PLS analysis of fMRI data on cognitive processes

Cognitive processes like memory and self-referential are known to be underlined by extended neural networks. To study these complex processes, multivariate methods appear as the methods of the first choice since they take into account the functional integration. Partial Least Square (PLS) was used to study neural networks related to memory and self-referential processing in Alzheimer’s disease patients (AD) and two examples were presented. In the first one, we investigated the metabolic correlates of two forms of memory (conjunctive and relational memory performances) in AD. PLS identified two different brain networks highlighting correlations of the two types of memory and the glucose metabolism. In the second example, we assessed brain regions engaged during self-referential processing of information in AD patients during a task related fMRI study. PLS identified a wide brain network showing the effect of self- vs. other-referential processing. In contrast to univariate methods, PLS showed to be suitable for the study of cognitive processes.