Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Decoding Abstract Relations: Investigating the Neural Representation of Relational Categories in the Human Brain

Poster Session B - Sunday, March 8, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom

Anthony Dunn¹ (), Katherine Alfred¹, Nicholas Ichien³, Brianna Aubrey¹, Sophia Baia², Silvia Bunge², David Kraemer¹; ¹Dartmouth College, ²University of California, Berkeley, ³University of California, Los Angeles

Analogical reasoning supports the human ability to derive generalizable knowledge about the world and, in turn, hinges on our ability to understand abstract relations between concepts. Revealing how these abstract relations are organized in patterns of neural activity is key to explaining how the brain implements analogical reasoning. For example, the meanings of the words “fruit” and “apple” are distinct from those of “car” and “sedan”—however, “fruit:apple” and “car:sedan” both instantiate the same abstract relation (category:exemplar). Evidence suggests that abstract relations are activated even when they are irrelevant to task goals. However, it is unclear whether such ‘implicit’ representations differ from those activated during explicit analogy judgments. Across two tasks, we modeled abstract relations using representational similarity analysis (RSA). Both tasks included word pairs that instantiate the same set of abstract relational categories (e.g., category:exemplar). Importantly, relational categories were incidental to the first task (judging semantic relatedness between individual words), allowing us to test whether abstract relations were activated implicitly. The second task required participants to judge analogies directly. Representational Similarity Analyses revealed a common set of parietal, ventral temporal, and anterior temporal regions that reflect abstract relational categories across both tasks. Additional regions reached significance in only one of the two tasks, suggesting task-dependent recruitment of regions that reflect abstract relational categories. Variability in representational strength was consistent with individual differences in analogical reasoning. These findings support the presence of a network that reflects abstract relations independent of explicit demands on analogical reasoning.

Topic Area: THINKING: Reasoning