1. Experimental Design in Functional Neuroimaging
Institute of Cognitive Neuroscience
University College London
Experimental Design in Functional Neuroimaging

2. Overview Subtractive designs Factorial designs
Functional specialisation and functional integration
Parametric designs
Psychophysiological interactions

3. Functional specialisation
Identification of regionally specific effects that can be attributed to changing stimuli or task conditions
Functional integration
Identification of interactions among specialised cortical areas and how these interactions depend upon context

4. Cognitive subtraction
Region(s) involved in the cognitive/ sensorimotor
process of interest
Activation task involving process of interest
Baseline task
identical to A except for process of interest

5. Biological motion MT / V5 STS Frontal eye fields Cerebellum
Parietal cortex

6. Assumptions of cognitive subtraction
1. The experimental task and baseline task must be identical in every way except for the process of interest
Many processes in addition to presence of biological motion in A including:
Visual motion
Eye movements

7. Superior temporal sulcus activated by biological motion minus random dot motion

8. 2. There must be no implicit processing of the component of interest in the baseline task
Activation task read words aloud
Baseline task
look at words
Regions involved in semantic processing?
motor, language and visual areas A B
Violates assumption 1: processes other than semantic processing in A
Violates assumption 2: difficult not to read (silently) a visually presented word, and if this is the case there will be semantic processing due to implicit reading of words in B

9. 3. Pure insertion
‘Pure insertion’ assumes that an extra cognitive component can be introduced without affecting the expression of existing components
There may be an interaction between the new component the old component
If visual cortex is activated more by reading words than by seeing words, it might be concluded that the visual cortex is involved in reading
But adding the cognitive component of reading might affect the processing or expression of seeing words
There may be a top-down effect of reading on the way words are processed in visual cortex - this interaction effect will be indistinguishable in the subtraction

10. Conjunctions
Cognitive conjunctions combine a series of subtractions with the aim of isolating a process that is common to two (or more) task pairs
The assumption of pure insertion can be avoided by extracting the presence of a main effect in the absence of an interaction

11. Conjunctions
Require two (or more) task pairs: the activation task of each pair must engage the component of interest, and each must have its own baseline task e.g. object processing: A1 B1
Activation task Name objects
Baseline task
Name colour X Y A2 B2
Activation task Looking at objects
Baseline task
Looking at colours Z

12. Conjunctions Baseline tasks can be high level or low level
The only restriction is that differences between the task pairs both contain the component of interest
The analysis results in any commonality in activation differences between the task pairs
The resulting region should be uniquely associated with the process of interest, not any interactions specific to each subtraction

13. Factorial designs
In factorial designs there are two or more factors and the different levels of each factor must be equated
The interaction identifies brain areas where the effect of one factor varies depending on the presence or absence of the other factor
This allows the effect of one factor on the expression of the other factor to be measured

14. Interaction effect (A x B)
With Factor A
Without Factor A
Words
Pictures
With Factor B 1 2
Semantic association
Without Factor B 3 4
Non-semantic association (size)
Interaction effect (A x B)
Left posterior inferior temporal sulcus is activated more by the semantic processing of words than by the semantic processing of pictures (Vandenberghe et al. 1996) A A 1 2 3 4
Condition
With factor B
Without factor B

15. Main effect of semantic association
600
Semantic RT
Non-semantic
Words
Pictures
Stimuli

16. Main effect of semantic association
BOLD signal in voxel Y
Non-semantic
Words
Pictures
Stimuli

17. Main effect of pictures
Semantic
Non-semantic
600 RT
Words
Pictures
Stimuli

18. Main effect of pictures
Semantic
Non-semantic
BOLD signal in voxel Y
Words
Pictures
Stimuli

19. Interaction
600 RT
Semantic
Non-semantic
Words
Pictures
Stimuli

20. Interaction BOLD signal in voxel Y Semantic Non-semantic Words
Pictures
Stimuli

21. Crossover interaction
Semantic
600 RT
Non-semantic
Words
Pictures
Stimuli

22. Crossover interaction
Semantic
BOLD signal in voxel Y
Non-semantic
Words
Pictures
Stimuli

23. SEMANTIC PROCESSING (1+2)-(3+4) = 1 1 -1 -1
Without Factor B 2 1
With Factor B
Without Factor A
With Factor A
Words
Non-semantic association (size)
Pictures
Semantic association
Main effect of:
SEMANTIC PROCESSING (1+2)-(3+4) =
Cond 1 2 3 4

24. Main effect of: WORDS (1+3)-(2+4) = 1 -1 1 -1 Cond 1 2 3 4 4 3
Without Factor B 2 1
With Factor B
Without Factor A
With Factor A
Words
Non-semantic association (size)
Pictures
Semantic association
Main effect of:
WORDS (1+3)-(2+4) =
Cond 1 2 3 4

25. Interaction between the factors: (1-2)-(3-4) = 1 -1 -1 1
Without Factor B 2 1
With Factor B
Without Factor A
With Factor A
Words
Non-semantic association (size)
Pictures
Semantic association
Interaction between the factors: (1-2)-(3-4) =
Cond 1 2 3 4

26. Interaction between the factors: (3-4)-(1-2) = -1 1 1 -1
Without Factor B 2 1
With Factor B
Without Factor A
With Factor A
Words
Non-semantic association (size)
Pictures
Semantic association
Interaction between the factors: (3-4)-(1-2) =
Cond 1 2 3 4

27. Parametric designs Process of interest is treated as a dimension
The covariate of interest (or regressor) can be:
manipulated externally by the experimenter (e.g. word presentation rate)
determined by subject (e.g. number of words correctly remembered)
Analysis reveals brain areas whose activity shows a significant regression on the covariate of interest
Positive or negative slope
Linear or non-linear

28. PET study Delays: 0 ms 100 ms 200 ms 300 ms Delay Subject’s right hand
robot
Tactile
stimulus
left hand
Left
Delay
Delays:
0 ms
100 ms
200 ms
300 ms

29. Right cerebellar cortex activity shows a positive regression on delay
Delay (ms)
Cerebellum rCBF

30. Blood flow (PET)
BOLD signal (fMRI)
Saturation effect

31. Psychophysiological interactions
Assess how interactions among brain regions depend upon context
Explain the physiological response in one part of the brain in terms of an interaction between a psychological process and activity in another part of the brain
Task A
BOLD signal in voxel Y
BOLD signal in voxel X
Task B

32. Cerebellum contributes of SI activity during self-produced but not externally produced touch
Self-produced touch
Externally produced touch