Stimulus Representation in the Interactions Between Multiple Brain Regions

Poster Session C - Sunday, April 14, 2024, 5:00 – 7:00 pm EDT, Sheraton Hall ABC

Paul C. Bogdan^1,2, Simon W. Davis^1,3, Roberto Cabeza^1,2; ¹Center for Cognitive Neuroscience, Duke University, Durham, NC 27708, USA, ²Department of Psychology and Neuroscience, Duke University, Durham, NC 27708, USA, ³Department of Neurology, Duke University, Durham, NC 27710, USA

Cognitive neuroscience research increasingly focuses on how cognitive functions are distributed across multiple brain areas and networks. Yet, studies on how individuals represent stimuli have largely examined neural populations and representation at local levels. The present study considered that stimuli may also be represented in terms of the interactions between multiple regions. For example, consider how age is represented: Analogous to how a voxel may show increased activation when viewing a child as opposed to an adult, a pair of regions may show increased connectivity when viewing a child relative to viewing an adult. To investigate the possibility that some areas of the brain represent stimuli in this distributed fashion, we adapted typical procedures for Representational Similarity Analysis (RSA). Traditional RSA involves isolating regions to examine the links between stimulus features and voxelwise activation patterns. By contrast, we examined groups of regions simultaneously and investigated how stimulus features are linked to the interactions between regions. Applying this strategy within an fMRI study, we demonstrate the existence of distributed stimulus representations and show how examining brain data from this perspective yields particular insights into how the prefrontal cortex represents conceptual features of stimuli. These findings have the potential to open new avenues for understanding how information is processed and encoded. Accordingly, our study also lays out methodological principles that other researchers can use to investigate this and related phenomena within their own data.

Topic Area: LONG-TERM MEMORY: Semantic