Hyperaligning Neural Representational Spaces

Abstract

Neural populations in different brain regions represent different domains of information, but accounting for how population responses in homologous regions in different brains encode the same fine distinctions has been elusive. Common models of cortical functional architectures based on anatomy account for coarse regional topography that encode coarse-scale information such as visual versus auditory stimulation or perception of animate versus inanimate entities but fail to account for fine-scale information that captures distinctions between two songs or two insects. We proposed a method of functional alignment called hyperalignment that aligned high-dimensional neural representational spaces to derive a new common model of cortical functional architecture. This model is based on a common representational space rather than a common cortical topography. By modeling functional topographies as weighted sums of overlapping topographic basis functions, our model also accounts for coarse-scale regional topography and goes further to capture fine-scale topographies that coexist with coarse topographies and carry finer distinctions. In this chapter we present steps for an experimenter to use hyperalignment in their own study to derive a common model representational space and perform analyses in that space.