experimental design tali sharot & christian kaul with slides taken from presentations by: tor...
TRANSCRIPT
Experimental Design
Tali Sharot & Christian Kaul
With slides taken from presentations by: Tor Wager
Christian Ruff
Part I
• Taxonomy of experimental design (Friston ’97)• Aim of design• Block Design• Event Related Design• Baseline / Control• Timing
General Taxonomy of Experimental Design(Friston, 1997)
• Categorical – activation in one task as compared to that in another task– categorical designs assume that the cognitive processes can be
dissected into sub-cognitive processes & that one can add or remove cognitive processes by “Pure insertion”.
• Factorial– Factorial designs involve combining two or more factors within a
task and looking at the effect of one factor on the response to other factor
• Parametric – systematic changes in the brain responses according to some
performance attributes of task can be investigated in parametric designs
Categorical experimental design
– Subtraction • Pure insertion: assumption that one can add or remove
cognitive processes without influencing others. • activation in one task as compared to that in another task
considering the fact that the neural structures supporting cognitive and behavioural processes combine in a simple additive manner
– Conjunction • Testing multiple hypotheses• several hypotheses are tested, asking whether all the
activations in a series of task pairs, are jointly significant
Factorial design - example
• A – Low attentional load, motion• B – Low attentional load, no motion• C – High attentional load, motion• D – High attentional load, no motion
A B
C D
LOW
LOAD
HIGH
MOTION NO MOTIONLoad task Rees, Frith & Lavie (1997)
Terminology
• Simple main effects
• Main effects
• Interaction terms
A B
C D
LOW
LOAD
HIGH
MOTION NO MOTION
SIMPLE MAIN EFFECTS
• A – B: Simple main effect of motion (vs. no motion) in the context of low load
• B – D: Simple main effect of low load (vs. high load) in the context of no motion
• D – C: ?• Simple main effect of no
motion (vs. motion) in the context of high load
A B
C D
LOW
LOAD
HIGH
MOTION NO MOTION
OR
The inverse simple main effect of motion (vs. no motion) in the Context of high load
MAIN EFFECTS
• (A + B) – (C + D): • the main effect of low load
(vs. high load) irrelevant of motion
Main effect of load
• (A + C) – (B + D): ?• The main effect of motion (vs.
no motion) irrelevant of load Main effect of motion
A B
C D
LOW
LOAD
HIGH
MOTION NO MOTION
INTERACTION TERMS
• (A - B) – (C - D): • the interaction effect of
motion (vs. no motion) greater under low (vs. high) load
• (B - A) – (D - C): ?• the interaction effect of no
motion (vs. motion) greater under low (vs. high) load
A B
C D
LOW
LOAD
HIGH
MOTION NO MOTION
Factorial design in SPM5
A B
C D
LOW
LOAD
HIGH
MOTION NO MOTION
• How do we enter these effects in SPM5?
• Simple main effect of motion in the context of low load:
• A vs. B or (A – B)A B C D
[1 -1 0 0]
Factorial design in SPM5
• Main effect of low load: • (A + B) – (C + D)
• Interaction term of motion greater under low load:
• (A – B) – (C – D)
A B C D
A B C D
[1 -1 -1 1]
[1 1 -1 -1]
Factorial design in SPM5
• Interaction term of motion greater under low load:
• (A – B) – (C – D) A B C D
Parametric experimental design
• What do we want to measure? systematic changes in the brain responses
according to some performance attributes of task can be investigated in parametric designs:
Part I
• Aim of design • Block Design• Event Related Design• Baseline / Control• Timing
What we want from a design
• Interpretability: Can I relate brain data to specific psychological events?– Memory retrieval and comparison processes associated with recognition
• Power: Can I detect results?
Experimental (A) - Control (B) Experimental (A) - Control (B)
Block Design• Similar events are grouped
….
pros • Avoid rapid task-switching (patients)• Fast and easy to run • Good signal to noise
Block design - some pros & cons
cons
• Expectation (cognitive set, attention, fatigue)
• Habituation (olfactory, emotional)
• Different trials according to subjects’ responses.
Event Related Design
• Events are mixed
• Encode:
Event Related Design
Recognition Test:
……..
Response: new old old new old
Category: CR HIT HIT MISS FA
Baseline?
Known: Queen! Unknown? Aunt Jenny?
• Different stimuli & Task
Queen! Female!?
• Same stimuli, different task
• Different stimuli similar Task
+
Queen! i - pod!Queen! Mmm..Whats for dinner?...
• Similar stimuli, same task
“Baseline” here corresponds to session mean
“Cognitive” interpretation hardly possible, but useful to define regions generally involved in the task
Baseline?
Timing : the long and the short of it
3 s picture viewing3 s picture viewing
Recognition: 250 msRecognition: 250 ms “What was he in?” ““I used to wear Batman I used to wear Batman PJs…”PJs…”
Timing : the long and the short of it
Autobiographical Memory RetrievalAutobiographical Memory Retrieval
Word RecognitionWord Recognition
“Friend”
Search Episodic Retrieval &ElaborationSearch Episodic Retrieval &Elaboration
~5sc ~14sc
Response
Part II
• ROIs in early visual cortex• Multivariate decoding
• Natural viewing• Individual differences
ROI – Regions of interest
• A) anatomically defined
• B) functionally defined
What are ROIs in early visual cortex?
Stimuli 3D representation Flatmesh
How do we get ROIs?
+
SPMmip
[6.4
2772
, -62
.907
, -24
.589
]
<
< <
SPM{T435
}
on>off
SPMresults: .\model smoothedHeight threshold T = 4.818653 {p<1e-006 (unc.)}Extent threshold k = 0 voxels
Design matrix5 10 15 20 25 30
50
100
150
200
250
300
350
400
450
contrast(s)
1
0
1
2
3
4
5
6
Meridian map
50 100 150 200 250
50
100
150
200
250
V1v
V2v
V1d
V2d
=
Meridian map
50 100 150 200 250
50
100
150
200
250
V1v
V2v
V1d
V2d
Meridian map
50 100 150 200 250
50
100
150
200
250
V1v
V2v
V1d
V2d
right hemisphere dot localiser retinotopic location of dot
exemplar ROI Result
• Extracting activity-values from ROIs for all conditions.
• Then compute interaction term for activity in V5/MT greater under motion (vs. no motion) under high versus low load
• (replication: Rees et al. ’97)
cont
rast
bra
in a
ctiv
ity
ROI
Load task
Multivariate pattern analysis?
• What is MVPA?
• Methodology in which an algorithm is trained to tell two or more conditions from each other.
• The algorithm is then presented with a new set of data and categorises/classifies it into the conditions previously learned.
What questions can (& cannot) be answered with multivariate pattern analysis?
• When conventional analysis is not feasible, multivariate analysis might be an option
but what are we actually measuring?
• Assumption:
• Feature sensitive information is present in BOLD signal
Haynes & Rees (2006)
Haynes & Rees (2005)
Mean
signal
LDA
What questions can (& cannot) be answered with multivariate pattern analysis?
• Feature sensitive information is present in BOLD signal (biased competition)
• Multivariate decoding extracts this info
• Thus feature selective processing promises new insights (i.e. towards a better understanding of neuronal population coding contained in the BOLD signal)
Haynes & Rees (2006)
Haynes & Rees (2005)
Mean
signal
LDA
Multivariate pattern analysis – how to design an experiment
• Other then with conventional analysis we are asking a different question:
Does the pattern of activity contain meaningful information we can extract?
Not the level of brain activity is addressed, but the pattern of information within the activity.
Experiment 1 Question:
• Does feature selective information (left vs. right tilted orientation as measured by decoding from BOLD signal) for the irrelevant annulus change between the two central load conditions?
• Prediction (by load theory): – Feature selective information will be
reduced in high load condition
Multivariate Decodingexample Result:
Number of voxels
% c
orre
ct d
ecod
ed
50 1001
Low
High
N voxels
Acc
urac
y
Result: Feature selective info present and decoded
expected actual
The reverse-correlation method
Hasson et al., (2004)
Individual Differences
Individual Differences
Post – Scanning Questionnaires/ Tests …
Select subjects that vary on a specific dimension
Thank you…
Between- Subject Correlation
Hasson et al., (2004)
Kahn et al., (2004)
Example: Subsequent Memory
Kahn et al., (2004)