Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks

Artifact concepts are more reliably represented than animal concepts across the cortex

There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.

Poster D24 in Poster Session D, Wednesday, October 25, 4:45 - 6:30 pm CEST, Espace Vieux-Port

Stephen Mazurchuk¹, Lisa L Conant¹, Jeffrey R Binder¹, Leonardo Fernandino¹; ¹Medical College of Wisconsin

Introduction: Brain damage can result in severely impaired ability to retrieve general knowledge about items belonging to a particular category (e.g., animals) in the face of relatively preserved knowledge for other categories (e.g., man-made artifacts, fruits/vegetables, and body parts). One observation that remains unexplained is that cases of disproportionate semantic impairment for animals are much more common than cases of disproportionate impairment for artifacts (Capitani et al., 2003). It has been proposed that this asymmetry is due to differences in the representational nature of these categories in the healthy brain, such that artifact representations rely on a broader range of semantic features, including function and motor schemas. We assessed this proposal using representational similarity analysis (RSA) of fMRI data. We hypothesized that if, relative to animal concepts, artifact concepts are represented in terms of a more diverse set of features, their underlying neural activity patterns would be more discriminable from one another compared to animal concepts, resulting in a more reliable representational geometry across participants. Methods: We recruited 11 participants. Eight participants completed both scanning days (six presentations of the stimulus set). Two participants completed one day of scanning (three stimuli presentations), and one participant was excluded due to errors during task administration. We used a rapid event related design with 150 animal concepts and 150 artifact concepts matched on 13 lexical variables. Image data were pre-processed using fMRIprep, and beta-values for each of the 300 concepts were estimated using 3dREMLfit. Our primary analysis used an ROI derived from a meta-analysis of semantic language processing, and follow-up analysis examined other ROIs. We assessed the strength of the semantic representations using two approaches. First, we calculated a consistency based Intraclass correlation coefficient for both animal and artifact neural RDMs. Second, using word vectors based on a distributional semantic model (GloVe), we compared the RSA correlations obtained for the two categories. Results: The reliability coefficient for the artifact neural RDMs (ICC = .0076, 95% CI = [.0047 - .011]) was larger than for RDMs containing only animal concepts (ICC = .0019, 95% CI = [-.001, .0047]). We also found that RSA correlation values were higher for artifact than for animal concepts (p < .05, Wilcoxon test). Follow-up analysis using a different semantic mask and Glasser parcels showed a similar pattern. Conclusion: We found that artifact concepts have a more reliable representational geometry across the cortex than animal concepts. Our findings lend support to the hypothesis that, relative to animal concepts, artifact concepts depend on a wider variety of semantic features and cortical areas.

Topic Areas: Meaning: Lexical Semantics, Disorders: Acquired