Neural representation of object category and viewpoint in the entopallium of pigeons.

Object recognition depends on the ability of the visual system to preserve stable category assignment despite variation in viewpoint. Although the mechanisms encoding object category and viewpoint have been extensively described in the primate ventral visual pathway, it remains unclear how the avian brain, which does not contain a layered cortical structure, carries out similarly complex visual computations. Here, we systematically investigated how neurons in the pigeon entopallium (ENTO), the terminal station of the tectofugal visual pathway, represent object identity and viewpoint. Large-scale electrophysiological recordings showed that ENTO neurons displayed moderate category selectivity and strong continuity in viewpoint tuning, and a subset of neurons expressed both tuning properties. At the population level, neural responses formed organized representational manifolds that supported categorical separation and viewpoint continuity at the same time. Moreover, ENTO neurons were strongly sensitive to color, and subpopulation analyses indicated that object representations in the ENTO were not confined to one processing level but jointly included low-level visual features (e.g., color and shape) and higher-level semantic information. These results suggest that the avian visual system can construct complex object representations. During this process, the ENTO may establish functionally specific neural representations through distributed coding based on sparse combinations of distinct neuronal subpopulations.