apical dendrite and theory of consiousness (la berge, 2007)
TRANSCRIPT
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 1/17
Neural Networks 20 (2007) 1004–1020www.elsevier.com/locate/neunet
2007 Special Issue
The apical dendrite theory of consciousness
David LaBerge∗, Ray Kasevich
Bard College at Simon’s Rock and Stanley Laboratory of Electrical Physics, Great Barrington, MA, 01230, United States
Abstract
The neural basis of consciousness is theorized here to be the elevated activity of the apical dendrite within a thalamocortical circuit. Both
the anatomical and functional properties of these two brain structures are examined within the general context of the cortical minicolumn, which
is regarded as the functional unit of the cerebral cortex. Two main circuits of the minicolumn are described: the axis circuit, which sustains
activity for extended durations and produces our sensory impressions, and the shell circuit, which performs input–output processing and produces
identifications, categorizations, and ideas. The apical dendrite operates within the axis circuit to stabilize neural activity, which enables conscious
impressions to be steady and to be sustained over long periods of time. In an attempt to understand how the conscious aspect of subjective
impressions may be related to apical dendrite activity, we examine the characteristics of the electric and magnetic fields during the movement of
charges along the apical dendrite. The physical correlate of consciousness is regarded here as the relatively intense electromagnetic field that is
located along the inside and the outside close to the surface of the active apical dendrite.c 2007 Elsevier Ltd. All rights reserved.
Keywords: Consciousness; Apical dendrite; Minicolumn; Electric and magnetic fields
1. Introduction
The special kind of activity in the brain that is presumed
to underlie consciousness continues to elude the grasp of
our scientific concepts. However, current theories of the
neural correlates of consciousness seem to agree on one
particular aspect of consciousness: its extendable duration.
Consciousness is an activity that is extended in time and
typically its duration continues from the time we awake to
the time we fall asleep. Grossberg (1999, 2005) describes
consciousness in terms of “resonance states” that occur in short-
term memory. When top-down expectations select consistentbottom-up signals and their reciprocal feedback activity
settles into a steady state, their circuits resonate together;
and while this resonance continues, consciousness exists,
according to his Adaptive Resonance Theory (ART). Taylor
(1999, 2005, 2007) describes consciousness as “temporally
extended neural activity” in a buffer store of working memory,
∗ Corresponding address: P.O. Box 301, So. Egremont, MA, 01258, UnitedStates. Tel.: +1 413 528 1303; fax: +1 413 644 9836.
E-mail address: [email protected] (D. LaBerge).
which is based on the corollary discharge of attention
movement (the CODAM model). According to the Theory
of Neuronal Group Selection (TNGS) of Edelman (2003),
consciousness is produced by reentrant interactions involving
the thalamocortical system, whose circuitry is dominated by
recurrent or looping circuits. Newman, Baars, and Cho (1997)
regard the activation loops of the thalamocortical circuit within
a global workspace model as crucial for generating conscious
states.
Unfortunately, mental activities that have extendable
durations, even of a few seconds, are not the kinds of
events that the traditional scientific framework of input–outputprocessing is structured to investigate and explain. Processing
of information is generally regarded as directed toward an
output, and not toward sustaining itself over prolonged periods
of time without apparent outputs. Therefore, if the input-
processing–output framework is the current reigning paradigm,
might the restrictive use of it hinder the quest for the neural
correlates of consciousness? In the present paper this paradigm
is modified to emphasize the input-processing components as
we seek to understand how consciousness may be generated at
the level of the neuron within neural circuits.
0893-6080/$ - see front matter c 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neunet.2007.09.006
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 2/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1005
2. The circuitry in minicolumns of the cortex
2.1. The pyramidal neuron in the minicolumn circuit structure
The theory of consciousness described in the present paper
centers on the activity of a particular kind of cortical neuron: the
pyramidal neuron. The defining feature of the pyramidal neuron(the pyramid) that separates it from other neurons of the brain
is its very long apical dendrite, which generates the special
kind of electrical activity that is presumed here to underlie the
subjective impressions we call consciousness.
To set the stage for the description of the pyramidal neuron
and its apical dendrite in the circuit activity of the minicolumn,
we show how the pyramidal neuron is distributed within the
cerebral cortex. In Fig. 1, the top two diagrams show how
the cortical fabric, which is extensively convoluted to fit into
the sphere-like shape of the skull, can be transformed into a
flat surface (Van Essen, Drury, Joshi, & Miller, 1998), whose
thickness is defined by the length of columns of neurons.
Fig. 1 indicates that each column of neurons consists of a cluster of many minicolumns, and the structure of the
minicolumn appears to be organized around the central core
of layer 5 pyramids, which are the pyramids that contain the
longest apical dendrites (Peters & Sethares, 1991). Layer 6
pyramids are not organized within the minicolumn but instead
are organized within the column, which is a cluster of
approximately 100 minicolumns (Mountcastle, 1998). The
minicolumn is regarded as the functional unit of the cortex
(Mountcastle, 1957), which is based on the observation that
neurons in a given minicolumn share many receptive field
properties. The centrality of the long apical dendrites of layer
5 pyramidal neurons in the structure of the minicolumn isapparent in the two diagrams in the lower part of Fig. 1.
Later sections of the present paper will address the possible
centrality of the layer 5 pyramidal neurons in the function of
the minicolumn.
The pyramidal neuron is shown diagrammatically in
Fig. 1 as a triangular shape, representing the soma, with
a relatively long line attached at the top, representing the
apical dendrite. Thus, the axon and the many basal dendrites
of the pyramids are omitted in these diagrams. The other
neurons shown in the figure are stellate neurons, whose
many radiating lines represent the relatively short dendrites
attached to the soma. Visual inspection of Fig. 1 gives the
impression that the considerable majority of neurons in thecerebral cortex are pyramidal neurons, which is confirmed
by Feldman’s (1984) measurements, which estimate their
percentage in the cortex as approximately 70%–80%. The
remaining percentages of stellate and inhibitory neurons
(in area V1) are approximately 5%–8% and 16%–20%,
respectively (Mountcastle, 1998). Under the area of a dime
placed on the human skull over the parietal cortex there are
approximately 18,000,000–20,000,000 pyramidal neurons, of
which approximately 22% or 4,000,000–4,400,000 are layer
5 pyramidal neurons (based a cell count of approximately 70
neurons beneath surface patches of 25 × 30 microns for the
monkey).
Fig. 1. Locations of the pyramidal neurons in the cerebral cortex. The
lateral diagram of the cortex within the skull (upper left) shows a highly
convoluted sheet, which is converted into a flat surface (upper right), whose
area is approximately three times that of the lateral view. This cortical fabric
has an average thickness of approximately 3 mm and is constructed of
neurons organized in columns, which are clusters of minicolumns. Within
each minicolumn are shown the major pyramidal neurons with their vertically
aligned apical dendrites. The star-shaped cells in the middle of the minicolumns
are stellate neurons. For clarity, inhibitory neurons, which make up 15%–20%
of the total number of minicolumn neurons, are omitted. Axons, which exit at
the bottom of the somas, are also omitted.
The average length of the layer 5 apical dendrite varies
with cortical area, with parietal and motor areas showing
20%–50% longer lengths than occipital V1 area, based on
cortical thickness data of Rockel, Hiorns, and Powell (1980).
Also, the average length of the layer 5 apical dendrite varies
across the mammalian species of mouse, rat, cat, monkey
and humans, with the mouse apical dendrite of area V1
measuring approximately 1/3 the length of the human apical
dendrite (LaBerge, 2005, Fig. 1). It seems difficult to avoid
the tantalizing question of why the apical dendrite varies
considerably in length across mammalian species and across
cortical areas within a species. Discovering the function that
apical dendrite activity serves in cortical processing could help
in answering that question.
2.2. The axis and shell circuits within the minicolumn
To explore apical dendrite activity in cortical circuitry, we
simplify the complex arrangements and interconnections of
neurons within a minicolumn by dividing the minicolumn
into two compartments, the axis and the shell. Fig. 2 shows
diagrammatically how the layer 5 pyramids, with their very
long apical dendrites, form the axis part of the minicolumn
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 3/17
1006 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
Fig. 2. Most of the neurons of a minicolumn are organized around the long
apical dendrites of the layer 5 pyramidal neurons (the axis), with apical
dendrites of layer 2/3 pyramidal neurons (the shell) arranged in a circular
pattern around the top sector of the apical dendrites of the layer 5 pyramidal
neurons.
while the layer 2 and layer 3 pyramids form the shell part
of the minicolumn. Fig. 3 shows the general structure of the
circuit within each of these two parts of the minicolumn.The axis circuit of the cortical minicolumn extends to the
subcortical thalamus in a reciprocally connected manner, such
that a recurrent or looping circuit exists between the layer
5 pyramid and a thalamic neuron. The axis circuit can be
characterized as an “input-stayput” circuit, because its apparent
function is not the processing of inputs into outputs, but instead
the holding of neural activity over time. The shell circuit is
shown as a one-directional input–output circuit that connects
one minicolumn with another, and the processing component
consists of connections between layer 2/3 pyramids within the
minicolumn. One could also characterize the axis circuit as
a “vertical” circuit, because it extends from the cortex to thesubcortical thalamus, and the shell circuit as a “horizontal”
circuit, because it extends within the cortex, linking one cortical
minicolumn to another.
2.3. The minicolumn circuitry of primary sensory cortical
areas
Traditionally, the study of connections of neurons in the six
layers of the neocortex has been aimed at the discovery of their
receptive field properties (e.g., Bolz, Gilbert, & Wiesel, 1989).
Here we study the interlaminar connections within the cortical
minicolumn to discover how they could support sustained
activity in recurrent circuits that are believed to underlie
Fig. 3. The two main kinds of minicolumn circuits defining the cortical
minicolumn. The recurrent circuit (left) contains the layer 5 pyramidal neuron
and connects with thalamic neurons; and the input–output circuit that contains
layer 2/3 pyramidal neurons (right) restricts its connections to the cortex. The
inhibitory neurons of thecortex andof thethalamicreticular nucleus (and dorsal
thalamus) are omitted for clarity.
consciousness. It is assumed in this paper that sustained
activity in a primary sensory area provides the ongoing
activity of background consciousness for that particular sense.
Elevated consciousness for selected aspects of background
consciousness is assumed to arise when sustained activity
of a primary sensory area is sent to higher sensory areas,
where a selected part of the sensory scene is amplified
by attentional activity controlled from the frontal lobes.
The elevated attentional activity of a part of the sensory
scene in higher sensory areas is regarded here as foreground
consciousness.
Fig. 4 shows a diagram of the major connections between
excitatory neurons in the primary visual area (V1), which
applies generally to primary auditory (A1) and primary
somatosensory (S1) areas as well. The circuit diagram is
adapted in part from diagrams in two publications by Jones(2002, 2007), which summarize what is known to date about the
reciprocal connections between a sensory area and the thalamic
neurons that serve that area.
Layer 5 and 6 pyramids participate in the two kinds of
corticothalamic loops. Apparently layer 2/3 pyramids do not
participate in major circuit loops; nevertheless they do have
apical dendrites of appreciable length. It has been suggested
(LaBerge, 2001, 2005) that these apical dendrites sustain
activity supplied by tonic input from the layer 5 circuit loop,
and that this sustained activity, operating at subthreshold firing
levels, modulates the input–output processing of basal dendrites
inputs within the soma of layer 2/3 neurons.
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 4/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1007
Fig. 4. Diagram of the major circuits in a minicolumn of a primary sensory area
(e.g., area V1), involving two kinds of thalamic relay neurons, Tmatrix and
Tcore. Inhibitory neurons are omitted for clarity. Adapted in part from Jones
(2002, 2007).
When the inputs to the layer 5 and layer 6 loops of Fig. 4
are compared, it is clear that the layer 6 loop is capable of
supplying large amounts of activity to the apical dendrite of
the layer 5 loop (via thalamic Tcore axons terminating on
the many layer 4 stellate neurons), while the layer 5 loop
delivers no activity to the apical dendrite of the layer 6 loop
(and apparently only minor activity to the basal dendrites
of the layer 6 pyramid). Thus, while the axons from the
two kinds of thalamic relay neurons drive activity in their
respective thalamocortical circuits, the total activity in the layer
5 circuit includes additional, and relatively strong, inputs from
the activity of the layer 6 pyramidal circuit. This preferential
convergence of activity on the apical dendrite of the layer 5
pyramid is consistent with the anatomical centrality of the layer
5 apical dendrite in the structure of the minicolumn.
2.4. The minicolumn circuitry of higher sensory cortical areas
The neurons in the primary sensory minicolumn send
activity to minicolumns of higher sensory areas along two
pathways: a direct axon pathway from the layer 2/3 pyramid
of the shell circuit, and an indirect axon pathway from the
layer 5 pyramid of the axis circuit that synapses with a relay
thalamic neuron before entering the higher sensory cortical
minicolumn. Fig. 5 shows the terminations of these pathways
Fig. 5. Diagram of the major circuits in a minicolumn of a higher sensory
area (e.g., area V4). Some of the inputs to the thalamic matrix neuron from
the frontal area connect with thalamocortical circuits in the parietal area before
terminating on the thalamic neuron serving this higher sensory area. Adapted
in part from Jones (2002, 2007).
in the higher sensory minicolumn, along with the pattern of
connecting circuitry of neurons.
Perhaps the most salient difference between the circuitry of
the higher sensory minicolumn and the circuitry of the primary
sensory minicolumn is the change in number of stellate neurons
in layer 4. The reduction of stellate neurons in the higher
sensory minicolumn is reflected by the shrinkage in thickness
of layer 4, which is produced also by a corresponding increase
in the thicknesses of layers 3, 5, and 6, which contain pyramidal
neurons that make up the vast majority of neurons of the cortex.
Another noticeable difference between the circuitry of primary
and higher sensory minicolumns is the increase in synaptic
connections of the thalamocortical axon (from the matrix relay
neuron) onto the distal region of the layer 5 and layer 2/3 apical
dendrites. Taken together, going from the primary to the higher
sensory minicolumn, these two changes in circuitry suggest that
the influence of the thalamic core (T core) neuron on activity of
the layer 5 apical dendrite is reduced, while the influence of the
thalamic matrix (T matrix) neuron is increased.
This shift in the influence of the two kinds of thalamic
neurons on activity of the layer 5 apical dendrite is parallel
to the shift in influence of bottom-up and top-down sources
of activity, globally considered. The many stellate neurons
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 5/17
1008 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
in layer 4 of the primary sensory minicolumn appear to
serve the strong registration in cortex of activity arising from
sensory receptors (retinal, auditory, somatosensory), but once
this registration takes place in the primary sensory area of the
cortex, the need for the special function of the stellate neuron
is apparently reduced. To summarize this section, there are two
major kinds of inputs to the higher sensory minicolumn: onefrom the stimulus registration activity of the primary sensory
minicolumn, and the other from frontal areas of the cortex.
2.5. Attentional activity in higher sensory minicolumns
How does the circuitry of the higher sensory minicolumn
operate on the information arising from the primary sensory
area and from the frontal areas? One answer is that the higher
sensory minicolumn serves as a site for attention, defined as
the elevation of circuit activity corresponding to a selected
part (typically a small part) of the total registration of the
sensory scene in V1, A1, or S1. The operations of attention
are presumed to involve both an enhancement of activity inthose minicolumns that correspond to, or code, the selected
locations and/or appearances of the scene, and a concurrent
suppression of minicolumns that code neighboring parts of the
scene (locations and appearances of distracters). For example,
under attentional conditions, an orientation-sensitive V4 neuron
shows amplification in responding, while the width and mean
of its tuning curve remains unchanged (McAdams & Maunsell,
1999). The control of selective amplification/suppression
during attentional activity is presumed to be produced by axons
of the frontal cortex that make contact with the thalamic relay
neurons that serve those minicolumns (see Fig. 5). Hence,
the thalamic neurons receive activation from two sources:frontal areas of attentional control and the primary sensory
minicolumns. The intensity of activity from the frontal area can
be varied, but the intensity of activity from the primary sensory
minicolumn would seem to remain relatively constant, under
typical conditions. When attentional manipulations produce
changes in V1 (McAdams & Reid, 2005; Silver, Ress, &
Heeger, 2007) and/or lateral geniculate nucleus (LGN) activity
(O’Connor, Fukui, Pinsk, & Kastner, 2002), the change would
seem to be minor relative to the change in activity that
attentional manipulations can produce in higher sensory areas,
given the circuit properties of these areas shown in Figs. 4
and 5. Thus, attention is regarded as the intensification of
a particular part of the sensory scene registered in primarysensory areas by means of the amplification of axis circuit
activities in minicolumns of higher sensory areas that code that
particular part of the sensory scene.
The higher sensory minicolumn is presumed to perform a
function in addition to the expression of attentional activity
in an axis circuit. The shell circuit of these minicolumns
receives inputs from shell circuits of the primary sensory
minicolumn, which are processed into an output that we
assume constitutes the identification of stimuli coded by that
minicolumn. Identification of input information is presumed to
be a brief event, relative to the typically prolonged sustaining
of attention to that event. But in a cluttered scene (visual,
auditory, or somatosensory), attentional enhancement of the
shell circuit activity may be required to produce accurate
identifications. It has been suggested (LaBerge, 2001, 2005)
that tonic enhancement of soma activity in layer 2/3 pyramids
can be produced by inputs from the layer 5 axis circuit, as
shown in Fig. 5. With activity already at an elevated level in
the soma of the layer 2/3 pyramid, the input to a basal dendritefrom a layer 2/3 pyramid in the primary sensory minicolumn
has a higher likelihood of being successfully processed into
an output. This issue will be addressed in more detail in
Section 4.4 of this paper.
3. Special functions of apical dendrite activity
3.1. The issue of stability in corticothalamic circuit activity
Under the assumption that activity in corticothalamic
circuits underlies the extended durations of consciousness,
it follows that the circuit activity must maintain stability
over these durations of time. The cortical environment, aswell as the circuit itself, contains sources of excitatory and
inhibitory noise. Wide fluctuations in voltage levels at the
soma of the pyramidal neuron can interrupt ongoing circuit
activity when intense voltage peaks distort circuit operations,
and when momentary drops in voltage shut down the circuit
operations. To maintain steady levels of soma activity some
means must be found to oppose the naturally noisy nature of
both the cortical background activity and the circuit operations
themselves (Tegner, Compte, & Wang, 2002).
Evidently, the brain has solved the stability problem
for apical dendrite activity, because synchronous oscillations
over extended durations have been recorded from implantedelectrodes and electrodes on the scalp while animals sustain
attention (e.g., Bouyer, Montaron, & Rougeul, 1981). Here we
describe one way that oscillatory activity in the apical dendrite
can become uniform so that it can promote stability in the
corticothalamic circuit of which it is a part.
Because stability in a recurrent circuit is related to the extent
or width of momentary fluctuation of its activity level, one
indicator of stability is the variability of successive electrical
events at a given location within the circuit. Therefore, to
increase the stability of circuit activity the variability of activity
level must decrease. The part of the corticothalamic circuit that
appears to be most variable is the pyramidal neuron, particularly
at synaptic sites along the apical dendrite where the almostsimultaneous discharge of many incoming axons induces
strong electric currents that result in large EPSPs (excitatory
postsynaptic potentials). By the time a EPSP propagates to the
soma, the variability of the EPSP must be substantially reduced,
so that a sequence of EPSPs produce a steady level of soma
activity that can generate a synchronous series of axon pulses.
In Fig. 6 are shown four curves that describe the decay of
four levels of initial EPSP voltages as a function of distance
traveled along the apical dendrite from the region of its
initiation to the soma. At any given location on the apical
dendrite the variability of voltage is indicated by the spread of
a Gaussian-shaped distribution. Almost all of the area under
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 6/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1009
Fig. 6. A family of theoretical exponential functions that describe the decay of
voltage with distance along the apical dendrite from the initiation of the EPSP.
Four levels of voltage, V , are shown at the initiating synapse, and the distance
to the soma is represented by d . The decay constant for this example is k , which
represents the average of the many active and passive membrane conductances.All voltage values are increments in voltage from baseline level. The left-to-
right sequence of Gaussian-like distributions describes the reduction of voltage
variability as the EPSP propagates toward the soma.
a distribution of voltages is assumed to lie below the firing
threshold of an action potential, for reasons indicated in the
next section of this paper. The voltage level at the soma and
its variability are shown to decrease with the distance the
EPSP travels to reach the soma. The exponential equation that
describes the rate of EPSP decay with distance was developed
by Rall (1989), and represents the passive decay of voltage
in a cable model of the dendrite. The average membrane
(outward) conductance is represented by the constant, k , andmore recent versions of the cable model assume that the
membrane conductances of dendrites in the awake animal are
dominated by active inhibitory (outward) currents, with an
inhibitory chem/excitatory ratio that may be as high as 5-to-1
(Destexhe, Rudolph, & Pare, 2003).
3.2. The give-and-take operations of apical dendrite activity
The dendrites of a typical pyramidal neuron contain as many
as 10,000 synapses (DeFelipe & Farina, 1992), and a typical
apical dendrite (of the rat) contains at least 3000 (excitatory)
synaptic spines on its major vertical shaft, with a similar
quantity of spines on oblique dendrites that are branches of the dendrite shaft (Larkman, 1991). Since the length of the
human apical dendrite is approximately 2 times that of the rat,
the estimated quantity of synaptic spines on the major shaft
of the human apical dendrite could be as high as 6000. The
locations of these ubiquitous dendritic spines are indicated in
an abbreviated manner along the apical dendrites shown in
Figs. 4 and 5. Aside from the synapses made by thalamic axons,
the vast majority of synapses are activated from other cortical
neurons (Braitenberg & Schuz, 1998), which provide most of
the background noise for the apical dendrite. Synaptic activity
along the apical dendrite produces a rich variety of intrinsic
electric and chemical effects (Llinas, 1988; Reyes, 2001; Yuste
& Tank, 1996), and the activity at the thousands of spine
synapses on the dendrite maintains membrane depolarization
at subthreshold firing levels. It is estimated that the thousands
of synapses on dendrites are responsible for approximately 80%
of the mainly inhibitory conductance of the dendritic membrane
(Azouz & Gray, 1999; Contreras, Timofeev & Steriade, 1996;
Nowak, Sanchez-Vives, & McCormick, 1997). The chaoticnature of this background input to the apical dendrite produces
fluctuations that sometimes generate action potentials that may
arise from transient calcium currents (Helmchen, Svoboda,
Denk, & Tank, 1999). Action potentials that originate in
dendrites do not always propagate to the soma (Jarsky, Roxin,
Kath, & Spruston, 2005; Larkum & Zhu, 2002), and may
participate in local changes of synaptic long-term potentiation
(Colbert, 2001; Goldberg, Holthoff, & Yuste, 2002). Recent
findings from intracellular recordings of action potentials in
vivo during Up states and during visual stimulation indicate that
action potentials are typically initiated in the axon (Shu, Duque,
Yu, Haider, & McCormick, 2007). The part of the axon that
appears to be the preferred site of action potential initiation isthe initial segment of the axon that typically lies 40–55 microns
from the soma.
Of particular relevance to the stabilizing of thalamocortical
circuits is the effect of the active membrane conductances
maintained by tonic activity in the thousands of synaptic spines
along the apical dendrite. This spine activity produces the
attenuation of the voltage of the EPSP as it propagates along
the apical dendrite shafts illustrated diagrammatically in Figs. 4
and 5. The decaying curves shown in Fig. 6 indicate the
attenuating effect of spinal activity between the site of EPSP
initiation and the soma.
Therefore, it could be said that the propagating EPSP of the apical dendrite is treated in a “give-and-take” manner.
The EPSP is “given” to the apical dendrite by axon activity
of groups of layer 4 stellate neurons and by activity of
thalamic axons. Immediately there ensues a “taking” of the
EPSP, which is the attenuation of the EPSP produced by the
outward conductances of the membrane produced by low-
level synaptic activity in the thousands of spines that dot the
dendritic shaft. As the EPSP approaches the soma, it decreases
in strength, but, importantly, the variability of the strength
also decreases. The average rate of decrease is described by
the exponential decay equation based on average membrane
conductance. The theoretical convergence of decreasing mean
voltage with distance can also be described by equations other
than the exponential equation (see, e.g., Koch (1998)).
The variance of the propagating EPSP at the soma is
assumed to be relatively high for the first cycle of electrical
activity through the recurrent thalamocortical circuit. But in
subsequent cycles, the EPSP at the soma moves more closely
to the mean voltage it will ultimately achieve, so that the
variability of the EPSP at the asymptote can be regarded as the
characteristic level of stability for this recurrent circuit, given
the particular conditions that are present at the time.
Conditions under which the thalamocortical circuit operates
can vary considerably. One important condition is the intensity
level of the afferent input to the thalamic relay neuron, which
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 7/17
1010 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
can be varied by the brightness contrast of an oriented line in
vision. Variability of input to the thalamic relay neuron can also
be produced by the focusing of attention to the oriented line in
the axons arising from frontal areas of attentional control. In
both of these cases the number of active thalamic relay neurons
can vary the level of EPSPs in a layer 5 apical dendrite (directly,
or via stellate neurons) by varying the number of active axonsthat contact that apical dendrite, while maintaining a constant
frequency of pulses in the input axons. This inference is based
on the finding that, in the cortex, a branching axon seldom
makes more than 10 contacts with any single neuron, and
frequently with only one or two neurons (Mountcastle, 1998).
When the magnitude of the EPSP that is “given” to the apical
dendrite varies, its variability will also change, and this change
will be maintained as it propagates to the soma (see Fig. 6).
When the variability of the EPSP at the soma changes, so will
the stability of the entire thalamocortical circuit. Therefore, to
maintain a high level of stability in the thalamocortical circuit,
some means must be found to make appropriate adjustments in
the trimming operation of the variable EPSP before it reachesthe soma.
3.3. Balancing the “give-and-take” operations in the apical
dendrite
One possible way to compensate for a large initial EPSP
“given” to the apical dendrite is to increase the synaptic
strengths of the thousands of spine synapses that “trim” the
EPSP as it propagates to the soma. Adjustments to the axon
inputs to these spines could be produced in two general ways.
The first way involves specific circuits that respond directly
to the intensity of the initial EPSP. For example, the layer 6pyramid, whose momentary activity presumably keeps pace
with the activity level of the layer 5 pyramid, sends collaterals
of its axons into layer 4, where they branch profusely. If enough
of these axon arborizations contact the spines of the layer 5
apical dendrite, a close link would be formed between the
momentary amount of the “give” operation to the consequent
amount of the “take” operation. As it turns out, however, the
supporting evidence for a high number of layer 6 axon synapses
on layer 5 apical dendrites (McGuire, Hornung, Gilbert, &
Wiesel, 1984) is not strong.
A second way that the “take” operation may be controlled
by the “give” operation is by means of a large and diffuse set
of circuits that operate in a more indirect way than the kind of specific circuit just described. When a large EPSP is delivered
to the apical dendrite by the thalamic relay neurons contacting
pyramidal and stellate neurons of the cortex, it is likely that
many local circuits of the local minicolumn increase the level
of their activity. If these local circuits supply most of the axon
input to the thousands of spines on an apical dendrite, then an
increase in the activity level in these many circuits will result
in an upward adjustment in the spine activity, which would
increase the local outward membrane conductance, thereby
increasing the trimming of the “take” operation. However,
because of the diffuse nature of the connectivity of the many
local minicolumn circuits, it is likely that their activity levels
may not change quickly. Hence, adjustments to the “take”
operations on the propagating EPSP in the dendritic spines may
lag well behind changes in the “give” operation of the initiating
EPSP. Therefore, it would seem that this diffuse circuit model
exhibits much less direct coupling between the “give” and the
“take” operations than would be shown by a specific circuit
model.The adjustment of the “give-and-take” treatment of variable
EPSPs of the apical dendrite may take place by means
other than those described in the foregoing paragraphs. When
attention to an object is intensified over longer time periods,
activity in a thalamocortical circuit is repeated with elevated
intensities of EPSPs occurring in the apical dendrite. As a
result, new spines may appear, which induce the apical dendrite
shaft to elongate to accommodate the additional spines. A
longer length of the spine-covered apical dendrite enables the
EPSP to undergo additional reduction in variability, which
compensates for the additional variability in the initial EPSP
that accompanies higher levels of intensity.
In view of these considerations, it might seem puzzling thatin area V1 the major input synapses of the stellate neurons
on the apical dendrite of layer 5 pyramids are located at the
middle part of the layer 5 pyramidal apical dendrite, and not at
the distal part, whence an initial EPSP could undergo a larger
amount of trimming before it reached the soma. One hypothesis
suggests that the activity in the lateral geniculate neurons
already possesses a moderate level of stability, because the
visual scene that stimulates the retina is typically quite stable.
In comparison, major synapses on apical dendrites of layer 5
pyramids in higher sensory minicolumns appear to be located
at more distal parts, while midsection stellate neurons supply
a minor portion of the synaptic input. The distal synapseson apical dendrites in the higher sensory areas are driven by
thalamic pulvinar neurons, which do not receive the major
part of their inputs from the retina. Therefore these EPSPs at
the distal apical dendrites exhibit a higher level of variability
and require more trimming before they reach the soma. Fig. 5
shows that the thalamic pulvinar neurons are activated from
two major sources: the frontal cortical areas, and area V1.
The variability from area V1 is presumed to be low, having
started at a moderate level (from the relatively stable retinal
source) and having received further refinement by the trimming
operation of the V1 layer 5 apical dendrite. But the source of
variability from frontal cortical areas is presumed to be quite
large, owing to presumed high fluctuations in frontal cortical
circuits. Also, the weight of the frontal input component to the
pulvinar neuron is presumed to be influenced by the intensity
level of attentional control. Therefore, the higher the intensity
of attentional control, the higher the variability of the activity
delivered to the pulvinar neuron. Hence the EPSP induced at
the distal part of the apical dendrite in the higher sensory area
is increased, and a commensurate increase in length of apical
dendrite is needed to reduce this increased variability before
the EPSPs reach the soma.
Therefore, these theoretical considerations predict that high
levels of attention directed to a particular object should, over
weeks and years, elongate the apical dendrites of layer 5
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 8/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1011
pyramidal neurons in the minicolumns (of higher sensory areas)
that code that object. One potential indicator of apical dendrite
length is cortical thickness, which has been measured with
variations of the fMRI technique. We turn now to an analysis of
the effectiveness of this measure for estimating apical dendrite
length.
3.4. Lengths of apical dendrites and cortical thickness
Many apical dendrites of layer 5 and layers 2/3 extend
to the top of layer 1 and spread laterally in tufts for a short
distance, apparently leaving little space in which to expand
upward. Therefore, an elongation of the shaft of the apical
dendrite should shift the soma in the other direction toward the
bottom of layer 6. However, in the case of stretching apical
dendrites of layer 5 pyramids, it would seem that there is
little space within layer 6 to accommodate the relative massive
somas of layer 5 pyramids, owing to the already high packing
density of neurons in a minicolumn (Hendry, Schwark, Jones,
& Yan, 1987; Rockel, Hiorns, & Powell, 1974). In view of theselimiting factors, it would seem that an elongation of the apical
dendrite shaft would result in an expansion of the individual
cortical layers through which the shaft passes. As a result the
total thickness of the cortex would increase in the local cortical
areas containing these elongated apical dendrites.
A fMRI study of meditation practice by Lazar et al. (2005)
found selected brain regions in which cortical thickness was
larger in participants who had extensive meditation experience
than in controls, and Makris et al. (2006) found smaller
cortical thickness in cortical areas underlying attention in
participants diagnosed with ADHD compared with controls. It
is tempting to infer that the observed change in thickness wasstrongly influenced by the changes in apical dendrite lengths
owing to meditation practice or attentional pathology. However,
even though increases in apical dendrite length could produce
increases in cortical thickness, would these measured increases
in cortical thickness necessarily imply an increase in apical
dendrite lengths? Elevated activity in minicolumns over time is
presumed to generate new synapses, and an increase in number
of synapses (and spines) along with their increased lengths
of local axon branches could also increase cortical thickness
owing simply to the overall increase in volume of neuronal
tissue. However, if an appreciable proportion of new synapses
appear on an apical dendrite, then it would seem that the
apical dendrite would have to elongate to accommodate theseadditional synapses. Also, data from the frontal cortex of rats
that undergo confinement each day for 21 days show that the
loss of spines on apical dendrites is accompanied by a 20%
reduction in the length of the apical dendrite (Radley et al.,
2006). Nevertheless, the more cautious interpretation is that
observed changes in cortical thickness could conceivably be
produced by an increase in synapses without an increase in
apical dendrite length. On this view, cortical thickness measures
by themselves are not an unambiguous indicator of underlying
apical dendrite lengths. What is clearly needed is a technique
that enables apical dendrite lengths to be measured directly in
live animals and humans.
3.5. Lengths of apical dendrites and the ERP
The ERP (event-related potential) measured at the scalp is
a transient EEG that is produced by a relatively abrupt onset
of stimulus whose amplitude is increased when its location
and attribute have been cued in advance of its appearance.
The cue is presumed to induce preparatory attention to thelocation and attributes of the target stimulus, which presumably
produces elevated activity in the columns that code the location
and attributes of the stimulus, relative to the activity levels
in neighboring columns that code for similar locations and
attributes. In the typical ERP task, both target and distracter
stimulus appear in a random sequence, with the target appearing
at a low frequency. The task for the observer is to count the
number of targets over a marked time period. The finding
of interest is that the ERP amplitude is higher at the cued
location of the stimulus attributes than at an uncued location
(e.g., DiRusso, Martinez, and Hillyard (2003)).
According to the present theory, when the cue induces
preparatory attention to a particular stimulus content, thecorresponding minicolumn axis circuits elevate their activity
levels, and they maintain or increase these levels throughout the
time period in which the target stimulus is expected to occur.
When a stimulus of the cued attribute and location appears,
the apical dendrites in the minicolumn axis circuits show a
transient increase in amplitude of electric fields. For vision, the
pathways that are likely to deliver this increase are the relatively
fast conducting Y -cell driven pathways in the cortex and in
the superior colliculus. Both pathways synapse with pulvinar
thalamic neurons before terminating within a minicolumn of
a higher sensory area (see Fig. 5). The Y -cells in the retina
respond only transiently to the onset of a visual stimulus, sothat the activity in their pathways presumably has a transient
effect on the apical dendrites where their pathways eventually
terminate.
In a study by Pantev et al. (1991) the repeated displays of the
target stimulus produced a transient increase in amplitude of the
scalp-recorded EEG 20–130 ms after stimulus onset (the MEG
recording was virtually identical in form to the EEG recording).
The short interval of amplitude increase was shown to consist
of 4 or more cycles of 40 Hz oscillations, which suggests that
the sustained preparatory attention was being produced by axis
oscillations at 40 Hz, and that when a stimulus appeared, the
oscillations were amplified for a fraction of a second.
Amplifications of the ERP are produced not only bymomentary attention but apparently also by long-term effects
of attention during training. Simple and complex tones produce
higher amplitudes of early ERPs in musicians who have had
many years of musical training compared with non-musicians,
even under passive attention conditions (Kuriki, Kanda, &
Hirata, 2006; Shahin, Bosnyak, Trainor, & Roberts, 2003;
Shahin, Roberts, Pantev, Aziz, & Picton, 2007). Also, trained
meditators show higher amplitude ERPs (and frequencies)
compared to controls (Lutz, Greichar, Rawlings, Ricard, &
Davidson, 2004).
The source of the electric fields that produce an ERP is
assumed to be the electric dipole, which is defined by the
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 9/17
1012 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
location of the two opposite electric charges along the active
portion of the apical dendrite. The intensity of the dipole,
termed the dipole moment, is influenced by two variables:
the charge at the two poles and the distance between them.
Therefore, an increase in ERP amplitude can be produced
by stronger charges of EPSPs or by a greater length of the
apical dendrite shaft traveled by the EPSP, or by both of these variables. The present theory suggests that the larger
EPSPs produced by strong sustained attention over years of
training induces an elongation of apical dendrite shafts, so that
distance as well as charge of the dipole would be affected
in expert musicians. However, in the short duration of a task
condition that instructs or otherwise induces the observer to
increase attention there is not enough time to elongate the apical
dendrites, so that the increase in dipole strength is apparently
solely due to increase in the charge produced by the larger
EPSPs of heightened attention.
The foregoing analysis suggests a way that training may
increase ERP amplitude by changes in apical dendrite length
over time. Does it follow that an increase in ERP amplitude overtime indicates that the apical dendrite length has increased?
Apparently not, because an observed increase in ERP amplitude
at the scalp could be produced simply by more minicolumns
being recruited, without any change in apical dendrite length
and without any change in EPSP amplitude induced on an
individual apical dendrite.
4. Implications of minicolumn circuitry for cognition
The two different kinds of minicolumn circuits make
possible two different kinds of cognitive activity, according to
the theory of cortical function described in the present paper.The shell circuits, which connect layer 2/3 pyramidal neurons
across minicolumns, operate as input–output mechanisms that
typically complete their processing in a short fraction of a
second. Shell circuits in early sensory areas that receive V1
inputs identify object appearances and locations, and later
sensory areas process these identifications into categories. In
the frontal areas categories are organized into propositions that
may be based mainly on activity in the shell circuitry spread
across many minicolumns. Fig. 7 shows a diagram of the global
shell pathway, along which early identifications of stimulus
objects and locations are processed into the categories and
propositions upon which the operations of intellectual thinking
operate.The second general kind of minicolumn circuit is the axis
circuit, which connects layer 5 pyramidal neurons between
minicolumns through intervening synapses with thalamic
neurons. Fig. 7 shows a diagram of the global axis pathway,
which begins with the initial cortical registration of sensory
inputs in primary sensory minicolumns, and extends into the
frontal cortex where, among other activities, axis circuits are
presumed to perform special supporting functions for the shell-
based processes of thinking. The axis circuits act as holding
circuits that sustain activity for extended periods of time.
When the intensity of the apical dendrite component of the
thalamocortical holding circuit is sufficiently high, it produces
Fig. 7. Global ascending pathways that connect posterior cortical columnswith anterior cortical columns in two ways. One way (the lower-tier pathway)
connects columns by their axis circuits, and the other way (the upper-tier
pathway) connects columns by their shell circuits.
the cognitive event that may be called “having an impression”
of something. The earliest of these subjective impressions takes
place in the primary sensory areas of the cortex where the
entire array of sensory input is first registered in the cortex
(see Fig. 4). We regard a scene that contains many impressions
as background consciousness, and the array of impressions is
typically sustained for very long periods of time while attention
selects impressions of specific objects in the scene. Examples
are visual scenes seen through a window, the ambient sounds of
traffic heard outside a city room, and the continuous sensations
of the somatosensory landscape of the body.
Foreground consciousness is produced by axis circuits that
produce impressions in higher sensory areas (see Fig. 5);
these axis circuits are presumed to be strongly influenced
by top-down frontal activations which elevate activity in
minicolumns that code a selected part of the total registered
input. This attentional activity can take on variable degrees
of spatial focus and intensity, depending on the manner
of control activity received from the frontal areas. Thus
the simultaneous impressions from both foreground and
background consciousness together constitute the momentary
content of consciousness.Fig. 8 shows diagrams of top-down global pathways for
both the shell and axis circuits. Two of the important cognitive
functions served by these pathways are attention and imaging.
In the case of attention, the top-down axis pathway provides one
of the two input sources that activate thalamic neurons serving
the minicolumns of higher sensory areas in the posterior cortex
(see Fig. 5). Thus, the sustaining of attention over extended
time periods is controlled by sustained activity in axis circuits
of frontal areas.
The top-down shell pathway may function to shift attention
rapidly among an array of stimulus items during search.
Axons from frontal attentional control areas are presumed
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 10/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1013
Fig. 8. Global descending pathwaysthat connect anterior cortical columns with
posterior cortical columns in two ways. One way (the lower-tier pathway)connects columns by their axis circuits, and the other way (the upper-tier
pathway) connects columns by their shell circuits.
to contact layer 2/3 pyramids (through stellate neurons),
whose axon collaterals produce strong lateral inhibition in
neighboring minicolumns that code for distracter locations and
appearances. Since the operations of inhibitory neurons are
typically very rapid compared to the operations of excitatory
neurons, the top-down shell pathway can support a series of
very rapid attentional selection of minicolumns that code for
locations. Identifications of each scanned item can be quickly
accomplished in the shell circuitry of attribute minicolumns
(selected by axons from location minicolumns), and becausethe “dwell time” of attention on any particular item is very brief,
the contribution of axis circuitry has a minimal contribution
to the operation of selecting minicolumns during the scanning
operations. Therefore the major circuit operations supporting
rapid search appear to be the lateral inhibitory connections
within shell circuits.
4.1. Attention and the binding of object attributes
It is generally agreed that the binding of component
attributes produces the unitary impression of an object. The
theory of minicolumn circuitries, described in the present paper,
suggests that the binding of attributes of an object occurs
by the set of impressions produced by the simultaneously
elevated apical dendrite activity in minicolumns that code
those attributes. When this set of minicolumns is linked
to minicolumns coding a common location, attention to
that location raises the activation of the connected attribute
minicolumns in a correlated manner. Therefore, the binding
could take place along the axis pathways shown in Fig. 8
that intersect a common location. The amplified activity in the
axis circuits containing the apical dendrites is based in part
on a common frequency of electric oscillations, and but also
on the amplitude of the EPSPs in the apical dendrites. Phase
locking does not seem crucial to binding, on this view, owing
to the variable lengths of axons that connect axis circuits of
participating minicolumns. When visual objects are displayed
very briefly, the axis circuits may have sufficient time to
produce accurate impressions of attributes but not sufficient
time to produce impressions of precise locations. As a result,
some attributes may be bound to incorrect locations and lose
their correct conjunctive relationships, so that the observerreports seeing objects whose conjunction of attributes is not
correct but “illusory”.
4.2. Holding circuits in frontal areas
Many of the potential cognitive aspects of axis activity in
frontal minicolumns have not yet been explored. Some of the
possible functions of the thalamocortical holding circuit are
suggested by the way the basal ganglia influences activity in
the frontal cortex. Axons from neurons of the basal ganglia do
not directly synapse on neurons of frontal cortex but instead
on thalamic neurons that are connected to frontal cortical
neurons (Alexander & Crutcher, 1990). One implication of this neuroanatomical arrangement is that the thalamocortical
holding circuit is involved when motivational activities in the
basal ganglia act upon the cognitive and motor actions that
are processed in frontal circuits. Activity held in these circuits
can then operate on shell circuits in a modulatory manner,
thereby influencing the way that input–output processing takes
place in these circuits. For example, holding circuit activity
may stabilize the input–output processes involved in choosing
among several tasks to perform in the Wisconsin Card Sorting
Task, for example, sorting by color, or shape, or number.
A thalamocortical holding circuit, controlled from the basal
ganglia, may also stabilize the processes involved in the earlytrials of performing a specific task, such as sorting by color.
An unstable holding circuit of a newly cued task may induce
the participant to revert to the previous task, which is a finding
frequently observed in schizophrenic participants.
Another subcortical circuit that may influence cortical
circuits through the thalamic holding circuit is the limbic
circuit. One cortical target of the limbic circuit is the anterior
cingulate area, which has been shown to be particularly active
during cognitive events that involve conflict. When conflict is
evoked in a task, it typically persists over a period of time; and
while the conflict activity is sustained, it is often accompanied
by “feelings” or impressions of the conflict. These observations
suggest that the holding circuit participates in maintainingstates of conflict. Evidence supporting this conjecture is
given by an fMRI study of attention (Buchsbaum et al.,
2006) in which elevation in activity of the anterior cingulate
and its connected anterior thalamic neurons implicates the
thalamocortical recurrent circuit that connects these two brain
structures.
Frontal holding circuits also would seem to be implicated
during the effort to recall an item from long-term memory (e.g.,
recalling an image of the Eiffel Tower). Although the typical
impression of the effort to recall is not as vivid as the sensory
impression of the item itself, or an image of the item, there
may be occasions in which the effort to recall an item believed
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 11/17
1014 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
to be “on the tip of the tongue” is indirectly experienced
as an impression of tension somewhere in the body. The
sustained activity of a bodily somatosensory indicator implies
that a holding circuit is maintaining this activity. Because the
occasion that gives rise to the effort is the attempt to recall
an item, it would seem that the holding circuit is located in
or close to the recall operation. However, one ordinarily isnot conscious of an impression of effort activity in these axis
circuits because the intensity of activity in these apical dendrites
is overshadowed by the characteristically strong intensities in
axis circuits of body feelings coded in SS2 minicolumns and
in axis circuits of minicolumns that code ongoing visual and
auditory impressions.
4.3. The continuity of successive conscious impressions
Considered as a unit of the impressions that constitute
consciousness, axis circuit activity can occur in many
minicolumns simultaneously, as is the case in primary sensory
areas where inputs from sensory receptors register manydetails of a scene in parallel. However, in higher sensory
areas, attentional processes typically operate to select parts
of the registered scene, by elevating the axis circuit activity
in minicolumns, which will continue at this higher level
of intensity as long as frontal attentional control sites send
activation to the thalamic neurons of these axis circuits. The
elevated activity in these axis circuits does not abruptly return to
baseline when attentional control is shifted to another part of the
sensory scene, but instead the activity decays gradually toward
baseline. Attentional effects observed in the primary visual area
apparently exhibit this overlap of activity at the destination
and origin sites of attentional shifts (Khayat, Spekreijse, &Roelfsema, 2006). This overlap of minicolumn axis activity
during shifts of attention produces continuity in the succession
of impressions that would not exist if each axis circuit shuts
down immediately when attentional activation is shifted away
from it. If attentional control induces relatively high levels
of intensity in a particular minicolumn (or a column cluster
of minicolumns), then the course of the decay process may
be sufficiently long that some elevated activity remains after
attention has shifted to another minicolumn and then returned
to the original minicolumn. This is the prediction derived and
tested for the case of shifting visual attention between spatial
locations (LaBerge, Carlson, Williams, & Bunney, 1997).
The continuity of successive impressions commonlyobserved during shifts of attention across visual, auditory and
tactual scenes contrasts with the discontinuities of successive
identifications of objects (attended or unattended) during
these shifts of attention. Units of processing are distinct and
separated, owing largely to their typically short durations
and their abrupt terminations when an output occurs. When
identifications evoke categories and ideas further along in the
shell global pathway, there may be overlap between ideas when
they are stored in working memory circuits. But, if working
memory circuits are exclusively shell-based, there will be no
impressions generated. But even if the working memory circuits
contained an axis component, the intensity of apical dendrite
activity within that axis circuit would seem to be of low
intensity so that it would be easily obscured by the typically
high intensities in axis circuits of sensory areas in the posterior
cortex.
Therefore, the appreciable duration of the “passing moment”
implies that axis activity is providing the principal neural
substrate, while the vanishingly small duration of the “presentmoment” implies that shell activity is providing the principal
neural substrate. Stated in another way suggested by the present
theory, the passing moment is an impression of which we are
conscious, and the present moment is a concept which we
process as an idea.
4.4. Interactions between axis and shell circuit activities
A cognitively important relationship between having an
impression of an object and having an idea about an object is
embodied in the connective links between the shell circuit and
the axis circuit within a minicolumn. The main links between
the axis circuit and the shell circuits are shown in Figs. 4 and 5.Thalamocortical axons from the thalamic matrix neurons send
collateral axons to the apical dendrites of layer 2/3 pyramidal
neurons, so that activity in the axis circuit produces and
maintains activity in the layer 2/3 apical dendrites. Therefore,
when minicolumns of higher sensory areas exhibit attentional
activity, the elevated activity in the axis circuits of layer 5
pyramids induces EPSPs in the layer 2/3 apical dendrites.
However, the amplitudes at the soma of layer 2/3 pyramids
of these EPSPs are assumed to remain below threshold firing
levels, so that the axons do not exhibit significant rates of
output pulses. For example, while a driver waits for a traffic
light to change from red to green and wishes to move quicklyahead at the onset of the green light, the driver may direct
preparatory attention to the location and color of the green
light during the time interval leading up to the appearance of
the green light. During this preparatory interval, the level of
attention corresponding to the amplitude of layer 2/3 apical
dendrite activity is elevated at the soma, but not to the level
that will produce axon output. When the green light appears,
axonal pulses from V1 minicolumns coding the location and
color of the green light produce EPSPs in the basal dendrites of
these layer 2/3 pyramids, which sum with the existing level of
activity in the soma and produce axon outputs. Fig. 9 shows two
diagrams of a layer 2/3 pyramidal neuron, one showing activity
in the apical dendrite (received from the layer 5 axis circuit),and the other showing no activity in the apical dendrite. When
axons from area V1 produce EPSPs in the basal dendrites of
these layer 2/3 pyramidal neurons, the pyramidal neuron with
the attentionally induced higher tonic activity at the soma will
emit a train of output pulses in the axon sooner than the other
pyramid, which represents a state of no preparatory attention.
In addition to a shortening of the output latency, the tonic soma
activation by the apical dendrite will increase the rate of pulses
in the output, so that, in effect, the signal-to-noise level of the
input pulses received at the basal dendrite is increased.
But if the level of apical dendrite activity delivered to
the soma of the layer 2/3 pyramid momentarily drifts above
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 12/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1015
Fig. 9. Apical dendrite activity modulates input–output processing in layer 2/3
pyramidal neurons. Here, the same train of input pulses contacts the basal
dendrites of both neurons, but the sustained elevated apical dendrite activity
in the neuron on the left increases excitability of the soma, so that fewer input
pulses are needed to produce an output pulse.
firing threshold, then axon output will occur, which typically
produces false positive responses, called “jumping the gun”.
Therefore, to produce axon outputs to the onset of the green
light with short latency and without momentary fluctuationsof apical dendrite activity at the soma that produce false
positive outputs, the mean level of apical dendrite activity
delivered to the soma must be elevated as close to the firing
threshold as the variability of the EPSP will allow. Thus, an
observer cannot effectively reduce the time to process the green
light (in the attention-based higher sensory area) simply by
increasing the amplitude of the EPSP delivered to the layer 2/3
apical dendrites unless the lengths of the apical dendrites are
sufficiently long to reduce the variability of the larger EPSP by
the time it reaches the soma. With training, one might expect the
lengths of the apical dendrites to lengthen, so the modulatory
activity delivered to the soma could be raised to progressively
higher levels while maintaining the steady stability of low EPSP
variability.
There are additional benefits for tasks of sustained attention
that are produced by high but steady activity in the apical
dendrites of layer 5 and layer 2/3 pyramids. These pyramids
send collaterals of their axons to neighboring columns, where
they synapse on inhibitory neurons that contact pyramids
and lower their responsiveness. Also, axis circuits that serve
neighboring columns inhibit each other in the thalamus by
their synapses on inhibitory neurons of the thalamic reticular
nucleus (Jones, 2007). When the activity levels of axis circuits
of a column cluster of minicolumns are elevated, the activity
in neighboring columns that code for potential distractingsignals is concurrently suppressed. When this state of affairs
is extended over durations of time, as in anticipation of
an upcoming green light, or in imaging the arrangement of
furniture in a familiar room, or activating a plan of action,
there is less likelihood that the onsets of distracting stimuli
will perturb the ongoing axis activity that supports preparatory
attention, imaging, and planning. Thus, high amplitudes of axis
activity can protect axis circuit activity from interference by
distracting events, much as increased rates of axon outputs can
increase the signal-to-noise ratio in input–output processing
and protect the output information from interference by ambient
noise.
As a general cognitive principle, shell circuit activity enables
the identification of the axis activity that is taking place in
the same minicolumn (or column cluster of minicolumns). The
attachment of a symbol or label to a stimulus input is based on
the process of identifying a stimulus input from the primary
sensory area and then categorizing it in successively higher
sensory areas. Thus, it could be said that the symbolizingprocess produces the basic materials for the construction
of ideas. But the symbolizing process does not capture the
subjective impression itself, because sustained axis activity
does not translate or transform into the shell activity by
input–output processing circuitry. However, the shell circuits
can respond to certain aspects of the axis activity in the
minicolumn they share. In many cases, shell processing will
occur or not occur according to the presence of axis activity,
so that an input to the shell circuit can act as a query about the
existence or non-existence of axis activity in that minicolumn
(see Fig. 9). A succession of these queries can therefore
provide a means of estimating the duration of axis activity
in a minicolumn or cluster of minicolumns. Also, since theintensity of axis activity affects the latency and strength of the
output signals in shell processing (described in the foregoing
example of anticipating a green traffic light), an input query to
the shell circuit provides a means of detecting the intensity of
axis activity in a minicolumn or cluster of minicolumns. Thus,
it could be said that the shell circuitry can have knowledge
about the duration and intensity of axis activity, but shell
circuitry does not have direct knowledge (e.g., knowledge
by acquaintance) of “what it is like” to have the impression
produced by an axis circuit. The sharp limitation of knowledge
about axis activity in a particular minicolumn appears to be
reflected in the manner in which the layer 2/3 pyramidalneurons are arranged around the outside of the apical dendrites
of the layer 5 pyramidal neurons. The etymological form of
the word “about”, prior to 900 A.D. was the word abutan,
which meant “on the outside of”. Thus, the shell circuits
may be viewed as exterior to the interior events of the axis
circuits and specialize in “aboutness”. The symbolic nature of
shell activity enables its outputs concerning this “aboutness”
to be communicated to others through language and gestures.
In contrast, the impressions of axis activity would seem to
constitute a subjective “inwardness” of the individual’s mental
life, and as such are not capable of being communicated to other
individuals. Furthermore, it could be said that the extended
duration of axis activity confers a concrete “existence” to
impressions compared to the specious existence of the fleeting
moments in which ideas are processed. In summary, the
distinctive structures of the axis and shell circuits within the
minicolumn appear to define two very different ways in which
the cerebral cortex responds to and constructs its knowledge of
the sensory inputs from objects of the external world.
4.5. The role of the self in consciousness
The global circuits connecting axis and shell circuits also
serve as a basis for distinguishing two selves, which, for
convenience, we label as the “thinking” self and the “feeling”
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 13/17
1016 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
self. The thinking self is a product of ideas produced by shell
circuits. It includes items of personal history and verbalized
personal characteristics that arise from the way we interpret
our behaviors in social situations. Gazzanga’s theory of the
“Interpreter” (Turk, Heatherton, Macrae, Kelley, & Gazzaniga,
2003) provides an example of the self as “thought about”. In
contrast, the feeling self is based on the sustained impressionsof the body’s landscape, whose intensities can be modulated
by the emotional activities of the body. The present notions
concerning the relationship of the feeling self to ongoing
emotional events in the internal organs are influenced by the
work of Damasio (1994). By analogy with descriptions of
activities in early cortical areas of vision, audition, and touch,
we conjecture that the ongoing inputs from the internal organs
and external bodily surfaces are registered as impressions in
axis circuits of somatosensory area S1 and selectively attended
in axis circuits of area S2.
The continuing inputs from our body provide impressions
that constitute background consciousness of our bodily feeling.
This ongoing feeling of our body provides the primitive senseof presence, which then can be attributed to objects of the world
that we imagine our body can touch. Thus, when the sustained
bodily impressions are paired in a coordinated manner with
the sustained impressions of objects seen, heard, and especially
touched, these objects also become present to us. Together with
the impression of the presence of the body, the impressions
of the presence of these objects provide our ongoing sense of
presence of ourselves in the world. For example, as we walk
along a path and approach a large tree, we seem to attend more
and more strongly to our body as well as to the tree. As we close
the distance between our body and the tree the impressions of
both body and tree seem to intensify, and this intensity peaksif a part of our body actually touches the tree. These sustained
impressions of the tree and our bodily self are likely to become
very intense if the tree we approach is a giant redwood.
A conductor of an orchestra can experience a similar
impression of the presence of both the self and an external
object, but in this case the looming object is the overall
orchestral sound as the conductor gradually directs a crescendo
from the level of a pianissimo to the level of fortissimo. In a
similar way, the sounds of music or the sounds of voices at a
cocktail party in the hotel room down the hall seem to require
a minimal level of intensity to give the impression of their
presence, but the additional requirement is the relationship to
our current impression of our body. It would seem, therefore,
that it is the attention directed to our body while we attend
to a suitably intense impression of an external object that adds
the impression of presence to the consciousness of that object.
It has been suggested that the term “awareness” should be
reserved for such cases of elevated consciousness that arise
when attention is directed to the body at the same time as
attention is directed to an object (LaBerge, 1997).
In view of these considerations, the present theory
of consciousness does not require the self, considered
either as the felt-body based on axis activity or as an
intellectual construction based on shell activity, as a necessary
component of consciousness. High levels of visual or auditory
consciousness produced by attention can occlude the ongoing
background feelings of the body, so that we can apparently
“forget ourselves” when we are beholding a sunset, or when
we are listening to a favorite piece of music, or when we are
reading a novel. Thus, the body-based or body feeling-based
impression of the self is based on background consciousness of
the current state of the body. As such, this consciousness maybe regarded as another kind of sensory-based consciousness,
with registration of ongoing bodily feelings taking place in the
primary somatosensory area, S1. Foreground consciousness of
a bodily feeling, then, is the elevated intensity of an axis circuit
coding an attentionally selected part of this registered array of
feelings in minicolumns of the higher sensory area S2.
5. Consciousness and the electromagnetic fields of apical
dendrites
The present paper proposes that sustained activity in the axis
circuit of minicolumns is a necessary condition for producing
and maintaining consciousness in the brain. The specific partof the axis circuit that underlies the subjective aspect of
conscious impressions is the electromagnetic activity of the
apical dendrite, according the present theory. In this final
section of the paper we examine in some detail the properties of
the electromagnetic activity of the apical dendrite to determine
what properties of conscious impressions they may underlie.
In particular, do the properties of electromagnetic activity of
the apical dendrite correspond to the properties of extended
duration and variations in intensity that we have attributed to
conscious impressions?
The electrical nature of the propagating EPSP in the
apical dendrite contrasts sharply with the electrical nature of the propagating action potential in the myelinated and un-
myelinated axon in several important and related respects.
Firstly, the initiating current and voltage of the apical dendrite
EPSP can vary considerably in amplitude while the current
and voltage characteristics of the action potential are virtually
constant (and relatively low) in a given axon. Secondly,
the EPSP of the apical dendrite decays substantially as it
propagates toward the soma, while the action potential decays
only very slightly within the inter-nodal segment of the
myelinated axon, except for the highest frequency components
of the Fourier spectrum of the action potential, which affect the
pulse rise-time only. Thirdly, owing to the clustering of several
apical dendrites aligned in parallel there is a strengtheningof the overall electric dipole effect, whereas sites of action
potentials are not consistently aligned within nerve bundles
of axons. We note that the superposition of electrical fields
from many parallel and closely spaced electric dipoles with
codirectional currents produces a local enhancement of the
electric fields in an axial direction along the inner and outer
surface of the apical dendrite membrane. This superposition
of fields concentrates the electromagnetic energy within the
cluster volume, so that the total field energy (inside and
outside the membrane) undergoes a gain that would not occur
in unclustered, well-separated apical dendrites. This gain or
enhancement in localized energy (or power) increases as the
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 14/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1017
Fig. 10. (Above) The decay in voltage, V , of an EPSP mass of charge shown at
three locations along the apical dendrite shaft as the EPSP mass propagates
toward the soma. (Below) At each of the three EPSP locations there is an
(appropriately small) electric dipole accompanied by its electromagnetic field.
square of the sum of the field amplitudes of individual apical
dendrites.To promote an understanding of how these fields could
relate closely to consciousness we describe the electromagnetic
characteristics of the transient fields in the extracellular region
produced by the propagating EPSP in the apical dendrite.
Because the present theory assumes that the time-dependent
charges decay as they propagate along the apical dendrite (seeFig. 6) we partition the apical dendrite into a collinear spatial
array of electric dipoles at any given instant of time. Each dipole
is a source of electromagnetic wave energy in the form of a
transient pulse of both localized electric field and magnetic
field energy (density) that radiates into the extracellular space
(see Fig. 10). Each electric dipole along the apical dendrite
shaft is derived from the time- and space-dependent current
of the moving EPSP waveform multiplied by an appropriately
chosen short length, d z, of the apical dendrite. These dynamic
dipoles are characterized by both magnetic and electric fields,
and their dipole moments depend on three variables: the charge
at the two poles, the velocity of the charges, and the distance
between them. When these component dipoles are superposed
they produce the overall dipole effect of the apical dendrite
activity.The time-dependent EPSP current at any given segment, d z,
along the dendrite is the excitation source of the dipole. The
pulse shape of the radiation field from this dipole is preserved
as it propagates away from the dendrite at a decreasing velocity.
This preservation of pulse shape results from the dominance
of conduction currents over Maxwellian displacement currents
in the extracellular medium (approximating that of seawater),
and from the low frequency characteristic of EPSPs. As
a consequence, the propagation of a radiating pulse may
be described by a low-frequency window (LFW), following
the analysis presented by Burrell and Peters (1979). When
propagation is in the LFW, as is the case for the EPSP or
action potentials, the extracellular pulse will propagate with
little distortion. The theoretical behavior of the transient pulse
or electromagnetic field propagation from an electrically small
electric dipole source is well-known for such lossy material as
extracellular fluids, and has been rigorously developed by Songand Chen (1993). For the known frequency content of EPSPs
and action potentials, the electric and magnetic field time
dependence at any localized point in the extracellular region
is identical to the spatially localized current time waveform
of the internal EPSP. Thus, for EPSP propagation along the
apical dendrite we can make the simplifying assumption that
the electric and magnetic fields in the extracellular medium
preserve the current waveform exhibited by the EPSP inside
the dendrite with little distortion.
Therefore, the attenuation of the extracellular electromag-
netic pulse depends only on the cube of the radial distance
from the apical dendrite and simple angular variables within
the LFW. We employ here a spherical coordinate system to de-scribe the field structure of the EPSP pulse, shown in Fig. 11.
Two electric field components (radial and tangential) and one
magnetic field component (circumferential) describe the total
instantaneous field at any point outside the apical dendrite. The
diagram shown in Fig. 11 describes the spatial distribution of
the electric field at a point in time for one current dipole, while
the magnetic field component is always circumferential about
the axis of the current dipole. The angular dependence of each
component specifies that at right angles to the axis of the api-
cal dendrite the tangential component of the electric field that is
parallel (or approximately parallel) to the dendrite is the dom-
inant electric field component. This parallel component willexhibit strong electrical coupling to a neighboring dendrite lo-
cated within several diameters of the dendrite (a typical apical
dendrite diameter is approximately 2 microns, and 6–8 apical
dendrite shafts lie within an area of approximately 12 × 12 mi-
crons (Peters & Sethares, 1991, Fig. 5)).
The equations that describe the near electric and magnetic
fields produced by the velocity-dependent dipole of the apical
dendrite are as follows.
E θ = ( I d z/4πσ r 3) sin θ, (1)
E r = ( I d z/2πσ r 3) cos θ, (2)
H φ = ( I d z/4πr
2
) sin θ, (3)ν ∼ 1/σµ0r , (4)
where E θ and E r are components of the electric field, H φ is
the magnetic field, ν is the field propagation velocity, I is the
current, d z is an appropriately small dipole length, r is the radial
distance from the dipole, θ is the angle defined by the radial
direction and the dipole axis, σ is the electrical conductivity
of the medium, and µ0 is the magnetic permeability of the
medium.Near the axis of the apical dendrite, the radial component
that points in the axial direction of EPSP propagation is the
dominant electric field component. Inside the apical dendrite
this axially directed component is at its strongest level, where
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 15/17
1018 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
Fig. 11. Detailed diagram (in a spherical coordinate system) of the radiating
electromagnetic field from an EPSP mass of charges at a particular location as
it moves along the apical dendrite. At a given point in space the electric field
intensity is E and the magnetic field intensity is H φ . The radial electric field
component, E r is highest in the interior of the dendritic shaft, and it decays very
rapidly in the radial direction as 1/r 3. For clarity, the circumferential magnetic
field distribution has been omitted from the diagram.
it is closest to the EPSP current flow mechanism. Eq. (4)
indicates that the electromagnetic pulse emitted by the dipole
propagates away from the apical dendrite at a velocity, ν, which
is proportional to the inverse product of radial distance andelectrical conductivity. In effect, the velocity of propagation is
close to the speed of light at the outer surface of the dendrite and
quickly decays with radial distance r to values several orders of
magnitude lower than the speed of light at distances beyond
about a millimeter, according to theoretical predictions. The
velocity decays as 1/r for a constant conductivity and magnetic
permeability.
The relationship between the EPSP voltage waveform and
current waveform can be derived from classical cable theory
using the well-known concept of wave impedance or cable
characteristic impedance. Figs. 10 and 11 contain examples
of EPSP voltage waveforms described as Gaussian pulse
shapes. The exact time dependence of the electromagnetic fieldpulse at a point in the extracellular fluid will depend on the
conductivity, dielectric properties, and thickness of the surface
layer of the apical dendrite. Clusters of apical dendrites in close
proximity therefore will develop electric field energy in their
core structures from the combined electric field components
parallel to the dendritic axes. The development of magnetic
field energy between parallel electric field dipoles is much less
significant compared to electric field energy coupling between
pairs of parallel apical dendrites with internal collinear dipoles
moving at the conduction velocity. According to the present
theory, it is this electromagnetic field intensity that underlies
our immediate conscious impressions.
6. Other ways of producing consciousness
This paper makes the strong claim that the physical correlate
of consciousness is the near electromagnetic field (see Eqs.
(1)–(3)) located along the surface of the neural membrane.
This field contains the “energy of consciousness” that has
the ability to do work on other electric charges and therebyinfluence other neural events (e.g., by modulation effects at the
soma); or it can simply be the state of “being conscious” of
the particular content coded by the apical dendrite (e.g., the
color red). Thus, a major physical property of the field
activity is the intensity of energy, which corresponds to
the cognitive property of intensifying a sensory impression
(e.g., concentrating attention more strongly to the redness of an
apple). Level of intensity determines what content dominates
consciousness, so that low intensity levels of active apical
dendrites in many areas of the brain (e.g., apical dendrites
that code for the ongoing impression of pressure of the feet
on the floor) will normally not be experienced as part of the
momentary content of consciousness. While the long apicaldendrite seems well-suited to provide variations in intensity and
other properties of consciousness (e.g., extendable duration),
we cannot rule out the possibility that other structures, organic
or inorganic, could also produce electric fields with these kinds
of properties. For example, action potentials in stable circuit
loops produce extended durations of electric field oscillations;
an example is the high spectral fields produced by spikes at the
axon initial segment of layer 5 pyramidal neurons within the
corticothalamic circuit loop. Since action potentials fire with
a constant voltage, variations in intensity would be based on
the number of participating axons in the near neighborhood;
however, action potentials at nodes of myelinated axons are notlikely to produce large summation of fields because the nodes
do not line up in close neighborhoods across axon clusters.
Another, more global, view combines the fields of all axons,
dendrites, and somas in subcortical and cortical regions of the
brain into one complex electromagnetic field, and regards this
total brain field as the basis of consciousness (McFadden, 2002;
Pockett, 2000). Finally, the electromagnetic fields produced by
inorganic structures such as wire antennas could, in principle,
serve as the basis of artificial consciousness, but it may
be somewhat difficult to infer the properties of conscious
activity in these structures without additional circuitry (e.g., for
accessing and identification) which the normal human brain
provides.
7. Tentative conclusions
In cortical circuits that operate by input–output processing,
the strengths of electrical signals typically are at relatively low
levels. In comparison, the electrical activity in apical dendrites
normally operates at relatively high levels, as suggested by
the stronger demands upon blood flow of local field potentials
compared to that of the multi-unit activity of action potentials
(Logothetis, Pauls, Augath, Trinath, & Oeltermann, 2001).
Also, EEG recordings, which are driven by electrical activity
of clusters of apical dendrites, can show exceptionally high
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 16/17
D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020 1019
amplitudes during meditation (e.g., Lutz et al. (2004)). Given
the level of electrical currents/voltage in an individual apical
dendrite, close clustering of apical dendrites increases the
concentration of field energies as the square of the number
of active apical dendrites within the cluster. Owing to the
length of the apical dendrites, these high levels of electrical
activity will be attenuated before they reach their somas,where they can modulate the input–output activity in circuits
that process signals. Thus, the extended length of the apical
dendrite provides one way to separate the pocket of intense
electrical activity in moderately distal sectors of the apical
dendrite from the input–output processing at the soma. These
pockets of intense electrical activity in the apical dendrite are
also separated from neighboring circuits in the intercellular
medium by the restriction of their high field strengths to small
regions very near the surface of the apical dendrite, because
the near electric field intensity falls off as 1/r 3, where r
is the radial distance from a point in the apical dendrite. It
would seem, therefore, that these small, relatively isolated
regions in the apical dendrites are able to increase their electricfield intensity to a relatively high level, and sustain it there
for extended durations without perturbing the input–output
electrical activities of neurons in neighboring circuits.
Therefore the following question arises: What function is
served by high electric field intensities and their extended
durations in apical dendrites? We have suggested that high
field intensities benefit input–output processing by their
stronger modulatory effects at the somas of pyramidal
neurons, with consequent increases in the strength of lateral
inhibitory connections that protect the axis circuit activity
from interference by neighboring minicolumn activity. But
these pockets of high electric field intensity are activated andsustained even when they are not modulating input–output
processing, for example, when we are attending to sunsets,
music, and other pleasurable activities. These localized high
electric field intensities may serve to raise our raw impressions
of the world to the level of consciousness, and thereby provide
our mental life with its special kind of existence.
References
Alexander, G. E., & Crutcher, M. D. (1990). Functional architecture of
basal ganglia circuits: Neural substrate of parallel processing. Trends in
Neuroscience, 13, 266–271.
Azouz, R., & Gray, C. M. (1999). Cellular mechanisms contributing to
response variability of cortical neurons in vivo. Journal of Neuroscience,
19, 2209–2223.
Bolz, J., Gilbert, C. D., & Wiesel, T. N. (1989). Pharmacological analysis of
cortical circuitry. Trends in Neuroscience, 12, 292–296.
Bouyer, J. J., Montaron, M. F., & Rougeul, A. (1981). Fast fronto-parietal
rhythms during combined focused attentive behavior and immobility in cat:
Cortical and thalamic localizations. Electroencephalography and Clinical
Neurophysiology, 51, 244–252.
Braitenberg, V., & Schuz, (1998). Cortex: Statistics and geometry of neuronal
connectivity (2nd ed.). Berlin: Springer.
Buchsbaum, M. S., Buchsbaum, B. R., Chokron, S., Tang, C, Wei, T. C.,
& Byne, W. (2006). Thalamocortical circuits: FMRI assessment of the
pulvinar and medial dorsal nucleus in normal volunteers. Neuroscience
Letters, 404, 282–287.
Burrell, G. A., & Peters, L., Jr. (1979). Pulse propagation in lossy media using
the low-frequency window for video pulse radar application. Proceedings
of IEEE , 67 , 981–990.
Colbert, C. M. (2001). Back propagating action potentials in pyramidal
neurons: A putative signaling mechanism for the induction of Hebbian
synaptic activity. Restoration of Neuroscience, 19, 100–211.
Contreras, D., Timofeev, I., & Steriade, M. (1996). Mechanisms of long lasting
hyperpolarizations underlying slow sleep oscillations in cat corticothalamic
networks. Journal of Physiology, 494, 251–264.
Damasio, A. R. (1994). Descartes’ error: Emotion, reason and the human
Brain. New York: Avon Books.
DeFelipe, J., & Farina, I. (1992). The pyramidal neurons of the cerebral cortex:
Morphological and chemical characteristics of the synaptic inputs. Progress
in Neurobiology, 3, 563–607.
Destexhe, A., Rudolph, M., & Pare, D. (2003). The high-conductance state of
neocortical neurons in vivo. Nature Reviews of Neuroscience, 4, 739–751.
DiRusso, F., Martinez, A., & Hillyard, S. A. (2003). Source analysis of event-
related cortical activity during visuo-spatial attention. Cerebral Cortex, 13,
486–499.
Edelman, G. M. (2003). Naturalizing consciousness: A theoretical framework.
Proceedings of the National Academy of Sciences of the United States of
America, 100, 5521–5524.
Feldman, M. L. (1984). Morphology of the neocortical pyramidal neuron. In A.
Peters, & E. G. Jones (Eds.), Cerebral cortex. Cellular components of the
cerebral cortex: Vol. 1 (pp. 123–200). New York: Plenum Press.
Goldberg, J., Holthoff, K., & Yuste, R. (2002). A problem with Hebb and local
spikes. Trends in Neuroscience, 25, 433–435.
Grossberg, S. (1999). The link between brain learning, attention, and
consciousness. Consciousness and Cognition, 8, 1–44.
Grossberg, S. (2005). Linking attention to learning, expectation, competition,
and consciousness. In L. Itti, G. Rees, & J. K. Tsotsos (Eds.), Neurobiology
of attention (pp. 652–662) (Chapter 107).
Helmchen, F., Svoboda, K., Denk, W., & Tank, D. W. (1999). In vivo
dendritic calcium dynamics in deep-layer cortical pyramidal neurons.
Nature Neuroscience, 2, 989–996.
Hendry, S. H. C., Schwark, H. D., Jones, E. G., & Yan, J. (1987). Numbers
and proportions of GABA-immunoreactive neurons in different areas of the
monkey cerebral cortex. Journal of Neuroscience, 7 , 1503–1519.Jarsky, T., Roxin, A., Kath, W. L., & Spruston, N. (2005). Conditional dendritic
spike propagation following distal synaptic activation of hippocampal CA1
pyramidal neurons. Nature Neuroscience, 8, 1667–1676.
Jones, E. G. (2002). Thalamic circuitry and thalamocortical synchrony.
Philosophical transactions of the Royal Society of London, B, 357 ,
1659–1673.
Jones, E. G. (2007). The Thalamus (2nd ed.). Cambridge (UK): Cambridge
University Press.
Khayat, P. S., Spekreijse, H., & Roelfsema, P. R. (2006). Attention lights
up new object representations before the old ones fade away. Journal of
Neuroscience, 26 , 138–142.
Koch, C. (1998). Biophysics of computation: Information processing in the
single neuron. New York: Oxford University Press.
Kuriki, S., Kanda, S., & Hirata, Y. (2006). Effects of musical experience on
different components of MEG responses elicited by sequential piano-tonesand chords. Journal of Neuroscience, 26 , 4046–4053.
LaBerge, D. (1997). Attention, awareness, and the triangular circuit.
Consciousness and Cognition, 6 , 149–181.
LaBerge, D. (2001). Attention, consciousness, and electrical wave activity
within the cortical column. International Journal of Psychophysiology, 43,
5–24.
LaBerge, D. (2005). Sustained attention and apical dendrite activity in recurrent
circuits. Brain Research Reviews, 50, 86–99.
LaBerge, D., Carlson, R. L., Williams, J. K., & Bunney, B. G. (1997). Shifting
attention in visual space: Tests of moving-spotlight models versus and
activity-distribution model. Journal of Experimental Psychology: Human
Perception and Performance, 23, 13801392.
Larkman, A. U. (1991). Dendritic morphology of pyramidal neurons of the
visual cortex of the rat: III. Spine distributions. Journal of Comparative
Neurology, 306 , 332–343.
8/7/2019 Apical dendrite and theory of consiousness (La Berge, 2007)
http://slidepdf.com/reader/full/apical-dendrite-and-theory-of-consiousness-la-berge-2007 17/17
1020 D. LaBerge, R. Kasevich / Neural Networks 20 (2007) 1004–1020
Larkum, M. E., & Zhu, J. J. (2002). Signaling of layer 1 and whisker-evoked
Ca2+ and Na+ action potentials in distal and terminal dendrites of rat
neocortical pyramidal neurons in vitroand in vivo. Journal of Neuroscience,
22, 6991–7005.
Lazar, S. W., Kerr, C. E., Wasserman, R. H., Gray, J. R., Greve, D. N.,
Treadway, M. T., et al. (2005). Meditation experience is associated with
increased cortical thickness. Neuroreport , 16 , 1893–1897.
Llinas, R. R. (1988). The intrinsic electrophysiological properties of
mammalianneurons: Insights into central nervous system function. Science,
242, 1654–1664.
Logothetis, N. K., Pauls, J., Augath, M., Trinath, T., & Oeltermann, A. (2001).
Neurophysiological investigation of the basis of the fMRI signal. Nature,
412, 150–157.
Lutz, A., Greichar, L. L., Rawlings, N. B, & Ricard, M. (2004). Long-term
meditators self-induce high-amplitude gamma synchrony during mental
practice. Proceedings of the National Academy of Science, USA, 101,
16369–16373.
Makris, N., Biederman, J., Valera, E. M., Bush, G., Kaiser, J., Kennedy, D.
N., et al. (2006). Cortical thinning of the attention and executive function
networks in adults with attention deficit/hyperactivity disorder. Cerebral
Cortex, 17 , 1364–1375.
McAdams, C. J., & Maunsell, J. H. (1999). Effects of attention on the reliability
of individual neuron in monkey visual cortex. Neuron, 23, 765–773.
McAdams, C. J., & Reid, R. C. (2005). Attention modulates the responses of
simple cells in monkey primary visual cortex. Journal of Neuroscience, 25,
11023–11033.
McFadden, J. (2002). The conscious electromagnetic information (cemi) field
theory: The hard problem made easy. Journal of Consciousness Studies, 9,
45–60.
McGuire, B. A., Hornung, J. P., Gilbert, C. D., & Wiesel, T. N. (1984). Patterns
of synaptic input to layer 4 of cat striate cortex. Journal of Neuroscience, 4,
3021–3033.
Mountcastle, V. B. (1957). Modality and topographic properties of single
neurons in cat’s somatic sensory cortex. Journal of Neurophysiology, 20,
408–434.
Mountcastle, V. B. (1998). The cerebral cortex. Cambridge (MA): Harvard
University Press.
Newman, J., Baars, B. J., & Cho, S-B. (1997). A neural global workshop modelof conscious attention. Neural Networks, 10, 1195–1206.
Nowak, L. G., Sanchez-Vives, M. V., & McCormick, D. A. (1997). Influence
of low and high frequency inputs on spike timing in visual cortical neurons.
Cerebral Cortex, 7 , 487–501.
O’Connor, D. H., Fukui, M. M., Pinsk, M. A., & Kastner, S. (2002). Attention
modulates responses in the human lateral geniculate nucleus. Nature
Neuroscience, 5, 1203–1209.
Pantev, C., Makieg, S., Hoke, M., Galambos, R., Hampson, S., & Gallen, C.
(1991). Human auditory evoked gamma-band magnetic fields. Proceedings
of the National Academy of Sciences. USA, 88, 8996–9000.
Peters, A, & Sethares, C. (1991). Organization of pyramidal neurons in area
17 of monkey visual cortex. Journal of Comparative Neurology, 306 ,
1–23.
Pockett, S. (2000). The nature of consciousness: A hypothesis. Lincoln (NE):
Writers Club Press.
Radley, J. J., Rocher, A. B, Miller, M., Janssen, W. G., Liston, C., Hof, P. R.,
et al. (2006). Repeated stress induces dendritic spine loss in the rat medial
prefrontal cortex. Cerebral Cortex, 16 , 313–320.
Rall, W. (1989). Cable theory for dendritic neurons. In C. Koch, & I. Segev
(Eds.), Methods in Neuronal Modeling (pp. 9–62). Cambridge (MA): MIT
Press.
Reyes, A. (2001). Influence of dendritic conductances on the input–output
properties of neurons. Annual Review of Neuroscience, 24, 653–675.
Rockel, A. J., Hiorns, R. W., & Powell, T. P. S. (1974). Numbers of neurons
through the full depth of the neocortex. Proceedings of the Anatomy Society
of Great Britain and Ireland , 118, 371.
Rockel, A. J., Hiorns, R. W., & Powell, T. P. S. (1980). The basic uniformity
in structure of the neocortex. Brain, 103, 221–244.
Silver, H. A., Ress, D, & Heeger, D. J. (2007). Neural correlates of sustained
spatial attention in human early visual cortex. Journal of Neurophysiology,
97 , 229–237.
Shahin, A., Bosnyak, D. J., Trainor, L. J., & Roberts, L. E. (2003).
Enhancement of neuroplastic P2 and N1c auditory evoked potential in
musicians. Journal of Neuroscience, 23, 5545–5552.
Shahin, A. J., Roberts, L. E., Pantev, C., Aziz, M., & Picton, T. W. (2007).
Enhanced anterior–temporal processing for complex tones in musicians.
Clinical Neurolphysiology, 118, 209–220.
Shu, Y., Duque, A., Yu, Y., Haider, B., & McCormick, D. A. (2007). Properties
of action potential initiation in neocortical pyramidal cells: evidence from
whole cell axon recordings. Journal of Neurophysiology, 97 , 746–760.
Song, J., & Chen, K. (1993). Propagation of EM pulses excited by an electric
dipole in a conducting medium. IEEE Transactions on Antennas and
Propagation, 41, 1414–1421.
Taylor, J. T. (1999). The race for consciousness. Cambridge (MA): MIT Press.
Taylor, J. G. (2005). Mind and consciousness: Towards a final answer? Physics
of Life Reviews, 2, 1–45.
Taylor, J. G. (2007). On the neurodynamics of the creation of consciousness.
Cognitive Neurodynamics, 1, 97–118.Tegner, J., Compte, A., & Wang, (2002). The dynamical stability of
reverberatory neural circuits. Biological Cybernetics, 87 , 471–481.
Turk,D. J., Heatherton, T. F., Macrae, C. N.,Kelley, W. M.,& Gazzaniga, M. S.
(2003). Out of contact, out of mind: The distributed nature of the self.
Annals of the New York Academy of Science, 1001, 65–78.
Van Essen, D. C., Drury, H. A., Joshi, S., & Miller, M. I. (1998). Functional and
structural mapping of human cerebral cortex: Solutions are in the surfaces.
Proceedings of the National Academy of Sciences, USA , 95, 788–795.
Yuste, R., & Tank, D. W. (1996). Dendritic integration in mammalian neurons,
a century after Cajal. Neuron, 16 , 701–71.