Distributed Representations for Cognitive Control in Frontal Medial Cortex

Abstract

In natural and artificial neural networks, modularity and distributed structure afford complementary but competing benefits. The former allows for hierarchical representations that can flexibly recombine modules to address novel problems, whereas the latter affords better generalization. Here we investigate these competing demands in the context of sequential behavior. First, we explore this by comparing the properties of several recurrent neural network models. We find that explicit hierarchical structure fails to provide an advantage for generalization above a "flat" model that does not incorporate hierarchical structure. However, hierarchy appears to facilitate cognitive control processes that support non-routine behaviors and behaviors that are carried out under computational stress. Second, we compare these models against functional magnetic resonance imaging (fMRI) data using representational similarity analysis. We find that a model that incorporates so-called wiring costs in the cost function, which produces a hierarchically-organized gradient of representational structure across the hidden layer of the neural network, best accounts for fMRI data collected from human participants in a previous study (Holroyd et al., 2018). The results reveal that the anterior cingulate cortex (ACC) encodes distributed representations of sequential task context along a rostro-caudal gradient of abstraction: rostral ACC encodes relatively abstract and temporally extended patterns of activity compared to those encoded by caudal ACC. These results provide insight into the role of ACC in motivation and cognitive control.