Dissociating learning-induced changes in fMRI signal from structural modifications: a comment on Dorjee and Bowers (2012).

In a recent commentary, Dorjee and Bowers (2012) raise several issues concerning the interpretation of functional MRI (fMRI) studies that examine learning in general and perceptual learning in particular. The authors rightly criticize an implicit assumption that differences in blood oxygen level dependent (BOLD) levels before and after training can be used to exhaustively determine the brain areas where learning leads to structural changes associated with information retention. However, there are several problematic aspects in Dorjee and Bowers’ line of reasoning. Here we would like to underline the importance of the issues raised by Dorjee and Bowers and at the same time prevent misunderstandings about the actual scope of the problem. One problem with the argument of Dorjee and Bowers (2012) is their claim that learning studies using fMRI usually assume BOLD-signal changes to reflect learning-related structural modifications. This statement is probably too general. Similarly, the authors’ distinction between learning studies and research on what they call “on-line processing” does not constitute a valid simplification, because any form of learning that is behaviorally relevant will affect neural structure as well as future on-line processing. Finally, an isolated discussion of retention aspects of learning may be counterproductive for assessing the relevance of BOLD-fMRI to the study of learning. Both human and animal research suggests that learning involves several steps, each characterized by dynamic interactions between many brain regions, including