Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Contextual role modulates conceptual representations in the human brain
Poster Session B, Wednesday, September 30, 4:30 - 6:30 pm, Wangari Maathai
Julien Dirani1, Shankar Chawla2, Leila Wehbe1,3, Brad Z. Mahon1,2; 1Neuroscience Institute, Carnegie Mellon University, 2Department of Psychology, Carnegie Mellon University, 3School of Computer Science, Carnegie Mellon University
The human brain represents concepts in a way that is both invariant across instances and flexible enough to support different contexts and tasks. Whether you recognize a knife sitting on the kitchen counter, have to name the knife, or reach and grab it to slice an apple, your brain retrieves the same underlying concept of what a knife is. Yet, the brain must also flexibly adapt these conceptual representations to support specific tasks and contexts. However, it remains unknown what aspects of object representations undergo such dynamic modulation. Here we combined fMRI with naturalistic movie viewing to investigate how object representations are modulated based on behavioral context. Participants viewed a first-person cooking movie showing a range of objects appearing in two contextual roles: passive elements of the scene or targets of goal-directed actions. Voxel-wise encoding models and representational similarity analysis (RSA) were used to characterize both which cortical regions preferentially encode objects in each context and how the geometry of object representations is modulated across object contexts. Objects recruited dissociable cortical networks across contexts, with action targets engaging an action-oriented network across postcentral gyrus and inferior parietal lobule, while passive objects recruited a distributed occipito-temporal network. Within the networks most strongly encoding objects in their respective contexts, representational geometry showed a double dissociation: target object representations were organized by action affordance and hand posture affordance structure, while passive object representations aligned with semantic similarity. A searchlight RSA further revealed that this dissociation is graded rather than absolute: affordance structure remained detectable for passive objects, but localized specifically to the somatosensory and parietal regions most strongly engaged by action targets, suggesting that context modulates the degree of engagement of these regions rather than the representational format they employ. This suggests that beyond context-specific brain networks, representational content was more context-invariant. Together these findings suggest that flexibility and invariance operate at different levels of the same representational system. Context modulates the degree of engagement of context-preferring networks, rather than the representational format they encode.
Topic Areas: Meaning: Lexical Semantics,