Schedule of Events | Symposia

Causal knowledge is embedded in semantic networks

Poster Session B - Sunday, March 30, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom
Also presenting in Data Blitz Session 4 - Saturday, March 29, 2025, 10:30 am – 12:00 pm EDT, Constitution B.

Miriam Hauptman¹ (mhauptm1@jhu.edu), Marina Bedny¹; ¹Johns Hopkins University

When reading something like, “Sam attended a busy conference. Now he has COVID,” we naturally infer a causal relationship between crowded spaces and the invisible transmission of illness. What neural mechanisms support such automatic causal inferences? We tested the pre-registered hypothesis that causal knowledge is embedded within distinct high-level semantic networks. Prior work suggests that thinking about living things (people, animals) and inanimate objects depends on partially distinct networks (e.g., Warrington & Shallice, 1984). The precuneus has been implicated in high-level representations of living things (e.g., Fairhall & Caramazza, 2013). Participants (n=32) undergoing fMRI read causal vignettes that encouraged biological inferences about illness (e.g., cancer, flu) or encouraged mechanical inferences about objects breaking down (e.g., teapots, houses). Non-causal control vignettes contained the same sentences but were not causally connected. All vignettes were about people and contained similar grammatical structure and lexical items. The same participants performed localizer tasks: theory of mind, language, and logical reasoning. Univariate and multivariate analyses revealed that biological causal inferences selectively recruit the precuneus. Within the precuneus, responses to biological inferences were ventral to individually localized responses to mental states, pointing to a neural distinction between causal inferences about the body and the mind. Visual regions in lateral ventral occipitotemporal cortex involved in the perception of living things did not exhibit sensitivity to biological causal inferences. Mechanical causal inferences recruited a distinct set of areas implicated in intuitive physics and place concepts. Together, these findings suggest that causal knowledge is distributed across distinct high-level semantic networks.

Topic Area: THINKING: Reasoning