Schedule of Events | Symposia

The cerebellar components of the human language network

Poster Session E - Monday, March 31, 2025, 2:30 – 4:30 pm EDT, Back Bay Ballroom/Republic Ballroom
Also presenting in Data Blitz Session 1 - Saturday, March 29, 2025, 10:30 am – 12:00 pm EDT, Grand Ballroom.

Colton Casto^1,2 (ccasto@mit.edu), Benjamin Lipkin², Hannah Small³, Moshe Poliak², Greta Tuckute², Anila D'Mello^4,5, Evelina Fedorenko^1,2; ¹Harvard University, ²MIT, ³Johns Hopkins University, ⁴University of Texas Southwestern, ⁵University of Texas at Dallas

The cerebellum’s capacity for neural computation is arguably unmatched. Yet despite ample evidence of cerebellar contributions to cognition, including language, its precise role in linguistic processing remains debated. One reason for this is that most prior studies have had to choose between cross-domain breadth—sampling a wide range inputs/tasks across domains to inform specificity—and domain-specific depth—sampling specific, theoretically-motivated tasks within a singular domain to inform function. Here, we undertake a large-scale evaluation of cerebellar language-responsive areas using precision fMRI that is both broad and deep with respect to the inputs/tasks that we consider. We identify four cerebellar regions that respond robustly during language processing across both auditory and written modalities (Experiments 1a-b, n=754). However, only one of these areas—spanning Crus I/II/lobule VIIb—appears to be selective for language relative to diverse motor, perceptual, and cognitive nonlinguistic tasks (Experiments 2a-e, n=776). Similar to the cortical language system, Crus I/II/VIIb supports semantic processing, in both comprehension and production, but it does not support word access or phrase structure building (Experiments 3a-b, n=111). Crus I/II/VIIb is also modulated by some, but not all, of the same sentence-level features that modulate cortical language regions (e.g., grammaticality and frequency; Experiment 3c, n=5). Finally, of the cerebellar language-responsive areas, Crus I/II/VIIb is the most functionally integrated with the cortical language system (Experiment 4, n=85), suggesting that it may receive information from the cortical language network for further semantic processing.

Topic Area: LANGUAGE: Other