Dendritic Computation and the Fine Structure of Receptive Fields: A Model of V1 Neurons

Dendrites are now recognized as nonlinear integrative compartments that can implement complex computations within single neurons, yet their specific contribution to cortical response diversity remains incompletely understood. In primary visual cortex (V1), neurons exhibit a broad range of response properties, including simple, complex, and end-stopped selectivity, but the mechanisms by which such diverse receptive field structures arise within a largely homogeneous population of pyramidal cells are still debated. This work presents a computational model in which these response classes emerge from dendritic integration of spatially organized excitatory and inhibitory inputs. In the model, synaptic inputs are distributed across dendritic branches according to their orientation preference and location within an underlying orientation map, and dendritic nonlinearities transform these inputs into somatic responses. By varying the relative distributions of excitatory and inhibitory inputs along dendrites, while keeping synaptic weights fixed, the model generates a continuum of response profiles ranging from phase-sensitive simple-like cells with segregated ON and OFF subfields to phase-invariant complex-like and end-stopped responses. The fine structure of model receptive fields, including the emergence of ON/OFF regions and end-stopping, arises from local differences in the balance and placement of excitatory and inhibitory inputs on dendritic branches. The model also predicts that modest changes in input organization can alter preferred orientation, consistent with reports of orientation plasticity in V1. These results provide a hypothesis-generating framework showing how dendritic input organization can constrain and shape V1 response properties within a single cell type, complementing classical feedforward and pooling-based accounts of these response classes.