Parahippocampal cortex integrates semantic and visual features of object images

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

Sophie Allen¹ (sda20a@fsu.edu), Ryan Kretschmar¹, Tyler Delmore², Morgan Barense^2,3, Chris Martin^1,4; ¹Department of Psychology, Florida State University, ²Department of Psychology, University of Toronto, ³Rotman Research Institute, Baycrest Hospital, Toronto, ⁴Program in Neuroscience, Florida State University

Recognizing objects and understanding how they relate to one another depends on distinctive and integrative coding of semantic and perceptual features. Evidence from prior neuroimaging research in which words were used as stimuli suggests that the semantic and visual features of objects are represented in distinct cortical areas, including lateral occipital cortex (LOC), the temporal pole (TP), and parahippocampal cortex (PHC), and that fully-specified object concepts reflect integrative coding of these features in perirhinal cortex (PRC). It is unclear, however, whether a similar representational landscape will emerge when images are used as stimuli. We asked this question using functional magnetic resonance imaging with healthy young adults. During scanning, participants made semantic and visual judgements about object images in separate task contexts. The stimulus set contained objects that were either visually similar or semantically similar, but not both. We used representational similarity analyses to determine whether the activity evoked by objects were predicted by behavior-based models that captured the semantic and visual similarities among the stimuli. Using this approach, we revealed evidence of visual similarity structure in LOC across both task contexts, which is consistent with our previous results obtained using words as stimuli. Activity in PRC expressed semantic similarity structure in the semantic task context only. Thus, this result diverges from the integrative coding we revealed using object concepts presented as words. Interestingly, PHC showed both semantic and visual similarity structure across task contexts, suggesting that it supports the integration of these different kinds of object features.

Topic Area: LONG-TERM MEMORY: Semantic