Abstract
Information that is shared across brains is encoded in idiosyncratic fine-scale functional topographies. Hyperalignment captures shared information by projecting pattern vectors for neural responses and connectivities into a common, high-dimensional information space, rather than by aligning topographies in a canonical anatomical space. Individual transformation matrices project information from individual anatomical spaces into the common model information space, preserving the geometry of pairwise dissimilarities between pattern vectors, and model cortical topography as mixtures of overlapping, individual-specific topographic basis functions, rather than as contiguous functional areas. The fundamental property of brain function that is preserved across brains is information content, rather than the functional properties of local features that support that content. In this Perspective, we present the conceptual framework that motivates hyperalignment, its computational underpinnings for joint modeling of a common information space and idiosyncratic cortical topographies, and discuss implications for understanding the structure of cortical functional architecture.
Highlights
Information encoded in cortex is organized at varying spatial scales
We have shown that reducing the dimensionality of the common model space using principal component analysis (PCA; Guntupalli et al, 2016; Haxby et al, 2011) can preserve or even improve performance, as indexed with between-subject multivariate pattern classification
With regularized Canonical Correlation Analysis (CCA) we find that a small relaxation of the orthogonality constraint enhances performance, as indexed with between-subject multivariate pattern classification
Summary
Information encoded in cortex is organized at varying spatial scales. The introduction of multivariate pattern analysis of fMRI data (MVPA; Haxby, 2012; Haxby et al, 2014; Haxby et al, 2011; Haxby et al, 2001) revealed that information can be decoded from fine-grained patterns of cortical activity (see Bzdok, 2017; Haynes, 2015; Haynes and Rees, 2006; Hebart and Baker, 2018; Norman et al, 2006; Pereira et al, 2009; Tong and Pratte, 2012). The vector geometry of information in cortical spaces is highly similar across brains (Connolly et al, 2016; Connolly et al, 2012; Connolly et al, 2011; Diedrichsen and Kriegeskorte, 2017; Guntupalli et al, 2018; Guntupalli et al, 2016; Kriegeskorte and Kievit, 2013; Nastase et al, 2017) Prior to hyperalignment, these functional pattern vectors are instantiated in the spatially organized cortical topography of a given individual. The goal of hyperalignment is to create a common information space in which the vectors that encode information that is shared across brains – representations of the same stimulus or cognitive process, functional connections to other cortical fields – are aligned. The individual transformation matrices afford projecting individual data embedded in idiosyncratic cortical topographies into the common space and, projecting group data embedded in the common space into the cortical topographies of individual brains
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
