Dual Computational Systems in the Development and Evolution of Mammalian Brains

Abstract

Analyses of brain sizes across mammalian taxonomic groups reveal a consistent pattern of covariation between major brain components, including a robust inverse covariation between the limbic system and the neocortex. To find the functional basis of this inverse relationship, we mapped the multidimensional representations of task-optimized machine learning systems onto two-dimensional surfaces resembling the forebrain cortices. We found that visual, somatosensory and auditory systems develop ordered spatiotopic maps where units draw information from localized regions of the sensory input. Olfactory and relational memory systems, in contrast, develop fractured maps with distributed patterns of information convergence. Evolutionary optimization of multimodal systems for varying task objectives results in inverse covariation between spatiotopic and disordered system components that compete for representational space. These results suggest that the observed pattern of covariation between brain components reflect an essential computational duality in brain evolution.