaral & Witter, 1995; Burwell, 2000). However because

rodents cannot fulfill a key requirement of the declarative memory system according to the classic

psychological definition, that is, they can’t declare something, it is often questioned whether they

really have this type of memory and serve as good models for its investigation (Tulving, 2001).

Rodents can however be tested for their memory of a prior event, and exhibit savings of what, when,

and where something took place (Eichenbaum et al., 2007; Knierim et al., 2013). Using this

definition there is ample evidence to support the notion that they do have what is commonly called

episodic-like memory or memory that is a model of episodic memory and that the encoding and

retrieval of this type of memory requires the hippocampal system, analogous to that in humans

(Aggleton & Pearce, 2001; Bunsey & Eichenbaum, 1996; Manns & Eichenbaum, 2006; Rolls, 2010;

Page 15: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

4

Squire, 1992; Weisz, Rios, & Argibay, 2012). For example, rodents can be observed to form very

accurate memories of a spatial layout (Moser et al., 2014; O’Keefe, 1983; Rolls, Stringer, &

Trappenberg, 2002), or of events that transpired in a particular context (LeDoux, 1995; Penick &

Solomom, 1991), the formation and retrieval of which is severely disrupted by hippocampal

dysfunction (Squire, 1992).

Though it has long been acknowledged that the medial temporal lobe (hippocampal system) are

involved in memory, the hippocampal formation has remained at the forefront of memory research

since the initial discovery that this specific region of the brain may be the seat of episodic memory

formation (Milner, 2005; Scoville & Milner, 1957; Squire, 2009). The hippocampus has since been

studied in exquisite detail, from its anatomy, connectivity and physiological properties to its

involvement in a variety of tasks (Manns & Eichenbaum, 2006; Morris, 2001; Squire, 1992). Cells

within the hippocampus have been revealed to encode locations (Eichenbaum, Stewart, & Morris,

1990; O’Keefe, 1983; Wilson & McNaughton, 1993), associations (Liu et al., 2012), objects

(Komorowski, Manns, & Eichenbaum, 2009), even time (MacDonald, Lepage, Eden, & Eichenbaum,

2011). It was within the hippocampus where researchers discovered place cells, neurons that fire in a

particular place in a particular environment, and are proposed to form mental representations of space

(O’Keefe, 1983). It was also in the hippocampus where researchers discovered engram cells, neurons

active during an episode whose specific inactivation prevents recall of that episode, and whose

specific activation triggers retrieval of the event (Liu et al., 2012; Ramirez, Tonegawa, & Liu, 2014).

1.1.1 Hippocampal index theory

At its core, the most widely accepted view of hippocampal encoding proposes that the hippocampus

forms an index of the cortical activity that was present during the actual experiencing of an event

(Teyler & DiScenna, 1986) and that the content of a memory is actually stored in these distributed

Page 16: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

5

cortical networks (Damasio, 1989). Recent advances in the neurobiological toolbox have lent strong

support to this theory, for example cells active during a memory encoding experience were tagged

and genetically driven to express a light controlled proton pump (ArchT). When animals underwent

retrieval of the memory, specific inhibition of CA1 cells active during memory encoding prevented

retrieval of the memory and importantly, prevented the activation of cells in the surrounding cortical

regions that were also active during encoding (Tanaka et al., 2014).

More recently, efforts have begun to shift some focus onto the cortices surrounding the hippocampus,

which serve as a key node between the hippocampus and the rest of the neocortex. These cortices

provide direct access from the neocortical sites presumed to store memory, to the hippocampal

indices presumed to encode their activity patterns. Cortical inputs into the hippocampus travel almost

exclusively through these extra-hippocampal regions, and likewise, outputs from the hippocampus

back to the cortex again traverse these regions (Burwell, 2000; Witter, 2007; Witter, Hoesen, &

Amaral, 1989). The extra-hippocampal structures, as they are often referred, consist of the perirhinal

cortex, the postrhinal cortex and the entorhinal cortex. Anatomically they can be observed to be

arranged in a hierarchical like system, with the hippocampus at the top and the entorhinal cortex right

below (see Figure 1). Together in the next level of the hierarchy are the perirhinal and postrhinal

cortices which receive a variety of largely segregated inputs from various cortical regions and in turn

largely project to separate regions of the entorhinal cortex, in addition to some direct projections to

the hippocampus (Burwell & Amaral, 1998b). The perforant and temporo-ammonic pathways, the

primary cortical projections into the hippocampus, originate from layers II and III, respectively, of

the entorhinal cortex. Cortical projections back out from the hippocampus largely originate in area

CA1 and the subiculum and target the deep layers of the entorhinal cortex, which in turn projects

back down the hierarchy and to the various cortical targets (Agster & Burwell, 2009; Burwell &

Amaral, 1998a; Canto, Wouterlood, & Witter, 2008). Although the precise contributions of these

surrounding cortical regions to memory formation and retrieval is still being elucidated, it is evident

Page 17: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

6

their presence is necessary for the formation and retrieval of many types of event memories (Burwell,

2004; Cho, Beracochea, & Jaffard, 1993; Cho & Jaffard, 1995; Herfurth, Kasper, Schwarz, Stefan, &

Pauli, 2010; Manes, Graham, Zeman, de Luján Calcagno, & Hodges, 2005; Suzuki, Zola-Morgan,

Squire, & Amaral, 1993; Thornton, Rothblat, & Murray, 1997; Zola-Morgan, Squire, Amaral, &

Suzuki, 1989).

As incoming information about the environment moves from the primary sensory cortices to the

association cortices greater degrees of sensory integration can be found, and it is believed to be the

hippocampus that is the highest point of integration. A nice example of this varying degree of

integration can be seen in the rodent object recognition task. The simplest form of the task, novel

object recognition, tests a rodent’s memory for a previously encountered object, taking advantage of

their preference for exploring a novel versus a previously encountered object. To successfully

discriminate between a new and previously explored object requires a stored representation of the

encountered object and integration of the various features of the two current objects to detect which

is novel and which has been experienced. Perirhinal cortex lesions severely impair this simple novel

object recognition, implying a certain degree of integration taking place prior to this information

arriving in the hippocampus (Ennaceur, Neave, & Aggleton, 1996; Mumby & Pinel, 1994). The task

can be made more difficult by using the same object but presenting it in different locations in an

environment or in different environmental contexts. This type of object-location or object-context

test requires the conjunctive representation of the object and either the location or environment in

which it was encountered. Whereas lesions of the entorhinal cortex do not affect the simpler novel

object recognition paradigm, they severely impair performance in novel object-location and novel

object-context tests, suggesting that this region is important for these more complex conjunctive

representations requiring a higher level of integration (Wilson, Langston, et al., 2013; Wilson,

Watanabe, Milner, & Ainge, 2013). Introducing long delays between exploring and testing

sessions,adding an element of temporal order or a representation of location or context to these tests

Page 18: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

7

requires the hippocampus for normal performance (Clark, West, Zola, & Squire, 2001; Clark, Zola,

& Squire, 2000; Mumby, Gaskin, Glenn, Schramek, & Lehmann, 2002)

An explanation for these findings can be found in the anatomical connectivity and following the

route of different types of information into the hippocampus. Following a more rudimentary view,

two primary streams of information can be observed, a what and a where stream. Information about

objects and events is sent to the perirhinal cortex from uni- and polymodal association cortices,

which then project most heavily to the lateral aspects of the entorhinal cortex (LEC) as part of the

what stream. On the other hand, visuospatial information originating in the visual cortices, the

retrosplenial cortex and the parietal cortex is sent to the postrhinal cortex, which then projects most

heavily to the medial aspects of the entorhinal cortex (MEC) as part of the where stream. Projections

from the LEC and MEC to the hippocampus converge on the same cells in the dentate gyrus and

CA3, and diverge onto separate cells in the CA1 of the hippocampus. Feedback from the

hippocampus to the entorhinal cortex and connections between the LEC and MEC allow integration

of this somewhat separate information (Knierim et al., 2013). Therefore the increasing levels of

integration of different types of information observed in the behavioural paradigms described above

are supported by the anatomical organization of the hippocampal system. The hippocampus,

receiving varying levels of integrated information, can ultimately place events within a spatio-

temporal contextual structure, linking events, context and related memories (Eichenbaum et al.,

2012; Komorowski et al., 2009).

Page 19: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

8

Figure 1. Overview of hippocampal-cortical patterns of connectivity.

A general overview of the flow of information into and out of the hippocampus from various neocortical sites. Lines between structures indicate connectivity, and the thickness of the line indicates the strength of connections. HPC hippocampus; LEC lateral entorhinal cortex; MEC medial entorhinal cortex; PER perirhinal cortex; POR postrhinal cortex; mPFC medial prefrontal cortex; Tev Ventral temporal association area; AUD/p/v primary, ventral, posterior auditory areal; Ald/v/p dorsal, ventral, posterior agranular insular cortex; gustatory cortex; VIC visceral area; RSPd dorsal retrosplenial cortex; RSPv ventral retrosplenial cortex; PRLp posterior parietal association cortex; SSs supplementary somatosensory area; SSp primary somatosensory areal; VISl lateral visual cortex; VISm medial visual cortex; VISp primary visual cortex. Adapted from Agster and Burwell (2009).

1.1.2 Theories of memory consolidation

The idea that the hippocampus may not always be involved in the retrieval of a hippocampal

dependent memory goes back as far as the original evidence that the hippocampus was crucial for

memory formation (Scoville & Milner, 1957). It had been reported that patients with medial temporal

lobe amnesia showed evidence of preserved memories for information acquired long-before the

Page 20: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

9

disruption (Milner, 1959; Scoville & Milner, 1957). This led to the theory that over-time, memories

undergo a process of consolidation, wherein connections between cortical regions in which the

memory is presumed to preside are strengthened, and the role of the hippocampus to initiate the

retrieval of the memory decreases (Alvarez & Squire, 1994; McClelland et al., 1995). This is known

as the standard model of consolidation, and it has remained a contentious area of brain memory

research. Patient studies have often produced conflicting reports on the state of a person’s memory

for previous events after medial temporal lobe damage, even in the same patients (see Bayley, Gold,

Hopkins, & Squire, 2005; Moscovitch & Nadel, 1998; Nadel & Moscovitch, 1997; Squire, 1992). In

animals, the presence of similarly conflicting findings is evident, with many studies revealing a clear

temporal gradient after disruption of the hippocampal system (Cho & Jaffard, 1995; Kim, Clark, &

Thompson, 1995; Kim & Fanselow, 1992; Winocur, 1990; S. M. Zola-Morgan & Squire, 1990) and

others displaying a flat gradient (Bolhuis, Stewart, & Forrest, 1994; D. G. Mumby, Astur, Weisend,

& Sutherland, 1999; Sutherland et al., 2001). In both mice and rats, lesions of the hippocampus

impairing memory retrieval immediately after learning, when produced at 3-4 weeks after learning

have been observed to spare memory in contextual fear conditioning (Frankland et al., 2006), context

discrimination (Szu-Han Wang, Teixeira, Wheeler, & Frankland, 2009), spatial five-arm maze task

(Maviel, Durkin, Menzaghi, & Bontempi, 2004), trace eyeblink conditioning (Kim et al., 1995; Kaori

Takehara, Kawahara, & Kirino, 2003), trace fear conditioning (Quinn, Ma, Tinsley, Koch, &

Fanselow, 2008) and paired-associate memory (Dorothy Tse et al., 2007). In contrast, in many tests

requiring spatial navigation, like the Morris Water Maze, impairments are typically observed after

hippocampal lesions regardless of how long prior the memory was formed (Broadbent, Squire, &

Clark, 2006; Clark, Broadbent, & Squire, 2005; Gervais, Barrett-Bernstein, Sutherland, & Mumby,

2014; Winocur, Sekeres, Binns, & Moscovitch, 2013)

Such conflicts led to an alternative explanation of the findings, the multiple trace theory proposed

that the hippocampus would always be required for the retrieval of a memory that required the

Page 21: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

10

hippocampus for its formation. It was proposed that the re-activation and re-experiencing of these

memories creates additional indices in the hippocampus. Therefore, the more indices within the

hippocampus, the greater the probability that a trace of this memory will survive hippocampal

disruption, explaining some of the conflicting findings in the literature (Nadel & Moscovitch, 1997).

It also suggested that the creation of multiple indices, each providing contextual information for each

episode, allows for the neocortical extraction of the abstract or overlapping features of these episodes

independent of context. This suggests that abstract and semantic information could be retrieved

without the aid of the hippocampus, whereas discreet, contextual or autobiographical information

would always require the hippocampus for successful retrieval (Nadel & Moscovitch, 1997). Later,

the transformation theory extended this postulate by suggesting that a cortical gist-like or abstract

memory and a hippocampal detailed contextual memory dynamically interact, and depending on the

strength and retrieval circumstance, the dominance of one over the other can change (Winocur,

Moscovitch, & Bontempi, 2010). Many studies support this view (see Moscovitch et al., 2005) but

conflicting findings are also present; for example mice were observed to be able to successfully

discriminate between two contexts without a hippocampus 42 days after the formation of the

memory, but not 1 day after (Szu-Han Wang et al., 2009).

The nature of the time-limited role of the hippocampus and the nature of what memories are like

without a hippocampus is still fairly uncertain. However, what does seem clear, is that certain

memories can be retrieved independent of the hippocampus that was once essential for their

formation, and that this process requires both time and cortical plasticity (Frankland & Bontempi,

2005). Regardless of the nature of the long-term memory, both theories suggest cortical restructuring

supports the consolidation of memory and it is these changes that permit the retrieval of the memory

without input from the hippocampus (Alvarez & Squire, 1994; McClelland et al., 1995; Nadel &

Moscovitch, 1997; Winocur et al., 2010). Indeed, cortical plasticity over the period presumed to

encompass these changes has been shown to be necessary for the successful retrieval of remote

Page 22: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

11

memory. For example, CamKII mutant mice that displayed intact hippocampal LTP but impaired

cortical LTP were able to learn and retrieve a memory shortly after learning, but retrieval of the

memory several weeks after learning was significantly impaired (Frankland, O’Brien, Ohno,

Kirkwood, & Silva, 2001).

1.2 Role of the medial prefrontal cortex in consolidated memory

In one of the first examples of specific brain regions involved in remote memory retrieval, an

intensive study mapped the networks involved in remote memory retrieval in mice using (14C)2-

deoxyglucose to measure regional levels of glucose metabolism during memory retrieval in a spatial

discrimination task either 5-days or 25-days after acquisition of the task (Bontempi, Laurent-Demir,

Destrade, & Jaffard, 1999). The study confirmed decreased activity of the hippocampus at the 25-day

retrieval time-point and also identified several regions that unexpectedly showed increased activity at

this time-point compared to retrieval 5 days after learning. Increased activity was observed at the 25-

day (remote) time-point in the frontal cortex, the temporal cortex, and the anterior cingulate cortex

(Bontempi et al., 1999). The increased recruitment of frontal regions for older memories had also

been reported in human memory experiments, particularly the medial prefrontal cortex (Fink et al.,

1996; Markowitsch, 1995; Takashima et al., 2006).

At this time it was beginning to appear that the mPFC was a playing a unique role in memory

retrieval, as earlier interest in the mPFC had linked its function to short-term memory maintenance

(Kolb, 1984) and also in the formation of memory (Buckner, Kelley, & Petersen, 1999; Chachich &

Powell, 1998; Kronforst-Collins & Disterhoft, 1998; Meunier, Jaffard, & Destrade, 1991; Weible,

Weiss, & Disterhoft, 2003). This led to an analysis of the necessity of the mPFC (consisting of the

anterior cingulate AC, prelimbic PrL, and infralimbic IL cortices) for memory retrieval at different

time-points in trace eyeblink conditioning (Takehara et al., 2003). These experiments revealed little

Page 23: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

12

effect of mPFC lesions shortly after memory acquisition, but when the mPFC was lesioned several

weeks after learning, severe memory impairments were observed (Kaori Takehara et al., 2003).

Shortly after these findings increased immediate early gene activity (Fos & Zif268) was observed

within mPFC sub-regions during remote memory retrieval (Frankland, 2004; Maviel et al., 2004). In

addition, targeted reversible inaction of these sub-regions revealed specific involvement of the AC in

contextual fear conditioning (Frankland, 2004), the PrL cortex in trace eyeblink conditioning

(Takehara-Nishiuchi, Nakao, Kawahara, Matsuki, & Kirino, 2006), and the prelimbic and AC in the

spatial five-arm maze (Maviel et al., 2004) for remote but not recent memory retrieval. Similar

findings have since been observed in the Morris Water Maze (Lopez et al., 2012; Teixeira, Pomedli,

Maei, Kee, & Frankland, 2006), trace fear conditioning (Quinn et al., 2008) and paired-associate

memory (Wang, Tse, & Morris, 2012). Together, these findings suggest that over the course of

several weeks after learning a change in the memory network responsible for the retrieval of the

memory undergoes a shift. This implies several key points, 1) that there is likely dynamic

communication taking place between regions of the neocortex and the hippocampus throughout this

consolidation process (Maviel et al., 2004), 2) that the mPFC, particularly the PrL and AC appear to

play an enhanced role in older more remote memories, and 3) that mPFC-hippocampal interactions

may be important for this shift (Frankland & Bontempi, 2005).

1.2.1 Sub-regions of the medial prefrontal cortex

The medial prefrontal cortex (mPFC) of the rodent typically refers to a set of structures comprising

the rostral cingulate cortex (see Figure 2). This encompasses the most anterior aspects of the dorsal

anterior cingulate cortex (dAC), the prelimbic cortex (PrL), and the infralimbic cortex (IL). The area

immediately posterior can be classified as the caudal cingulate cortex and contains the caudal aspects

of the dorsal anterior cingulate, and the ventral anterior cingulate (vAC). Finally, the posterior

aspects of the cingulate consist of the dorsal and ventral retrosplenial cortices (dRS & vRS). Regions

Page 24: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

13

of the mPFC are divided based on cytoarchitectural and morphological differences (Jones & Witter,

2007), differences in efferent and afferent projections (Hoover & Vertes, 2007; B. F. Jones & Witter,

2007; Vertes, 2004), and differential effects of their disruption on behaviour (Baldi & Bucherelli,

2015). For example PrL but not IL disruption impairs trace fear conditioning memory formation and

reconsolidation, whereas IL but not PrL disruption impairs the extinction of this memory (reviewed

in Baldi & Bucherelli, 2015). Though there are clear differences in patterns of inputs and outputs of

the sub-regions of the mPFC, there are also many overlapping similarities (Hoover & Vertes, 2007;

Jones & Witter, 2007). For example both the IL and PrL send projections to the amygdala, but the IL

targets the medial, basomedial, cortical and central nuclei of the amygdala, whereas the PrL only

projects to the central and basolateral amygdala (Vertes, 2004). Both regions send fairly dense

bilateral projections to the midline thalamus, including the paratenial, paraventricular,

medial/intermedial dorsal nucleus, interanteromedial nucleus, central medial nucleus, and nucleus

reuniens. Similarities also exist in cortical projections, both the PrL and IL project to the anterior

piriform cortex, the orbitomedial, insular cortex, and the entorhinal cortices (Vertes, 2004).

The mPFC also communicates strongly with the hippocampus. A direct monosynaptic unidirectional

pathway exists between the ventral hippocampus and mPFC (Hoover & Vertes, 2007; Vertes, 2004)

and a very recent study identified a direct pathway from the AC to the dorsal hippocampus that was

shown to have the capacity to drive memory retrieval (Rajasethupathy et al., 2015). The mPFC is

also strongly connected with the rhinal cortices, particularly the entorhinal cortex (EC) and perirhinal

cortex (PER) (Jones & Witter, 2007). The AC, PrL, and IL project primarily to the lateral aspects of

the EC, and to the perirhinal cortex, with very weak connectivity with the medial aspects of the EC

and the postrhinal cortex. The remaining structures of the caudal anterior cingulate and posterior

cingulate on the other hand show the inverse, with much stronger connections with the medial

entorhinal and postrhinal cortices than with the lateral and perirhinal cortices, implying their

differential involvement in the two primary streams of information what and where discussed earlier

Page 25: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

14

(Jones & Witter, 2007). These connections and interactions with the hippocampal system are

proposed to be key to the process of consolidation, but the specific interactions driving this process

are still not clear (Frankland & Bontempi, 2005; Insel & Takehara-Nishiuchi, 2013; Preston &

Eichenbaum, 2013).

Figure 2. Flat-map of the Cingulate cortex Defining Sub-regions.

Illustration of a flat-map of the cingulate cortex cytoarchitectonic sub-divisions based on findings of Jones and Witter (2007). Colours designate common divisions of the medial prefrontal (pink) caudal anterior cingulate (purple) and retrosplenial cortices (blue). Abbreviations: vAC, ventral anterior cingulate cortex; dAC, dorsal anterior cingulate cortex; fx, fornix; hipp, hippocampus; IL, infralimbic cortex; PrL, prelimbic cortex; vRS, ventral retrosplenial cortex (RSv-a, RSvb: parts a and b respectively); dRS, dorsal retrosplenial cortex. Figure adapted from Jones and Witter (2007), Insel and Takehara-Nishiuchi (2012).

1.2.2 Medial prefrontal cortex and consolidated memory

Investigations into a shift in the memory network with consolidation has confirmed that plasticity

and cortical restructuring is taking place within the mPFC during a several week period after

learning, and that these changes rely on an intact hippocampus (Restivo, Vetere, Bontempi, &

Ammassari-Teule, 2009). Importantly, these changes have also been observed to be necessary for

successful memory retrieval at remote time-points (Takehara-Nishiuchi et al., 2006; Vetere et al.,

Page 26: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

15

2011). In trace eyeblink conditioning, NMDA dependent plasticity mechanisms specifically within

the PrL cortex of the mPFC were shown to be essential for remote retrieval of memory during a two-

week period immediately after learning, but not after this time-point (Takehara-Nishiuchi et al.,

2006). Subsequently it was shown that dendritic spine growth (spinogenesis) within the AC shortly

after learning but not 6 weeks after learning was essential to remote memory retrieval in contextual

fear conditioning (Vetere et al., 2011). These findings reveal that not only are structural changes

taking place within mPFC sub-regions during the period presumed to be when consolidation of the

memory takes place, but that these changes are necessary for the formation of a long-term

representation of the memory in a state that can be successfully retrieved.

Within the mPFC, there are also many overlapping effects of lesions and inactivation of specific sub-

regions, likely because of the similar connection patterns just described. Studies of the effects of

lesions or inactivation of the mPFC on behaviour have produced somewhat variable results, likely a

result of differences in task features, compensation from spared mPFC regions, and inconsistencies

and imprecision in sub-region targeting (reviewed in Cassady 2014). This is confounded with the fact

that the mPFC seems to be important in many different tasks. Electrophysiological studies are

similarly confounded, as single neurons in the mPFC are observed to respond to many different

features in many different tasks, to varying degrees (Rigotti et al., 2013).

Differences between mPFC sub-regions in their role in the consolidation of memory and its long-

term retrieval has also remained fairly nebulous. For example, the necessity of the PrL but not the

AC region for remote memory in trace eyeblink conditioning has been demonstrated (Oswald,

Maddox, Tisdale, & Powell, 2010; Takehara-Nishiuchi et al., 2005, 2006). In contextual fear

memory, the AC but not PrL, was found to be important for remote memory retrieval (Frankland,

2004) and in a spatial memory task both PrL and AC inactivation produced impairments in remote

memory retrieval (Maviel et al., 2004). In trace fear conditioning, non-specific lesions of the mPFC

Page 27: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

16

impair remote memory retrieval (Beeman, Bauer, Pierson, & Quinn, 2013; Quinn et al., 2008).

Disruption of the PrL specifically impairs remote retrieval in trace fear conditioning (Runyan,

Moore, & Dash, 2004) though to my knowledge the specific contributions of the AC and IL for

remote retrieval in this task has not been directly assessed. The relative contribution of these sub-

regions though still somewhat unclear, seem to be task specific, and the AC and PrL in particular

appear to be essential to the consolidation process.

The idea that a specific region of the brain would acquire a role in remote memory retrieval was not

originally predicted by theories of memory consolidation (Alvarez & Squire, 1994; McClelland et

al., 1995). Under this original view, cortico-cortical connections of the networks representing the

memory are slowly strengthened over time (McClelland et al., 1995). However the idea of the mPFC

becoming an essential node in this remote memory network suggests a revision to this model. It has

been proposed that at this more remote time-point of a memories representation in the brain that the

mPFC may be fulfilling a similar function as to that ascribed to the hippocampus for memory

formation and initial retrieval according to hippocampal index theory (Frankland & Bontempi, 2005;

Takehara-Nishiuchi & McNaughton, 2008).

An example of the benefit of such a system was recently demonstrated in studies of schema memory.

The original theory describing a rapid hippocampal driven memory acquisition system and a slower

neocortical consolidation system argued that the advantage of this system was to prevent the

‘catastrophic interference’ of existing information represented in neocortical synaptic connections if

new information was suddenly added to this neocortical network (McClelland, McNaughton, &

O'Reilly, 1995). However this idea was recently challenged in experiments on schema learning. A

schema is a framework of knowledge or information upon which new learning can be built (Bartlett

& Bartlett, 1995; Piaget, 1929). Rats trained to associate different flavours with the different

locations of rewards in an event arena developed a schema like memory structure (Tse et al., 2007).

Page 28: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

17

Learning and retrieval of memories in this task required the hippocampus but after a typical

consolidation phase were shown to be retrievable without an intact hippocampus. When rats were

trained with a new cue-location pair that was similar to an existing consolidated schema, this new

information was observed to undergo a rapid consolidation process, in that the hippocampus, while

required for its acquisition, was only necessary for the retrieval of the memory for up to 2 day after

learning (Dorothy Tse et al., 2007). This finding suggested that new similar information can be added

rapidly to an existing memory framework. A recent extension of the original complementary learning

system theory (McClelland et al., 1995) shows that the rapid incorporation of similar information to

an existing memory framework is compatible with the original theory and does not result in

catastrophic interference (McClelland, 2013). Interestingly, this rapid incorporation was further

shown to depend on the mPFC (Tse et al., 2011; Wang et al., 2012). Therefore, this points to the

mPFC as a structure that is important for the long-term retrieval of memories and plays an important

role in building schemas or incorporating new information into existing knowledge structures

(Preston & Eichenbaum, 2013). This also points to the potential necessity of such a structure for the

long-term maintenance of memories, due to the sheer size and weak interconnectivity among

neocortical regions, for stored memories to be efficiently cued, retrieved, separated and updated,

likely requires the presence in this network of a brain region that can serve these functions over the

long-term.

This adds to growing evidence that suggests that newly formed memories are not merely transferred

to the cortex, but instead become integrated into existing neocortical memory networks or schemas,

through the modification and reorganization of these networks (McKenzie et al., 2014; Szu-Han

Wang & Morris, 2010a). If the mPFC becomes the central node in these networks, possibly retaining

an index code of neocortical memory patterns and building and updating schematic frameworks, the

question remains, is the reactivation of this now mPFC guided memory or schema the same as the

original hippocampal guided memory? Many have argued that it is not (Winocur & Moscovitch,

Page 29: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

18

2011; Winocur et al., 2010). Whether the contents of a schema include only the abstract features of

common elements among integrated events (van Kesteren, Ruiter, Fernández, & Henson, 2012) or

whether a schema can contain detailed features specific to discrete events represented within a

related event network is not exactly clear (Preston & Eichenbaum, 2013). The mPFC has often been

linked to categorical, or rule based encoding (Durstewitz, Vittoz, Floresco, & Seamans, 2010; Rich &

Shapiro, 2009; Wallis, Anderson, & Miller, 2001). This may be accomplished by a more semantic-

like encoding, or extracting consistencies across events with the more distinctive aspects of each

event being lost (Richards et al., 2014).

On the other hand, the development of organized mental representations that share the properties of

schemas have been observed that contain details of individual events. For example, the Tse et al.

(2007) study showed that prior learning facilitates subsequent memory for details of new events that

fit existing knowledge, the incorporation and retrieval of which depends on the mPFC. Similarly,

mice trained in the Morris water maze where the platform locations were systematically changed

across training sessions were better able to incorporate representations of new distinct platform

locations once the original schematic framework was given enough time to develop (Richards et al.,

2014). Again this incorporation relied on processing within the mPFC (Richards et al., 2014). These

examples demonstrate that new information is not simply absorbed into the existing information in

an abstract manner but that details specific enough to solve a spatial maze task are preserved.

Additionally, the phenomenon of reconsolidation suggests that memories can be modified and

updated throughout their lifetime (Nader, 2015; K. Nader, Schafe, & Le Doux, 2000). In

reconsolidation, a previously consolidated memory, through its retrieval, enters a state of lability

which requires another round of consolidation to be restabilized. It has been suggested that

reconsolidation may reflect the reorganization of a schema to incorporate new information

(McKenzie & Eichenbaum, 2011). This opens up the possibility of memory updating, possibly

through prefrontal-hippocampal interactions to support, build and adapt overlapping mental

Page 30: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

19

representations which contain enough details of specific events to detect conflicts between new and

existing information (McKenzie et al., 2014; Preston & Eichenbaum, 2013; Richards et al., 2014).

Trying to understand what features of a memory are represented by the mPFC and how consolidation

may change these representations were a main goal of my project. These previous investigations

examine memory representation indirectly based on behavioural observation. Here I will

systematically examine the level of specificity and abstraction of the prefrontal memory code directly

through the use of electrophysiological recordings.

1.2.3 Activity of single prefrontal neurons

One common overlapping feature across studies of the role of the mPFC sub-regions in memory

retrieval highlights its necessity when information must be bridged across a temporal delay. For

example, associating two discontiguous stimuli in trace conditioning, the PrL cortex is typically

seen to be important (Fuster, 2001; Gilmartin, Miyawaki, Helmstetter, & Diba, 2013). The PrL

region has been linked to working memory processes, and electrophysiological recordings from

this region reveal that PrL neurons will display persistent firing between two temporally

separated stimuli (Gilmartin & McEchron, 2005; Takehara-Nishiuchi & McNaughton, 2008). A

similar phenomenon is observed in neurons located in the primate dorsolateral prefrontal cortex

(dlPFC) during a delay period (Fuster, 1973). It is suggested that the PrL cortex of the rodent

which is more similar to the primate ventromedial pfrefrontal cortex (vmPFC) from an

anatomical standpoint, shares many similarities in function with the primate dlPFC and may be

viewed as a more primitive version of this primate frontal region (Seamans, Lapish, &

Durstewitz, 2008)

In trace fear conditioning acquisition, PrL neurons are observed to increase firing to the CS with

learning and exhibit persistent responses throughout the trace interval, IL neurons on the other

Page 31: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

20

hand show learning related decreases in activity to the stimuli (Gilmartin & McEchron, 2005).

PrL neurons show persistent firing in in trace eyeblink conditioning as well, with the strength of

their activity and selectivity for the memory association gradually increasing across learning and

consolidation (Hattori, Yoon, Disterhoft, & Weiss, 2014; Takehara-Nishiuchi & McNaughton,

2008). Interestingly, the changes observed in PrL firing overlaps with the timing of severe

impairments of prelimbic inactivation on retrieval (Takehara-Nishiuchi et al., 2005) and with the

period during which NMDA receptor activity is necessary for successful long-term retrieval

(Takehara-Nishiuchi et al., 2006).

In addition to the association selective activity just described, single neurons in the mPFC have

been observed to encode objects within an environment (Weible, Rowland, Pang, & Kentros,

2009) and also to the place in which an object was previously presented (Weible, Rowland,

Monaghan, Wolfgang, & Kentros, 2012). In the novel-object recognition task, some single

anterior cingulate neurons that were responsive to an object in an environment, when placed

back in the environment in the object’s absence, continued to fire in the location where the object

had been. This effect increased with familiarization with the object and was still evident one-

month later (Weible et al., 2012), suggesting that these neurons also maintain long-term traces of

experience within environments.

In primates, medial prefrontal neurons have been shown to respond to many different stimuli in

many different tasks (di Pellegrino & Wise, 1993; Fuster, 1973; Fuster, Bodner, & Kroger, 2000;

Rao, Rainer, & Miller, 1997; Watanabe, 1996). Single prefrontal neurons have been shown to

encode concrete rules between specific stimuli and specific responses (White & Wise, 1999),

abstract rules (Wallis et al., 2001), categorical representations (Freedman, Riesenhuber, Poggio,

& Miller, 2001), and response strategies (Genovesio, Brasted, Mitz, & Wise, 2005). In many

Page 32: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

21

cases this has been linked to a proposed role for this region in flexible cognition and decision

making. Likewise in rats, prelimbic mPFC neurons have been shown to show activity selective

for behavioural strategies, displaying activity dependent on the strategy used to solve a task

despite identical behaviour (Rich & Shapiro, 2009). Such encoding may be related to the

proposed abstraction of medial prefrontal memory representations and schema formation. In

these examples, many prefrontal neurons encode task relevant features in an abstract, categorical,

or rule-based manner, but others display stimulus selective activity, and some even show highly

selective conjunctive encoding (Wallis et al., 2001). In addition, all of these studies record from

animals well-trained with the stimuli and features of the employed task. It is not clear therefore,

if these abstract representations develop with repeated experience or if these generalized

representations are simply an intrinsic property of medial prefrontal encoding.

1.2.4 Networks of long-term memory

Besides unraveling the features of memory encoded by the mPFC and better understanding the state

of consolidated memory, another big question involves how the mPFC interacts with the

hippocampus and other regions of the episodic memory network to form and consolidate these

memories. The pathway between the hippocampus and prefrontal cortex is well known and proposed

to be the main driver behind the consolidation of hippocampal dependent memories (Frankland &

Bontempi, 2005; Preston & Eichenbaum, 2013; Simons & Spiers, 2003; Takehara-Nishiuchi, 2014).

However, how interactions between the prefrontal cortex, the hippocampus, and the

extrahippocampal structures determines this process is still uncertain. It is these interactions in

particular that have interested researchers and are proposed to be a possible mechanisms through

which hippocampal-prefrontal interactions supporting memory consolidation may occur (Insel &

Takehara-Nishiuchi, 2013; Preston & Eichenbaum, 2013; Takehara-Nishiuchi, 2014).

Page 33: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

22

1.2.4.1 Entorhinal Cortex

The entorhinal cortex (EC) sits as the primary interface between the hippocampus and the neocortex,

receiving a vast amount of cortical inputs, partially through strong connectivity with the perirhinal

and postrhinal cortices and also directly from an array or mono and polymodal association and

sensory cortices (Rebecca D. Burwell & Amaral, 1998a; Insausti, Herrero, & Witter, 1997).

Projections from many neocortical areas arrive in the superficial layers of the EC, this information is

then transferred into the hippocampus through direct connections from these same layers to the

dentate gyrus/CA3 and also the CA1. Projections from the CA1 and subiculum to the deep layers of

the EC in turn maintain the primary output pathway from the hippocampus back to the neocortex

(Kerr, Agster, Furtak, & Burwell, 2007; Witter, 2007; Witter et al., 1989). The EC is also

reciprocally connected with the mPFC, maintaining the same projection patterns as the entorhinal-

hippocampal connections. Projections to the mPFC originate in the superficial layers of the EC,

whereas output from the mPFC to the EC targets the deep layers of the EC (Apergis-Schoute, Pinto,

& Pare, 2006; Hoover & Vertes, 2007). Thus, anatomically the EC appears to be in a prime position

to interact with both the hippocampus and mPFC. Numerous findings demonstrate the role of the

entorhinal cortex in memory acquisition and retrieval (reviewed in Buzsáki & Moser, 2013;

Coutureau & Di Scala, 2009; Morrissey & Takehara-Nishiuchi, 2014). It was also noted that in

patients in which medial temporal lobe damage extends into the rhinal cortices much more extensive

retrograde amnesia is often observed (Bayley et al., 2005; S. Zola-Morgan et al., 1989). Similarly,

extended temporal gradients have been observed after entorhinal lesions in rodents (Cho et al., 1993;

Remondes & Schuman, 2004).

Our own work looked at the role of the entorhinal cortex in memory retrieval focusing on the

functions of entorhinal subdivisions. Based on structural anatomy and patterns of connectivity the EC

can be divided into medial (the more caudomedial aspects, MEC), and lateral (the rostrolateral

aspects, LEC) sub-regions (Canto et al., 2008). As briefly described earlier, the major inputs into the

Page 34: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

23

EC provide a major source of division with the more dense inputs arriving from the nearby perirhinal

(PER) and postrhinal (POR) cortices. The POR maintains connections most heavily with visual and

spatial cortical areas such as the visual, posterior parietal and retrosplenial cortices and primarily

targets the medial portions of the entorhinal cortex. Whereas the PER tends to receive inputs from a

wide range of association areas such as the piriform, insular, cingulate, frontal, and temporal cortices,

and primarily targets the lateral portions of the entorhinal cortex (Rebecca D. Burwell & Amaral,

1998a). These anatomical patterns of connectivity can also be seen to continue as a topological

arrangement of connections between the MEC, LEC and the hippocampal formation (Canto et al.,

2008; Hargreaves, 2005).

These projection patterns have led to the suggestions that these sub-regions of the EC may serve

functionally distinct roles in hippocampal dependent memory (Hargreaves, Rao, Lee, & Knierim,

2005; Knierim et al., 2013). The most striking division between the MEC and LEC from a functional

standpoint comes from electrophysiological recordings during exploration in rodents where neurons

located within the most caudodorsal regions of the MEC are observed to fire in a grid-like pattern

within a spatial environment (grid cells) (Hafting, Fyhn, Molden, Moser, & Moser, 2005).

Conversely a near absence of such spatially responsive cells is observed in the LEC (Hargreaves et

al., 2005). These results suggest that the MEC but not the LEC, contributes to spatial memory

representation, they also raise a possibility that the LEC, but not MEC, would contribute to non-

spatial memory. To address this point, we assessed the relative contributions of the LEC and MEC to

the retrieval of both recent and remote memory in trace eyeblink conditioning, a non-spatial

associative memory paradigm (Morrissey, Maal-Bared, Brady, & Takehara-Nishiuchi, 2012). Here

we saw that LEC but not MEC inactivation impaired memory retrieval both 1-day and 4-weeks after

learning (Morrissey et al., 2012). It is thought that the LEC may play an important role in bridging

the gap between the CS and US, as persistent firing to a stimulus has been observed as an intrinsic

property of cells in this region (Tahvildari, Fransén, Alonso, & Hasselmo, 2007). In addition, lateral

Page 35: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

24

entorhinal lesions were not observed to impact trace eyeblink acquisition with a shorter trace free

interval (Suter, Weiss, & Disterhoft, 2013). Using a longer trace interval, investigators showed that

LEC inactivation prior to conditioning impeded the acquisition of the association (Tanninen et al.,

2015).

These findings point to a role for the LEC in associative memory retrieval prior to and after the

consolidation of this memory, supporting the idea of the entorhinal cortex as an intermediary

between the hippocampus and mPFC cortex. Interestingly, EC-prefrontal connectivity patterns show

that the PrL is strongly connected with the LEC (and the perirhinal cortex), and sparsely to the MEC

(Jones & Witter, 2007). Simultaneous local field potential recordings from the hippocampus, LEC

and PrL during memory acquisition and consolidation in associative memory also reveal differences

in communication patterns across learning and remote retrieval (Takehara-Nishiuchi, Maal-Bared, &

Morrissey, 2012). Learning correlated synchronized oscillations in the theta frequency between the

LEC and hippocampus were high in the initial days of learning and decreased with retention. In

contrast, theta synchrony between the LEC and PrL was observed to gradually increase over-time,

remaining high during post-consolidation retrieval (Takehara-Nishiuchi et al., 2012). It has been

suggested therefore, that a key aspect of memory consolidation may be network modifications

between these circuits, with the entorhinal cortex providing neocortical access to the hippocampus

for memory formation and to the mPFC for the consolidation and long-term retrieval of these

memories (Insel & Takehara-Nishiuchi, 2013; Takehara-Nishiuchi, 2014).

1.2.4.2 Perirhinal cortex

The perirhinal cortex (PER) provides the LEC with the most dense input of all cortical regions

(Kerr et al., 2007) and like the LEC is also reciprocally connected with the prelimbic mPFC.

Fewer studies have investigated the role of the PER in memory consolidation, however plasticity

in and communication between the PER and mPFC was shown to be necessary for long-term

Page 36: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

25

memory retrieval in object recognition memory (Barker & Warburton, 2008). In addition,

disruption of the PER was also shown to impair fear context memory up to 100 days after

training, suggesting that similar to the LEC, the PER may continue to play an active role in the

retrieval of consolidated memory (Burwell, 2004). The perirhinal cortex (PER) has also recently

been observed to be important in trace eyeblink memory acquisition (Suter et al., 2013).

Perirhinal interactions with the mPFC through electrophysiological recordings suggest the

communication between these regions is important in memory formation (Hannesson, Howland,

& Phillips, 2004; Paz, Bauer, & Pare, 2007). The role of the PER in the long-term memory

network is not exactly clear, however the current evidence suggests it may play an important

function.

1.2.4.3 Ventral hippocampus

Another important pathway that has suggested to be important for the role of the mPFC in memory

retrieval is the direct mPFC projection observed from the ventral hippocampus (vHPC) and

subiculum (Hoover & Vertes, 2007; Jones & Witter, 2007). vHPC-mPFC synchrony has been

observed in anxiety evoking situations (Adhikari, Topiwala, & Gordon, 2010) and in spatial memory

encoding (Spellman et al., 2015). The vHPC has been shown to be involved in acquisition and

retrieval of trace and contextual fear memory (Gilmartin, Kwapis, & Helmstetter, 2012; Raybuck &

Gould, 2010; Wang et al., 2013; Zhu et al., 2014). Few studies have directly examined the role of the

ventral hippocampus in long-term memory, but it was observed that trace fear conditioning memory

retrieval was not affected by ventral hippocampal lesions, but contextual fear memory retrieval

seemed to be similarly disrupted at both recent and remote time-points (Beeman et al., 2013). The

vHPC and the vHPC-mPFC interaction was also shown to be important for contextual fear memory

generalization. At remote memory time-points, fear memory generalization was related to mPFC as

well as vHPC activity, and disrupting the function of these regions impaired contextual

Page 37: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

26

generalization (Cullen, Gilman, Winiecki, Riccio, & Jasnow, 2015). This finding is supported by

recordings from vHPC neurons, which in contrast to neurons in the dorsal hippocampus, gradually

develop context selective activity that forms a more generalized representation of contextual meaning

(Komorowski et al., 2013). Theta synchrony between the mPFC and dorsal hippocampus, a well

established property of hippocampal-prefrontal interactions(Adhikari et al., 2010;Jones & Wilson,

2005), has also been revealed to be modulated by ventral hippocampal activity (Adhikari, Topiwala,

& Gordon, 2011). Therefore it has been proposed that these ventral hippocampal-mPFC interactions

may support prefrontal dependent memory by building schema-like contextual representations

(Preston & Eichenbaum, 2013).

Figure 3. Connectivity and Organization of a Circuit that may support Memory

Consolidation and Retrieval.

Illustration of the connectivity of regions involved in memory formation, recent and remote memory retrieval linking the medial prefrontal cortex (mPFC), the hippocampal system and the various sensory and association cortices.

Page 38: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

27

1.3 Dissertation Objectives

1.3.1 Questions

A strong case is being made for the importance of the mPFC in the consolidation of event memory

and as a crucial node in the long-term memory network (Frankland & Bontempi, 2005; Insel &

Takehara-Nishiuchi, 2013; Preston & Eichenbaum, 2013; Simons & Spiers, 2003). Uncovering what

the mPFC is encoding to permit the successful retrieval and consolidation of episodic memory will

form an important piece of our deepening understanding of the networks of memory in the brain.

Within the prelimbic cortex in particular, memory selective activity has been shown to gradually

develop over the course of learning and consolidation, remaining high during remote memory

retrieval (Hattori et al., 2014; Takehara-Nishiuchi & McNaughton, 2008). However, the following

questions remain.

1) What features of the memory are these cells encoding?

2) How does their encoding for particular features change over the course of learning and

consolidation?

3) How does their encoding of past situations influence their encoding of a similar new

situation?

4) How is their encoding shaped (or modulated) by incoming inputs from the hippocampal

system over the course of learning and consolidation ?

To address these questions, I have conducted a series of experiments in which I applied in vivo

electrophysiological approaches to a trace eyeblink conditioning paradigm in rats.

Page 39: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

28

1.3.2 In-vivo electrophysiology

Electrophysiological techniques offer a unique opportunity to study brain functions by

monitoring actual computations taking place within and between brain networks in line with the

temporal scale in which they occur. Placing very small electrodes into a brain regions of interest

allows for the monitoring of extracellular potential changes within the vicinity of electrode tips.

This technique, the monitoring of extracellular activity, can capture the activity of neurons in two

forms. The first type is highly dependent on electrode size, by using a wire thin enough that the

tip can sit between small groups of cells, action potentials (spikes or firings) of single neurons

can be detected. The second type detects the oscillatory activity of neuron populations, typically

the larger the electrode the larger the population. Action potentials reflect a neuron’s output, and

their frequency and temporal pattern allow for decoding types of information that each neuron

represents and computes. By comparing the frequency and timing of action potentials across

conditions in which task features are systematically manipulated, I will be able to decode in my

paradigm the particular task features a particular neuron is responsive to.

Oscillatory activity, on the other hand, reflects fluctuations in the excitability of local neurons

(that is, fluctuations in membrane potentials), thereby affecting the timing of local neurons’

firing (i.e., their output) as well as their sensitivity to incoming inputs. The oscillatory activity

can be decomposed to two dimensions, amplitude and phase. The amplitude of the oscillation

reflects the local synchronization of large populations of neurons , whereas the phase of the

oscillation, reflects the degree of excitability of the neurons, closer to or further away from

threshold, and influences the spike timing of the cells in the network ( Elbert, 1987). The

oscillatory activity is useful to probe over all activity of a recorded region as well as to probe

communication between distant brain regions. In particular, the correlated fluctuation of

Page 40: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

29

oscillatory amplitude (coherence) and the synchronization of oscillatory phase (phase

synchronization) between two regions are proposed to indicate temporal coupling of neuronal

processing by bringing these regions into the depolarization phase at the same time (Fell &

Axmacher, 2011). As the timing of a neuron spiking in one region can be modulated by the

particular phase of an oscillation in another region, neuron spiking can be locked to the excitable

phase of afferent inputs. Here the particular phase of an ongoing oscillations in one region can

affect the spiking of neurons in an afferent region. This phase-locking of neurons to oscillations

or spike-field coupling, has been a proposed mechanism for the flexible routing of neuronal

signals (Burchell et al., 1998), cell assembly formation (Singer, 1999), and coding (Konig et al.,

1995).

Oscillatory activity, measured as local field potentials (LFPs), can be decomposed to oscillations

at different frequency bands. Previous studies have assigned different functions to oscillations at

specific frequency bands. Among them, theta rhythms, 4-12 Hz oscillations, are prominent

within the hippocampus, have been consistently implicated in complex behaviours requiring

mnemonic processing in rodents, primates, and humans (Kahana, Sekuler, Caplan, Kirschen, &

Madsen, 1999; Lee, Simpson, Logothetis, & Rainer, 2005; O’Keefe & Recce, 1993; Skaggs,

McNaughton, Wilson, & Barnes, 1996; Vanderwolf, 1969) and are within the frequency range

proposed to synchronize areas across large distances (Buzsáki, 2002). Theta synchrony between

regions, when two regions’ ongoing theta oscillations become locked, has also been observed

between the hippocampus and many connected regions like the entorhinal cortex (Frank, Brown,

& Wilson, 2001), striatum (Berke, Okatan, Skurski, & Eichenbaum, 2004), amygdala

(Seidenbecher, Laxmi, Stork, & Pape, 2003), cerebellum (Hoffmann & Berry, 2009; Wikgren,

Nokia, & Penttonen, 2010) and prefrontal cortex (Anderson, Rajagovindan, Ghacibeh, Meador,

& Ding, 2010; Jones & Wilson, 2005). In trace eyeblink conditioning, 7-11 Hz synchronized

Page 41: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

30

theta oscillations between the hippocampus and the lateral entorhinal cortex and prelimbic cortex

revealed a network shift relevant to memory acquisition and expression (Takehara-Nishiuchi et

al., 2012). Additionally, prelimbic neurons have been observed to display phase-locked activity

to theta oscillations in both the dorsal (Siapas, Lubenov, & Wilson, 2005) and ventral (Adhikari

et al., 2010, 2011) hippocampus. Here I focused on the phase-locking of prelimbic neurons to the

ongoing theta oscillations (7-11 Hz) from several connected regions of the hippocampal system.

The electrophysiological approach is advantageous for my purposes for several reasons.

Importantly, with these data, I can observe the types of features single neurons in the prelimbic

medial prefrontal cortex may be encoding by actually monitoring their responses in real-time. By

sampling the activity of several hundred individual medial prefrontal neurons at different time-

points of learning, consolidation and retrieval in a memory task, I can obtain a snapshot of

memory relevant representations within these cells activity and observe how they may change

across these time-periods in the same animals. I can also relate this activity to the simultaneous

oscillations in several different brain regions to assess patterns of communication and how they

may support or be affected by memory consolidation.

1.3.3 Trace eyeblink conditioning

Classical eyeblink conditioning is a well-established model of associative memory. As a model it has

several advantageous features for the purposes of my study from both a technical and functional

standpoint. Eyeblink conditioning involves the repeated pairing of a conditioned stimulus (CS; e.g.

tone) with stimulation of the eyelid (unconditioned stimulus, US; e.g. airpuff or electrical

stimulation). Gradually over repeated pairings, presentation of the CS triggers the subject to produce

an adaptive eyeblink response (conditioned response, CR) in anticipation of the upcoming US. Delay

eyeblink conditioning consists of the paired presentation of an overlapping CS and US: the US-onset

Page 42: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

31

is slightly delayed from the CS-onset but the two co-terminate. The neural circuitry driving this

simple form of Pavlovian conditioning has been described in exquisite detail, requiring the brainstem

and the cerebellum for the timing of the CR and association of the CS and US (reviewed in Freeman

& Steinmetz, 2011; Thompson et al., 1997). However, the introduction of a small stimulus free trace

interval (several hundred milliseconds) in between the CS and US drastically changes the neural

circuitry involved in linking the two together (Christian & Thompson, 2003; Krupa, Thompson, &

Thompson, 1993; Woodruff-Pak & Disterhoft, 2008). In trace eyeblink conditioning (TEBC), as it is

known, making the CS and US temporally discontiguous requires the recruitment of higher order

brain structures to form the association. In this paradigm the brain must maintain a memory ‘trace’ of

the CS during this CS-US interval to enable the association of the CS with the upcoming US.

The hippocampus has been shown to be essential for the formation of TEBC, as hippocampal lesions

severely disrupt acquisition of the association between the CS and US in the trace paradigm (Beylin

et al., 2001; Moyer, Deyo, & Disterhoft, 1990; Solomon & Vander Schaaf, 1986; Weiss,

Bouwmeester, Power, & Disterhoft, 1999). Hippocampal lesions also severely impair the behavioural

expression of an already acquired association between the CS and US, provided the lesion is

produced shortly after acquisition (Kim et al., 1995; Takehara et al., 2003; Takehara, Kawahara,

Takatsuki, & Kirino, 2002). Similarly, when the lesion is produced immediately after acquisition, the

animal is also unable to reacquire the association (Takehara et al., 2003). It believed that the

hippocampus is essential in binding these temporally discontinuous stimuli, and it has been

demonstrated that the severity in impairment after hippocampal lesions is dependent upon the length

of the temporal interval between the CS and US, with greater impairments observed with longer

intervals (Tseng, Guan, Disterhoft, & Weiss, 2004; Walker & Steinmetz, 2008).

Among hippocampus-dependent memory paradigms used in rodents, TEBC is particularly

suitable for neurophysiological investigation and has many advantages as a model to study the neural

Page 43: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

32

mechanisms of learning and memory encoding in the brain. For one, it is not easily acquired in

rodents, requiring on average 6+ days with 100 trials per day to achieve asymptotic learning with

responses in over 60% of trials. This is advantageous for electrophysiological investigation as it

greatly increases the sample size compared to trace fear conditioning for example, in which learning

can take place in one session with only a few trials. Additionally, TEBC also offers the opportunity

to observe subtle changes in the encoding of the memory as the association is gradually acquired,

compared with memory paradigms in which memory formation occurs very rapidly. These

advantages are also afforded by the fact that TEBC contains precise time-stamped events (CS-

presentation) to easily relate neural responses to. In fact, many studies have successfully used this

paradigm to uncover neuronal representation in the hippocampus formed during and after trace

eyeblink conditioning (McEchron, Weible, & Disterhoft, 2001; McEchron & Disterhoft, 1997;

Munera, Gruart, Munoz, Fernandez-Mas, & Delgado-Garcia, 2001; Weible et al., 2006; Weiss,

Kronforst-Collins, & Disterhoft, 1996).

Furthermore, this paradigm can be directly applicable to human subjects. Importantly, in

humans, the acquisition of the association in TEBC also relies on proper hippocampal functioning

(McGlinchey-Berroth, Carrillo, Gabrieli, Brawn, & Disterhoft, 1997; Woodruff-Pak, 1993). Another

key feature in humans is that the successful acquisition of the trace procedure depends on an

awareness of the temporal relationship between the CS and US, awareness being an analogue of

conscious knowledge and a key defining feature of declarative or episodic memory (Clark & Squire,

1999). Considering this evidence trace eyeblink conditioning in rodents has been suggested to be a

good model of episodic memory and therefore useful to dissect the neural circuitry and

representations of episodic memory in the brain (Woodruff-Pak & Disterhoft, 2008).

In addition to the hippocampus, the number of forebrain regions are involved in memory

acquisition and later expression in trace eyeblink conditioning. These include the rhinal cortices

Page 44: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

33

(Ryou et al., 2001; Suter et al., 2013; Tanninen et al., in press), various sensory cortices (Galvez,

Weible, & Disterhoft, 2007; Steinmetz, Harmon, & Freeman, 2013), the caudate nucleus (Flores

& Disterhoft, 2009 & 2013), and several thalamic nuclei (Oswald, Knuckley, Maddox, &

Powell, 2007; Powell & Churchwell, 2002). The medial prefrontal cortex is also involved in TEBC

memory formation, as its disruption can slow acquisition of the association (Kronforst-Collins &

Disterhoft, 1998; McLaughlin, Skaggs, Churchwell, & Powell, 2002; Takehara-Nishiuchi et al.,

2005; Weible, McEchron, & Disterhoft, 2000), and it has been used to differentiate functional roles

among many prefrontal regions. For example, in contrast to arrested acquisition after mPFC lesions,

lesions of the caudal anterior cingulate (cAC) are shown to severely disrupt TEBC acquisition

(Kronforst-Collins & Disterhoft, 1998; Weible et al., 2000). The importance of several of these

frontal regions in this task have also been shown to be memory stage dependent. Lesions to the

anterior portions of the anterior cingulate (dAC) and the prelimbic mPFC (PrL) which only mildly

disrupt the speed of acquisition, severely impair expression of the acquired memory, particularly

when the damage is induced several days to several weeks after learning (Oswald, Maddox, &

Powell, 2008; Oswald et al., 2010; Takehara-Nishiuchi et al., 2006). Interestingly, lesions to cAC

shown to severely disrupt acquisition, had no impairment when performed one-week after learning

(Oswald et al., 2010).

Single neuron recordings from these regions in TEBC have further advanced our understanding of

their functioning in this task, as reviewed in Chapter 1.2.3. Neurons located within the cAC show

strong firing changes to the CS in TEBC during the initial acquisition sessions, however this

activity is observed to decrease with learning (Hattori et al., 2014; Weible et al., 2003). Firing

patterns from neurons located within the rostral dAC and PrL on the other hand, show strong

changes to the CS that is maintained across the trace interval (Hattori et al., 2014; Siegel,

Kalmbach, Chitwood, & Mauk, 2012; Siegel & Mauk, 2013; Takehara-Nishiuchi &

Page 45: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

34

McNaughton, 2008). Such persistent activity continues to the onset of the US across the temporal

gap in which no stimuli are presented and is observed to become stronger as the CR is acquired

and consolidated (Hattori et al., 2014; Siegel et al., 2012; Siegel & Mauk, 2013; Takehara-

Nishiuchi & McNaughton, 2008). Prelimbic neurons on the other hand show persistent responses

during the trace interval between the CS and US in trace eyeblink conditioning, and the

magnitude of this response increases and becomes more selective to the memory association over

the course of the consolidation period (Hattori et al., 2014; Takehara-Nishiuchi & McNaughton,

2008), very similar to the changes observed in the dorsal hippocampus during the initial

acquisition of the memory (Weible, O’Reilly, Weiss, & Disterhoft, 2006).

In a similar study also using trace eyeblink conditioning, neurons located in the prelimbic, rostral

dorsal anterior cingulate (rAC) and caudal dorsal anterior cingulate (cAC) were recorded

simultaneously (Hattori et al., 2014). Clear differences emerged across time in the response

profiles of the neurons from the different regions. Prelimbic neurons, consistent with Takehara &

McNaughton (2008), gradually acquired sustained firing rate changes during the trace interval,

presumably bridging the gap between the CS and US. In contrast, cAC neurons displayed strong

robust responses to the CS early in learning, but these tended to fade with learning and

consolidation. Neurons in the rAC displayed decreases in responsivity within learning sessions,

something not observed in the PrL or cAC. The authors suggest that in this task, the different

regions of the mPFC may work together during the formation and retrieval of the memory, with

the rAC providing salience signals to ready the animal and the rest of the mPFC network for the

association to be formed, the cAC driving attention and stimulus selection, and PrL cells in this

task may be encoding the temporal or associative features of the memory eventually becoming

the key node in retrieval (Hattori et al., 2014). In my project I focus specifically on the prelimbic

Page 46: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

35

region of the mPFC based on the previous findings in the trace eyeblink conditioning task, the

same task employed here.

1.3.4 Specific aims

To examine which features of a memory prelimbic mPFC cells are encoding, I have conducted

multiple analyses of single prelimbic neuron activity as well as their population activity with two

specific aims. The first aim was to identify which aspects of the memory these neurons are selective

for and how these representations evolve across the learning and consolidation phases of event

memory (Chapter 3). I compared firing changes across the conditions in terms of the number of

selective neurons as well as the degree of selectivity in each neuron. I also applied correlation

analyses and multivariate analyses on firings of neuron populations to quantify the degree of

selectivity of population firing patterns. These analyses revealed that the prelimbic ensemble consists

of neurons with a range of selectivity, some highly selective for task features while others more

generalized for stimulus meaning, and a similar proportion of each was observed at all time-points

studied. The information encoded within the population of neurons recorded, however, did appear to

change across time, from highly selective to discrete features early in learning, to more generalized

response patterns post consolidation.

The second aim was to investigate how interactions with other cortical regions involved in

this memory may shape these prelimbic representations, and if these interactions encourage changes

in prelimbic representations throughout learning and consolidation (Chapter 4). To address this I

additionally recorded local field potential activity from the lateral entorhinal cortex, perirhinal cortex,

and ventral hippocampus simultaneously with the single prelimbic recordings during our memory

paradigm. I was then able to examine whether prelimbic neurons firing was modulated by oscillatory

activity in these different brain regions, whether this changed across the different stages of the

memory, and whether particular regions support different types of prelimbic selectivity. These

Page 47: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

36

analyses revealed that fairly distinct populations of prelimbic neurons display event related activity

that is modulated by the local oscillatory activity in each of the targeted regions. Again the

proportions of these populations did not change across the course of the learning paradigm, but their

population activity revealed changes in the information encoded by these neurons depending on their

preferred contact and the state of the memory across the phases of learning and consolidation.

Specifically, neurons modulated by lateral entorhinal, perirhinal, or ventral hippocampal theta

oscillations became less selective for discrete stimulus features across time, but neurons modulated

by the ventral hippocampus in particular became more selective for the association over-time.

Collectively I aimed to and revealed prelimbic mPFC encoding of memory features, expanding

knowledge of how this region may support consolidation and post-consolidation retrieval and how

other cortical regions may factor in to this role by monitoring actual computations in the neocortex.

Page 48: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

37

Chapter 2

Materials and Methods

2.1 Animals

All experiments were performed on male Long-Evans rats (Charles River Laboratories, St. Constant,

QC, Canada) between 16-25 weeks old at the time of surgery. A total of 15 rats underwent the

experimental procedure, however 9 had to be removed from the final analysis due to various technical

and mechanical issues. Rats were housed individually in Plexiglass cages and maintained on a reversed

12-hour light/dark cycle. Water and food was available ad libitum. All methods were approved by the

Animal Care and Use Committee at the University of Toronto.

2.2 Electrode Construction

2.2.1 Electrodes for local field potential recording

Electrodes were made in-house using Teflon-coated stainless steel wire (AM Systems) and 26 gauge

stainless steel cannula (Amazon Supply) and tailored for each brain target. Two wires were inserted

into the stainless steel cannula offset in the dorsal-ventral axis by .45-.7 mm depending on the target

and secured with cyanoacrylate.

2.2.2 Electrodes for single unit recording

Tetrodes were made in-house by twisting together four 12 μm polyimide coated nichrome wires

(Sandvik). To permit independently adjustable tetrode depths, each tetrode was housed inside a screw

operated microdrive. The complete Microdrive-array consisted of a bundle of 12 microdrives, each

guiding a tetrode, contained within a 3d printed plastic base (Kloosterman et al., 2009). The

Microdrive-array also enclosed the Electrode Interface Board (EIB-54-Kopf, Neuralynx) to which all

Page 49: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

38

electrodes were connected and served as the interface between the recording and stimulating electrodes

and the recording system. Prior to implantation, the impedance of the nichrome tetrode wires were

reduced to ~250 khOms by electroplating them with gold. Tetrodes were then drawn inside a 1.8 mm

diameter stainless steel cannula at the Micro-drive array base and a small drop of sterilized mineral oil

was added to ensure smooth movement of the tetrodes after implantation.

2.3 Surgical Procedures

Following guidelines set by the Institutional Animal Care Committee at the University of Toronto, all

surgeries were conducted under aseptic conditions in a sterile surgical suite. For the chronic

implantation of electrodes and the Microdrive Array, rats were anaesthetized under isoflurane (1–

1.5% by volume in oxygen at a flow rate of 1.5 L/min; Holocarbon Laboratories, River Edge, NJ)

and placed in a stereotaxic holder with the skull surface in the horizontal plane. For post-operative

analgesia rats were given a subcutaneous injection of ketoprofen (1.5mg/kg). The shaved head was

cleaned with 70% ethanol followed by Betadine and the skull surface exposed with an incision in the

anterior-posterior direction along the midline with a scalpel blade.

2.3.1 Local field potential electrode implantation

To chronically implant electrodes for long-term LFP recordings small burr holes were opened in the

skull using a micro-drill (.6mm) to target the lateral entorhinal cortex (LEC), perirhinal cortex (PER)

and ventral hippocampus (vHPC) at the following coordinates relative to bregma: for LEC, 6.45 mm

posterior, 5.4 mm lateral; for PER, 6.0 mm posterior/ 6.65 mm lateral; for vHPC, 5.15 mm posterior/

4.8 mm lateral. Several more burr holes were created around the skull to hold stainless steel screws

(Morris Screw) to anchor the implantations. For each brain region the electrodes were carefully

lowered into the brain at each drill site. LEC electrodes were lowered at a 10° lateral angle to 9.95

mm ventral to bregma, PER electrodes lowered straight to 7.75 mm ventral to bregma, and vHPC

electrodes were lowered to 8.45 mm ventral to bregma. Electrodes were held in place using self-

Page 50: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

39

curing dental acrylic. For referencing of local field potential activity, a stainless steel screw wired to

the electrode interface board was implanted in the surface of the cerebellum.

2.3.2 Microdrive-array and tetrode implantation

All tetrodes were targeted to the prelimbic medial prefrontal cortex (PrL mPFC). To implant the

tetrode bundle, a craniotomy was opened over the PrL mPFC at 3.2 mm anterior and 1.4 mm lateral

to bregma and the dura matter removed. The Micro-drive array was then lowered at a 9.5° medial

angle until the base made contact with the surface of the brain. The craniotomy was then sealed with

Kwik-Sil and the array was held in place with self-curing dental acrylic.

2.4 Adjustment of tetrode locations

Immediately after the surgery, all tetrodes were lowered 1 mm into the brain. For the next 3-4 weeks

the rat was connected to the system each day to visualize the quality of activity and monitor

movement of the tetrodes. Each tetrode was lowered slightly each day (75-125 μm) over the course

of this 3-4 week period to target tetrodes tips to the PrL mPFC at 3.0-4.0 mm ventral from the brain

surface. This procedure increases the yield of neurons and stability of the signal during the recording.

One tetrode was positioned superficially in the cortex (1 mm below brain surface) to serve as a

reference electrode for single-unit activity. Once the recordings began, tetrode position was adjusted

only as necessary to obtain good quality high yield recordings. It is therefore possible that recordings

were repeated for some neurons across days. Tetrode position adjustments were only made after a

given recording session providing ~ 24 hours for the tetrode to stabilize before the next recording,

2.5 Data Acquisition

All rats experienced the same general experimental procedure. Beginning 3-4 weeks following

Micro-drive array implantation, when stable single unit recordings were achieved and tetrodes were

Page 51: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

40

positioned within the PrL mPFC, rats were subjected to daily conditioning in the trace eyeblink

paradigm.

2.5.1 Trace eyeblink conditioning

Rats were placed in a large dark rectangular box, fitted with an LED light source and speaker. Within

the box rats were enclosed in a square plexiglass container, fitted with holes on one side to enable

sound-waves from the speaker to enter the enclosure. The conditioned stimulus (CS) was presented

for 100-msec and consisted of an auditory stimulus (85-dB, 2.5-kHz tone) or a visual stimulus (white

LED light blinking at 50 Hz). The unconditioned stimulus (US) was a mild electrical shock to the

eyelid (100 Hz square pulse, 0.3-2.0 mA, US), and the intensity carefully monitored via webcam and

adjusted to ensure a proper eyeblink/headturn response.

Conditioning took place over two sessions, each with 100 trials, separated by a 10 minute rest period.

Each session consisted of 20 presentations of the CS alone, followed by 80 trials in which the CS is

paired with the US, separated by a stimulus-free interval of 500 msec (see Figure 4). The first session

used only one of the two CS (e.g. tone), and the second session used the other CS (e.g. light), with

the CS order and schedule pseudorandomized across days and across rats. This design provided four

conditions for comparison: Tone-Alone, Tone-Paired, Light-Alone, and Light-Paired. Before and

after each conditioning session the rat was placed in a comfortable rest box separate from the

conditioning box for 10 minutes.

Page 52: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

41

Figure 4. Experimental Procedure and Trace Eyeblink Conditioning.

Diagram of the experimental paradigm (left) wherein rats underwent two conditioning sessions daily in the same conditioning chamber (Box A). Each session consisted of 100 trials of conditioned stimulus (CS) presentations (Light or Tone) in which the first 20 trials the CS was presented alone and the last 80 trials the CS was paired with the unconditioned stimulus (US). On the right, a diagram of the trace eyeblink conditioning procedure wherein the offset of the CS and the onset of the US were separated by a 500 millisecond stimulus-free interval. The bottom figure displays an example of an adaptive eyeblink conditioned response (CR) monitored with electromyogram (EMG) eyelid activity observed as an increase in signal amplitude immediately before the US onset. Artefact from the US was ignored in the analysis of EMG activity.

2.5.2 Context Selectivity

In several animals (n = 3), upon completion of the full conditioning procedure, I investigated whether

prelimbic neurons would show a high degree of selectivity for contextual features. Animals

underwent a similar conditioning procedure over three days in which the conditioning context was

manipulated but the conditioning stimulus remained the same. In two sessions, rats underwent tone

trace eyeblink conditioning with the first 20 trials being Tone-Alone and the last 80 trials Tone-

Paired. One session was run in the same context as the previous 30+ days of conditioning (Context

A) in a dark box with brown floors and plain walls, in the other session the conditioning took place in

a box in which the visual and textile features were manipulated (Context B; light box, stripped walls,

white floor) (see Figure 15). The session order was pseudorandomized across rats. Firing rates were

compared and analyzed in the same manner as described above, but here looking at selectivity for the

environmental context, the relationship, or their conjunction.

2.5.3 EMG, LFP, and Unit recording

During the daily conditioning sessions, action potentials from individual neurons in the prelimbic

medial prefrontal cortex, local field potentials (LFPs) in the lateral entorhinal cortex, perirhinal

cortex, ventral hippocampus and mPFC, and electromyogram (EMG) activity from the eyelid were

simultaneously recorded. Action potentials were captured using the tetrode technique, which allows

Page 53: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

42

for recording the activity of many individual neurons per recording session (Gray, Maldonado,

Wilson, & McNaughton, 1995; Wilson & McNaughton, 1993). Experimental rats were connected to

the system through an Electrode Interface Board (EIB-54-Kopf, Neuralynx, Bozeman, MT)

contained within the Micro-drive array fixed to the animals head. The EIB was connected to a

headstage (HS-54, Neuralynx, Bozeman, MT) and signals were acquired through the Cheetah Data

Acquisition System (Digital Lynx and Cheetah Software, Neuralynx, Bozeman, MT). A single

tetrode consists of 4 individual micro-wires spun together each of which measures rapid changes in

extracellular electrical potential. A threshold voltage was set at 40-50 mV, and if the voltage on any

channel exceeded this threshold, activity would be collected from all four channels. Spiking activity

of single neurons was sampled for 1 ms at 32 kHz and signals were amplified and filtered between

600- 6000 Hz. LFP activity and EMG activity was continuously sampled and filtered between .1-

400 Hz and 300-3000 Hz, respectively.

2.5.4 Schedule

Recordings and data acquisition began on the first day of conditioning and continued for 30+ days.

During the first 3 days of the experimental procedure the rat was habituated to the apparatus, with

one session each on the first 2 days and the full procedure on the third day, all done with no

presentation of either CS or US. These habituation sessions also served to assess spontaneous

eyeblinks in the absence of conditioning stimuli.

2.6 Data Analysis

2.6.1 Behaviour analysis

The adaptive conditioned eyeblink response (CR) which represents the learning of the association

between the conditioned (CS) and unconditioned (US) stimuli was assessed through the analysis of

electromyogram (EMG) activity recorded from the upper left eye-lid muscle. Each trial was assessed

offline with custom codes (Matlab) for the presence of a conditioned response (CR) detected as a

Page 54: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

43

significant increase in eyelid EMG amplitude immediately before US onset (Morrissey et al., 2012;

Takehara-Nishiuchi & McNaughton, 2008). Specifically, EMG activity was sampled around the

presentation of the conditioned stimulus in each trial and the instantaneous amplitude of the signal

was calculated as the absolute value of the Hilbert transform of the filtered signal (using the

hilbert function in Matlab). For each trial, the average amplitude during a 300-ms period

immediately before CS-presentation was defined as the Pre-Value and the averaged amplitude during

a 200-ms period immediately before US-presentation was defined as the CR-Value. A Threshold

value was set as the averaged Pre-Value plus 2 standard deviations. For a given trial, if the CR-Value

exceeded the Pre-Value and the Threshold, that trial was classified as containing a CR. However if

the Pre-Value for a given trial exceeded the Threshold value, the trial was classified as hyperactive

and discarded. The ratio of trials containing a CR within each condition to the total number of valid

trials within each condition represented the CR% for a given condition (CS-Alone, CS-Paired) for

each session (Tone, Light).

To assess changes in neuron activity across the various stages of learning and long-term memory

storage, time-points were defined within each individual rat according to their acquisition and

expression of the conditioned eyeblink response (CR). These criteria were selected based on

observation of general patterns of CR acquisition and expression across many animals. In the first

few days of training, rats show very few trials in which they exhibit the CR, I define this period as

Before Learning, i.e. before the animal has begun to associate the CS with the US. However once rats

begin to form this association the percentage of trials in which they exhibit a CR rapidly increases

but can fluctuate greatly across days, I define this period as During Learning. Eventually the rats

reach a point in which their responding plateaus and reaches asymptote, from this point on I

generally see small fluctuations in response rate across days but rarely see large deviations. This

point defines the end of this During Learning period. All days past this point are defined in weeks

Page 55: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

44

Post Learning. To operationally define these learning stages I set a threshold of responding. All days

prior to the rat displaying a conditioned response in 30% of trails are defined as Before Learning. All

days following two consecutive days of the rat displaying a CR in 60% of trials are defined in weeks

Post Learning. All days in between Before Learning and the beginning of the Post Learning stage are

defined as During Learning (see Figure 8).

2.6.2 Spike sorting

Putative single neurons were isolated offline using a specialized software package in Matlab

(KlustaKwik, author: K.D. Harris, Rutgers, The State University of New Jersey, Newark, NJ;

MClust, author: D.A. Redish, University of Minnesota, Minneapolis, MN; Waveform Cutter,

author: S.L. Cowen, University of Arizona, Tucson, AZ). Before the activity of individual

neurons can be analyzed, signal contributions from background noise and other nearby neurons

must be disentangled in a process known as spike sorting. The software used to achieve this

combines both automatic spike-sorting and manual sorting allowing for sorting clusters based on

the relative action potential amplitudes on the different tetrode channels and various other

waveform parameters including peak/valley amplitudes, energy, and waveform principle

components (Gray et al., 1995). The final result was a collection of time stamps associated with

each action potential from a given neuron. Only neurons with < 1% of inter stimulus intervals

distribution falling within a 2 msec refractory period were used in the final analysis.

2.6.3 Firing rate analysis

Activity from tetrodes was processed offline using spike sorting algorithms to isolate the activity

of putative single neurons (units). The result was a collection of putative single neurons for a

given days recording, each represented by timestamps corresponding to action potentials for the

entire length of the recording. The number of action potentials (spikes) from a given neuron

Page 56: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

45

during a series of 1-msec bins which cover a 4-second period (from 2 seconds before CS-onset to

2-seconds after CS-onset) was counted in each trial for each of the four conditions. Activity

during the presentation of the US and shortly after was ignored due to electrical artifacts from the

stimulating electrodes. If a neuron did not show a spike in more than 15 trials in either the tone

or light paired conditions, the neuron was removed from further analyses.

To determine the selectivity of prelimbic neurons I directly compared firing rate changes relative

to baseline across the four conditions within each individual neuron. For each neuron, mean

spike counts across trials during a baseline period (500-msec immediately before CS-onset) in

the tone and light paired conditions determined the baseline firing rates. A neuron was classified

as responsive if the averaged firing rate during the CS-period (100-msec) or during the trace

interval (TI: 500-msec period between CS and US presentations) was different from baseline

using the students t test with Bonferroni correction (p < .025) in either the tone or light

conditions. Neurons whose firing rate during the CS or trace interval did not significantly change

from baseline in Paired trials were classified as Non-Responsive.

To determine selectivity, firing rates during the CS and trace intervals were compared across the

four conditions (Tone: CS-alone, CS-paired; Light: CS-alone, CS-paired) for responsive neurons.

To correct for differences in baseline activity between the four conditions, firing rates were

standardized to mean baseline firing rates and converted to z-scores: �̃�𝑟𝑒𝑠𝑝−�̃�𝑏𝑎𝑠𝑒

𝑆𝐷𝑏𝑎𝑠𝑒 where �̃�𝑟𝑒𝑠𝑝

equals the mean of the firing rate during CS or the TI from a given condition, �̃�𝑏𝑎𝑠𝑒 equals the

mean of the firing rate during the 500 msec period immediately prior to CS onset and 𝑆𝐷𝑏𝑎𝑠𝑒 is

the standard deviation of the firing rate during this Pre-CS baseline period in the same condition.

These standardized firing rates were then compared across conditions to determine if the neuron

shows selectivity for the CS-US association (Relationship, CS-Alone ≠ CS-Paired), for a

Page 57: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

46

particular CS (Identity, Light ≠ Tone) and any combination of the two (Conjunctive, RxI). Again

these comparisons were done using paired-sample t-tests with Bonferroni correction. The results

of this characterization will provide information about the overall selectivity of neurons within

the prelimbic mPFC during a memory task. The proportion of each type of selectivity was then

calculated and compared across the learning stages with the chi-square test.

2.6.4 Mutual information

Mutual information (MI) gives us a measure of how much one variable tells us about another

variable (Wallisch, 2014). For example, whether the firing rate of a neuron tells us about a

particular stimulus. In this manner it allows us to define the information that the firing rates of a

neuron at different time-windows contains about the stimulus or condition. For each neuron, I

evaluated its degree of selectivity along a continuum for the Relationship, Identity, and

Conjunction of the two. MI represents the entropy of the stimulus, minus the entropy of the

stimulus given the response (Shannon & Weaver, 1949). Here the stimulus represents the

particular condition, and the response represents the neurons firing rate. For my purposes, I

calculated the MI represented in each neurons firing rate for the three features of selectivity by

calculating the joint and marginal probability distributions of each stimulus (trial type) and

response (firing rate) for each neuron with Alone vs. Paired trials (Relationship), Tone vs. Light

trials (Identity), and Tone-Alone vs. Tone-Paired vs. Light-Alone vs. Light-Paired trials

(Conjunction).

Page 58: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

47

In the preceding equation, for Relationship, I represents the mutual information of the binned

firing rate (R, 10 bins) for two trial types, Paired and Alone (S). P(s,r) is the joint probability of a

particular trial type and a particular firing rate. Whereas P(s) and P(r) are the marginal

probabilities of each trial type (Paired or Alone) and each response (firing rate), respectively.

In this example, a higher MI value for a neuron indicates a higher degree of selectivity for the

Relationship, relational stimulus features. MI values were calculated during 200 msec time

windows with a 50% overlap for a 2 second period centered on the CS for each neuron

separately for Relationship, Identity, and Conjunction estimates. To prevent biasing the estimates

(Panzeri, Senatore, Montemurro, & Petersen, 2007; Wallisch, 2014), MI at chance level was

estimated by randomly shuffling the trial identities. This process was repeated 100 times, and the

mean of this null distribution from shuffled data was subtracted from the actual MI value

(Hatsopoulos, Ojakangas, Paninski, & Donoghue, 1998; Panzeri et al., 2007). Neurons with a

corrected MI value above 0.4 were considered ‘selective’.

Binning can have a big impact on the estimates produced, for spike counts using larger bins loses

information (for example 20 spikes may end up in the same bin as 10 spikes) whereas using very

small bins introduces a larger bias (Wallisch, 2014). For my data binning firing rates into 10 bins

was observed to produce enough separation to preserve information while avoiding significant

bias.

2.6.5 Population firing rate matrix correlation

To examine the similarity between firings rates of population of neurons across the different

conditions, I constructed a population firing rate matrix which contains the binned firing rate of

all recorded neurons during a 2 second period centered on the CS from Tone-Paired trials. Firing

rates for each neuron were normalized to mean baseline firing rates during a 500 msec window

Page 59: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

48

before CS onset. I then sorted these neurons based on their change in firing rate during the trace

interval (TI) relative to baseline with those neurons with a maximal increase at one end and those

with a maximal decrease at the other.

𝑇𝑃 = [

𝑥11 ⋯ 𝑥1𝑚

⋮ ⋱ ⋮𝑥𝑛1 ⋯ 𝑥𝑛𝑚

] 𝑇𝐴 = [

𝑦11 ⋯ 𝑦1𝑚

⋮ ⋱ ⋮𝑦𝑛1 ⋯ 𝑦𝑛𝑚

] 𝐿𝑃 = [

𝑧11 ⋯ 𝑧1𝑚

⋮ ⋱ ⋮𝑧𝑛1 ⋯ 𝑧𝑛𝑚

]

The matrix TP stores the firing rate activity during the Tone-Paired condition in Cells (n) x Time

(m) bins. Each row contains the firing rate of a single neuron in 50-msec bins for a period of 2-

seconds centered on the CS-onset, averaged across trials and the sorted from a maximal increase

(n=1) to a maximal decrease (n=n) in firing rate during the TI relative to baseline. Matrices TA

and LP store the firing rate activity in the same cell order (1:n) from TP for Tone-Alone and

Light-Paired conditions, respectively. The Pearson R correlation statistic between the template

matrix TP, and matrices TA and LP provide an indication of whether prelimbic neurons show

more similar patterns of activity to the same stimulus that differed in its relational features

(Tone-Alone vs. Tone-Paired) or to a different stimulus that shared the same relational features

(Tone-Paired vs. Light-Alone).

2.6.6 Support vector machine learning

Subsequently I sought to examine whether a supervised machine learning algorithm could

decode memory features from prelimbic neuronal population activity. Support Vector Machine

(SVM) is a supervised learning model that looks for patterns in data to find the best way to

separate data sets. SVM produces a model from training data which then predicts the target

values of test data given only the test data attributes (Hsu, Chen & Lin, 2010). For the current

study, the attributes were the normalized firing rates of a population of neurons and the target

Page 60: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

49

values were the conditions from which they were sampled. SVM is advantageous because in

addition to completing linear classification, using the kernel trick, SVM can perform non-linear

classification by mapping inputs into a higher dimensional feature space.

For a 2-second period centered on the onset of the CS, firing rates during 200 msec windows

were sampled with a 50% overlap and were linearly scaled by dividing each spike count by the

maximum in each window: 𝑆𝑖 =𝑆𝑖

𝑆𝑚𝑎𝑥 where Si equals the spike count for the ith trial and Smax

equals the maximum spike count from a trial during that window. This procedure produced

values ranging from 0-1. All algorithms were run in Matlab using the freely accessible LIBSVM

library (Chang & Lin, 2011). I used the radial-basis-function (RBF) kernel to classify the four

conditions with a multi-class SVM. To prevent over-fitting, the parameters C and γ were selected

beforehand by performing a grid-search using cross-validation on exponentially growing

sequences of C and γ. The pair with the best cross-validation accuracy was then selected (Chang

& Lin, 2011). Using these parameters, the SVM was trained with the concatenated firing rates of

a population of neurons from 10 randomly chosen trials in each condition. A separate set of data

from 10 different trials was then used to test the classification accuracy. Each SVM was repeated

200 times with different samplings of 10 training trials from each condition and 10 separate test

trials. The significance of the classification performance was assessed by running the same SVM

procedure with the identities of the trials randomly shuffled and repeated 200 times. Chance

level of prediction given 4 trial types was ¼ = .25.

To look at specific classification errors made, confusion matrices were created indicating the

percentage of classifications in which a response vector belonging to condition x was classified

as condition y from a 200 msec time window during the TI between CS-offset and US onset.

These percentages were computed from the classifications of the test trials across the 200

Page 61: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

50

resampling’s. The accuracy of classification for each condition was summed to produce accuracy

estimates for the stimulus Relationship (Alone vs. Paired), Identity (Light vs. Tone), and the

conjunction of the two. The accuracy percentage of each type of selectivity was then computed

from these values along with 99% confidence intervals.

2.6.7 Phase-locked firings

Analyses of local field potential (LFP) activity was done offline using custom written codes in

Matlab (Mathworks,Natick,MA,USA). Movement artefact was observed as clipping in the LFP

signal. Prior to any analyses, trials containing clipping during a period spanning from 1 second prior

to the CS to the onset of the US were removed.

To quantify the degree of phase-locking of firings of each neuron, phase data from theta frequency

filtered LFP activity from each trial centered on the CS-presentation was extracted using the Hilbert

Transform. I then looked at the consistency of individual spike-LFP phases from these theta filtered

signals (Vinck, Battaglia, Womelsdorf, & Pennartz, 2012). For each neuron, a phase-value from the

ongoing theta oscillation was assigned every-time the neuron fired an action potential, I then used

this to determine if the neuron displayed a preference to fire an action potential at a particular theta

phase. Significance was determined by the Rayleigh test for circular uniformity. The Rayleigh test is

equivalent to a likelihood ratio test, and the test statistic 2𝑅2

𝑛⁄ , where R= resultant length, n =

sample size, follows an asymptotic chi-square distribution with 2 degrees of freedom (Siapas,

Lubenov, & Wilson, 2005b; Sirota et al., 2008) and tests of significance using the variance-stabilized

log(Z), 𝑍 = 𝑅2

𝑛⁄ (Siapas et al., 2005b; Sirota et al., 2008). The proportion of neurons whose

distribution of collected phase values was significantly different from the uniform distribution (p <

.05) were then totaled for each region, and the log(Z) of these neurons indicated the degree of phase-

locking.

Page 62: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

51

2.7 Statistical Analyses

Statistical analyses were performed using SPSS and Matlab, where appropriate. In Chapter 3,

ANOVA was used to compare changes in behavioural responding across time and between

sessions. Follow-up tests used the students t test with Bonferonni correction. Paired-samples t

tests were used to compare firing rate changes to baseline and between conditions with

Bonferonni corrections for multiple comparisons.

To compare correlations between the population matrices, Steiger’s Z-test for correlated

correlations within a population. Here 𝑍 = [𝑍12 − 𝑍13] ∗ √[𝑁−3]

√2∗[1−𝑟23]∗ℎ.

Where Z12 and Z13 are the Fischer’s Z transformations of r12 and r13, respectively, and

ℎ =1−[𝑓∗𝑟𝑚2]

1−𝑟𝑚2 where 𝑓 =

1−𝑟23

2∗[1−𝑟𝑚2] and 𝑟𝑚2 =

𝑟122 +

𝑟132

2.

In Chapters 3 and 4, proportions were compared using the chi-square analysis. Follow-up tests of

significant chi-square tables compared individual chi-square cells using the z-test (Marascuilo &

Serlin, 1988; Sharpe, 2015). Here, 𝑧 = (𝛹 − 0)/𝑆𝐸𝛹. This test compares individual cells from

the chi-square table, where 𝛹 = 𝜌1− 𝜌2 represents the comparison of interest with ρ1 being the

proportion in the first contrast cell relative to its column marginal and ρ2 being the proportion in

the second contrast cell relative to its column marginal. 𝑆𝐸𝛹 = √(1)2𝑆𝐸2𝜌1 + (−1)2𝑆𝐸2

𝜌2

represents the standard error of that contrast. The obtained z value was then compared against the

square root of the chi-square critical value for the entire contingency table to determine

significance (Marascuilo & Serlin, 1988). Confidence intervals were calculated using the

Clopper-Pearson method with the binofit Matlab function.

Page 63: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

52

2.8 Histology

Upon completion of all recordings, the location of electrodes were marked by electrolytic lesions.

Rats were first injected intraperitoneally with an overdose of sodium pentobarbital. For tetrodes, 5

μA was passed through one wire of each tetrode for 20 seconds (positive to electrode, negative to

animal ground), for LFP electrodes 20 μA was passed for 45 seconds (positive to electrode, negative

to animal ground). Rats were then perfused intracardially with 0.9% saline followed by 10% buffered

formalin. The brain was removed from the skull and stored in 10% formalin for several days. To

prepare the brain for cryogenic sectioning, the tissue was infiltrated with 30% sucrose solution. The

brain was then frozen and sectioned in a cryostat (Leica) at 50 μm. Sectioned tissue was stained with

cresyl violet and imaged under a light microscope to locate electrode locations. Only recordings from

tetrodes located in the prelimbic mPFC were used for single unit analysis. For LFP analysis only

activity from electrodes located in the target regions were used.

Page 64: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

53

Chapter 3

Prelimbic Representations of Associative Memory Throughout Learning and Consolidation

3.1 Introduction

The medial prefrontal cortex (mPFC) has emerged as an important node among the network

of brain regions responsible for the formation, consolidation, and retrieval of memories of events or

episodes (Frankland & Bontempi, 2005; Insel & Takehara-Nishiuchi, 2013; Szu-Han Wang &

Morris, 2010). Whereas its exact role in these different states of a memories existence is uncertain,

there is strong evidence to suggest that for memory retrieval it plays a much more significant role

after the consolidation phase than before (Frankland et al., 2004; Lopez et al., 2012; Maviel et al.,

2004; Quinn et al., 2008; Takehara et al., 2003; Teixeira et al., 2006). This functional niche of the

mPFC has been demonstrated through various experimental approaches (see Chapter 1.2 for

discussion). For associative memory in particular, both functional and electrophysiological

experiments converge on the finding that the prelimbic (PrL) mPFC seems to be intimately involved

in the consolidation of a memory and its long-term retrieval (Hattori et al., 2014; Runyan et al., 2004;

Stern, Gazarini, Vanvossen, Hames, & Bertoglio, 2014; Takehara-Nishiuchi & McNaughton, 2008;

Takehara-Nishiuchi et al., 2006). Electrophysiological investigations offer an excellent complement

to functional experiments and provide the means to assess actual neuronal computations taking place

in real-time. In associative memory, electrophysiological recordings of PrL neurons have revealed

that these neurons display activity selective for a memory association, both in their individual and

population activity patterns. In trace eyeblink conditioning this selectivity was observed as persistent

changes in firing during the stimulus-free trace interval which increased in strength with learning and

consolidation of the association (Hattori et al., 2014; Takehara-Nishiuchi & McNaughton, 2008).

However these studies leave open questions about the encoding of memory in the PrL, for example,

Page 65: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

54

which features of a memory are maintained in this network, and how does the content of mPFC

memory representations change as it engages in the long-term consolidation of the memory?

The mPFC has been linked to schema formation and targeted as a potential key brain region in the

building, maintenance and updating of memory frameworks (Preston & Eichenbaum, 2013; Richards

et al., 2014; Wang et al., 2012). How these frameworks are maintained, and the detail to which

specific events are represented within the mPFC is not clear. Here, I attempt to uncover how features

of a memory event are represented within the PrL. Specifically, I looked at the selectivity of

prelimbic neurons to the distinct features of a memory, and how this selectivity and encoding

changes throughout learning, consolidation and retrieval of the memory. I simultaneously recorded

the action potentials of many individual neurons in the rat prelimbic mPFC while rats received daily

conditioning in trace eyeblink conditioning. The paradigm employed was designed to manipulate the

identity and relationship of the conditioned stimulus to permit the examination of which features of

an episodic memory prelimbic neurons were responsive to. Rats were conditioned in this paradigm

daily over the course of 4+ weeks to allow for the direct comparison of prelimbic activity before

learning, during learning, and during post-consolidation retrieval.

Here I reveal the presence of neurons in the prelimbic cortex exhibiting varying degrees of selectivity

for either the relational features of the conditioned stimulus (CS), the identity of the CS, or a

conjunction of the two. At the single neuron level there is a greater proportion of responsive cells that

do not show selectivity and those selective for the relationship of the CS than cells that display more

highly selective activity. Also at the single neuron level the proportions of these neurons across time

does not significantly change. However, I show here that in looking at activity at the population

level, we see a shift across time from higher selectivity in the early phases or learning to more

generalized selectivity in the post-consolidation retrieval stages of the paradigm.

Page 66: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

55

3.2 Results

Experiments were conducted on 4-6 month old Long-Evans rats (n=5) with functioning tetrodes

located in the prelimbic medial prefrontal cortex. Animals were surgically implanted with a chronic

Micro-drive array (see Chapter 2.3 for detailed methods on surgical procedure and the tetrode

method). Animals underwent daily conditioning in the trace eyeblink paradigm for 30+ days (see

Chapter 2.5.1for a clear overview of the procedure). Briefly, the paradigm employed the use of two

different conditioned stimuli (Tone & Light) in two different relational conditions (Alone & Paired).

Each day, each stimuli was presented in one of two 100 trial sessions, the first 20 trials of which the

stimulus was presented alone, and the last 80 trials the stimulus was paired with eyelid shock (US).

This resulted in 4 conditions that differed in their combination of the identity and relationship of the

stimulus (Tone-Alone; Tone-Paired; Light-Alone; Light-Paired). Electrophysiological signals were

recorded from tetrodes located in the prelimbic region throughout the procedure and were sorted

offline to isolate stable single units for analyses.

3.2.1 Acquisition and expression of the conditioned response

Rats underwent trace eyeblink conditioning wherein two different stimuli (a tone and a light)

predicted the same outcome (eyelid shock). This was conducted daily in two conditioning

sessions, the first with one CS (e.g. tone) the second with the other CS (e.g. light), with session

order determined pseudo-randomly (see Figure 5). Each conditioning session consisted of 100

trials, the first 20 of which the CS was presented alone (CS-alone) and in the following 80 trials

the CS was paired with the US (CS-paired). Animals gradually acquired the adaptive conditioned

eyeblink response (CR) over the course of training during the Tone-Paired and Light-Paired

conditions, and learned to not respond during the Tone-Alone and Light-Alone conditions

(Figure 5). Repeated measures ANOVA revealed a significant Condition Day interaction

(F(93,393) = 4.702, p < .001). Follow up tests reveal that for Tone sessions CR% significantly

Page 67: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

56

differs between Alone and Paired trials (F(1,131) = 1227.241, p < .001) and a significant

Condition x Day interaction was detected (F(31,131) = 8.282 p < .001). For the Paired condition

there was a significant effect of Day (F(31,162) = 8.418, p < .001), that was not observed in the

Alone conditioned (F(31,162) = .931, p = .576). Similarly, for the Light sessions, CR%

significantly differs between Alone and Paired trials (F(1,131) = 1605.870, p < .001, and there

was a significant Condition Day interaction (F(31,131) = 7.61, p < .001). For the Paired

condition there was a significant effect of Day (F(31,162) = 5.32, p < .001), that was not observed

for the Alone condition (F(31,162) = .901, p = .620). There was no difference between Alone

conditions from the Tone and Light sessions. However, for Paired conditions CR% in Light

sessions was slightly higher than Tone conditions (F(1,131) = 7.968, p = .006, M = 55.25 vs.

51.10), but there was no Condition Day interaction. This difference between Tone and Light

Paired conditions was only evident during early acquisition and disappeared after Day 12 of

conditioning (F(1,67) = 1.014, p = .317, M = 65.93 vs. 64.32).

Page 68: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

57

Figure 5. Acquisition and Expression of the Conditioned Eyeblink Response.

A) Illustration of the experimental paradigm. B) Averaged acquisition and expression of the conditioned eyeblink response (CR) across animals (n = 5), across days, in each of the four conditions. Rats gradually acquired the conditioned response in both the Tone-Paired (TP, red) and Light-Paired (LP, blue) conditions while not responding during the Tone-Alone (TA, magenta) and Light-Alone (LA, turquois) conditions.

3.2.2 Single unit activity patterns

My first aim was to determine the selectivity of prelimbic neurons for the various features of the

memory. For each individual neuron, the averaged firing rate across trials during the CS-period

and TI-period was compared to the averaged firing rate during a 500-msec period immediately

before CS-onset during Tone-Paired and Light-Paired conditions. A significant change from

baseline activity in either the Tone or Light Paired conditions determined if a neuron was

responsive to the CS or not. The total percentage of responsive neurons from all sampled neurons

Page 69: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

58

(n = 1288) reveals that prelimbic neurons are more responsive during the TI period (70.89%)

than during the CS period (31.06%) (X2(1, N = 2572) = 413.043, p < .001). (Figure 6).

Figure 6. Histology and Single Unit Isolation.

A) Representative image of a tetrode tip located within the prelimbic medial prefrontal cortex (top). Final locations of all usable tetrodes targeted at the prelimbic cortex (black dots, bottom). Images and coordinates adapted from Paxinos and Watson (2007, 6th edition). B) Example of cluster sorting in MClust (left) and illustrative waveforms of isolated single units (right) from a single tetrode from one recording session.

Of these responsive neurons, firing rates were normalized to baseline activity in each condition

and within each neuron, activity was compared across the four conditions. Student’s paired

sample t-tests with Bonferroni correction revealed 4 different types of selectivity 1) Non-Selective

(CS-alone = CS-paired & Light = Tone); 2) Relationship-Selective (CS-alone ≠ CS-paired & Light =

Tone); Identity-Selective (CS-alone = CS-paired & Light ≠ Tone); 4) Conjunctive (CS-alone ≠ CS-

paired & Light ≠ Tone) (Figure 7). The total percentage of each type of selectivity reveals the type of

information being represented by prelimbic neuron activity. During the TI period the largest

proportion observed was Non-Selective (37.35%). Of the other types of selectivity I observed a

Page 70: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

59

substantial representation of each, Identity (21.25%), Relationship (23.22%), and the Conjunction of

the two (18.18%) (Figure 7). These findings suggest that within the prelimbic region there is a

diversity of neuronal responses to the various features of a memory episode, encompassing very

unselective stimulus activity to highly selective activity to a particular stimulus in a particular

condition.

Page 71: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

60

Figure 7. Single Neuron Selectivity for Task Features.

A) Representative raster plots and peri-stimulus time histograms (PSTHs) from prelimbic units smoothed with a 50-msec Hanning window for each type of observed selectivity of responsive neurons. Selectivity was defined by firing rate differences between the four conditions: Tone-Alone (TA, magenta), Light-Alone (LA, turqouis), Tone-Paired (TP, red), Light-Paired (LP, blue). This revealed 4 types of selectivity: Non-selective: CS-Alone = CS-Paired & Tone = Light; Relationship: CS-Alone ≠ CS-Paired & Tone = Light; Identity: CS-Alone = CS-Paired & Light ≠ Tone; Conjunctive RxI: CS-Alone ≠ CS-Paired & Light ≠ Tone. Examples of each type of selectivity are shown with raster plots, top, showing spiking activity (dots) during a 1-second period centered on the onset of the conditioned stimulus (CS, 0) in the Tone (left) and Light (session) and PSTH (bottom) showing the smoothed averaged firing rate for each condition. B) Proportion of responsive (left) and each type of selective (right) neurons observed by looking at the 100-msec period when the CS is presented (top) and the 500-msec trace interval (TI) between CS and US (bottom) of all neurons recorded. More responsive neurons were observed during the ISI (70.89%) than during the CS (31.06%). Of responsive neurons during the ISI period, similar proportions of each type of selectivity were observed.

Page 72: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

61

3.2.3 Changes in selectivity across time

The necessity of the prelimbic region for memory retrieval changes over time (Maviel et al.,

2004; Quinn et al., 2008; Takehara-Nishiuchi et al., 2005; Takehara-Nishiuchi et al., 2006). This

suggests that changes in connectivity and possibly changes in encoding among neurons within

the prelimbic region may occur over the course of this time period. Once I had established the

range of selectivity of prelimbic neurons, I sought to examine whether these representations at

the single neuron level change across time. The first way to examine this was to look at the

proportions of each response type at different time-points throughout the conditioning procedure

to see if any differences emerge. As described in Chapter 2.6.1 neuronal data collected for each

rat was divided into individual behaviorally defined time-periods of conditioning: Before

learning period, prior to acquisition of the CS-US association as evidenced by a very low

percentage of trials exhibiting the adaptive conditioned response (CR); During learning, as the

animal begins to acquire the association gradually with fluctuating performance across days; Post

leaning period, once responding reaches an asymptotic stable level all time-points beyond this

period are defined in weeks Post learning (see Figure 8).

Page 73: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

62

Figure 8. Behaviourally Defined Learning Stages.

Example of acquisition and expression of the conditioned response in one rat displaying how learning stages were individually defined based on behavioural data. Conditioned response rates during Tone-Paired (TP, red) and Light-Paired (LP, blue) indicate periods prior to acquisition of CR in 30% of trials (Before Learning, light grey), period between this point and asymptotic responding (During Learning, medium grey) and all days following (Post Learning, dark grey). An epoch 3 weeks into the Post Learning period (3w-Post) is highlighted as this was a major comparison in subsequent analyses.

I focused my comparisons on neurons sampled from three time-points, the Before learning

period (n = 149), the During learning period (n = 487), and the 3 week post-learning period (3w-

Post, n = 188) consisting of conditioning days taking place 3 weeks after asymptote had been

reached. The proportion of neurons exhibiting each type of response selectivity was calculated

across animals (n = 5) for these different time periods (Figure 9). A chi-square analysis revealed

no differences between the three time-points in the proportion of each type (X2(8, n = 822) =

11.753, p = .163).

Page 74: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

63

Figure 9. Selectivity Across Learning Stages.

Neurons categorized by their response patterns during the task were separate and their relative proportions compared across the behaviourally defined learning stages before learning (Before, light grey), during learning (During, medium grey) and three weeks after asymptotic responding (3w-Post, black) as described in Section 2.6.1. No differences in the proportions were revealed between time periods.

While the mean comparison approach is revealing, it has its drawbacks in that we are imposing a

threshold to classify selectivity to a particular feature. Alternatively, mutual information (MI)

allows us to quantify selectivity along a continuum by calculating the information contained

between the stimulus features and the responses of neurons. The MI between variables is

expressed in bits with 0 being no information shared, or completely independent variables. I

calculated the MI separately for the stimulus relational features (Relationship: Alone vs. Paired),

stimulus identity features (Identity: Tone vs. Light), and the conjunctional stimulus features

(Conjunction: Relationship vs. Identity) to provide the degree of selectivity of each neurons

response rate. Again this was done at the Before learning, During learning, and 3w-Post learning

time-points. Figure 10 illustrates a scatterplot of this data in which it can again be observed that

more information about the stimulus features is evident during the trace interval (TI) than during

Page 75: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

64

the CS presentation. MI values were shuffle-corrected using the mean of MI values from 100

repetitions of shuffled data. All neurons with a shuffle-corrected MI value above 0.4 were

counted as selective. Chi-square analysis on the number of selective neurons during the CS and

TI confirmed more selective encoding of the stimulus during the TI (X2(3, n = 2398) = 563.743,

p < .001). I then looked at the proportion of selective neurons based on MI estimates for

selectivity to the Relationship, Identity, and Conjunction of the two during the TI across the three

learning periods. Results support the previous analysis, revealing no differences in the proportion

of neurons selective for the stimulus features across the different learning stages (X2(6, n = 2398)

= 10.306, p = .112) (Figure 10).

Using both mean comparison and mutual information I have shown that PrL neurons firing rates

exhibit more selective activity during the TI than the CS, similar to previous reports (Takehara-

Nishiuchi & McNaughton, 2008; A. P. Weible, Weiss, & Disterhoft, 2003). Both estimates of

selectivity also reveal no differences at the single neuron level in the proportion of selective PrL

neurons from periods prior to learning the association, during its gradual acquisition, and three

weeks after stable memory expression.

Page 76: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

65

Figure 10. Mutual Information defined Selectivity.

A) scatterplots showing neurons (blue-dots) mutual information for the Identity (y-axis) and the Relationship (x-axis) of the stimulus during CS-presentation (CS) and during the inter-stimulus interval (Trace Interval) across three stages of learning (Before, During, 3w-Post). Again more selective activity is observed during the trace interval compared to the CS-period. B) Proportion of neurons whose corrected mutual information during the trace interval for the Identity, Relationship and their Conjunction (RxI) was greater than 0.4 across three stages of learning. No differences in the proportions across the learning stages was observed.

3.2.4 Population activity patterns

Selectivity at the single neuron level informs us of the type of information that is encoded

within a brain region. Here we see a diversity of single neuron response profiles within the prelimbic

mPFC (PrL), from responsive regardless of the identity and relational features of a stimulus to highly

selective to a specific stimulus in a specific relational condition. However neurons do not function

individually in the brain, they function as part of a network. Therefore it can be informative to look at

Page 77: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

66

activity at the population level. A previous study recording from many single neurons from the same

brain region in trace eyeblink conditioning also revealed no changes in the proportion of memory

selective neurons across time, but at the population level an increase in memory selective activity

was observed over a time-frame consistent with the consolidation period (Takehara-Nishiuchi &

McNaughton, 2008). Here I looked at selectivity at the population level of activity and whether

differences could be observed across time in the representation of different features of the memory

association. As a first course of investigation I constructed firing rate population matrices containing

the firing rates of all neurons during a 2-second period centered on the CS-onset. Normalized firing

rates averaged across trials from the Tone-Paired (TP) condition were sorted based on change in

firing rate during the trace interval (TI) relative to baseline in order from a maximal increase to a

maximal decrease (Figure 11). Taking this same order of neurons, matrices were created containing

the firing rates during the Tone-Alone (TA) condition and Light-Paired (LP) condition.

Page 78: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

67

Figure 11. Prelimbic Neuron Population Matrices.

Populations of neurons recorded from the prelimbic region before acquisition (Before, left), during learning (During, middle), and 3 weeks after asymptote (3w-Post, during) in the Tone-Paired condition (top) were sorted based on their change in firing rate after the CS relative to baseline from the largest decrease (cell #1) to the largest increase. This same order of neurons was then plotted during the Tone-Alone condition (middle) and Light-Paired condition (bottom). Here you can see the similarity between the conditions with the same stimulus, different relationship (Tone-Paired vs. Tone-Alone) and a different stimulus, same relationship (Tone-Paired vs. Light-Paired) change across the learning stages.

Page 79: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

68

The correlation between these matrices provides an indication of the similarity of activity across

neurons for the same stimulus in a different relational condition, or for a different stimulus with the

same relational features. Figure 11 displays this data for periods Before, During and 3w-Post learning

where it can be seen that the similarity of the population activity between TP and TA becomes less

similar as the animals progress through the learning stages to the 3w-Post time period. However the

similarity between TP and LP remains high throughout with a slight increase at the 3w-Post period.

To assess the evolution of these changes I then calculated the correlation between the TP and the

TA/LP matrices across all time periods including the 1 week and 2 week Post learning periods. The

correlation between TP and TA/LP was significant at all times-points (Table 1). I subtracted the

correlations with TA from LP to provide an indication of the changes in the difference between the

two across time. This revealed a pattern where correlations in population activity were different

during the first few days of conditioning but became more similar as learning progressed (During &

1w-Post). However by 2 and 3 weeks after learning (2w/3w-Post) this correlated activity began to

again separate and then became maximally different (Figure 12). Using Steiger’s Z-test of correlated

correlations within a population (Steiger, 1980), I directly compared correlations between TP and

TA/LA at each time-period. With a Z-critical set at 2.58 for p < .01 to correct for multiple

comparisons, the correlation between TP-LP was significantly greater than TP-TA for all time-points

except 1w-Post learning, corroborating the observed differences (Figure 12).

Table 1. Correlations between population firing rates across conditions

Before During 1w-Post 2w-Post 3w-Post

Tone-Paired

vs.

r Z r Z r Z r Z r Z

Tone-Alone 0.654* 3.843*

0.727* 3.435* 0.754* 1.4 0.642* 4.388* 0.616* 4.94*

Light-Paired 0.815* 0.802* 0.798* 0.808* 0.828*

Page 80: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

69

Neurons (n) 156 461 193 186 165

Note: Pearson correlation (r) and Steiger’s Z. * p < .05

Figure 12. Matrix Correlation Differences Between Relational and Identity Features.

The difference in correlation between the Tone-Paired/Light-Paired and Tone-Paired/Tone-Alone population matrices. Difference in Pearson r (bars) and Stieger’s Z value of correlated correlations reveal that conditions sharing the same relational features are more similar than conditions sharing the same stimuli early in learning (Before) and after the consolidation phase 2 weeks and 3 weeks after asymptote (2wP & 3wP). During acquisition (During) and in the initial week after asymptote (1wP), the population activity became more similar. Dotted line represents Stieger’s Z critical value at p < .01. * p < .01.

3.2.5 Decoding Prelimbic population activity

I show that activity appears to become less similar at the population level for the same stimulus when

its relational features differ over the same time-course that the consolidation of this memory is

presumed to take place (Takehara-Nishiuchi & McNaughton, 2008). I next investigated whether a

supervised learning algorithm could be trained with population firing rate data to accurately predict

trial conditions given only population activity. I used a machine learning classifier to decode

prelimbic neuron population activity, the successes and failures of which would indicate the amount

Page 81: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

70

of information represented for the various memory features and how their representation may change

with learning and consolidation. Firing rates from a population of neurons were fed into the Support

Vector Machine (SVM) classifier using 10 randomly selected trials to trial the classifier, and 10

independent trials to test the trained model from each condition in a multi-class classification across

the 4 conditions. Accurate prediction was also observed using 19 training trials and 1 test trial, and

using 15 training and 5 test trials, however the 10-10 method produced more consistent and less

variable results.

Figure 13. Support Vector Machine Classification Accuracy.

A) Condition classification accuracy from a population of neurons recorded during the 3 week period after asymptotic responding (Real, red line) and accuracy when trial identities were shuffled during model training (black line, Shuffled) with the support vector machine decoder. Accuracy was calculated in 200 msec windows with a 50% overlap for a 2-second period centered on the onset of the

Page 82: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

71

conditioned stimulus (CS). Accuracy can be seen to increase after the CS-onset and peak during the inter-stimulus interval. B) Confusion matrix showing the probability of accurate classification for each condition with the predicted trial condition along the x-axis and the actual trial condition along the y-axis. All entities along the diagonal indicate perfect classification whereas entities off the diagonal indicate errors. Here we see the presence of some errors confusing Tone-Alone and Light-Alone trials as well as Tone-Paired and Light-Paired trials indicating less accurate decoding of the stimulus Identity and more accurate decoding of the stimulus Relationship from a population of neurons at this time-point.

Figure 13B shows the prediction accuracy from a population of neurons (n = 156) recorded during

the 3w-Post learning period (Real) compared to the accuracy when the identities of the trials were

shuffled (Shuffled). In this example the trained model was observed to be overall very accurate

(maximum = 97.5% classification accuracy from 200 repetitions). When trial ID’s were shuffled

performance in accuracy was at the chance level for four conditions 1/4 = .25 (shuffled average =

25.52 % correct). It also reveals that classification performance increases during the CS and then

peaks and remains high throughout the TI (average 80.63% correct during 500 ms TI vs. average

48.12% during 500 ms Pre-CS period). The accuracy of all time-points from CS-offset to US-onset

were significantly higher than the chance level of prediction as revealed by paired sample t-tests (all

p’s < .0001). This shows that features of the memory, including the stimulus relational features and

the stimulus identity are encoded by prelimbic neuron population firing rates during the stimulus free

trace interval between the CS and US.

Next I looked at trial specific errors using a confusion matrix to show the proportion of trials in

which condition x was classified as condition y across the 200 resamplings.

Page 83: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

72

Figure 14. SVM Classification Accuracy changes for Stimulus Features Across Time.

A) Percentage of accurate classification for the different features of the task, Paired vs. Alone (Relationship), Tone vs. Light (Identity) and the conjunction of the two (RxI) across the different learning stages. Accuracy in classification of the stimulus relationship gradually increased from before acquisition (Before) to the acquisition period (During) divided in first and second halves (Early and Late, respectively), and peaking 3 weeks after asymptote (3w-Post). The opposite was observed for classification of the identity of the stimulus, which was high Before and dropped significantly to 3w-Post. Classification of the conjunction of Relationship and Identity (RxI) was low initially, peaked during Early learning and then dropped by the 3w-Post period. * p < .01 using the z test to compare chi-square cells. Error bars indicate 99% Clopper-Pearson confidence intervals. n = 156 for all time-periods.

Figure 13 shows the confusion matrix from the concatenated firing rates of a population of neurons

from the prelimbic mPFC recorded 3 weeks after learning. Here I show accurate classification in

each condition (entities along the diagonal) and the minor presence of errors, which appear specific

to confusion of the stimulus identity, where tone trials are mistaken for light trials and vice-versa. To

quantify this, the proportion of correct classifications was calculated between conditions differing in

relational features (Relationship: Tone-Alone vs. Light-Alone + Tone-Paired vs. Light-Paired),

conditions differing in stimulus features (Identity: Tone-Alone vs. Tone-Paired + Light-Alone vs.

Light-Paired) and the conjunction of relationship and identity (RI: Tone-Alone vs. Tone-Paired vs.

Light-Alone vs. Light-Paired). This quantification reveals the information represented about the

Page 84: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

73

stimulus identity, stimulus relationship, and their conjunctive representation within population firing

rates during the TI. I then ran the SVM analysis on separate groups of neurons recorded across the

learning stages to look at differences in these representations across time. Because SVM can be

highly dependent on the number of inputs, the number of neurons in a population across the different

stages was fixed to the lowest number sampled (n = 156, Before Learning). From all other periods

neurons were randomly drawn from the total recorded to match this set. For the During learning

period, in which many more days are included and therefore a much larger population of cells (n =

471), I divided this period in half and defined an Early (n = 235) and Late (n = 236) phase containing

neurons recorded during days in the first and second half of the During learning period, respectively.

This was done to reduce the variability of classification based on random sampling. This analysis

reveals a change in the representation of different types of information at the population level of

neuronal activity across time (x2(10) = 73.93, p < .001). From the chi-square, specific components

were compared using the z test (Marascuilo & Serlin, 1988; Sharpe, 2015) with p < .005 to correct

for multiple comparisons.

In Figure 14 I show in the proportion of correctly classified cases that decoding relational stimulus

features (Relationship) increases over time, from Before to Early learning (z = 6.15, p < .001) and

from Early learning to 3w-Post learning (z = 14.27, p < .001). Conversely, decoding of the stimulus

itself (Identity) becomes less accurate over time, from Before to Late learning (z = -10.11, p < .001)

and from Late to 3w-Post learning (z = -8.84, p < .001). These results suggest that at the population

level, information about the stimulus relationship gradually increases with learning and remains high

after memory consolidation, whereas specific information about the stimulus identity (Tone or Light)

while initially high, gradually decreases with learning and consolidation. The representation of the

conjunction between the stimulus and relationship shows an initial increase from Before learning to

During learning (Early z = 5.03, p < .005) but returns to a similar level as Before learning at all other

time-points.

Page 85: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

74

With SVM I show that representations at the population level change throughout the course of

learning and the memory’s presumed consolidation. Prelimbic neuron populations contain more

information about the CS-US relationship and less information about the particular CS as learning

progresses and the memory becomes consolidated. This supports the results from the population

firing correlations wherein I showed that activity between the two conditions sharing the same

relational features are more similar than the activity between two conditions sharing the same

stimulus, and that this difference becomes progressively larger as the memory consolidates. Here,

using the SVM classifier, the amount of information being represented was interpreted based on the

performance of the classifier to accurately predict trials, given only the firing rates of a population of

neurons in those trials. In the days prior to acquisition of the memory association, the classifier was

more accurate in decoding the identity of the stimulus (Tone vs. Light conditions) than the

relationship of the stimulus (Alone vs. Paired conditions). In the later stages of learning and by 3

weeks post learning the pattern flipped and the classifier was more accurate in decoding the

relationship of the stimulus than the identity. These findings may reflect a network property of the

prelimbic cortex with population activity becoming more generalized (i.e. less specific information

about the CS details) as the memory becomes firmly established in the prelimbic memory network.

3.2.6 Contextual encoding of a consolidated associative memory

Following the 30+ day conditioning procedure, several animals underwent a modified version of the

conditioning paradigm using the same stimulus across two sessions (Tone) but modifying the

conditioning context in one session (Context B). Behaviourally, the conditioned response (CR) rates

remained high during the Paired conditions and low during the Alone conditions and did not differ

between contexts or across days (Figure 15). Repeated measures ANOVA revealed a main effect of

condition (F(3,18) = 51.429, p < .001), and no effect of Day or a Day by Condition interaction.

Context had no effect on CR in Alone or Paired trials and CR percentage was significantly higher in

Paired vs. Alone trials (Context A: t(8) = 8.214, p < .001; Context B: t(8) = 8.454, p < .001).

Page 86: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

75

Figure 15. Context Manipulation Paradigm and Behavioural Results.

A) Design of context manipulation carried out over 3 days after completion of the paradigm previously described. Using the same design as before 20 CS-Alone trials and 80 CS-paired trials, the same stimulus (Tone) was used in both sessions however the environmental context was changed between the two sessions (Context A, original & Context B). B) Conditioned response rates in each of the 4 conditions over the 3 days, Context-A Tone-Alone (Cxt A-TA, magenta), Context-A Tone-Paired (Cxt A-TP, red), Context-B Tone-Alone (Cxt B-TA, turquois) and Context-B Tone-Paired (Cxt B-TP, blue). Conditioned response rates remained high in the Paired conditions versus Alone conditions, however there was no difference in responding between the two contexts.

Mean comparison of firing rate changes relative to baseline again revealed that a majority of neurons

are responsive during the trace interval (TI) (75% Responsive vs. 25% Nonresponsive). Looking at

the selectivity of Responsive neurons I observed the presence of context selective neurons,

suggesting that the context manipulation was effective, but a much smaller proportion than those

selective for the stimulus relationship was observed (Context 13.56%; Relationship 28.81%). I

compared this distribution of selectivity to the distribution of selectivity several days earlier at the

3w-Post time-point in the original conditioning paradigm. By equating Context selective to Identity

Page 87: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

76

selective response proportions, I observed no difference between the two (X(4) = 3.64, p = .456),

suggesting that the change in context had not substantially changed the representation (Figure 16).

Figure 16. Single Neuron and Population Selectivity for Contextual Features.

A) Proportion of responsive (left) and each type of selective (right) prelimbic neurons recorded during context manipulation sessions. A similar ratio of Responsive to Non Responsive neurons to the previous conditioning paradigm. Of responsive neurons, a greater proportion of neurons was observed to be selective for the Relationship (Tone vs. Paired) of the stimulus than the Context (Context A vs. Context B). B) Populations of neurons recorded during the three context manipulation days in the Context A

Page 88: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

77

Tone-Paired condition were sorted based on their change in firing rate after the conditioned stimulus (CS) relative to baseline from the largest decrease (cell #1) to the largest increase. This same order of neurons was then plotted during the Context-A (CxtA) Tone-Alone condition (bottom left) and Context-B (CxtB) Tone-Paired condition (bottom right). Here you can see the similarity between the conditions with the same relationship and different context (CxtA Tone-Paired vs. CxtB Tone-Paired) compared to conditions with the same Context and different Relationship (CxtA Tone-Paired vs. CxtA Tone-Alone).

I then looked at this activity at the population level. First, firing rate population matrices, as

described in the previous section, revealed that population activity was more similar (p < .01)

between Tone-Paired in Context A and Tone-Paired in Context B (r = .738) than between Tone-

Paired in Context A and Tone-Alone in Context A (r = .534) (see Figure 16). Suggesting that there is

stronger encoding of the relationship of the stimulus than the context, as observed at the 3w-Post

time-point for the stimulus identity.

I also ran this population of neurons in the SVM classifier to look at classification accuracy across

the 4 conditions with context being a variable. In the confusion matrix in Figure 16 I show accurate

classification from this population of neurons, however it reveals the presence of specific errors.

Errors appear to be made by the model in classifying activity from Context A with Context B and

vice versa. Quantifying the proportion of correct classification for the Relationship, Context, and

their conjunction revealed very accurate decoding of the stimulus relationship similar to what I

observed at the 3w-Post time-point. Decoding of the Context and the conjunction of Context and

Relationship were both significantly less accurate (X2(2) = 43.79, z=-17.78, -22.55, respectively, p’s

<.001). This suggests that at this time-point when the population encoding of this memory appears

more generalized, the encoding of a new feature, Context, may become embedded in this already

generalized memory code. However because I did not run this context manipulation prior to

completion of the original memory paradigm it is possible that the level of contextual encoding

observed is simply intrinsic to the prelimbic network and not contingent upon a consolidated

representation. Evidence that this alternative interpretation may not be the case can be found by

Page 89: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

78

closer inspection of the confusion matrix data. The errors made in Context classification are not

uniform across all conditions. Quantification of Context classification errors across the four

conditions reveals that errors seem to be specific to the CS-Paired conditions. The model confuses

Context-A: Tone-Paired with Context-B: Tone-Paired trials but makes very few errors between

Context-A: Tone-Alone and Context-B: Tone-Alone trials (see Figure 17). This suggests that the

population firing rates of these prelimbic neurons do convey accurate contextual information as the

model is accurately able to classify trials between the two contexts during the CS-Alone conditions.

It is possible that given the Context is not adding any useful or relevant information for this firmly

embedded CS-US association, contextual information is less accurately encoded in this

representation.

Figure 17. SVM decoding of Contextual Features.

Page 90: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

79

A) Confusion matrix shows accurate decoding as classification probability values along the diagonal with the predicted trial condition along the x-axis and the actual trial condition along the y-axis. Errors can be observed confusing Context A and B CS-Alone trials and Context A & B CS-Paired trials. B) Quantification of correct classification of the relational features (Relationship, Alone vs. Paired), contextual features (Context, A vs. B) and the conjunction of the two RxC, indicates better classification of the relational than contextual task features. C) Examination of specific contextual classification errors reveals few errors in classification of Context A and B in CS-Alone conditions, and greater percentage of errors in classification of Context A and B in CS-Paired conditions.

3.3 Discussion

3.3.1 Summary

Here I report the activity of multiple single neurons in the rat prelimbic medial prefrontal cortex

recorded across several weeks during a trace eyeblink conditioning procedure in which the stimulus

identity (Tone or Light) and stimulus relationship (Alone or Paired) differed across trials. Comparing

firing rates across these four conditions demonstrated the selectivity of these neurons for the various

features of the memory. A diversity of selectivity was observed, from neurons that are responsive

during the task but do not show selectivity for particular features, responding to both stimuli

regardless of whether it was predictive or not (Paired vs. Alone conditions), to neurons that were

selective to a particular stimulus in a particular condition (e.g. Tone-Paired). The proportions of these

different types of responsive neurons were calculated and were not observed to change at the single

neuron level from a period prior to acquisition of the memory association to a period 3 weeks after

the association was firmly expressed. However, changes were observed when I looked at the activity

at the population level. I showed that the population of prelimbic neurons is initially highly selective

for the particular stimulus features and less selective for the relational features. As the memory

association is gradually acquired this difference becomes less pronounced. The population of neurons

recorded 3 weeks after stable memory expression show the inverse relationship, being less selective

for the particular stimulus features and highly selective for the relational features. Finally, when a

new dimension was added to this stable memory through the manipulation of the context, these

Page 91: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

80

neurons did not show a rebound in highly selective encoding of this feature and instead displayed

less selective representations of this new information.

3.3.2 Single Neuron Selectivity

The mPFC is a brain region that has been extensively studied and in recent years its role in memory

has been advanced by many groups (Frankland & Bontempi, 2005, 2006; Insel & Takehara-

Nishiuchi, 2013; Szu-Han Wang & Morris, 2010; Weiss & Disterhoft, 2011).Whereas its role in

memory retrieval, especially at time-points remote to its formation, is convincing, its role in memory

formation is still fairly nebulous (Bero et al., 2014; Einarsson & Nader, 2012; Takehara-Nishiuchi et

al., 2006). It is also still very much unclear what type of information this region encodes as a memory

trace, how selective this region is to the specific features of a memory event and how this changes

with learning and memory consolidation. I observed the presence of neurons exhibiting a diversity

of selective responsiveness, the proportion of which did not change from the first few days of

conditioning to 30+ days later. The same result was obtained using both mean comparison and

mutual information. A similar result at the single neuron level has also been described previously,

that the proportion of memory selective neurons in the prelimbic cortex does not change throughout

learning, consolidation, and post-consolidation retrieval (Hattori et al., 2014; Takehara-Nishiuchi &

McNaughton, 2008).

3.3.3 Population level encoding

Although as the single neuron level, proportions of selective neurons do not appear to differ what

does appear to change is the magnitude of their response with learning and consolidation, and thus

the relative activity of a population of memory selective neurons increases (Hattori et al., 2014;

Takehara-Nishiuchi & McNaughton, 2008). I found a similar finding here, whereas no change was

observed in the percentage of selective neurons across time, at the population level differences began

to emerge. The activity of a population of neurons during the Tone-Paired condition was similar in

Page 92: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

81

both the Tone-Alone and Light-Paired conditions during the learning of the association, but by 3

weeks post learning, Tone-Paired activity was more similar to Light-Paired than Tone-Alone activity.

This suggests that at this 3 week post learning period, the population response is more selective for

the relationship of the stimulus (Alone vs. Paired) than for the particular physical features of the

stimulus (Tone vs. Light). This 3 week post learning time-point overlaps with the time point after

learning where prefrontal lesion and inactivation produces major impairments in memory retrieval

(Frankland et al., 2004; Maviel et al., 2004; Takehara et al., 2003; Takehara-Nishiuchi et al., 2005)

and after the time-period where mPFC plasticity is necessary for later memory retrieval (Takehara-

Nishiuchi et al., 2005, 2006; Teixeira et al., 2006; Gisella Vetere et al., 2011). To further elucidate

the differences in population activity across time I used a machine learning algorithm, support vector

machine (SVM). The algorithm was consistently accurate across all time-points, however differences

emerged in the classification of specific trial features. Correct classification of the identity of the

stimulus was very high early in the conditioning paradigm and gradually decreased throughout

learning and remained lower at 3 weeks post learning. Accurate classification of the relationship of

the stimulus followed the inverse, lower initially and gradually increasing up to the 3 week post

period. Correct classification of the conjunction of the identity and relationship peaked during the

learning phase and then decreased by 3 weeks post learning. These results imply that at the

population level, prelimbic neurons are initially highly selective for discrete stimulus features and as

the association is formed, the population becomes less responsive for these specific features and

becomes more generalized. At the same time, their response for the relevant more general relational

features of the stimulus increases.

3.3.4 Implications of the varying degrees and changes in selectivity observed

It was initially surprising to find the existence of neurons in the prelimbic mPFC that display such

highly selective activity for the conjunctive representation of a particular stimulus in a particular

Page 93: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

82

condition. The mPFC is hypothesized to be important for schema generation (Preston & Eichenbaum,

2013; Richards et al., 2014; van Kesteren et al., 2012; Wang et al., 2012), encoding rules (Rich &

Shapiro, 2009), and representing abstract features (Wallis et al., 2001). The changes in population

representations reported here generally support this view, showing that over the same time course

that the prelimbic mPFC becomes necessary for the retrieval of this memory, prelimbic population

activity becomes less selective for detailed physical stimulus properties and more selective for their

relational properties. However, while this pattern of changes is apparent, even at the remote time-

point assessed here, single prelimbic neurons still displayed highly selective stimulus responses and

the population could still be accurately decoded for both stimulus and relational properties. It is

unlikely that these results can be explained by the conditioning paradigm, in which training/retrieval

of memory took place daily over this entire period. A previous study directly compared prelimbic

responses in animals that were trained and then tested 4 weeks after learning versus animals that

underwent daily conditioning over the same time-period and revealed no differences in neuron

selectivity between the two cases (Takehara-Nishiuchi & McNaughton, 2008). Therefore the current

findings highlight an interesting property of the prelimbic cortex, a network of neurons containing

both highly selective and highly generalized responses within a brain region.

That the prelimbic regions contains both highly selective and generalizing neurons seems to counter

the hypothesis of the transformation of memory through the process of consolidation, wherein the

memory loses its discrete details, becoming more of an abstraction (Winocur et al., 2010). This

process is proposed to take place gradually, as related information becomes integrated within

knowledge frameworks (McClelland et al., 1995). However the population activity of the prelimbic

region does identify a possible neural correlate of this schematization or generalization of associative

memory that takes place within the window of presumed consolidation. Here the population activity

becomes less selective for the discrete stimulus features, generalizing across irrelevant dimensions of

information. At the same time the population becomes more selective for the relevant relational

Page 94: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

83

features of the stimuli. This suggests that the population contains the capacity to become more or less

selective for particular features of the memory over the course of consolidation driven by the

meaning and relevance of the features.

I observed a greater proportion of neurons in this network to be either non-selective or selective for

the relational stimulus features, than those that were selective for a particular stimulus or the

conjunction of a stimulus and its association. Similarly in primates, a majority of prefrontal neurons

were shown to be responsive to the rule of the task (Wallis et al., 2001). However, although not

highlighted, the same study does report the existence of prefrontal neurons selective for specific

stimuli, and the conjunction of a specific stimulus and a rule (Wallis et al., 2001), as was observed

here. The function of such highly selective cells is not clear, but I propose that this property may

support the formation, consolidation and maintenance of event memories within this network. It has

been demonstrated that incorporating new information into an existing long-term memory or schema,

occurs rapidly and requires the mPFC (Richards et al., 2014; Tse et al., 2007; Wang et al., 2012). It is

possible the highly selective activity observed here may support the incorporation of new information

into firmly embedded more generalized network activity by allowing the accurate comparison of new

information with previous experience.

Similarly it has been proposed that the mPFC is always involved in memory retrieval, playing a key

role in the representation of context, and in the authors’ case, context referring to any information

relevant for the current experience to guide adaptive behaviour (Euston, Gruber, & McNaughton,

2012). They hypothesize that for the retrieval of recently acquired memory, the role of mPFC is to

represent context, events and responses whereas the hippocampus stores the index linking these

features together. However, consolidation reorganizes these networks, and the mPFC acquires a more

important role in the storage of the memory, both representing and storing context-event-response

indices for remote memory retrieval (Euston et al., 2012). The theory accounts for the discrepancy in

Page 95: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

84

behavioural impairments from mPFC disruption at recent and remote time-points by suggesting that

compensation from other prefrontal regions and stronger hippocampal indices at the recent time-

point may permit memory retrieval without the mPFC, but a mPFC consolidated memory would

always require the mPFC for remote retrieval (Euston et al., 2012). Alternatively, according to this

model, where the mPFC both represents and stores contexts and events for remote memory, here the

presence of highly selective cells may represent mPFC indices whereas the presence of more

generalized activity may serve as the contextual representations.

However the evidence presented here shows the presence of both types of selectivity in the initial

stages of memory acquisition, and if these highly selective neurons represent cortical indices, they

are not gradually acquired as is suggested in the model. It is also possible that these neurons reflect

hippocampal computations in the initial acquisition of the memory, and their role in indexing or

representing cortical activation patterns is acquired through gradual consolidation. Both of these

postulates remain hypothetical, but present avenues of future investigation.

Here I show that as the memory is strengthened and consolidated the mPFC population code

becomes more generalized, less selective for behaviourally irrelevant features (Tone vs. Light) and

more selective for behaviourally relevant features (Alone vs. Paired). This generalization likely

reflects the extraction of the relevance among overlapping experiences. I also report that when

adding new information to the well-established associative memory, the network appears to encode

this information in a more generalized way, matching the present existing encoding strategy.

Following 30+ days of the trace eyeblink paradigm, I ran several animals in a similar procedure that

used the same stimulus for both sessions but manipulated the conditioning context. Here I showed

less selective encoding of the context specifically in the CS-Paired conditions. However because I

did not run a similar procedure before the original conditioning paradigm I cannot rule out that the

data observed during context manipulation is not simply the default code by which features of the

Page 96: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

85

environmental context are represented within the prelimbic region. I propose that given contextual

decoding errors were specific to the CS-paired conditions and I observed accurate decoding of

environmental contextual features in the CS-Alone conditions, information about the environmental

context can be represented by highly selective prelimbic activity. This suggests that the observed

results were not simply a default property of the prelimbic region but were a function of the existing

coding strategy. Again this may represent the extraction of relevance among the prelimbic ensemble.

These results fit the views of Euston et al. (2012) in that the extraction of information has less to do

with the particular mode of the information and more to do with its relevance for guiding adaptive

responding. Here the prelimbic population code became less selective for the identity of the stimuli

because the identity did not provide relevant information, and became more selective for the meaning

of the stimulus within a session. Changing the environmental context occurred when this pattern was

already established, and it is possible that the prelimbic code quickly recognized that this new

environment was not adding any new information to the relevance of the stimuli. Despite a new

environment, the same relationship existed within a session, first the stimulus was not relevant

(Alone trials), then suddenly it became relevant (Paired trials). Accordingly, if the relevance of the

stimulus did depend on the particular identity, or environmental context in which it was presented, I

predict that the medial prefrontal code would not generalize across these dimensions. This suggests

that the prefrontal code is selective for meaning, i.e. whether the current situation is relevant or not, a

proposition supported by findings that prefrontal neurons can develop selectivity to cues signaling

both reward (Mulder, Nordquist, Örgüt, & Pennartz, 2000; Peters, O’Donnell, & Carelli, 2005) and

punishment (Gilmartin & McEchron, 2005; Takehara-Nishiuchi & McNaughton, 2008). This

contrasts sharply from hippocampal encoding, which is selective for contextual details regardless of

their behavioural relevance (Wilson & McNaughton, 1993).

Page 97: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

86

These results together describe encoding properties of the prelimbic mPFC and highlight changes in

prelimbic representations that may reflect the reorganization of long-term memory networks during

the process of memory consolidation.

Page 98: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

87

Chapter 4

Prelimbic-cortical communication shaping changes in the selectivity of the prelimbic ensemble code with consolidation

4.1 Introduction

In the previous chapter I presented evidence that with learning and consolidation, the prelimbic

ensemble code gains a stronger selectivity for relational stimulus features while losing the

selectivity for physical stimulus features. Here, I sought to gain some insight into the network

mechanisms underlying these observed changes in encoding.

Over the same time-course in which I showed changes in prelimbic memory encoding, our lab

has previously shown that local field potentials (LFPs) in the lateral entorhinal (LEC) and

prelimbic medial prefrontal cortices (PrL mPFC) exhibit patterns of oscillatory activity

suggesting that their interaction is strengthened and becomes correlated with memory expression

over the course of learning and consolidation in trace eyeblink conditioning (Takehara-Nishiuchi

et al., 2012). This, along with the necessity of the LEC for the expression of old, consolidated

memory (Morrissey et al., 2012) suggest that the interaction between the LEC and mPFC may

play a critical role in memory consolidation and expression. Moreover, anatomically, the LEC

sits as the key transitional structure between the neocortex and hippocampus, and through its

reciprocal connectivity with the mPFC, may serve a similar function between the neocortex and

mPFC (Jones & Witter, 2007; Menno P. Witter, 2007). The LEC may, therefore, be optimally

positioned as an intermediary between a hippocampal driven memory acquisition network and a

mPFC driven consolidated memory network (Insel & Takehara-Nishiuchi, 2013; Takehara-

Nishiuchi, 2014). Together, these findings raise the possibility that the incoming inputs from the

Page 99: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

88

lateral entorhinal cortex provide neurons in the prelimbic cortex with necessary information to

generate and shape the memory-selective ensemble code during learning and consolidation.

Although our previous studies support the critical involvement of the LEC in consolidated

memory, the LEC is one of many regions that project to the prelimbic region. Therefore, it is

possible that afferent regions other than the LEC may also contribute to the shaping of prelimbic

ensemble codes with consolidation. For example, the ventral hippocampus (vHPC) sends a direct

mono-synaptic connection to the prelimbic region, and it has been shown in several studies that

their interaction increases during states of high anxiety (Adhikari et al., 2010, 2011), but also

during tasks requiring mnemonic demands, such as spatial working memory (Spellman et al.,

2015). Based on these findings, one theory posits that the connection between the ventral

hippocampus (vHPC) and mPFC is important for memory consolidation and to guide dorsal

hippocampal-mPFC communication in the development of mPFC memory representations

(Preston & Eichenbaum, 2013).

Another important afferent region is the perirhinal cortex (PER), which maintains very similar

patterns of connectivity with the prelimbic region as the LEC, showing strong direct reciprocal

connectivity with the prelimbic and most of the medial prefrontal cortex, but very little with the

caudal anterior and posterior cingulate areas (Hoover & Vertes, 2007; Jones & Witter, 2007).

The PER provides the LEC with the most dense input of all cortical regions (Kerr et al., 2007).

The PER has also recently been observed to be important in trace eyeblink memory acquisition

(Suter et al., 2013). Fewer studies have investigated the role of the PER in memory

consolidation, however plasticity in and communication between the PER and mPFC was shown

to be necessary for long-term memory retrieval in conjunctive recognition memory (Barker &

Warburton, 2008). In addition, disruption of the PER was also shown to impair fear context

Page 100: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

89

memory up to 100 days after training, suggesting that similar to the LEC, the PER may continue

to play an active role in the retrieval of consolidated memory (Burwell, 2004). Recordings of

these regions have also revealed that medial prefrontal-perirhinal learning dependent interactions

promotes enhanced perirhinal-entorhinal communication, an important pathway for the transfer

of hippocampal activity to neocortical sites (Paz et al., 2007).

Here I sought to systematically examine the modulation of prelimbic neuron firings by these

afferent regions by looking at the phase-locked firing of prelimbic neurons to local field potential

oscillations in these afferent regions. Previous studies reveal that oscillatory neuronal activity

reflects fluctuations in the excitability of local neurons (that is, fluctuations in membrane

potential shifts), thereby affecting the timing of local neurons’ firing (i.e., their output) as well as

their sensitivity to incoming signals (Buzsaki, 2004; Siapas et al., 2005; Sirota et al., 2008).

Therefore, if a prelimbic neuron is modulated by incoming inputs from an afferent region, it

should preferentially fire when the oscillation in the afferent region reaches the depolarized state

in which afferent neurons are most likely to fire. Based on this, I quantified the degree of

modulation on each neuron by examining the phase-locked firing to theta oscillations in the

lateral entorhinal, perirhinal cortex, and ventral hippocampus. Theta oscillations were targeted

because of our observation of learning related interactions between the LEC, mPFC, and dorsal

hippocampus at this frequency (Takehara-Nishiuchi et al., 2012) in addition to other studies

reporting phase-locked firings of mPFC neurons to ventral hippocampal theta oscillations

(Adhikari et al., 2011).

I first categorized prelimbic neurons based on the particular afferent region that modulated their

firing activity. This analysis revealed that individual prelimbic neurons show preferential phase-

locked activity to particular afferent regions. Based on this finding, I categorized neurons into

Page 101: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

90

“functional cell types”, such as LEC-, PER- and vHPC-modulated cells. Subsequently, I then

performed the same sets of analysis used in the previous Chapter to these cell types separately, to

examine whether memory representations and its changes with consolidation differ depending on

the functional cell type.

4.2 Results

Animals were rats described in Chapter 3 with functioning tetrodes located in the prelimbic

medial prefrontal cortex and functioning LFP electrodes located in the lateral entorhinal cortex

(LEC, n=4), perirhinal cortex (PER, n=4), and ventral hippocampus (vHPC, n=4) (see Figure

18).

Rats underwent daily conditioning in the trace eyeblink paradigm for 30+ days (see Chapter

2.5.1 section for a clear overview of the procedure) with activity recorded throughout from

tetrodes and LFP electrodes (see Chapter 2.5.3). The paradigm used two different conditioned

stimuli (Tone & Light) in two different relational conditions (Alone & Paired) to allow

comparisons across 4 conditions that differed in their combination of the identity and

relationship of the stimulus (Tone-Alone; Tone-Paired; Light-Alone; Light-Paired).

4.2.1 Prelimbic neuron phase-locking to the rhinal cortices and hippocampus

To examine the coupling of prelimbic neuron spiking activity and local field potential (LFP)

activity from the various brain regions I looked at the phase-modulated firing (phase-locking) of

prelimbic (PrL) neurons to theta oscillations recorded from the lateral entorhinal cortex (LEC),

perirhinal cortex (PER) and ventral hippocampus (vHPC). By combining spike and LFP data

together I can assess how the activity of individual neurons within the PrL is modulated by the

oscillatory activity in other brain regions involved in this task. This phase-locking of neurons to

Page 102: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

91

oscillations has been a proposed mechanism for the coordinated communication between

connected brain regions (Vinck et al., 2012) and may be important for cell assembly formation

(Canolty et al., 2010; Singer, 1999) and information coding mechanisms. This analysis allowed

me to quantify the proportion of PrL neurons that are significantly phase-locked to theta

oscillations in the three brain regions sampled, and the degree to which they are modulated.

Looking specifically at the interval between the CS and US, I observed phase-locking in PrL

neurons to all three brain regions (see Figure 18).

Similar proportions of prelimbic neurons phase-locked to theta oscillations in each region were

observed (LEC: 26.8%; PER: 27.4%; vHPC: 25.3%, see Figure 18). It was possible that there

was considerable overlap between regions and many of the same cells were showing significant

phase-locking to more than one region. To look at selectivity directly, instead of comparing

significant versus non-significant, I quantified the degree of phase-locking with the Rayleigh z

value. This is the log transformed mean resultant vector length (MRL). MRL is the sum of the

vectors representing the phases at which each spike occurred, divided by the number of spikes.

This value quantifies the degree to which a cells spikes occur at a particular phase of the

oscillation, or the strength of locking. Therefore I calculated the mean Rayleigh z value of

significantly locked cells for each region, and that cell’s Rayleigh z value with the other two

regions. For example, if cell 1 was significantly phase-locked to theta oscillations in region A, I

took the Rayleigh z value of cell 1 with region A as well as the Rayleigh z value of cell 1 with

regions B and C. Likewise, if cell 2 was significantly phase-locked to theta oscillations in region

B, I took the Rayleigh z value of cell 2 with region B, as well as region C and A. This was done

for all cells that displayed significant phase-locking with each region and the averages were

plotted. These data are represented in Figure 19, which reveals a high degree of specificity for

which regions theta oscillations a particular cells firing activity was modulated by. PrL neurons

Page 103: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

92

phase-locked to LEC theta showed on average significantly weaker modulation to PER and

vHPC theta. The same pattern was observed for PrL neurons phase-locked to PER theta and

vHPC theta.

Together these data show that the activity of neurons in the PrL are modulated by the oscillatory

activity in the LEC, PER, and vHPC in this task. Additionally, individual PrL neurons display a

preference for which region’s oscillatory activity they are phase-locked to. This leads to the

possibility that fairly distinct populations of neurons within the PrL synchronize with different

cortical structures. If this is the case, these distinct populations may be encoding different

information relevant for the task.

Page 104: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

93

Figure 18. Histology and examples of prefrontal locking and the proportion of locked cells.

A) Representative example of local field potential (LFP) electrode tips in their target locations (left column) and illustrations of all verified electrode positions in the lateral entorhinal cortex (LEC, top), the perirhinal cortex (PER, middle), and the ventral hippocampus (vHPC, bottom). On the right are example LFP traces from each region (black) overlayed with the filtered theta (7-11 Hz) trace (blue line) from a 2 second period centered on the onset of the conditioned stimulus (CS). Black bar indicates the stimulation from the unconditioned stimulus (US) during which data was not analyzed due to noise artifact from the shock. On the far right are three representative examples of different individual prelimbic neurons that were significantly phase-locked to theta oscillations from the three regions as demonstrated by spike counts across the different phases of the oscillations. B) The proportion of neurons that were significantly phase-locked to each of the three regions theta oscillations.

Page 105: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

94

4.2.2 Prelimbic phase-locking is not feature specific

To assess whether the different populations of neurons were differentially encoding task relevant

information, I looked at patterns of phase-locking in individual neurons categorized by their

selectivity as described in detail in Chapter 2.6.3. These categories defined responsive cells

selectivity to the different task features in the 4 conditions (1-Alone: CS only trials; 2-Paired:

CS-US trials; 3-Tone: Tone CS trials; 4-Light: Light CS trials) resulting in 4 types of selectivity

(1-Non-selective: Alone=Paired & Tone= Light; 2-Identity-selective: Alone=Paired & Tone ≠

Light; 3-Relationship-selective: Alone ≠ Paired & Tone = Light; 4-Conjunctive RxI: Alone ≠

Paired & Tone ≠ Light) plus cells that were not responsive (Non-responsive). Taking the

populations of cells phase-locked to each region, I calculated the proportions of the different

types of selectivity observed within these populations. For example, of all cells phase-locked to

region A, how many showed Identity-selective activity out of all Identity-selective cells? These

analyses revealed no differences in the type of information neurons phase-locked to the different

regions were selective for (see Figure 19). The only categories whose proportions were

significantly greater than Non-responsive cells were found to be locked to the LEC (Non-

selective, Identity, and Conjunctive) and PER (Conjunctive). However there were no differences

among the selectivity of responsive cells or between proportions locked to the different regions.

Here we do not see any differences in phase-locking based on single neurons selectivity for task

features, and at least at the single neuron level, neurons coding a particular feature of the task

were no more likely to be phase-locked to a particular cortical region.

Page 106: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

95

Figure 19. Prelimbic selectivity of phase-locking to specific regions and type of information

encoded by these cells.

A) Left column, each row displays that the averaged Rayleigh z value of prelimbic neurons significantly phase-locked to theta oscillations in a specific region (lateral entorhinal cortex LEC-modulated, top; perirhinal cortex PER-modulated, middle; ventral hippocampus vHPC-modulated, bottom) were

Page 107: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

96

significantly greater than the averaged Rayleigh z value of each of these cells phase modulation to the other two regions. Dotted lines indicate the typical threshold for Rayleigh z values critical values at p < .05 (z = 2.9) and p < .01 (z = 4.5). Error bars represent standard error B) Right column, shows the proportion of prelimbic neurons selective for the different task features that were phase-locked to the different theta oscillations (lateral entorhinal cortex LEC, top; perirhinal cortex PER, middle; ventral hippocampus vHPC, bottom). Error bars represent 95% Copper-Pearson confidence intervals. Red dotted line signifies the cut-off for difference compared to the proportion of phase-locked neurons that were Non-responsive in the task.

4.2.3 Regional phase-locking does not change across the learning stages

Though these previous analyses looked at all neurons recorded across learning, consolidation and

post-consolidation retrieval, it is possible that differences may emerge when looking at specific

time-points along this continuum of the memories existence. This would offer potential

mechanisms through which PrL representations change as the memory becomes consolidated

within the PrL network. Such differences may be related to the shift I observed in encoding

within the PrL population reported earlier. Differences in phase-locking across time was assessed

by calculating the proportion of neurons phase-locked to theta oscillations in the different

regions across the behaviourally defined learning periods reported earlier (see Chapter 2.6.1).

Figure 20 displays the proportion of total neurons recorded in each phase of learning, from prior

to acquisition of the association (Before), to during learning (During), to 3 weeks after stable

responding (3w-Post), that were significantly phase-locked to each region. Chi-Square analysis

revealed no differences across time, although there does appear to be a slight decrease in PrL

neurons locked to LEC and PER theta at the 3w-Post time-point compared to during acquisition.

Page 108: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

97

Figure 20. Phase-locking across learning stages.

The proportion of prelimbic neurons phase-locked to the lateral entorhinal cortex (LEC), perirhinal cortex (PER) ventral hippocampus (vHPC) calculated at three behaviourally defined periods of learning. There were no differences in the proportions of neurons locked to the different regions between all days prior to displaying a conditioned response (CR) in 30% of trials (Before), to all days in between Before and when the rat displayed a CR in 60% of trials for 2 consecutive days (During), to the period encompassing three weeks following the end of the During stage (3w-Post). Error bars indicate 95% Clopper-Pearson confidence intervals.

4.2.4 Decoding of region-specific phase-locking populations across time

Earlier I reported that at the single neuron level, the proportion of neurons selective for specific

task features does not change across the different learning and retrieval stages. However at the

population level, differences began to emerge. Therefore it was not entirely surprising that at

least proportionally, differences were not evident in the selectivity of phase-locked neurons.

However the possibility still remains that differences may emerge when taking these region-

specific phase-locked neurons as a population. Here I used a supervised machine learning

algorithm, support vector machine (SVM), to decode the information contained within

populations of cortical theta phase-locked prelimbic neurons separated by regional modulation

(LEC-modulated, PER-modulated, vHPC-modulated) and the learning stage in which they were

Page 109: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

98

recorded. Populations of neurons phase-locked to LEC, PER, and vHPC theta recorded from

different stages of the memory were fed independently into the SVM decoder and the ability of

the decoder to correctly classify trial conditions given only the information contained within

these different populations was assessed. Because a fair sample size is required to obtain

consistent classification with this data (n > 50 from earlier analyses), time-periods of learning

had to be divided more liberally than in previous analyses to ensure enough neurons were being

sampled from each time-point (fewer neurons from Before period) to achieve accurate decoding

and to prevent significant sample-bias when down-sampling from larger population time-periods

(many neurons from During period). In previous SVM analyses ( see Chapter 3.2.5) the During

learning period was divided in half with Early learning comprising the first half, and Late

learning comprising the second half. The Early period was more similar to the Before period, and

the Late period was more similar to the 1w-Post period (1 week after asymptote). Similarly, data

from 2 and 3 weeks after asymptote (2w-Post & 3w-Post) produced similar patterns of results

(see Figures 12 & 14). Therefore I combined the Before and the Early phase of the During

learning stage into one population (Early). I combined the Late phase of During learning with

1w-Post into another population (Late), and I combined the 2w-Post and 3w-Post into a third

population (Post). I then ran SVM on these three time-periods with data from prelimbic neurons

phase-locked to the different regions with the same number of neurons selected from each region

for each time-point (n = 80).

Figure 21 reveals the quantification of the accuracy of classification from these populations for

the different features of the task (Relationship, Identity, Conjunction RxI) visualized across

learning stages (Figure 21, A) and across regions (Figure 21, B) relative to the chance level of

prediction (50% Relationship/Identity; 25% for RxI).

Page 110: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

99

First, separate chi-square analyses were run on each type of information, Relationship (X2(6) =

56.459, p < .001), Identity (X2(6) = 61.165, p < .001) and RxI (X2(6) = 113.472, p < .001), and

were all significant. Post-hoc analyses compared individual proportions within each region

across time. For classification of the Relationship, LEC- and PER-modulated cells significantly

decrease from Early to Late and Post learning (LEC z’s = -6.78 & -11.11, p’s < .001,

respectively; PER z’s = -6.03 & -10.67, p’s <.001, respectively). In contrast, neurons phase-

locked to vHPC theta significantly increase Relationship classification accuracy Post learning

compared to Early and Late learning (z’s = 23.63 & 24.32, p’s < .001, respectively).

For classification of the Identity, LEC- and PER-modulated cells significantly decrease from

Early to Late learning and remain lower during Post learning (LEC z’s = -26.56 & -13.43, p’s <

.001, respectively; PER z’s = -7.76 & -7.90, p’s <.001, respectively). vHPC-modulated cells also

significantly decrease in Identity classification accuracy from Early to Late learning (z = -5.01, p

< .001).

For the conjunction of Relationship and Identity the exact same patterns was observed as for

Identity classification, with LEC- and PER-modulated neurons decreasing in accuracy from

Early to Late learning, and remaining lower Post learning (LEC z’s = -34.86 & 24.62, p’s <.001,

respectively; PER z’s = -6.41 & -7.27, p’s <. 001, respectively). vHPC-modulated neurons

classification accuracy of the RxI conjunction also decreased from Early to Late learning (z = -

6.31, p < .001) but return to Early learning levels by Post learning.

These results compared the selectivity of the population code across memory stages in each

functional cell-type, and observed that with learning and consolidation, the selectivity for the

physical features of the stimuli (their identity), was reduced. In parallel, the selectivity for the

relational stimulus features (their relationship) was reduced in LEC- and PER-modulated cells

Page 111: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

100

while it was increased in vHPC-modulated cells. The selectivity for specific trial types, the

conjunction of the stimulus relationship and its identity, was also reduced in the LEC- and PER-

modulated cells while it was maintained and strengthened in vHPC-modulated cells.

These results suggest that the enhanced selectivity for relational features was driven by sub-

groups of prelimbic neurons whose firings are modulated by the activity of the ventral

hippocampus. In contrast, the weakened selectivity for physical stimulus features was observed

in all functional cell types, indicating that it may result from local network changes rather than

the interaction with specific afferent regions.

Next, separate chi-square analyses were run on each time-period, Early (X2(4) = 15.645, p =

.004), Late (X2(4) = 65.468, p < .001) and Post (X2(4) = 14.905, p = .005), and all were

significant. Post-hoc analyses compared selectivity between neurons phased-locked to the

different regions. For the Early phase of learning, LEC- and PER-modulated cells were both

more selective for the Relationship of the stimulus than vHPC-modulated cells (z’s = 14.55 &

18.15, p’s < .001), whereas LEC-modulated cells were more selective than both PER- and

vHPC-modulated cells for the Identity of the stimulus (z’s = 9.64 & 12.53, p’s < .001). LEC- and

PER-modulated cells were similarly more selective for the conjunction of Relationship and

Identity (RxI) than vHPC-modulated cells (z’s = 21.59 & 16.64, p’s < .001).

Comparison across the Late learning phase again revealed higher accuracy among LEC- and

PER-modulated cells for the stimulus Relationship versus vHPC-modulated cells (z’s = 7.76 &

5.64, p’s < .001). However here, among LEC-modulated cells, selectivity for the Identity was

significantly lower than PER- and vHPC-modulated cells (z’s = -14.11 & -10.39, p’s < .001).

PER-modulated cells represented the highest selectivity for the RxI conjunction compared to

LEC- and vHPC-modulated cells (z’s = 19.93 & 10.52, p’s < .001).

Page 112: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

101

Finally, at the Post learning period vHPC-modulated cells were more selective than LEC- and

PER-modulated cells for the Relationship (z’s = 9.62 & 5.10, p’s < .001), Identity (z’s = 9.54 &

5.49, p’s < .001) and their conjunction (RxI, z’s = 21.12 & 8.02, p’s < .001).

These results compared population selectivity across functional cell-types in each stage, and

reveal that during the early stages of learning, selectivity for the stimulus relationship was higher

in the LEC- and PER-modulated cells than in vHPC-modulated cells while the selectivity for

identity was the highest among LEC-modulated cells. Therefore the highest accuracy rate in

decoding specific trial types (RxI) was observed in LEC-modulated cells, followed by PER-and

then vHPC-modulated cells. During the late stages of learning, the selectivity for relationship

remained higher in the LEC- and PER-modulated cells than in the vHPC-modulated cells. The

selectivity for the stimulus identity was observed to significantly drop in LEC-modulated cells

but remained in PER- and vHPC-modulated cells. This resulted in a higher accuracy rate in

decoding specific trial types (RxI) in the PER-modulated cells than the other cell-types.

After consolidation, the selectivity for relationship significantly increased in the vHPC-

modulated cells, making them more selective than the other cell types, whereas the selectivity for

identity was reduced in general across the three cell-types. This resulted in a higher accuracy rate

in decoding specific trial types (RxI) in the vHPC-modulated cells, followed by PER- and then

LEC-modulated cells.

Together these results highlight network changes across learning and consolidation that appear to

support the changes in population encoding revealed in Chapter 3 across time.

Page 113: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

102

Figure 21. SVM Classification Accuracy for Stimulus Features Across Time as a Function

of Functional Cell Types.

A) Percentage of accurate classification for the different features of the task, Paired vs. Alone (Relationship), Tone vs. Light (Identity) and the conjunction of the two (RxI) across the defined learning stages. Accuracy in classification of the stimulus Relationship (top left) increased from Late (medium grey) to Post learning (dark grey) in neurons phase-locked to ventral hippocampal theta oscillations (vHPC). In contrast accuracy in classification of the stimulus Identity (top right) decreased from Early (light grey) to Late learning in neurons phase-locked to the lateral entorhinal cortex (LEC) and perirhinal cortex (PER). Classification accuracy of the conjunction of Relationship and Identity (RxI) mirrored these findings, decreasing from Early to Late learning in LEC- and PER-modulated neurons, with no change from Early to Post learning in vHPC- modulated neurons but a decrease during Late learning. B) Percentage of accurate classification for the different features of the task across the defined functional cell types. During Early learning (left), LEC-(green) and PER-modulated(blue) cells were more selective for the Relationship of the stimulus. LEC-modulated cells were highly selective for the Identity and the conjunction of the two (RxI). By Late learning (middle) LEC- modulated cells lost selectivity for the Identity, and PER- modulated were the most accurate for specific stimulus features (RxI). During Post learning periods (right), vHPC-modulated (purple) cells were more selective for all features, particularly the Relationship. * p < .005 using the z test to compare chi-square cells. Error bars indicate 99% Clopper-Pearson confidence intervals. n = 80 for each cell type and time-point.

Page 114: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

103

4.3 Discussion

4.3.1 Summary

Here I report theta modulated activity in single prelimbic medial prefrontal cortex (PrL mPFC)

neurons to oscillations in the lateral entorhinal (LEC), perirhinal (PER), and ventral

hippocampus (vHPC) throughout the learning, consolidation and retrieval of an associative

memory in trace eyeblink conditioning (Figure 18). I reveal the presence of selective populations

of PrL neurons modulated by theta oscillations in these different afferent regions. Based on this

finding, neurons were categorized into distinct functional cell types created from their

preferential phase-locking as LEC-modulated, PER-modulated, or vHPC-modulated (Figure 19).

The proportions of each cell type (Figure 18) and the types of selectivity of single neuron activity

within each type were comparable (Figure 19). However the population activity of these distinct

cell populations revealed changes in selectivity for particular features of the memory throughout

learning and consolidation (Figure 20). This was assessed by the ability of a machine learning

algorithm to accurately classify trial conditions given only the firing activity of discrete sets of

populations of single neurons. Here accurate classification implied that enough information

about the various task features was present within the population spiking activity, whereas errors

implied missing information, and taken together they reveal a way of assessing the

representations actually contained within these networks.

These findings highlight a potential dynamic through which prelimbic representations of the

associative memory evolve over its learning and consolidation in a long-term network. Neurons

phase-locked to LEC and PER theta oscillations show very accurate classification in the Early

phases of learning for Relational and Conjunctive features of the stimulus, with LEC-modulated

cells also strongly selective for features of the stimulus Identity. In contrast neurons phase-

Page 115: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

104

locked to vHPC theta oscillations provide more accurate classification of the stimulus

Relationship at the Post learning period, overlapping with the point at which this memories

consolidation is presumed to be stabilized. These changes were not observed when looking at

single neuron selectivity proportions, but emerged when taking the population activity from

neurons phase-locked to particular regions. These interactions could serve as a potential

mechanism through which this memory becomes more generalized over-time, more selective for

the predictive features of the stimuli and less-selective for the particular details of the stimuli by

changes in patterns of communication with the ventral hippocampus and rhinal cortices.

Together these findings point to a network dynamic that may shape PrL memory representations

as part of a long-term memory framework.

4.3.2 Functionally defined cell types in the prelimbic cortex

Phase-locking of individual neurons spiking patterns to the oscillations of a different brain region

has been proposed to support the flexible routing of information across regions by coordinating

spike time and aiding in the creation of functional cell assemblies (Buzsaki, 2004; Fell &

Axmacher, 2011; Fries, 2005; Harris & Gordon, 2015; Harris, Csicsvari, Hirase, Dragoi, &

Buzsaki, 2003; Singer, 1999). Here I report that the spiking activity of many neurons in the PrL

is correlated with the phase of ongoing oscillations in the LEC, PER and vHPC during learning

and retrieval. Synchronized mPFC spiking patterns to hippocampal and ventral hippocampal

theta oscillations has been well established (Adhikari et al., 2010, 2011; Siapas et al., 2005), and

the proportion of phase-locked PrL neurons reported here is similar to what has been described

(Siapas et al., 2005). A similar proportion of PrL neurons were found to be phase-locked to the

three targeted brain regions which raised the possibility that the same cells were locked to theta

oscillations in all three regions. However looking at the degree of phase-locking revealed what

Page 116: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

105

appeared to be distinct populations of PrL neurons preferentially phase-locked to different

regions. This finding suggested that prelimbic neurons could be categorized into distinct

functional cell types based on their preferential phase-locking to theta oscillation in a particular

afferent region.

Previous studies have categorized cells for example based on differences in neurotransmitter

release or the expression of particular proteins. To my knowledge, this study is the first to

categorize prelimbic neurons based on which afferent region they are modulated by. This

complements recent studies that define the function of cells based on their projection targets

(Courtin et al., 2014; Rajasethupathy et al., 2015; Senn et al., 2014). It is important to note that

our approach here was to categorize neurons based on the incoming inputs, whereas these other

works categorize neurons based on their output targets. Therefore, here we are examining how

the local computation/representation is modulated by incoming inputs, rather than how prefrontal

computations affect the efferent targets.

When looking at these defined sub-populations, there did not appear to be differences in terms of

the type of information they were selective for when looking at the proportions of individual

neuron selectivity. There was also no change in the relative proportions of phase-locked neurons

to each region across learning, consolidation and over-training. However the population activity

of these different ensembles were observed to represent varying degrees of selectivity for the

different task features that changed across the different learning stages.

The fact that at the single neuron level, the proportions of cells phase-locked to the different

regions did not change across the learning stages was initially surprising, given for example the

evidence for prefrontal-entorhinal interactions in long-term retrieval of event memories and

possibly their consolidation (Paz et al., 2007; Takehara-Nishiuchi et al., 2012). However there is

Page 117: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

106

likely more dynamic and complex interactions taking place between these regions than can be

captured simply by the number of neurons observed to be phase-locked, as was discovered in the

previous Chapter when looking at selectivity. The previous finding that LEC-mPFC interactions

increase over consolidation also revealed that they still interact even early in learning,

demonstrated through theta synchronization (Takehara-Nishiuchi et al., 2012). However it was

the relative strength of this synchrony during trials in which the memory was expressed

compared to those in which it was not that increased with learning, and not simply their

interaction in general (Takehara-Nishiuchi et al., 2012). Therefore this does not necessarily

expect that the proportion of neurons phase-locked to LEC theta would change across time.

That the proportions of neurons selective for particular features of the task did not differ within

populations of prelimbic neurons phase-locked to the different regions was also not particularly

surprising for the same reason. I showed in Chapter 3 that despite no changes in the proportion of

selective neurons across time, changes in encoding at the population level were evident. Here,

despite similar proportions of neurons selective for different task features, the population activity

of these different functional cell types did differ, both between groups and across time. The

population of LEC-modulated cells was shown to contain highly accurate information about the

identity of particular stimuli as well as their relational meaning in a given trial in the very early

stages of learning. This selectivity was lost by the later stages of learning and post-consolidation,

particularly for the physical identity features. In contrast, populations of cells modulated by

vHPC theta oscillations (vHPC-modulated) contained much weaker selectivity for these features

very early on in learning, but acquired strong selectivity for the relational features of the stimuli

after the consolidation of the memory. Populations of PER-modulated cells showed similarities

to both LEC- and vHPC-modulated cells, like LEC-modulated cells they appeared to contain

initially high selectivity for relational features which decreased with time, but their selectivity for

Page 118: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

107

identity features more resembled that of vHPC-modulated cells, decreasing slightly with learning

and then remained at a stable level. These changes may reflect the shifting networks responsible

for the retrieval of the memory, particularly as hippocampal output becomes less important for

the memory retrieval across time.

Looking at the encoding of discrete memory features, I observed that the information about

specific conditions (memory with specific details) represented within the activity of different

phase-locked cell populations shifted among these populations across time. Specifically, the

ensemble of LEC-modulated cells could be accurately decoded for specific trial types in the early

days of the learning paradigm, with high selectivity for both the Relationship and Identity of the

stimuli. By the post-consolidation phase, the population of vHPC-modulated cells became the

most selective for specific trial conditions, driven mainly by their increased selectivity for the

Relationship of the stimuli. During the transition period between these two stages, LEC-

modulated cells had significantly decreased the selectivity for both Identity and Relationship

features while vHPC-cells had yet to acquire stronger Relationship selectivity. Here PER-

modulated cells showed the highest selectivity among the three cell-types despite lower

selectivity for the Identity and the Relationship compared to the early phase.

It is important to note the high accuracy rate across regions at the different time-points was

achieved by different sources. The high accuracy rate of LEC-modulated cells was driven by

their selectivity for both relational and physical features while that of vHPC-modulated cells was

driven by the selectivity for relational features alone.

Page 119: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

108

4.3.3 Implications of the role of each afferent input during memory acquisition and retrieval

The prelimbic mPFC is shown to be involved in event memory from its initial formation,

however the ensemble code for an association is observed to develop gradually over-time

(Hattori et al., 2014; Takehara-Nishiuchi & McNaughton, 2008). The information contained

within populations of prelimbic (PrL) neurons phase-locked to the different regions of the

hippocampal system targeted here, the lateral entorhinal (LEC), perirhinal (PER) and ventral

hippocampus (vHPC), was observed to change across behaviourally defined periods of learning

and retrieval.

The present study demonstrated that the input from each afferent region may evoke prelimbic

ensemble activity which contains differing levels of selectivity for specific features of experience

at different time-points. This point may provide insight into neurophysiological underpinnings of

changes in memory representation through memory consolidation. Previous studies show that

specific memory is mediated by the hippocampus, but that more generalized memory is mediated

by the neocortex (see McClelland et al., 1995; Winocur et al., 2010). Our study overall indicates

that both specific and generalized representations exist within the ensemble code in the prelimbic

region, though the specificity is mainly driven by the behaviourally relevant features.

In the initial days of acquisition in this paradigm, a time-point when hippocampal computations

are essential to the formation of the memory (Moyer et al., 1990; Weible et al., 2006; Weiss et

al., 1996; Weiss et al., 1999), LEC-modulated prelimbic ensemble activity is highly selective to

all task features. This may reflect highly selective dorsal hippocampal encoding and output

through the entorhinal region during the initial formation of the memory association. As the

association becomes firmly and stably acquired, the particular features of the stimuli become less

relevant, as they do not contain any unique predictive information about the outcome. At this

Page 120: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

109

time-point LEC-modulated cells sharply decline in the information about the particular stimuli

encoded in their population activity, mirroring the decreased selectivity observed overall within

the prelimbic population reported in Chapter 3. Although decreased, identity information is still

maintained in PER- and vHPC-modulated cells at this time, as is relational selectivity by LEC-

and PER-modulated cell groups. This time-point represents a very important stage in the

memory’s representation. It overlaps with what can be described as a transitional period of the

memory networks responsible for retrieval of the memory as retrieval becomes less dependent on

hippocampal function and more dependent on mPFC functioning (Takehara et al., 2003).

In the several weeks following this period, the memory is presumed to be consolidated within the

mPFC network, as hippocampal dysfunction has no impact on retrieval, and mPFC dysfunction

has devastating effects (Kim et al., 1995; Oswald et al., 2010; Kaori Takehara et al., 2003). The

disruption of NMDA-receptor mediated plasticity mechanisms within the prelimbic region also

no longer has any effect on memory expression at this time-point (Takehara-Nishiuchi et al.,

2006). Here the major change observed was an increase in selectivity of vHPC-modulated cells

to the relational features of the memory, while again selectivity of stimulus identity features

remained relatively lower within all three cell populations. These changes appear to reflect the

overall encoding of the prelimbic ensemble which becomes more generalized for specific

stimulus features at this time-point and more selective for their relevance.

It is therefore possible that the specific representation of experience here was driven by the

incoming inputs from the lateral entorhinal cortex early in learning, and maintained thereafter by

perirhinal input. Whereas the acquired generalized representation was driven by decreased

selectivity from lateral entorhinal input as the hippocampus becomes less engaged in the retrieval

of the memory. Conversely, the input from the ventral hippocampus may serve to sharpen the

Page 121: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

110

selectivity for the behaviorally relevant dimension of the memory after the consolidation process

has been completed.

4.3.4 Differences in selectivity between functional cell-types

I show that mPFC neurons phase-locked to ventral hippocampal theta during over-training

become more selective for the relational features of the stimuli. At this same time-point the

prelimbic population code displayed a similar pattern, suggesting that this population of vHPC-

modulated PrL cells may be driving this shift. This population of neurons seem to be important

for sharpening PrL selectivity to behaviourally relevant task features, distinguishing Alone from

Paired trials. The activity of ventral hippocampal neurons has been implicated in the formation

of schema like representations, generalizing across related events but distinguishing between

relevant contexts (Komorowski et al., 2013). Such input to the mPFC could presumably shape

activity towards the relevance of the stimuli (relational features) as opposed to their discrete

details. Considerable evidence also links ventral hippocampal activity to anxiety (reviewed in

Bannerman et al., 2004), ventral hippocampal-prefrontal synchrony has been observed during

anxiety behaviours (Adhikari et al., 2010) and is implicated in signaling the salience of a context.

Here the vHPC may be providing an anxiety signal to focus attention to the stimulus during the

behaviourally relevant Paired trials. However anxiety should be reduced rather than heightened

when the shock becomes predictable. So this explanation does not fit the finding of a

strengthened selectivity for the relationship of the stimulus from vHPC afferents after the

consolidation of the memory.

On the other hand, the Komorowski et al. (2013) findings demonstrate that vHPC neurons form

generalized representations across meaningfully distinct spatial contexts, suggesting this activity

is not merely an anxiety signal but may generate contextual representations signaling meaning.

Page 122: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

111

In the task employed in my thesis, the context may be the temporal order of the predictive value

of the CS, not predictive in the first 20 trials, and suddenly predicting the US in the latter 80

trials.

In contrast to vHPC-modulated population activity, neurons phase-locking to the rhinal cortices

(LEC and PER) became more generalized for all task features with over-training, but were

highly selective early in learning. However there were also differences observed between these

neuron populations. LEC-modulated cells were selective for all task features early in learning,

whereas PER-modulated cells did not show this same level of selectivity, particularly for the

Identity and Conjunctive trial features (RxI). It is possible that the high selectivity among LEC-

modulated cells directly reflects dorsal hippocampal computations and output during this early

phase of acquisition. The lateral entorhinal cortex receives direct projections from these

hippocampal computations through CA1 output targeting the deep entorhinal layers. This same

layer of entorhinal cells then forms the major source of projections to the PrL cortex (Hoover &

Vertes, 2007; Insausti et al., 1997; van Strien, Cappaert, & Witter, 2009). The perirhinal cortex

on the other hand, while receiving some direct hippocampal output, receives most of this output

indirectly through strong connectivity with the lateral entorhinal cortex (Burwell & Amaral,

1998b). Perirhinal-prelimbic afferents may therefore contain less detailed information than

lateral entorhinal-prelimbic afferents in this early stage of acquisition. These lateral entorhinal-

prelimbic afferents may therefore be important for prelimbic acquisition of the association, as it

is proposed that prelimbic memory representations extending beyond the working memory

capacity require hippocampal input during encoding (Euston et al., 2012). As the association is

gradually acquired and stabilized, hippocampal output becomes less important (Takehara et al.,

2003), and here I show that prelimbic representations become more selective for meaningful task

dimensions and more generalized for regularities among non-relevant dimensions. The decreased

Page 123: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

112

selectivity among all cell groups at this time-point reflects this, and particularly the decreased

selectivity of LEC-modulated cells may directly represent this transition in the network.

However perirhinal afferents may be less affected by this shift than LEC afferents and continue

to provide trial specific information.

Together I reveal changes in the feature encoding of populations of PrL neurons phase-locked to

different connected structures that mirrors the changes in the general prefrontal population code

over the course of learning and consolidation.

Page 124: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

113

Discussion

5.1 Summary

The role of the medial prefrontal cortex (mPFC) in the consolidation, long-term storage, and

retrieval of event memory in the brain has become an important area of focus in the neuroscience

of memory. Evidence of the importance of this brain region for the retrieval of old but not newly

acquired memories highlighted this collection of structures as a unique node in the episodic

memory network (Frankland et al., 2004; Maviel et al., 2004; Takehara et al., 2003). The mPFC

has also been implicated in incorporating new similar information into existing memory

frameworks (Wang et al., 2012) suggesting that this region may help form and update schema-

like mental representations of our past, and how the present fits with these existing

representations. However, what exactly is being encoded within the mPFC is not clear, neither is

our understanding of how these representations evolve in the brain throughout memory

consolidation. Put simply, part 1 of my thesis attempted to address what features of a memory

are actually represented by mPFC activity, and how the encoding of these features changes over

the course of the memory’s incorporation into a stable long-term network.

Part 2 addressed a related but different outstanding problem. Namely, how does the mPFC

acquire these representations, and what drives their evolution? To attempt to examine this

problem I looked at communication between the mPFC and several connected brain regions also

implicated in the episodic memory network, the lateral entorhinal cortex (LEC), the perirhinal

cortex (PER) and the ventral hippocampus (vHPC).

Using the trace eyeblink conditioning paradigm in which the identity of the stimuli (Tone or

Light) and relationship of the stimuli (Alone or Paired with US) was manipulated (Figure 4), I

revealed at the single neuron level the existence within the prelimbic mPFC (PrL) of both highly

Page 125: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

114

selective neurons, responsive to discrete features of the memory task, and highly generalized

neuronal responses (Figure 7). The proportions of this mixture surprisingly did not change over

the course of acquisition of the memory association or after several weeks of over-training

(Figure 9). Taking the activity of these single neurons together as a population however, I was

able to uncover differences in encoding across the time-course of acquisition and consolidation. I

first revealed that the population of activity became less similar during trials in which the

identity of the stimulus was the same (Tone-Alone and Tone-Paired) after the consolidation of

this memory compared to during its acquisition (Figure 11; Table 1). Further, using populations

of PrL neurons recorded at different stages of the memories acquisition and consolidation I was

able to train a supervised machine learning algorithm to accurately classify trial conditions

(Figure 13). Whereas accuracy overall did not change across the learning stages, patterns of

decoding did change. Specifically, the information represented in these populations became less

specific for the particular stimuli and more selective for their relational properties over the same

time-course that the consolidation of this memory has been observed (Takehara-Nishiuchi et al.,

2005, 2006). Together these data highlight characteristics of the PrL that could serve to support

its role in the long-term storage and maintenance of event memories in the brain; a structure

whose population responses become more generalized for overlapping and pertinent memory

features over-time but also still maintains discrete details of the memory within the network.

Secondly, by combining this data with the local oscillatory dynamics of other connected brain

structures, I was able to identity changes in the patterns of communication between these regions

and the PrL that may shape or reflect the changes in encoding just described. Through

simultaneous recordings of local field potentials from the LEC, PER and vHPC I was able to

observe distinct populations of PrL neurons modulated by these distant oscillations (Figure 18).

Similar to previous reports, I confirmed the presence of a population of PrL neurons phase-

Page 126: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

115

locked to theta oscillations in the vHPC during memory encoding (Adhikari et al., 2011) and also

reveal the presence of fairly distinct populations of PrL neurons modulated by LEC or PER theta

rhythms. The proportions of these populations (LEC-modulated, PER-modulated, vHPC-

modulated) were fairly similar and did not appear to change across time. However, looking at the

population activity of these different cell groups across time did reveal changes that may reflect

network interactions relevant to the consolidation of this memory. PrL neurons modulated by

rhinal theta rhythms (LEC and PER), but the LEC in particular, were highly selective for discrete

details of the memory during the early stages of its acquisition. However, by the time this

memory was well acquired and several weeks later, this population became far less selective.

Likewise, populations phase-locked to ventral hippocampal theta rhythms (vHPC-modulated)

though less selective overall, became more selective for behaviourally relevant features in the

several weeks following stable memory expression.

5.2 Conclusions

A greater understanding of the features of a memory encoded by the mPFC and how they change

through the process of memory consolidation are important aspects of our unraveling of the

networks and process by which memories are stored and maintained in the brain. Here I advance

this front by showing the level of detail represented within the prelimbic mPFC (PrL) and how it

changes throughout learning and over-training. I also highlight a potential process by which

these changes may take place, by observing differences in encoding across time in populations

linked to distinct parts of the memory network.

5.2.1 Role of medial prefrontal cortex in memory

The mPFC has emerged as a key node in the long-term memory network and there are many

different theories of how it fulfills this function, and of its role in the brain in general. For

Page 127: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

116

example, much of the focus of the role of the mPFC in the brain, particularly in human and

primate studies focus on its role in decision making in various shapes and forms (for e.g. conflict

monitoring: Botvinick, Cohen, & Carter, 2004; error detection: Holroyd, Coles, & Nieuwenhuis,

2002; self-regulation: Posner, Rothbart, Sheese, & Tang, 2007). Findings of its role in memory

requires the difficult task of incorporating these different properties together to form a coherent

theory of mPFC functioning. Many of these theories have overlapping features that seem to

converge on the idea of the medial prefrontal cortex providing adaptability. Whether it be

through emotional outcome expectancy (Fellows, 2007) or the maintenance of goal

representations relevant to the current situation (Miller & Cohen, 2001) a common theme is the

top-down control of behaviour given the present context. The link between the role of the mPFC

in memory and its role in decision making does not require far-reaching connections given that

proper decision making requires some knowledge of past experience (or memory) to properly

guide behaviour in the current situation.

The role of the mPFC in memory has often been confined to the long-term maintenance of

memory, or permitting the retrieval of remotely acquired memories given evidence of more

severe retrieval impairments from mPFC disruption at remote versus recent time-points

(Frankland et al., 2004; Maviel et al., 2004; Takehara et al., 2003). However it has also been

proposed that the mPFC is an active participant in all stages of learning and memory providing

contextual meaning and comparing and integrating new events with previously acquired

representations of experience (Euston et al., 2012; Preston & Eichenbaum, 2013). This is

supported by the recent finding that a direct medial prefrontal-hippocampus projection can

trigger the retrieval of a recently formed memory of a dangerous place (Rajasethupathy et al.,

2015). The medial prefrontal cortex is also proposed to play an essential role in the consolidation

of long-term memory throughout neocortical networks and to become an essential node in the

Page 128: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

117

long-term retrieval of memories that no longer require the hippocampus (Frankland & Bontempi,

2005; Insel & Takehara-Nishiuchi, 2013). Here it is typically suggested that the medial

prefrontal representation is similar to that suggested of cortical memory representations in

general, existing in a more abstract form containing gist-like details of the experience (Euston et

al., 2012; McClelland et al., 1995; Winocur et al., 2010).

Previous investigations of memory representations in the mPFC throughout learning and

consolidation reveal that these neurons gradually acquire activity selective for a memory

association (Hattori et al., 2014; Takehara-Nishiuchi & McNaughton, 2008). Whereas

investigations simply of the features of memory encoded by this region suggest that they are

predominantly encoding task rules or categorical features (Rich & Shapiro, 2009; Wallis et al.,

2001). However no study has looked at the evolution of the encoding of different memory

features across learning and consolidation. Here I reveal prelimbic neurons exhibiting an array of

varying selectivity for task features, the proportions of which are similar from very early on in

learning to several weeks after stable memory expression. This suggests that the content of

medial prefrontal representations in single neurons does not change across learning and

consolidation. This also supports the notion that mPFC computations are important in memory

acquisition. However the presence of neuron responses highly selective for particular task

features suggests that representations in this region are not merely abstractions.

The mPFC is thought to be involved in schema generation and has been proposed to be involved

in maintenance through the updating of existing knowledge structures in the presence of new or

conflicting information (McKenzie & Eichenbaum, 2011; Richards et al., 2014). I propose that

the selectivity within single neurons observed across time in the current study supports this

function allowing the mPFC to accurately identify conflicts between new and old information

Page 129: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

118

and to decipher the relevance of current experience within the context of information provided

by past experience.

In the Euston et al. (2012) model, mPFC memory functions change across the consolidation

phase of the memory, initially representing contextual meaning and later both representing

meaning and also storing cortical indices of the memory. Here I show that despite conserved

proportions of neurons with varying degrees of feature selectivity across time, the population

activity of prelimbic neurons does appear to change across consolidation relevant time-periods.

Specifically, the population activity became more generalized for task irrelevant features and

more selective for meaningful features. This finding fits better with the notion of a more abstract

memory representation in the mPFC that seems to be captured by looking at the network

computations within this region. Whereas the presence of highly selective neurons remain, in a

task in which the features go unchanged and no new information is available, the global activity

of these neuron populations highlight the meaningful dimensions and become less discriminating

of non-informative features. This is highlighted by the fact that when new information was

eventually added in the form of a change of environmental features, because this information did

not change the underlying representations, this information was represented within the existing

generalized dimension. I propose that had this new information changed the underlying

representation and provided contextual meaning to the association, the prelimbic code would

have rapidly acquired highly selective activity for this feature. This supports the crux of theories

of mPFC function, wherein memories are represented within meaningful contexts to guide

behaviour in the current situation and support adaptability. In my data the prefrontal population

acquires selectivity for meaningful and predictive features but also maintains individual

selectivity for discrete event features. This would permit that given a sudden change in the

Page 130: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

119

underlying representations, for example if suddenly light but not tone predicted the shock, the

population could rapidly adapt, ramping up its selectivity for the identity of the stimulus.

Both of the proposed functions of the mPFC in memory by the Euston et al. (2012) model are

proposed to depend on interactions with the hippocampus. With hippocampal-prefrontal

interactions initially enabling rapid mPFC associative learning to permit the formation of the

memory (Wise & Murray, 2000), and later through reply mechanisms (Peyrache, Khamassi,

Benchenane, Wiener, & Battaglia, 2009), driving the consolidation of the memory to permit the

long-term storage of cortical indices (Euston et al., 2012). In my experiment, I revealed that

distinct populations of prelimbic neurons modulated by inputs from the lateral entorhinal cortex

show highly selective activity early in learning that becomes more generalized with learning and

consolidation that was not observed in populations modulated by the perirhinal and ventral

hippocampal inputs. This initial selectivity and later generalization may reflect the interactions

proposed, with the entorhinal cortex providing immediate access to rapid hippocampal

computations creating highly selective encoding in prelimbic neurons and more generalized

patterns as the hippocampus is gradually disengaged. However in this case rather than simply

highly selective population activity, I would expect that a large proportion of entorhinal

modulated prelimbic neurons would show highly selective activity early in learning as opposed

to the varying selectivity observed that was comparable to other phase-locked populations.

Therefore selectivity at the single neuron level may also be acquired through other means,

possibly through connectivity with midline-thalamic regions (Vertes, Hoover, Szigeti-Buck, &

Leranth, 2007) or other connected regions of the cingulate (Hattori et al., 2014), whereas these

hippocampal inputs instead may direct these cells activity into functional ensembles.

Page 131: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

120

Similar to the schematic theory of mPFC memory function, it has also been proposed that a key

feature of mPFC memory storage is that it creates a more efficient index code of the memory

patterns across the neocortex (Frankland & Bontempi, 2005; Takehara-Nishiuchi &

McNaughton, 2008). It has recently been suggested that neurons showing mixed selectivity

(what I called here Non-selective) as a population can be decoded for many of the individual

features of the memory (Rigotti et al., 2013). This type of encoding is proposed to have a

computational advantage over more highly selective responses and may represent higher

dimensional encoding important in cognition. Here these Non-selective cells or cells showing

mixed-selectivity were the largest proportion of cells I observed in the PrL. If Rigotti et al.,

(2013) are correct in their proposition of the importance of these cells, they may serve as the

hallmark of a more efficient index of the memory association.

Together this study provides a neurophysiological account of memory representation within the

prelimbic mPFC and reveals a potential neural signature of episodic memory consolidation in the

brain. I report actual changes in prelimbic encoding of memory features across time and

highlight several cortical interactions that appear to drive these changes. Overall these results

seem to fit the current evolving theories of medial prefrontal memory function and provide

insight into several areas that have remained uncertain.

5.2.2 Future directions

This thesis looked at the computations taking place within the prelimbic mPFC across learning

and consolidation and highlighted several brain regions that may help shape these

representations. However a drawback of this study is that while the results are informative, they

are correlative and are interpreted based on the numerous functional investigations already done

on the role of the mPFC in memory. My findings highlight several properties of prelimbic

Page 132: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

121

encoding that require further study to uncover their functional significance. Future studies

combining pathway specific control and electrophysiological recordings would provide

substantial insight into the functional significance of the current findings. For example,

optogenetic methods can be used to target the LEC-mPFC projection or the mPFC-LEC

projection to demonstrate the functional significance of each of these for behaviour and also for

the formation and/or maintenance of mPFC representations and how they differ across learning,

consolidation and remote retrieval. Targeting pathways with methods that allow their direct

control would identify when and how these projections are shaping the state of memory in these

networks. For example during initial learning when hippocampal computations are essential for

memory formation, does the LEC drive mPFC activity and is this essential for the

representations of the memory created within the mPFC population? In contrast, after the

memory has been consolidated, does the directionality reverse, with the mPFC driving LEC?

Further the extent to which the mPFC stores cortical indices can be examined by adapting

methods previously used to reveal the engram like properties of hippocampal cells. Particularly,

mapping the effect of silencing memory activated neurons in the mPFC not only on behaviour,

but also on the memory relevant activity in other cortical structures, and how this differs before

and after consolidation of the memory. Systematically assessing these effects and their

differences across some of the important sub-regions of the mPFC would also be very revealing.

The combination of electrophysiological monitoring and functional interrogation can be very

revealing. The interaction between the mPFC and the other major nodes of the remote memory

network layout several possibilities of their functional significance, questions tailor-made for

such an approach. We are at an exciting time in the science of memory, a time when our

technical toolbox to manipulate and observe neural activity is starting to catch up to our

Page 133: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

122

understanding of how these circuits and networks mediate our cognitive and mental capacities.

New frontiers are being opened to investigate the neural workings of the brain and they can help

build upon the ingenious groundwork already laid. The field of memory research has taken great

leaps in the last couple decades and there is great optimism for what lies ahead.

Page 134: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

1

References

Adhikari, A., Topiwala, M. A., & Gordon, J. A. (2010). Synchronized Activity between the

Ventral Hippocampus and the Medial Prefrontal Cortex during Anxiety. Neuron, 65(2),

257–269. http://doi.org/10.1016/j.neuron.2009.12.002

Adhikari, A., Topiwala, M. A., & Gordon, J. A. (2011). Single Units in the Medial Prefrontal

Cortex with Anxiety-Related Firing Patterns Are Preferentially Influenced by Ventral

Hippocampal Activity. Neuron, 71(5), 898–910.

http://doi.org/10.1016/j.neuron.2011.07.027

Aggleton, J. P., & Pearce, J. M. (2001). Neural systems underlying episodic memory: insights

from animal research. Philosophical Transactions of the Royal Society of London B:

Biological Sciences, 356(1413), 1467–1482. http://doi.org/10.1098/rstb.2001.0946

Agster, K. L., & Burwell, R. D. (2009). Cortical efferents of the perirhinal, postrhinal, and

entorhinal cortices of the rat. Hippocampus, 19(12), 1159–1186.

http://doi.org/10.1002/hipo.20578

Alvarez, P., & Squire, L. R. (1994). Memory consolidation and the medial temporal lobe: a

simple network model. Proceedings of the National Academy of Sciences, 91(15), 7041–

7045.

Amaral, D., & Witter, M. (1995). Hippocampal formation. In G. Paxinos (Ed.), The Rat Nervous

System (pp. 443–486). San Diego, CA: Academic Press. Retrieved from http://ac.els-

cdn.com/B9780125476386500225/3-s2.0-B9780125476386500225-

main.pdf?_tid=975cb388-0fa7-11e5-a369-

00000aacb362&acdnat=1433964997_e477c8a20b9fd54725de05d21ad20810

Page 135: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

124

Anderson, K. L., Rajagovindan, R., Ghacibeh, G. A., Meador, K. J., & Ding, M. (2010). Theta

oscillations mediate interaction between prefrontal cortex and medial temporal lobe in

human memory. Cerebral Cortex (New York, N.Y.: 1991), 20(7), 1604–1612.

http://doi.org/10.1093/cercor/bhp223

Apergis-Schoute, J., Pinto, A., & Pare, D. (2006). Ultrastructural organization of medial

prefrontal inputs to the rhinal cortices. Eur J Neurosci, 24(1), 135–44.

http://doi.org/10.1111/j.1460-9568.2006.04894.x

Baldi, E., & Bucherelli, C. (2015). Brain sites involved in fear memory reconsolidation and

extinction of rodents. Neuroscience & Biobehavioral Reviews, 53, 160–190.

http://doi.org/10.1016/j.neubiorev.2015.04.003

Bannerman, D. M., Rawlins, J. N. P., McHugh, S. B., Deacon, R. M. J., Yee, B. K., Bast, T., …

Feldon, J. (2004). Regional dissociations within the hippocampus--memory and anxiety.

Neuroscience and Biobehavioral Reviews, 28(3), 273–283.

http://doi.org/10.1016/j.neubiorev.2004.03.004

Barker, G. R. I., & Warburton, E. C. (2008). NMDA Receptor Plasticity in the Perirhinal and

Prefrontal Cortices Is Crucial for the Acquisition of Long-Term Object-in-Place

Associative Memory. Journal of Neuroscience, 28(11), 2837–2844.

http://doi.org/10.1523/JNEUROSCI.4447-07.2008

Bartlett, S. F. C., & Bartlett, F. C. (1995). Remembering: A Study in Experimental and Social

Psychology. Cambridge University Press.

Page 136: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

125

Bayley, P. J., Gold, J. J., Hopkins, R. O., & Squire, L. R. (2005). The Neuroanatomy of Remote

Memory. Neuron, 46(5), 799–810. http://doi.org/10.1016/j.neuron.2005.04.034

Bédard, P., & Sanes, J. N. (2014). Brain representations for acquiring and recalling visual-motor

adaptations. NeuroImage, 101, 225–235.

http://doi.org/10.1016/j.neuroimage.2014.07.009

Beeman, C. L., Bauer, P. S., Pierson, J. L., & Quinn, J. J. (2013). Hippocampus and medial

prefrontal cortex contributions to trace and contextual fear memory expression over time.

Learning & Memory, 20(6), 336–343. http://doi.org/10.1101/lm.031161.113

Berke, J. D., Okatan, M., Skurski, J., & Eichenbaum, H. B. (2004). Oscillatory entrainment of

striatal neurons in freely moving rats. Neuron, 43(6), 883–896.

http://doi.org/10.1016/j.neuron.2004.08.035

Bero, A. W., Meng, J., Cho, S., Shen, A. H., Canter, R. G., Ericsson, M., & Tsai, L.-H. (2014).

Early remodeling of the neocortex upon episodic memory encoding. Proceedings of the

National Academy of Sciences, 111(32), 11852–11857.

http://doi.org/10.1073/pnas.1408378111

Beylin, A. V., Gandhi, C. C., Wood, G. E., Talk, A. C., Matzel, L. D., & Shors, T. J. (2001). The

role of the hippocampus in trace conditioning: temporal discontinuity or task difficulty?

Neurobiol Learn Mem, 76(3), 447–61. http://doi.org/10.1006/nlme.2001.4039

Bolhuis, J. J., Stewart, C. A., & Forrest, E. M. (1994). Retrograde amnesia and memory

reactivation in rats with ibotenate lesions to the hippocampus or subiculum. The

Page 137: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

126

Quarterly Journal of Experimental Psychology. B, Comparative and Physiological

Psychology, 47(2), 129–150.

Bontempi, B., Laurent-Demir, C., Destrade, C., & Jaffard, R. (1999). Time-dependent

reorganization of brain circuitry underlying long-term memory storage. Nature,

400(6745), 671–675. http://doi.org/10.1038/23270

Botvinick, M. M., Cohen, J. D., & Carter, C. S. (2004). Conflict monitoring and anterior

cingulate cortex: an update. Trends in Cognitive Sciences, 8(12), 539–546.

http://doi.org/10.1016/j.tics.2004.10.003

Broadbent, N. J., Squire, L. R., & Clark, R. E. (2006). Reversible hippocampal lesions disrupt

water maze performance during both recent and remote memory tests. Learning &

Memory (Cold Spring Harbor, N.Y.), 13(2), 187–191. http://doi.org/10.1101/lm.134706

Buckner, R. L., Kelley, W. M., & Petersen, S. E. (1999). Frontal cortex contributes to human

memory formation. Nature Neuroscience, 2(4), 311–314. http://doi.org/10.1038/7221

Bunsey, M., & Eichenbaum, H. (1996). Conservation of hippocampal memory function in rats

and humans. Nature, 379(6562), 255–257. http://doi.org/10.1038/379255a0

Burwell, R. D. (2000). The parahippocampal region: corticocortical connectivity. Annals of the

New York Academy of Sciences, 911(1), 25–42.

Burwell, R. D. (2004). Perirhinal and Postrhinal Contributions to Remote Memory for Context.

Journal of Neuroscience, 24(49), 11023–11028.

http://doi.org/10.1523/JNEUROSCI.3781-04.2004

Page 138: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

127

Burwell, R. D., & Amaral, D. G. (1998a). Cortical afferents of the perirhinal, postrhinal, and

entorhinal cortices of the rat. The Journal of Comparative Neurology, 398(2), 179–205.

Burwell, R. D., & Amaral, D. G. (1998b). Perirhinal and postrhinal cortices of the rat:

Interconnectivity and connections with the entorhinal cortex. The Journal of Comparative

Neurology, 391(3), 293–321. http://doi.org/10.1002/(SICI)1096-

9861(19980216)391:3<293::AID-CNE2>3.0.CO;2-X

Buzsáki, G. (2002). Theta Oscillations in the Hippocampus. Neuron, 33(3), 325–340.

http://doi.org/10.1016/S0896-6273(02)00586-X

Buzsaki, G. (2004). Neuronal Oscillations in Cortical Networks. Science, 304(5679), 1926–1929.

http://doi.org/10.1126/science.1099745

Buzsáki, G., & Moser, E. I. (2013). Memory, navigation and theta rhythm in the hippocampal-

entorhinal system. Nature Neuroscience, 16(2), 130–138. http://doi.org/10.1038/nn.3304

Canolty, R. T., Ganguly, K., Kennerley, S. W., Cadieu, C. F., Koepsell, K., Wallis, J. D., &

Carmena, J. M. (2010). Oscillatory phase coupling coordinates anatomically dispersed

functional cell assemblies. Proceedings of the National Academy of Sciences, 107(40),

17356–17361. http://doi.org/10.1073/pnas.1008306107

Canto, C. B., Wouterlood, F. G., & Witter, M. P. (2008). What Does the Anatomical

Organization of the Entorhinal Cortex Tell Us? Neural Plasticity, 2008, 1–18.

http://doi.org/10.1155/2008/381243

Chachich, M., & Powell, D. A. (1998). Both medial prefrontal and amygdala central nucleus

lesions abolish heart rate classical conditioning, but only prefrontal lesions impair

Page 139: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

128

reversal of eyeblink differential conditioning. Neuroscience Letters, 257(3), 151–154.

http://doi.org/10.1016/S0304-3940(98)00832-5

Chang, C.-C., & Lin, C.-J. (2011). LIBSVM : a library for support vector machines. ACM

Transactions on Intelligent Systems and Technology, 2(27), 1–27.

Charlesworth, J. D., Warren, T. L., & Brainard, M. S. (2012). Covert skill learning in a cortical-

basal ganglia circuit. Nature, 486(7402), 251–255. http://doi.org/10.1038/nature11078

Cho, Y. H., Beracochea, D., & Jaffard, R. (1993). Extended temporal gradient for the retrograde

and anterograde amnesia produced by ibotenate entorhinal cortex lesions in mice. The

Journal of Neuroscience, 13(4), 1759–1766.

Cho, Y. H., & Jaffard, R. (1995). Spatial location learning in mice with ibotenate lesions of

entorhinal cortex or subiculum. Neurobiology of Learning and Memory, 64(3), 285–290.

Christian, K. M., & Thompson, R. F. (2003). Neural substrates of eyeblink conditioning:

acquisition and retention. Learn Mem, 10(6), 427–455. http://doi.org/10.1101/lm.59603

Clark, R. E., Broadbent, N. J., & Squire, L. R. (2005). Impaired remote spatial memory after

hippocampal lesions despite extensive training beginning early in life. Hippocampus,

15(3), 340–346. http://doi.org/10.1002/hipo.20076

Clark, R. E., & Squire, L. R. (1999). Human eyeblink classical conditioning: Effects of

manipulating awareness of the stimulus contingencies. Psychol Sci, 10(1), 14–18.

http://doi.org/10.1111/1467-9280.00099

Page 140: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

129

Clark, R. E., West, A. N., Zola, S. M., & Squire, L. R. (2001). Rats with lesions of the

hippocampus are impaired on the delayed nonmatching-to-sample task. Hippocampus,

11(2), 176–186. http://doi.org/10.1002/hipo.1035

Clark, R. E., Zola, S. M., & Squire, L. R. (2000). Impaired Recognition Memory in Rats after

Damage to the Hippocampus. The Journal of Neuroscience, 20(23), 8853–8860.

Courtin, J., Chaudun, F., Rozeske, R. R., Karalis, N., Gonzalez-Campo, C., Wurtz, H., … Herry,

C. (2014). Prefrontal parvalbumin interneurons shape neuronal activity to drive fear

expression. Nature, 505(7481), 92–96. http://doi.org/10.1038/nature12755

Coutureau, E., & Di Scala, G. (2009). Entorhinal cortex and cognition. Progress in Neuro-

Psychopharmacology and Biological Psychiatry, 33(5), 753–761.

http://doi.org/10.1016/j.pnpbp.2009.03.038

Cullen, P. K., Gilman, T. L., Winiecki, P., Riccio, D. C., & Jasnow, A. M. (2015). Activity of the

anterior cingulate cortex and ventral hippocampus underlie increases in contextual fear

generalization. Neurobiology of Learning and Memory, 124, 19–27.

http://doi.org/10.1016/j.nlm.2015.07.001

Damasio, A. R. (1989). Time-locked multiregional retroactivation: A systems-level proposal for

the neural substrates of recall and recognition. Cognition, 33(1–2), 25–62.

http://doi.org/10.1016/0010-0277(89)90005-X

di Pellegrino, G., & Wise, S. P. (1993). Visuospatial versus visuomotor activity in the premotor

and prefrontal cortex of a primate. The Journal of Neuroscience: The Official Journal of

the Society for Neuroscience, 13(3), 1227–1243.

Page 141: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

130

Durstewitz, D., Vittoz, N. M., Floresco, S. B., & Seamans, J. K. (2010). Abrupt Transitions

between Prefrontal Neural Ensemble States Accompany Behavioral Transitions during

Rule Learning. Neuron, 66(3), 438–448. http://doi.org/10.1016/j.neuron.2010.03.029

Eichenbaum, H., Sauvage, M., Fortin, N., Komorowski, R., & Lipton, P. (2012). Towards a

functional organization of episodic memory in the medial temporal lobe. Neuroscience &

Biobehavioral Reviews, 36(7), 1597–1608.

http://doi.org/10.1016/j.neubiorev.2011.07.006

Eichenbaum, H., Stewart, C., & Morris, R. G. (1990). Hippocampal representation in place

learning. The Journal of Neuroscience, 10(11), 3531–3542.

Eichenbaum, H., Yonelinas, A. P., & Ranganath, C. (2007). The Medial Temporal Lobe and

Recognition Memory. Annual Review of Neuroscience, 30(1), 123–152.

http://doi.org/10.1146/annurev.neuro.30.051606.094328

Einarsson, E. Ö., & Nader, K. (2012). Involvement of the anterior cingulate cortex in formation,

consolidation, and reconsolidation of recent and remote contextual fear memory.

Learning & Memory (Cold Spring Harbor, N.Y.), 19(10), 449–452.

http://doi.org/10.1101/lm.027227.112

Ennaceur, A., Neave, N., & Aggleton, J. P. (1996). Neurotoxic lesions of the perirhinal cortex do

not mimic the behavioural effects of fornix transection in the rat. Behavioural Brain

Research, 80(1-2), 9–25.

Page 142: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

131

Euston, D. R., Gruber, A. J., & McNaughton, B. L. (2012). The Role of Medial Prefrontal Cortex

in Memory and Decision Making. Neuron, 76(6), 1057–1070.

http://doi.org/10.1016/j.neuron.2012.12.002

Fell, J., & Axmacher, N. (2011). The role of phase synchronization in memory processes. Nature

Reviews Neuroscience, 12(2), 105–118. http://doi.org/10.1038/nrn2979

Fellows, L. K. (2007). Advances in understanding ventromedial prefrontal function: the

accountant joins the executive. Neurology, 68(13), 991–995.

http://doi.org/10.1212/01.wnl.0000257835.46290.57

Fink, G. R., Markowitsch, H. J., Reinkemeier, M., Bruckbauer, T., Kessler, J., & Heiss, W.-D.

(1996). Cerebral Representation of One’s Own Past: Neural Networks Involved in

Autobiographical Memory. The Journal of Neuroscience, 16(13), 4275–4282.

Flores, L. C., & Disterhoft, J. F. (2013). Caudate Nucleus in Retrieval of Trace Eyeblink

Conditioning after Consolidation. The Journal of Neuroscience, 33(7), 2828–2836.

http://doi.org/10.1523/JNEUROSCI.2326-12.2013

Frankland, P. W. (2004). The Involvement of the Anterior Cingulate Cortex in Remote

Contextual Fear Memory. Science, 304(5672), 881–883.

http://doi.org/10.1126/science.1094804

Frankland, P. W., & Bontempi, B. (2005). The organization of recent and remote memories.

Nature Reviews Neuroscience, 6(2), 119–130. http://doi.org/10.1038/nrn1607

Frankland, P. W., & Bontempi, B. (2006). Fast track to the medial prefrontal cortex. Proceedings

of the National Academy of Sciences of the United States of America, 103(3), 509–510.

Page 143: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

132

Frankland, P. W., Bontempi, B., Talton, L. E., Kaczmarek, L., & Silva, A. J. (2004). The

Involvement of the Anterior Cingulate Cortex in Remote Contextual Fear Memory.

Science, 304(5672), 881–883. http://doi.org/10.1126/science.1094804

Frankland, P. W., Ding, H.-K., Takahashi, E., Suzuki, A., Kida, S., & Silva, A. J. (2006).

Stability of recent and remote contextual fear memory. Learning & Memory (Cold Spring

Harbor, N.Y.), 13(4), 451–457. http://doi.org/10.1101/lm.183406

Frankland, P. W., O’Brien, C., Ohno, M., Kirkwood, A., & Silva, A. J. (2001). α-CaMKII-

dependent plasticity in the cortex is required for permanent memory. Nature, 411(6835),

309–313. http://doi.org/10.1038/35077089

Frank, L. M., Brown, E. N., & Wilson, M. A. (2001). A comparison of the firing properties of

putative excitatory and inhibitory neurons from CA1 and the entorhinal cortex. Journal of

Neurophysiology, 86(4), 2029–2040.

Freedman, D. J., Riesenhuber, M., Poggio, T., & Miller, E. K. (2001). Categorical representation

of visual stimuli in the primate prefrontal cortex. Science (New York, N.Y.), 291(5502),

312–316. http://doi.org/10.1126/science.291.5502.312

Freeman, J. H., & Steinmetz, A. B. (2011). Neural circuitry and plasticity mechanisms

underlying delay eyeblink conditioning. Learning & Memory (Cold Spring Harbor,

N.Y.), 18(10), 666–677. http://doi.org/10.1101/lm.2023011

Fries, P. (2005). A mechanism for cognitive dynamics: neuronal communication through

neuronal coherence. Trends in Cognitive Sciences, 9(10), 474–480.

http://doi.org/10.1016/j.tics.2005.08.011

Page 144: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

133

Fuster, J. M. (1973). Unit activity in prefrontal cortex during delayed-response performance:

neuronal correlates of transient memory. Journal of Neurophysiology, 36(1), 61–78.

Fuster, J. M. (2001). The Prefrontal Cortex—An Update: Time Is of the Essence. Neuron, 30(2),

319–333. http://doi.org/10.1016/S0896-6273(01)00285-9

Fuster, J. M., Bodner, M., & Kroger, J. K. (2000). Cross-modal and cross-temporal association in

neurons of frontal cortex. Nature, 405(6784), 347–351. http://doi.org/10.1038/35012613

Galvez, R., Weible, A. P., & Disterhoft, J. F. (2007). Cortical barrel lesions impair whisker-CS

trace eyeblink conditioning. Learn Mem, 14(1-2), 94–100.

http://doi.org/10.1101/lm.418407

Genovesio, A., Brasted, P. J., Mitz, A. R., & Wise, S. P. (2005). Prefrontal cortex activity related

to abstract response strategies. Neuron, 47(2), 307–320.

http://doi.org/10.1016/j.neuron.2005.06.006

Gervais, N. J., Barrett-Bernstein, M., Sutherland, R. J., & Mumby, D. G. (2014). Retrograde and

anterograde memory following selective damage to the dorsolateral entorhinal cortex.

Neurobiology of Learning and Memory, 116, 14–26.

http://doi.org/10.1016/j.nlm.2014.07.012

Gilmartin, M. R., Kwapis, J. L., & Helmstetter, F. J. (2012). Trace and contextual fear

conditioning are impaired following unilateral microinjection of muscimol in the ventral

hippocampus or amygdala, but not the medial prefrontal cortex. Neurobiology of

Learning and Memory, 97(4), 452–464. http://doi.org/10.1016/j.nlm.2012.03.009

Page 145: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

134

Gilmartin, M. R., & McEchron, M. D. (2005). Single neurons in the medial prefrontal cortex of

the rat exhibit tonic and phasic coding during trace fear conditioning. Behav Neurosci,

119(6), 1496–510. http://doi.org/10.1037/0735-7044.119.6.1496

Gilmartin, M. R., Miyawaki, H., Helmstetter, F. J., & Diba, K. (2013). Prefrontal Activity Links

Nonoverlapping Events in Memory. The Journal of Neuroscience, 33(26), 10910–10914.

http://doi.org/10.1523/JNEUROSCI.0144-13.2013

Gray, C. M., Maldonado, P. E., Wilson, M., & McNaughton, B. (1995). Tetrodes markedly

improve the reliability and yield of multiple single-unit isolation from multi-unit

recordings in cat striate cortex. Journal of Neuroscience Methods, 63(1-2), 43–54.

Hafting, T., Fyhn, M., Molden, S., Moser, M.-B., & Moser, E. I. (2005). Microstructure of a

spatial map in the entorhinal cortex. Nature, 436(7052), 801–806.

http://doi.org/10.1038/nature03721

Hannesson, D. K., Howland, J. G., & Phillips, A. G. (2004). Interaction between Perirhinal and

Medial Prefrontal Cortex Is Required for Temporal Order But Not Recognition Memory

for Objects in Rats. The Journal of Neuroscience, 24(19), 4596–4604.

http://doi.org/10.1523/JNEUROSCI.5517-03.2004

Hargreaves, E. L. (2005). Major Dissociation Between Medial and Lateral Entorhinal Input to

Dorsal Hippocampus. Science, 308(5729), 1792–1794.

http://doi.org/10.1126/science.1110449

Page 146: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

135

Hargreaves, E. L., Rao, G., Lee, I., & Knierim, J. J. (2005). Major dissociation between medial

and lateral entorhinal input to dorsal hippocampus. Science, 308(5729), 1792–4.

http://doi.org/10.1126/science.1110449

Harris, A. Z., & Gordon, J. A. (2015). Long-Range Neural Synchrony in Behavior. Annual

Review of Neuroscience, 38(1), 171–194. http://doi.org/10.1146/annurev-neuro-071714-

034111

Harris, K. D., Csicsvari, J., Hirase, H., Dragoi, G., & Buzsaki, G. (2003). Organization of cell

assemblies in the hippocampus. Nature, 424(6948), 552–6.

http://doi.org/10.1038/nature01834

Hatsopoulos, N. G., Ojakangas, C. L., Paninski, L., & Donoghue, J. P. (1998). Information about

movement direction obtained from synchronous activity of motor cortical neurons.

Proceedings of the National Academy of Sciences, 95(26), 15706–15711.

http://doi.org/10.1073/pnas.95.26.15706

Hattori, S., Yoon, T., Disterhoft, J. F., & Weiss, C. (2014). Functional Reorganization of a

Prefrontal Cortical Network Mediating Consolidation of Trace Eyeblink Conditioning.

The Journal of Neuroscience, 34(4), 1432–1445.

http://doi.org/10.1523/JNEUROSCI.4428-13.2014

Herfurth, K., Kasper, B., Schwarz, M., Stefan, H., & Pauli, E. (2010). Autobiographical memory

in temporal lobe epilepsy: role of hippocampal and temporal lateral structures. Epilepsy

& Behavior: E&B, 19(3), 365–371. http://doi.org/10.1016/j.yebeh.2010.07.012

Page 147: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

136

Hikosaka, O., Nakamura, K., Sakai, K., & Nakahara, H. (2002). Central mechanisms of motor

skill learning. Current Opinion in Neurobiology, 12(2), 217–222.

Hoffmann, L. C., & Berry, S. D. (2009). Cerebellar theta oscillations are synchronized during

hippocampal theta-contingent trace conditioning. Proceedings of the National Academy

of Sciences of the United States of America, 106(50), 21371–21376.

http://doi.org/10.1073/pnas.0908403106

Holroyd, C. B., Coles, M. G. H., & Nieuwenhuis, S. (2002). Medial Prefrontal Cortex and Error

Potentials. Science, 296(5573), 1610–1611. http://doi.org/10.1126/science.296.5573.1610

Hoover, W. B., & Vertes, R. P. (2007). Anatomical analysis of afferent projections to the medial

prefrontal cortex in the rat. Brain Structure and Function, 212(2), 149–179.

http://doi.org/10.1007/s00429-007-0150-4

Insausti, R., Herrero, M., & Witter, M. P. (1997). Entorhinal cortex of the rat: cytoarchitectonic

subdivisions and the origin and distribution of cortical efferents. Hippocampus, 7(2),

146–183.

Insel, N., & Takehara-Nishiuchi, K. (2013). The cortical structure of consolidated memory: A

hypothesis on the role of the cingulate–entorhinal cortical connection. Neurobiology of

Learning and Memory, 106, 343–350. http://doi.org/10.1016/j.nlm.2013.07.019

Jones, B. F., & Witter, M. P. (2007). Cingulate cortex projections to the parahippocampal region

and hippocampal formation in the rat. Hippocampus, 17(10), 957–976.

http://doi.org/10.1002/hipo.20330

Page 148: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

137

Jones, M. W., & Wilson, M. A. (2005). Theta Rhythms Coordinate Hippocampal–Prefrontal

Interactions in a Spatial Memory Task. PLoS Biology, 3(12), e402.

http://doi.org/10.1371/journal.pbio.0030402

Kahana, M. J., Sekuler, R., Caplan, J. B., Kirschen, M., & Madsen, J. R. (1999). Human theta

oscillations exhibit task dependence during virtual maze navigation. Nature, 399(6738),

781–784. http://doi.org/10.1038/21645

Kerr, K. M., Agster, K. L., Furtak, S. C., & Burwell, R. D. (2007a). Functional neuroanatomy of

the parahippocampal region: The lateral and medial entorhinal areas. Hippocampus,

17(9), 697–708. http://doi.org/10.1002/hipo.20315

Kerr, K. M., Agster, K. L., Furtak, S. C., & Burwell, R. D. (2007b). Functional neuroanatomy of

the parahippocampal region: The lateral and medial entorhinal areas. Hippocampus,

17(9), 697–708. http://doi.org/10.1002/hipo.20315

Kim, J. J., Clark, R. E., & Thompson, R. F. (1995). Hippocampectomy impairs the memory of

recently, but not remotely, acquired trace eyeblink conditioned responses. Behav

Neurosci, 109(2), 195–203.

Kim, J. J., & Fanselow, M. S. (1992). Modality-specific retrograde amnesia of fear. Science

(New York, N.Y.), 256(5057), 675–677.

Knierim, J. J., Neunuebel, J. P., & Deshmukh, S. S. (2013). Functional correlates of the lateral

and medial entorhinal cortex: objects, path integration and local-global reference frames.

Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1635),

20130369–20130369. http://doi.org/10.1098/rstb.2013.0369

Page 149: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

138

Kolb, B. (1984). Functions of the frontal cortex of the rat: a comparative review. Brain Research,

320(1), 65–98.

Komorowski, R. W., Garcia, C. G., Wilson, A., Hattori, S., Howard, M. W., & Eichenbaum, H.

(2013). Ventral Hippocampal Neurons Are Shaped by Experience to Represent

Behaviorally Relevant Contexts. Journal of Neuroscience, 33(18), 8079–8087.

http://doi.org/10.1523/JNEUROSCI.5458-12.2013

Komorowski, R. W., Manns, J. R., & Eichenbaum, H. (2009). Robust Conjunctive Item–Place

Coding by Hippocampal Neurons Parallels Learning What Happens Where. The Journal

of Neuroscience, 29(31), 9918–9929. http://doi.org/10.1523/JNEUROSCI.1378-09.2009

Kronforst-Collins, M. A., & Disterhoft, J. F. (1998). Lesions of the caudal area of rabbit medial

prefrontal cortex impair trace eyeblink conditioning. Neurobiol Learn Mem, 69(2), 147–

62. http://doi.org/10.1006/nlme.1997.3818

Krupa, D. J., Thompson, J. K., & Thompson, R. F. (1993). Localization of a memory trace in the

mammalian brain. Science, 260(5110), 989–91.

LeDoux, J. E. (1995). Emotion: Clues from the Brain. Annual Review of Psychology, 46(1), 209–

235. http://doi.org/10.1146/annurev.ps.46.020195.001233

Lee, H., Simpson, G. V., Logothetis, N. K., & Rainer, G. (2005). Phase locking of single neuron

activity to theta oscillations during working memory in monkey extrastriate visual cortex.

Neuron, 45(1), 147–156. http://doi.org/10.1016/j.neuron.2004.12.025

Page 150: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

139

Liu, X., Ramirez, S., Pang, P. T., Puryear, C. B., Govindarajan, A., Deisseroth, K., & Tonegawa,

S. (2012). Optogenetic stimulation of a hippocampal engram activates fear memory

recall. Nature. http://doi.org/10.1038/nature11028

Lopez, J., Herbeaux, K., Cosquer, B., Engeln, M., Muller, C., Lazarus, C., … de Vasconcelos, A.

P. (2012). Context-dependent modulation of hippocampal and cortical recruitment during

remote spatial memory retrieval. Hippocampus, 22(4), 827–841.

http://doi.org/10.1002/hipo.20943

MacDonald, C. J., Lepage, K. Q., Eden, U. T., & Eichenbaum, H. (2011). Hippocampal “Time

Cells” Bridge the Gap in Memory for Discontiguous Events. Neuron, 71(4), 737–749.

http://doi.org/10.1016/j.neuron.2011.07.012

Manes, F., Graham, K. S., Zeman, A., de Luján Calcagno, M., & Hodges, J. R. (2005).

Autobiographical amnesia and accelerated forgetting in transient epileptic amnesia.

Journal of Neurology, Neurosurgery, and Psychiatry, 76(10), 1387–1391.

http://doi.org/10.1136/jnnp.2004.036582

Manns, J. R., & Eichenbaum, H. (2006). Evolution of declarative memory. Hippocampus, 16(9),

795–808. http://doi.org/10.1002/hipo.20205

Marascuilo, L. A., & Serlin, R. C. (1988). Statistical methods for the social and behavioral

sciences. New York: W.H. Freeman.

Markowitsch, H. J. (1995). Which brain regions are critically involved in the retrieval of old

episodic memory? Brain Research Reviews, 21(2), 117–127. http://doi.org/10.1016/0165-

0173(95)00007-0

Page 151: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

140

Martinez, S. (2014). The cerebellum: from development to structural complexity and motor

learning. Frontiers in Neuroanatomy, 8, 118. http://doi.org/10.3389/fnana.2014.00118

Maviel, T., Durkin, T. P., Menzaghi, F., & Bontempi, B. (2004). Sites of Neocortical

Reorganization Critical for Remote Spatial Memory. Science, 305(5680), 96–99.

http://doi.org/10.1126/science.1098180

McClelland, J. L. (2013). Incorporating rapid neocortical learning of new schema-consistent

information into complementary learning systems theory. Journal of Experimental

Psychology. General, 142(4), 1190–1210. http://doi.org/10.1037/a0033812

McClelland, J. L., McNaughton, B. L., & O’Reilly, R. C. (1995). Why there are complementary

learning systems in the hippocampus and neocortex: Insights from the successes and

failures of connectionist models of learning and memory. Psychological Review, 102(3),

419–457. http://doi.org/10.1037/0033-295X.102.3.419

McCormick, D. A., & Thompson, R. F. (1984). Cerebellum: essential involvement in the

classically conditioned eyelid response. Science, 223(4633), 296–9.

McEchron, M. D., & Disterhoft, J. F. (1997). Sequence of single neuron changes in CA1

hippocampus of rabbits during acquisition of trace eyeblink conditioned responses. J

Neurophysiol, 78(2), 1030–44.

McEchron, M. D., Weible, A. P., & Disterhoft, J. F. (2001). Aging and learning-specific changes

in single-neuron activity in CA1 hippocampus during rabbit trace eyeblink conditioning.

J Neurophysiol, 86(4), 1839–1857.

Page 152: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

141

McGlinchey-Berroth, R., Carrillo, M. C., Gabrieli, J. D., Brawn, C. M., & Disterhoft, J. F.

(1997). Impaired trace eyeblink conditioning in bilateral, medial-temporal lobe amnesia.

Behavioral Neuroscience, 111(5), 873–882.

McKenzie, S., & Eichenbaum, H. (2011). Consolidation and Reconsolidation: Two Lives of

Memories? Neuron, 71(2), 224–233. http://doi.org/10.1016/j.neuron.2011.06.037

McKenzie, S., Frank, A. J., Kinsky, N. R., Porter, B., Rivière, P. D., & Eichenbaum, H. (2014).

Hippocampal Representation of Related and Opposing Memories Develop within

Distinct, Hierarchically Organized Neural Schemas. Neuron, 83(1), 202–215.

http://doi.org/10.1016/j.neuron.2014.05.019

McLaughlin, J., Skaggs, H., Churchwell, J., & Powell, D. A. (2002). Medial prefrontal cortex

and pavlovian conditioning: trace versus delay conditioning. Behav Neurosci, 116(1), 37–

47.

Meunier, M., Jaffard, R., & Destrade, C. (1991). Differential involvement of anterior and

posterior cingulate cortices in spatial discriminative learning in a T-maze in mice.

Behavioural Brain Research, 44(2), 133–143.

Miller, E. K., & Cohen, J. D. (2001). An Integrative Theory of Prefrontal Cortex Function.

Annual Review of Neuroscience, 24(1), 167–202.

http://doi.org/10.1146/annurev.neuro.24.1.167

Milner, B. (1959). The memory defect in bilateral hippocampal lesions. Psychiatric Research

Reports, 11, 43–58.

Page 153: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

142

Milner, B. (2005). The Medial Temporal-Lobe Amnesic Syndrome. Psychiatric Clinics of North

America, 28(3), 599–611. http://doi.org/10.1016/j.psc.2005.06.002

Morris, R. G. (2001). Episodic-like memory in animals: psychological criteria, neural

mechanisms and the value of episodic-like tasks to investigate animal models of

neurodegenerative disease. Philosophical Transactions of the Royal Society of London.

Series B, Biological Sciences, 356(1413), 1453–1465.

http://doi.org/10.1098/rstb.2001.0945

Morrissey, M. D., Maal-Bared, G., Brady, S., & Takehara-Nishiuchi, K. (2012). Functional

Dissociation within the Entorhinal Cortex for Memory Retrieval of an Association

between Temporally Discontiguous Stimuli. The Journal of Neuroscience, 32(16), 5356–

5361. http://doi.org/10.1523/JNEUROSCI.5227-11.2012

Morrissey, M. D., & Takehara-Nishiuchi, K. (2014). Diversity of mnemonic function within the

entorhinal cortex: A meta-analysis of rodent behavioral studies. Neurobiology of

Learning and Memory, 115, 95–107. http://doi.org/10.1016/j.nlm.2014.08.006

Moscovitch, M., & Nadel, L. (1998). Consolidation and the hippocampal complex revisited: in

defense of the multiple-trace model. Current Opinion in Neurobiology, 8(2), 297–300.

http://doi.org/10.1016/S0959-4388(98)80155-4

Moscovitch, M., Rosenbaum, R. S., Gilboa, A., Addis, D. R., Westmacott, R., Grady, C., …

Nadel, L. (2005). Functional neuroanatomy of remote episodic, semantic and spatial

memory: a unified account based on multiple trace theory. Journal of Anatomy, 207(1),

35–66. http://doi.org/10.1111/j.1469-7580.2005.00421.x

Page 154: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

143

Moser, E. I., Roudi, Y., Witter, M. P., Kentros, C., Bonhoeffer, T., & Moser, M.-B. (2014). Grid

cells and cortical representation. Nature Reviews Neuroscience, 15(7), 466–481.

http://doi.org/10.1038/nrn3766

Moyer, J. R., Deyo, R. A., & Disterhoft, J. F. (1990). Hippocampectomy disrupts trace eye-blink

conditioning in rabbits. Behavioral Neuroscience, 104(2), 243–252.

Mulder, A. B., Nordquist, R., Örgüt, O., & Pennartz, C. M. A. (2000). Plasticity of neuronal

firing in deep layers of the medial prefrontal cortex in rats engaged in operant

conditioning. In B.-P. in B. Research (Ed.), (Vol. 126, pp. 287–301). Elsevier. Retrieved

from http://www.sciencedirect.com/science/article/pii/S0079612300260202

Mumby, D. G., Astur, R. S., Weisend, M. P., & Sutherland, R. J. (1999). Retrograde amnesia and

selective damage to the hippocampal formation: memory for places and object

discriminations. Behavioural Brain Research, 106(1-2), 97–107.

Mumby, D. G., Gaskin, S., Glenn, M. J., Schramek, T. E., & Lehmann, H. (2002). Hippocampal

damage and exploratory preferences in rats: memory for objects, places, and contexts.

Learning & Memory (Cold Spring Harbor, N.Y.), 9(2), 49–57.

http://doi.org/10.1101/lm.41302

Mumby, D. G., & Pinel, J. P. (1994). Rhinal cortex lesions and object recognition in rats.

Behavioral Neuroscience, 108(1), 11–18.

Munera, A., Gruart, A., Munoz, M. D., Fernandez-Mas, R., & Delgado-Garcia, J. M. (2001).

Hippocampal pyramidal cell activity encodes conditioned stimulus predictive value

during classical conditioning in alert cats. J Neurophysiol, 86(5), 2571–82.

Page 155: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

144

Nadel, L., & Moscovitch, M. (1997). Memory consolidation, retrograde amnesia and the

hippocampal complex. Current Opinion in Neurobiology, 7(2), 217–227.

http://doi.org/10.1016/S0959-4388(97)80010-4

Nader, K. (2015). Reconsolidation and the Dynamic Nature of Memory. Cold Spring Harbor

Perspectives in Biology, 7(10). http://doi.org/10.1101/cshperspect.a021782

Nader, K., Schafe, G. E., & Le Doux, J. E. (2000). Fear memories require protein synthesis in the

amygdala for reconsolidation after retrieval. Nature, 406(6797), 722–726.

http://doi.org/10.1038/35021052

O’Keefe, J. (1983). Two Spatial Systems in the Rat Brain – Implications for the Neural Basis of

Learning and Memory. In J. G. M. Imbert and F. E. Bloom J.-P. Changeux (Ed.),

Progress in Brain Research (Vol. 58, pp. 453–464). Elsevier. Retrieved from

http://www.sciencedirect.com/science/article/pii/S0079612308600485

O’Keefe, J., & Recce, M. L. (1993). Phase relationship between hippocampal place units and the

EEG theta rhythm. Hippocampus, 3(3), 317–330. http://doi.org/10.1002/hipo.450030307

Oswald, B. B., Knuckley, B., Maddox, S. A., & Powell, D. A. (2007). Ibotenic acid lesions to

ventrolateral thalamic nuclei disrupts trace and delay eyeblink conditioning in rabbits.

Behav Brain Res, 179(1), 111–117. http://doi.org/10.1016/j.bbr.2007.01.016

Oswald, B. B., Maddox, S. A., & Powell, D. A. (2008). Prefrontal control of trace eyeblink

conditioning in rabbits: Role in retrieval of the CR? Behav Brain Res, 122(4), 841–848.

http://doi.org/10.1037/0735-7044.122.4.841

Page 156: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

145

Oswald, B. B., Maddox, S. A., Tisdale, N., & Powell, D. A. (2010). Encoding and retrieval are

differentially processed by the anterior cingulate and prelimbic cortices: A study based on

trace eyeblink conditioning in the rabbit. Neurobiology of Learning and Memory, 93(1),

37–45. http://doi.org/10.1016/j.nlm.2009.08.001

Panzeri, S., Senatore, R., Montemurro, M. A., & Petersen, R. S. (2007). Correcting for the

Sampling Bias Problem in Spike Train Information Measures. Journal of

Neurophysiology, 98(3), 1064–1072. http://doi.org/10.1152/jn.00559.2007

Paz, R., Bauer, E. P., & Pare, D. (2007). Learning-Related Facilitation of Rhinal Interactions by

Medial Prefrontal Inputs. Journal of Neuroscience, 27(24), 6542–6551.

http://doi.org/10.1523/JNEUROSCI.1077-07.2007

Penick, S., & Solomom, P. R. (1991). Hippocampus, context, and conditioning. Behav Neurosci,

105(5). Retrieved from http://psycnet.apa.org/journals/bne/105/5/611/

Peters, Y. M., O’Donnell, P., & Carelli, R. M. (2005). Prefrontal cortical cell firing during

maintenance, extinction, and reinstatement of goal-directed behavior for natural reward.

Synapse, 56(2), 74–83. http://doi.org/10.1002/syn.20129

Peyrache, A., Khamassi, M., Benchenane, K., Wiener, S. I., & Battaglia, F. P. (2009). Replay of

rule-learning related neural patterns in the prefrontal cortex during sleep. Nature

Neuroscience, 12(7), 919–926. http://doi.org/10.1038/nn.2337

Piaget, J. (1929). The Child’s Conception of the World. Littlefield, Adams.

Page 157: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

146

Posner, M. I., Rothbart, M. K., Sheese, B. E., & Tang, Y. (2007). The anterior cingulate gyrus

and the mechanism of self-regulation. Cognitive, Affective & Behavioral Neuroscience,

7(4), 391–395.

Powell, D. A., & Churchwell, J. (2002). Mediodorsal thalamic lesions impair trace eyeblink

conditioning in the rabbit. Learn Mem, 9(1), 10–7. http://doi.org/10.1101/lm.45302

Preston, A. R., & Eichenbaum, H. (2013). Interplay of Hippocampus and Prefrontal Cortex in

Memory. Current Biology, 23(17), R764–R773. http://doi.org/10.1016/j.cub.2013.05.041

Quinn, J. J., Ma, Q. D., Tinsley, M. R., Koch, C., & Fanselow, M. S. (2008). Inverse temporal

contributions of the dorsal hippocampus and medial prefrontal cortex to the expression of

long-term fear memories. Learning & Memory, 15(5), 368–372.

http://doi.org/10.1101/lm.813608

Rajasethupathy, P., Sankaran, S., Marshel, J. H., Kim, C. K., Ferenczi, E., Lee, S. Y., …

Deisseroth, K. (2015). Projections from neocortex mediate top-down control of memory

retrieval. Nature. http://doi.org/10.1038/nature15389

Ramirez, S., Tonegawa, S., & Liu, X. (2014). Identification and optogenetic manipulation of

memory engrams in the hippocampus. Frontiers in Behavioral Neuroscience, 7, 226.

http://doi.org/10.3389/fnbeh.2013.00226

Rao, S. C., Rainer, G., & Miller, E. K. (1997). Integration of what and where in the primate

prefrontal cortex. Science (New York, N.Y.), 276(5313), 821–824.

Page 158: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

147

Raybuck, J. D., & Gould, T. J. (2010). The role of nicotinic acetylcholine receptors in the medial

prefrontal cortex and hippocampus in trace fear conditioning. Neurobiology of Learning

and Memory, 94(3), 353–363. http://doi.org/10.1016/j.nlm.2010.08.001

Remondes, M., & Schuman, E. M. (2004). Role for a cortical input to hippocampal area CA1 in

the consolidation of a long-term memory. Nature, 431(7009), 699–703.

http://doi.org/10.1038/nature02965

Restivo, L., Vetere, G., Bontempi, B., & Ammassari-Teule, M. (2009). The Formation of Recent

and Remote Memory Is Associated with Time-Dependent Formation of Dendritic Spines

in the Hippocampus and Anterior Cingulate Cortex. The Journal of Neuroscience, 29(25),

8206–8214. http://doi.org/10.1523/JNEUROSCI.0966-09.2009

Richards, B. A., Xia, F., Santoro, A., Husse, J., Woodin, M. A., Josselyn, S. A., & Frankland, P.

W. (2014). Patterns across multiple memories are identified over time. Nature

Neuroscience, 17(7), 981–986. http://doi.org/10.1038/nn.3736

Rich, E. L., & Shapiro, M. (2009). Rat Prefrontal Cortical Neurons Selectively Code Strategy

Switches. Journal of Neuroscience, 29(22), 7208–7219.

http://doi.org/10.1523/JNEUROSCI.6068-08.2009

Rigotti, M., Barak, O., Warden, M. R., Wang, X.-J., Daw, N. D., Miller, E. K., & Fusi, S. (2013).

The importance of mixed selectivity in complex cognitive tasks. Nature, 497(7451), 585–

590. http://doi.org/10.1038/nature12160

Rolls, E. T. (2010). A computational theory of episodic memory formation in the hippocampus.

Behavioural Brain Research, 215(2), 180–196. http://doi.org/10.1016/j.bbr.2010.03.027

Page 159: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

148

Rolls, E. T., Stringer, S. M., & Trappenberg, T. P. (2002). A unified model of spatial and

episodic memory. Proceedings of the Royal Society B: Biological Sciences, 269(1496),

1087–1093. http://doi.org/10.1098/rspb.2002.2009

Runyan, J. D., Moore, A. N., & Dash, P. K. (2004). A Role for Prefrontal Cortex in Memory

Storage for Trace Fear Conditioning. The Journal of Neuroscience, 24(6), 1288–1295.

http://doi.org/10.1523/JNEUROSCI.4880-03.2004

Scoville, W. B., & Milner, B. (1957). Loss of Recent Memory After Bilateral Hippocampal

Lesions. Journal of Neurology, Neurosurgery & Psychiatry, 20(1), 11–21.

http://doi.org/10.1136/jnnp.20.1.11

Seamans, J. K., Lapish, C. C., & Durstewitz, D. (2008). Comparing the prefrontal cortex of rats

and primates: Insights from electrophysiology. Neurotoxicity Research, 14(2-3), 249–

262. http://doi.org/10.1007/BF03033814

Seidenbecher, T., Laxmi, T. R., Stork, O., & Pape, H.-C. (2003). Amygdalar and hippocampal

theta rhythm synchronization during fear memory retrieval. Science (New York, N.Y.),

301(5634), 846–850. http://doi.org/10.1126/science.1085818

Senn, V., Wolff, S. B. E., Herry, C., Grenier, F., Ehrlich, I., Gründemann, J., … Lüthi, A. (2014).

Long-range connectivity defines behavioral specificity of amygdala neurons. Neuron,

81(2), 428–437. http://doi.org/10.1016/j.neuron.2013.11.006

Sharpe, D. (2015). Your Chi-Square Test is Statistically Significant: Now What? Practical

Assessment, Research & Evaluation, 20. Retrieved from

http://search.ebscohost.com/login.aspx?direct=true&profile=ehost&scope=site&authtype

Page 160: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

149

=crawler&jrnl=15317714&AN=108545923&h=%2FVdvTtD22c937Au4ZQKZt5y7%2B

h%2F0NXDUJl7smPIDlppxtrndLSNc%2B7BZeaojvZK%2FUyhzUMg69M4IPRs6kTt

WxQ%3D%3D&crl=c

Siapas, A. G., Lubenov, E. V., & Wilson, M. A. (2005a). Prefrontal Phase Locking to

Hippocampal Theta Oscillations. Neuron, 46(1), 141–151.

http://doi.org/10.1016/j.neuron.2005.02.028

Siapas, A. G., Lubenov, E. V., & Wilson, M. A. (2005b). Prefrontal Phase Locking to

Hippocampal Theta Oscillations. Neuron, 46(1), 141–151.

http://doi.org/10.1016/j.neuron.2005.02.028

Siegel, J. J., Kalmbach, B., Chitwood, R. A., & Mauk, M. D. (2012). Persistent activity in a

cortical-to-subcortical circuit: bridging the temporal gap in trace eyelid conditioning. J

Neurophysiol, 107(1), 50–64. http://doi.org/10.1152/jn.00689.2011

Siegel, J. J., & Mauk, M. D. (2013). Persistent Activity in Prefrontal Cortex during Trace Eyelid

Conditioning: Dissociating Responses That Reflect Cerebellar Output from Those That

Do Not. J Neurosci, 33(38), 15272–15284. http://doi.org/10.1523/JNEUROSCI.1238-

13.2013

Simons, J. S., & Spiers, H. J. (2003). Prefrontal and medial temporal lobe interactions in long-

term memory. Nature Reviews Neuroscience, 4(8), 637–648.

http://doi.org/10.1038/nrn1178

Singer, W. (1999). Neuronal Synchrony: A Versatile Code for the Definition of Relations?

Neuron, 24(1), 49–65. http://doi.org/10.1016/S0896-6273(00)80821-1

Page 161: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

150

Sirota, A., Montgomery, S., Fujisawa, S., Isomura, Y., Zugaro, M., & Buzsáki, G. (2008).

Entrainment of Neocortical Neurons and Gamma Oscillations by the Hippocampal Theta

Rhythm. Neuron, 60(4), 683–697. http://doi.org/10.1016/j.neuron.2008.09.014

Skaggs, W. E., McNaughton, B. L., Wilson, M. A., & Barnes, C. A. (1996). Theta phase

precession in hippocampal neuronal populations and the compression of temporal

sequences. Hippocampus, 6(2), 149–172. http://doi.org/10.1002/(SICI)1098-

1063(1996)6:2&lt;149::AID-HIPO6&gt;3.0.CO;2-K

Solomon, P. R., & Vander Schaaf, E. R. (1986). Hippocampus and trace conditioning of the

rabbit’s classically conditioned nictitating membrane response. Behav Neurosci, 100(5),

729–744.

Spellman, T., Rigotti, M., Ahmari, S. E., Fusi, S., Gogos, J. A., & Gordon, J. A. (2015).

Hippocampal-prefrontal input supports spatial encoding in working memory. Nature,

522(7556), 309–314. http://doi.org/10.1038/nature14445

Squire, L. R. (1992). Memory and the hippocampus: A synthesis from findings with rats,

monkeys, and humans. Psychological Review, 99(2), 195–231.

http://doi.org/10.1037/0033-295X.99.2.195

Squire, L. R. (2009). Memory and Brain Systems: 1969-2009. Journal of Neuroscience, 29(41),

12711–12716. http://doi.org/10.1523/JNEUROSCI.3575-09.2009

Steiger, J. H. (1980). Tests for Comparing Elements of a Correlation Matrix. Psychological

Bulletin, 87(2), 245–252.

Page 162: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

151

Steinmetz, A. B., Harmon, T. C., & Freeman, J. H. (2013). Visual cortical contributions to

associative cerebellar learning. Neurobiol Learn Mem, 104, 103–9.

http://doi.org/10.1016/j.nlm.2013.06.005

Stern, C. A. J., Gazarini, L., Vanvossen, A. C., Hames, M. S., & Bertoglio, L. J. (2014). Activity

in prelimbic cortex subserves fear memory reconsolidation over time. Learning &

Memory, 21(1), 14–20. http://doi.org/10.1101/lm.032631.113

Suter, E. E., Weiss, C., & Disterhoft, J. F. (2013). Perirhinal and postrhinal, but not lateral

entorhinal, cortices are essential for acquisition of trace eyeblink conditioning. Learning

& Memory, 20(2), 80–84. http://doi.org/10.1101/lm.028894.112

Sutherland, R. J., Weisend, M. P., Mumby, D., Astur, R. S., Hanlon, F. M., Koerner, A., …

Hoesing, J. M. (2001). Retrograde amnesia after hippocampal damage: recent vs. remote

memories in two tasks. Hippocampus, 11(1), 27–42. http://doi.org/10.1002/1098-

1063(2001)11:1<27::AID-HIPO1017>3.0.CO;2-4

Suzuki, W. A., Zola-Morgan, S., Squire, L. R., & Amaral, D. G. (1993). Lesions of the perirhinal

and parahippocampal cortices in the monkey produce long-lasting memory impairment in

the visual and tactual modalities. The Journal of Neuroscience: The Official Journal of

the Society for Neuroscience, 13(6), 2430–2451.

Tahvildari, B., Fransén, E., Alonso, A. A., & Hasselmo, M. E. (2007). Switching between “On”

and “Off” states of persistent activity in lateral entorhinal layer III neurons.

Hippocampus, 17(4), 257–263. http://doi.org/10.1002/hipo.20270

Page 163: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

152

Takashima, A., Petersson, K. M., Rutters, F., Tendolkar, I., Jensen, O., Zwarts, M. J., …

Fernández, G. (2006). Declarative memory consolidation in humans: A prospective

functional magnetic resonance imaging study. Proceedings of the National Academy of

Sciences of the United States of America, 103(3), 756–761.

http://doi.org/10.1073/pnas.0507774103

Takehara, K., Kawahara, S., & Kirino, Y. (2003). Time-Dependent Reorganization of the Brain

Components Underlying Memory Retention in Trace Eyeblink Conditioning. The Journal

of Neuroscience, 23(30), 9897–9905.

Takehara, K., Kawahara, S., Takatsuki, K., & Kirino, Y. (2002). Time-limited role of the

hippocampus in the memory for trace eyeblink conditioning in mice. Brain Res, 951(2),

183–90.

Takehara-Nishiuchi, K. (2014). Entorhinal cortex and consolidated memory. Neuroscience

Research, 84, 27–33. http://doi.org/10.1016/j.neures.2014.02.012

Takehara-Nishiuchi, K., Kawahara, S., & Kirino, Y. (2005). NMDA receptor-dependent

processes in the medial prefrontal cortex are important for acquisition and the early stage

of consolidation during trace, but not delay eyeblink conditioning. Learning & Memory,

12(6), 606–614. http://doi.org/10.1101/lm.5905

Takehara-Nishiuchi, K., Maal-Bared, G., & Morrissey, M. (2012). Increased entorhinal–

prefrontal theta synchronization parallels decreased entorhinal–hippocampal theta

synchronization during learning and consolidation of associative memory. Frontiers in

Behavioral Neuroscience, 5, 90. http://doi.org/10.3389/fnbeh.2011.00090

Page 164: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

153

Takehara-Nishiuchi, K., & McNaughton, B. L. (2008). Spontaneous Changes of Neocortical

Code for Associative Memory During Consolidation. Science, 322(5903), 960–963.

http://doi.org/10.1126/science.1161299

Takehara-Nishiuchi, K., Nakao, K., Kawahara, S., Matsuki, N., & Kirino, Y. (2006). Systems

Consolidation Requires Postlearning Activation of NMDA Receptors in the Medial

Prefrontal Cortex in Trace Eyeblink Conditioning. The Journal of Neuroscience, 26(19),

5049–5058. http://doi.org/10.1523/JNEUROSCI.4381-05.2006

Tanaka, K. Z., Pevzner, A., Hamidi, A. B., Nakazawa, Y., Graham, J., & Wiltgen, B. J. (2014).

Cortical Representations Are Reinstated by the Hippocampus during Memory Retrieval.

Neuron, 84(2), 347–354. http://doi.org/10.1016/j.neuron.2014.09.037

Tanninen, S. E., Yu, X., Giritharan, T., Tran, L., Bakir, R., Volle, J., … Takehara-Nishiuchi, K.

(2015). Cholinergic, but not NMDA, receptors in the lateral entorhinal cortex mediate

acquisition in trace eyeblink conditioning. Hippocampus, n/a–n/a.

http://doi.org/10.1002/hipo.22466

Teixeira, C. M., Pomedli, S. R., Maei, H. R., Kee, N., & Frankland, P. W. (2006). Involvement

of the Anterior Cingulate Cortex in the Expression of Remote Spatial Memory. The

Journal of Neuroscience, 26(29), 7555–7564. http://doi.org/10.1523/JNEUROSCI.1068-

06.2006

Teyler, T. J., & DiScenna, P. (1986). The hippocampal memory indexing theory. Behavioral

Neuroscience, 100(2), 147–154. http://doi.org/10.1037/0735-7044.100.2.147

Page 165: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

154

Thomas Elbert, B. R. (1987). Threshold regulation: A key to understanding the combined

dynamics of EEG and event-related potentials. Journal of Psychophysiology, 1(4), 317–

333.

Thompson, R. F., Bao, S., Chen, L., Cipriano, B. D., Grethe, J. S., Kim, J. J., … Krupa, D. J.

(1997). Associative learning. International Review of Neurobiology, 41, 151–189.

Thornton, J. A., Rothblat, L. A., & Murray, E. A. (1997). Rhinal cortex removal produces

amnesia for preoperatively learned discrimination problems but fails to disrupt

postoperative acquisition and retention in rhesus monkeys. The Journal of Neuroscience,

17(21), 8536–8549.

Tse, D., Langston, R. F., Kakeyama, M., Bethus, I., Spooner, P. A., Wood, E. R., … Morris, R.

G. M. (2007). Schemas and Memory Consolidation. Science, 316(5821), 76–82.

http://doi.org/10.1126/science.1135935

Tse, D., Takeuchi, T., Kakeyama, M., Kajii, Y., Okuno, H., Tohyama, C., … Morris, R. G. M.

(2011). Schema-Dependent Gene Activation and Memory Encoding in Neocortex.

Science, 333(6044), 891–895. http://doi.org/10.1126/science.1205274

Tseng, W., Guan, R., Disterhoft, J. F., & Weiss, C. (2004). Trace eyeblink conditioning is

hippocampally dependent in mice. Hippocampus, 14(1), 58–65.

http://doi.org/10.1002/hipo.10157

Tulving, E. (1987). Multiple memory systems and consciousness. Human Neurobiology, 6(2),

67–80.

Page 166: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

155

Tulving, E. (2001). Episodic memory and common sense: how far apart? Philosophical

Transactions of the Royal Society of London B: Biological Sciences, 356(1413), 1505–

1515. http://doi.org/10.1098/rstb.2001.0937

Vanderwolf, C. H. (1969). Hippocampal electrical activity and voluntary movement in the rat.

Electroencephalography and Clinical Neurophysiology, 26(4), 407–418.

van Kesteren, M. T. R., Ruiter, D. J., Fernández, G., & Henson, R. N. (2012). How schema and

novelty augment memory formation. Trends in Neurosciences, 35(4), 211–219.

http://doi.org/10.1016/j.tins.2012.02.001

van Strien, N. M., Cappaert, N. L. M., & Witter, M. P. (2009). The anatomy of memory: an

interactive overview of the parahippocampal–hippocampal network. Nature Reviews

Neuroscience, 10(4), 272–282. http://doi.org/10.1038/nrn2614

Vertes, R. P. (2004). Differential projections of the infralimbic and prelimbic cortex in the rat.

Synapse, 51(1), 32–58. http://doi.org/10.1002/syn.10279

Vertes, R. P., Hoover, W. B., Szigeti-Buck, K., & Leranth, C. (2007). Nucleus reuniens of the

midline thalamus: link between the medial prefrontal cortex and the hippocampus. Brain

Res Bull, 71(6), 601–9. http://doi.org/10.1016/j.brainresbull.2006.12.002

Vetere, G., Restivo, L., Cole, C. J., Ross, P. J., Ammassari-Teule, M., Josselyn, S. A., &

Frankland, P. W. (2011). Spine growth in the anterior cingulate cortex is necessary for

the consolidation of contextual fear memory. Proceedings of the National Academy of

Sciences, 108(20), 8456–8460. http://doi.org/10.1073/pnas.1016275108

Page 167: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

156

Vetere, G., Restivo, L., Cole, C. J., Ross, P. J., Ammassari-Teule, M., Josselyn, S. A., &

Frankland, P. W. (2011). Spine growth in the anterior cingulate cortex is necessary for

the consolidation of contextual fear memory. Proceedings of the National Academy of

Sciences, 108(20), 8456–8460. http://doi.org/10.1073/pnas.1016275108

Vinck, M., Battaglia, F. P., Womelsdorf, T., & Pennartz, C. (2012). Improved measures of

phase-coupling between spikes and the Local Field Potential. Journal of Computational

Neuroscience, 33(1), 53–75. http://doi.org/10.1007/s10827-011-0374-4

Walker, A. G., & Steinmetz, J. E. (2008). Hippocampal lesions in rats differentially affect long-

and short-trace eyeblink conditioning. Physiol Behav, 93(3), 570–578.

http://doi.org/10.1016/j.physbeh.2007.10.018

Wallisch, P. (2014). Chapter 20 - Information Theory. In P. Wallisch, M. E. Lusignan, M. D.

Benayoun, T. I. Baker, A. S. Dickey, & N. G. Hatsopoulos (Eds.), MATLAB for

Neuroscientists (Second Edition) (pp. 317–327). San Diego: Academic Press. Retrieved

from http://www.sciencedirect.com/science/article/pii/B9780123838360000187

Wallis, J. D., Anderson, K. C., & Miller, E. K. (2001). Single neurons in prefrontal cortex

encode abstract rules. Nature, 411(6840), 953–956. http://doi.org/10.1038/35082081

Wang, M. E., Fraize, N. P., Yin, L., Yuan, R. K., Petsagourakis, D., Wann, E. G., & Muzzio, I.

A. (2013). Differential roles of the dorsal and ventral hippocampus in predator odor

contextual fear conditioning. Hippocampus, 23(6), 451–466.

http://doi.org/10.1002/hipo.22105

Page 168: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

157

Wang, S.-H., & Morris, R. G. M. (2010a). Hippocampal-Neocortical Interactions in Memory

Formation, Consolidation, and Reconsolidation. Annual Review of Psychology, 61(1),

49–79. http://doi.org/10.1146/annurev.psych.093008.100523

Wang, S.-H., & Morris, R. G. M. (2010b). Hippocampal-Neocortical Interactions in Memory

Formation, Consolidation, and Reconsolidation. Annual Review of Psychology, 61(1),

49–79. http://doi.org/10.1146/annurev.psych.093008.100523

Wang, S.-H., Teixeira, C. M., Wheeler, A. L., & Frankland, P. W. (2009). The precision of

remote context memories does not require the hippocampus. Nature Neuroscience, 12(3),

253–255. http://doi.org/10.1038/nn.2263

Wang, S.-H., Tse, D., & Morris, R. G. M. (2012). Anterior cingulate cortex in schema

assimilation and expression. Learning & Memory, 19(8), 315–318.

http://doi.org/10.1101/lm.026336.112

Watanabe, M. (1996). Reward expectancy in primate prefrontal neurons. Nature, 382(6592),

629–632. http://doi.org/10.1038/382629a0

Weible, A. P., McEchron, M. D., & Disterhoft, J. F. (2000). Cortical involvement in acquisition

and extinction of trace eyeblink conditioning. Behav Neurosci, 114(6), 1058–67.

Weible, A. P., O’Reilly, J.-A., Weiss, C., & Disterhoft, J. F. (2006). Comparisons of dorsal and

ventral hippocampus cornu ammonis region 1 pyramidal neuron activity during trace eye-

blink conditioning in the rabbit. Neuroscience, 141(3), 1123–1137.

http://doi.org/10.1016/j.neuroscience.2006.04.065

Page 169: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

158

Weible, A. P., Rowland, D. C., Monaghan, C. K., Wolfgang, N. T., & Kentros, C. G. (2012).

Neural Correlates of Long-Term Object Memory in the Mouse Anterior Cingulate

Cortex. The Journal of Neuroscience, 32(16), 5598–5608.

http://doi.org/10.1523/JNEUROSCI.5265-11.2012

Weible, A. P., Rowland, D. C., Pang, R., & Kentros, C. (2009). Neural Correlates of Novel

Object and Novel Location Recognition Behavior in the Mouse Anterior Cingulate

Cortex. Journal of Neurophysiology, 102(4), 2055–2068.

http://doi.org/10.1152/jn.00214.2009

Weible, A. P., Weiss, C., & Disterhoft, J. F. (2003). Activity Profiles of Single Neurons in

Caudal Anterior Cingulate Cortex During Trace Eyeblink Conditioning in the Rabbit.

Journal of Neurophysiology, 90(2), 599–612. http://doi.org/10.1152/jn.01097.2002

Weible, A. P., Weiss, C., & Disterhoft, J. F. (2003). Activity profiles of single neurons in caudal

anterior cingulate cortex during trace eyeblink conditioning in the rabbit. J Neurophysiol,

90(2), 599–612. http://doi.org/10.1152/jn.01097.2002

Weiss, C., Bouwmeester, H., Power, J. M., & Disterhoft, J. F. (1999). Hippocampal lesions

prevent trace eyeblink conditioning in the freely moving rat. Behavioural Brain

Research, 99(2), 123–132. http://doi.org/10.1016/S0166-4328(98)00096-5

Weiss, C., & Disterhoft, J. F. (2011). Exploring prefrontal cortical memory mechanisms with

eyeblink conditioning. Behavioral Neuroscience, 125(3), 318–326.

http://doi.org/10.1037/a0023520

Page 170: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

159

Weiss, C., Kronforst-Collins, M. A., & Disterhoft, J. F. (1996). Activity of hippocampal

pyramidal neurons during trace eyeblink conditioning. Hippocampus, 6(2), 192–209.

http://doi.org/10.1002/(SICI)1098-1063(1996)6:2<192::AID-HIPO9>3.0.CO;2-R

Weisz, V. I., Rios, M. B., & Argibay, P. F. (2012). Episodic-like memory: New perspectives

from a behavioral test in rats. Journal of Integrative Neuroscience, 11(01), 1–15.

http://doi.org/10.1142/S021963521250001X

White, I. M., & Wise, S. P. (1999). Rule-dependent neuronal activity in the prefrontal cortex.

Experimental Brain Research, 126(3), 315–335.

Wiestler, T., & Diedrichsen, J. (2013). Skill learning strengthens cortical representations of

motor sequences. eLife, 2, e00801. http://doi.org/10.7554/eLife.00801

Wikgren, J., Nokia, M. S., & Penttonen, M. (2010). Hippocampo-cerebellar theta band phase

synchrony in rabbits. Neuroscience, 165(4), 1538–1545.

http://doi.org/10.1016/j.neuroscience.2009.11.044

Wilson, M. A., & McNaughton, B. L. (1993). Dynamics of the hippocampal ensemble code for

space. Science, 261(5124), 1055–8.

Winocur, G. (1990). Anterograde and retrograde amnesia in rats with dorsal hippocampal or

dorsomedial thalamic lesions. Behavioural Brain Research, 38(2), 145–154.

http://doi.org/10.1016/0166-4328(90)90012-4

Winocur, G., & Moscovitch, M. (2011). Memory Transformation and Systems Consolidation.

Journal of the International Neuropsychological Society, 17(05), 766–780.

http://doi.org/10.1017/S1355617711000683

Page 171: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

160

Winocur, G., Moscovitch, M., & Bontempi, B. (2010). Memory formation and long-term

retention in humans and animals: Convergence towards a transformation account of

hippocampal–neocortical interactions. Neuropsychologia, 48(8), 2339–2356.

http://doi.org/10.1016/j.neuropsychologia.2010.04.016

Winocur, G., Sekeres, M. J., Binns, M. A., & Moscovitch, M. (2013). Hippocampal lesions

produce both nongraded and temporally graded retrograde amnesia in the same rat.

Hippocampus, 23(5), 330–341. http://doi.org/10.1002/hipo.22093

Wise, S. P., & Murray, E. A. (2000). Arbitrary associations between antecedents and actions.

Trends in Neurosciences, 23(6), 271–276.

Witter, M. P. (2007). The perforant path: projections from the entorhinal cortex to the dentate

gyrus. In Helen E. Scharfman (Ed.), Progress in Brain Research (Vol. Volume 163, pp.

43–61). Elsevier. Retrieved from

http://www.sciencedirect.com/science/article/pii/S0079612307630039

Witter, M. P., Hoesen, G. V., & Amaral, D. G. (1989). Topographical organization of the

entorhinal projection to the dentate gyrus of the monkey. The Journal of Neuroscience,

9(1), 216–228.

Woodruff-Pak, D. S. (1993). Eyeblink classical conditioning in H.M.: delay and trace paradigms.

Behavioral Neuroscience, 107(6), 911–925.

Woodruff-Pak, D. S., & Disterhoft, J. F. (2008). Where is the trace in trace conditioning? Trends

in Neurosciences, 31(2), 105–112. http://doi.org/10.1016/j.tins.2007.11.006

Page 172: Memory representation in the medial prefrontal cortex...the meaning of the stimulus post-consolidation. I discuss how these findings fit with existing theories of the mPFC role in

161

Zhu, H., Pleil, K. E., Urban, D. J., Moy, S. S., Kash, T. L., & Roth, B. L. (2014). Chemogenetic

Inactivation of Ventral Hippocampal Glutamatergic Neurons Disrupts Consolidation of

Contextual Fear Memory. Neuropsychopharmacology, 39(8), 1880–1892.

http://doi.org/10.1038/npp.2014.35

Zola-Morgan, S. M., & Squire, L. R. (1990). The primate hippocampal formation: evidence for a

time-limited role in memory storage. Science (New York, N.Y.), 250(4978), 288–290.

Zola-Morgan, S., Squire, L. R., Amaral, D. G., & Suzuki, W. A. (1989). Lesions of perirhinal

and parahippocampal cortex that spare the amygdala and hippocampal formation produce

severe memory impairment. The Journal of Neuroscience, 9(12), 4355–4370.