PROJECT SUMMARY/ABSTRACT
Over the past two decades, neuroimaging studies of human episodic memory have established that
remembering an event from the past involves reactivation of activity patterns initially evoked during the
perceptual experience of that event. Indeed, reactivation has proven to be an incredibly valuable tool for studying
episodic memory as it, in turn, has been related to many behavioral expressions of memory. Yet, whereas
reactivation, by definition, is a measure of the similarity between perceptual and memory representations, there
is recent and accumulating evidence for systematic differences in how—or where—perceptual representations
and memory representations are expressed. At present, these systematic differences between perception and
memory are not well understood and there are many fundamental, open questions concerning the functional
significance of these differences and how to reconcile these differences with the phenomenon of reactivation.
The conceptual idea guiding the current proposal is that memory representations can be formally described as
transformed versions of perceptual experience. In our first aim, we will use innovative fMRI methods to learn
transformation functions describing the relationship between perceptual and memory representations. More
specifically, by measuring perceptual representations in sensory cortex, we will attempt to predict subsequent
expressions of memory in different brain regions (parietal cortex). In aim 2, we will use a combination of natural
language processing algorithms, image processing techniques, and voxel-wise encoding models to rigorously
characterize information content in neural expressions of memory. This will allow us to characterize how
information content changes with perception-to-memory transformations and whether transformation or
reactivation better predict behavioral expressions of memory. In aim 3, we will test whether perception-to-
memory transformations represent a more general example of transformation from external-to-internal
representations. Specifically, we will compare perceptual representations with representations during episodic
recall (of the past), episodic simulation (of the future), and working memory maintenance. The long-term goal of
this program of research is to better understand how memories are expressed in the brain. This goal has broad
relevance to basic science and health. While neuroimaging studies of reactivation have substantially advanced
the field, there is an opportunity for further advance by understanding how, why, and when perception-to-memory
transformations occur.