Abstract
In the fifth lecture, we examined the classic question of the temporal, rather than spatial, breakdown of brain activity.
Every cognitive operation, no matter how simple, such as reading a letter, involves a whole series of information-processing steps. To analyze them, mental chronometry, with Franciscus Donders, Saul Sternberg and Michael Posner, has developed a whole range of behavioral methods for breaking down response times (pure insertion ; additive factor method ; factorial plans, etc.). However, these methods remain indirect and do not directly separate the different execution stages of a cognitive process.
In the lecture, we saw how EEG and MEG can be exploited, in combination with an experimental factorial design, to break down a sequence of brain activity into stages. Multivariate decoding of EEG and MEG signals provides a wealth of information on the temporal unfolding of mental representations. By training a decoder to identify when brain signals contain significant information about a particular cognitive variable, and then examining whether the same decoder can generalize to another point in time, or to another experimental condition, we can determine the path of information in the human brain. The generalizationacross time(GAT) matrix provides detailed information on the duration of each step, its stability over time, and its recurrence at different times, possibly with a different delay or duration.
