Abstract
Visualizing brain activity is not enough to understand how this activity encodes mental representations, or how these are transformed into thoughts or behavior. Cognitive neuroscience is looking for the laws of passage from the neurophysiological to the psychological level - from brain activity to mental state. In the second lecture, we looked at strategies for discerning, in the abundance of brain activity that occurs at every moment, the neuronal states that code for mental states.
The subtraction method consists in contrasting the brain activity evoked by two minimally different experimental conditions (for example, naming a number versus subtracting it from 10 ; or naming a number versus naming a letter). It can be used to identify which brain circuit is involved in a specific stage of a cognitive task, but suffers from a reverse inference problem : while each cognitive operation activates a circuit that is often replicable, the observation that a given circuit is active does not tell us what cognitive function or code is being deployed.
Multivariate pattern analysis(MVPA) takes things a step further : it dissects the neural codes within a given region. We examine the extent to which two experimental conditions lead to similar activation maps, and deduce the extent to which these two experimental conditions call upon similar representations. In the laboratory, for example, we have used this method to show that the mathematician, who thinks about highly abstract mathematical objects, continues to use brain circuits activated in all of us during elementary arithmetic operations (2+3) or even the simple vision of a number. Thus, high-level mathematics " recycles " circuits from elementary arithmetic. Similarly, in children acquiring reading skills, multivariate analysis shows that certain sectors of the visual cortex become partially specialized for the recognition of written words, while retaining their previous selectivity for object recognition.
