Localising cognitive functions: Kanwisher et al. (1997) and the Fusiform Face Area (FFA). What was it contrasted with?
Are their findings good?
Greater FFA activation for faces compared to:
- Objects
- Scrambled Faces
- Houses
- Hands
- Great example for a well controlled experiment. Reliable in most participants and replicable.
What other modules were found other than FFA?
Parahippocampal Place Area (PPA)
- Houses or Places
Extrastriate Body Part Area (EBA)
- Body
What is the 1st Argument against Modularity: Gauthier and Tarr (1997)?
- After greeble expertise, FFA responded to greebles.
- FFA reflecting expertise
- Brain will run out of space if modular
What is the 2nd Argument against Modulaity: Malach et al. (2002)?; What did he suggest that organisation in the visual cortex follows?
- Visual system not by specific object categories; but by where in objects are usually encountered
According to “Eccentricity”, how is the visual cortex organised?
Ventral visual cortex organised by cortical topography
According to “Eccentricity”, how is coding driven?
Coding driven by resolution need
(1) High Resolution
* Centre = Faces = FFA
(2) Peripheral Resolution
* House/Place = PPA (that’s where places/houses usually are encountered in our visual field )
Modularity, Expertise and Eccentricity: Which one is right?
Evidence for all 3 - true to some extent.
fMRI signal might therefore reflect a mixture of all three coding schemes
What is reverse inference?
Drawing conclusions about cognitive processes from the presence of activation
What is the steps/logic in reverse inference?
- This study
- Task A, Brain Region Z active
- Other study
- Cognitive process X, Brain Region Z active
- Hence, in our study,
- Activation of Brain Region Z = Engagement of Cognitive Process X
Where is the first problem in reverse inference?
Steps 2: Cognitive Process X may not be exclusive
- Brain Region Z activate for many other cognitive processes/tasks
- If Brain Region Z is activated by many cognitive functions, we learn very little from observing activation in those areas
Example of Reverse Inference Problem: Frontal Cortex. What has been suggested of anterior, posterior, dorsal and ventral areas?
- Anterior regions (front)
- Abstract Information
- Posterior regions (back)
- Specific content
- Dorsal Axis
- Abstractness of Rules
- Ventral Axis
- Abstractness of Memory
Example of Reverse Inference Problem: Pre-Frontal Cortex. What have other studies found? What did Duncan suggest?
- Frontal cortex activated in lots of tasks
- Duncan: Frontal cortex reflect relative specialization instead of absolute specialization.
- Thus, frontal cortex is recruited “more strongly” as task difficult increases
What is the “multi-demand network”?; If we find a region of activation, can we know what the region is doing?
- Multi-demand network: A bunch of regions co-activated
- We don’t know what the specific region is doing.
Where is the second problem in reverse inference?
Steps 1 and 3: We need to know how good task A actually is for understanding cognitive process X
According to Poldrack (2006), what are the 2 things experiments must ensure?
Probability that cognitive process X is involved depends on
- 1.) Quality of task to measure this cognitive process
- 2.) Specificity of region for this cognitive process
What is the third problem in fMRI studies?
Over-interpretation of null results
If Region A was not significantly activated for A compared to B, we can conclude that…
We don’t know
Why can’t we interpret null results in fMRI?
- ) Statistical tests are designed to make it difficult for the H1
- ) Method might just not be sensitive enough to detect small differences (e.g. 3x3x3mm. Might not pick up layer-dependent activity)
