Tracking Thoughts with Functional MRI

Most fMRI studies are aimed at establishing a mapping between cognitive functions and neural structures. Over the past several years, my colleagues and I have been developing an complementary approach to fMRI data analysis. Instead of focusing on structure-function mapping, we use pattern classification methods (applied to distributed patterns of neural data) to identify the neural signatures of particular thoughts and memories. Once we have trained the classifier to recognize a particular thought, we can track the comings and goings of that thought over time. Variants of these methods can also be used to explore the similarity structure of cognitive representations, and how this similarity structure changes as a function of learning. In this talk, I will provide an overview of this program of research: I will discuss some successful applications of these “thought-tracking” methods to studying memory retrieval. I will also discuss challenges that we have faced (and continue to face) in getting these methods to reach their full potential.