Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session C: Friday, August 15, 2:00 – 5:00 pm, de Brug & E‑Hall

Common representations underlie idiosyncratic neural topographies

Bogdan Petre¹, Martin A. Lindquist, Tor D. Wager; ¹Dartmouth College

Presenter: Bogdan Petre

Primary sensory brain areas contain innate stereotyped topographies like retinotopy and somatotopy, but association cortex shows more idiosyncratic organization. Using a twin sample from the human connectome project and functional magnetic resonance imaging, we test the hypothesis that idiosyncrasies are the byproduct of convergent learning and subserve common functional representations. We combined representational similarity analysis, full brain neuromaps and twin heritability models to investigate the spatial profiles and sources of representational and topographic similarity across diverse task conditions and at rest. We found common representational geometry with idiosyncratic topographies in both task evoked responses and resting state network organization. Common representations with idiosyncratic topographies were especially common in transmodal brain areas, late in the cortical hierarchy, but did not show consistent associations with neuromaps of genetic or developmental markers. Additionally, while topography was heritable, response geometry was not, indicating it was learned. These findings are consistent with experience dependent but convergent circuit organization during development, and echoes learning principles in artificial neural networks where circuit weights are always idiosyncratic but learned representations are nevertheless predictable. This shows how similar principles affect brain organization. Differing topographies in association cortex are at least in part subtle implementation differences that underlie shared representations.

Topic Area: Brain Networks & Neural Dynamics

Extended Abstract: Full Text PDF