HyperBase
We are creating a standardized template for hyperalignment along with a normative fMRI dataset and turnkey software to allow others to hyperalign their own data to the standardized template and estimate multiple functional topographies in the brains of their participants. Use of this resource will require only a limited fMRI dataset collected during movie viewing or rest, with no need to collect functional localizer data. This resource will be called, “HyperBase”. The development of HyperBase is supported by an NIMH grant (1 R01 MH127199-01A1, “Infrastructure for hyperaligning fMRI data and estimating functional topographies”, MPIs Haxby and Gobbini). Xiaoxuan Fan is the project coordinator for HyperBase. Feilong Ma and Jiahui Guo are postdoctoral fellows working on this project.
Estimating functional topographies in individuals with HA
We have shown that individual-specific functional topographies for visual-category-selectivity and retinotopy can be estimated with high fidelity using hyperalignment transformation matrices, even when those matrices are calculated from data collected while watching different movies using CHA. We are extending this work to functional topographies for many other functions, e.g. language, voices, visual motion, calculations, and working memory, and will incorporate localizer data and software for estimating topographies in HyperBase.
· Modeling naturalistic human face perception with DCNNs
· Modeling individual differences with INT
· Hyper-HMM
· Monkey Kingdom
· Functional architecture in congenitally blind and deaf
· Face perception system· Representation of actions