Simulations support the simple hypothesis that persistent coupling of electrochemical activity in recurrent network neurons is an objective signature of visual object unity

Existing evidence supports the proposal that visual object unity can be identified with an epsilon-ball of recurrent synaptic conductance vectors (subsequently referred to as a connected open set). Simulations of a small retina and recurrent network demonstrate that such connected open sets persist despite temporal fluctuations in participating neurons, mimic the visual phenomena of grouping by proximity and object constancy over changes of image size and orientation, and behave similarly to human fMRI data from primary visual cortex during real and apparent motion. Simulations also demonstrate the nonlinear phenomenon of hysteresis in the appearance and disappearance of connected open sets, and human participants were subsequently found to demonstrate hysteresis in the perception of visual object unity. This report provides a physical foundation for these phenomena in the persistent coupling of differential equations that describe the behavior of electrochemical quantities in neurons. It also extends earlier work by improving temporal resolution, by employing clusters of neurons to promote redundancy, and by demonstrating that information on object unity is present in the frequencies of action potentials that can be transmitted to other networks by adjacent recurrent network neurons. Implications for the identification of objective signatures of aspects of experience in formal neural network models and applications to visual-cortical prostheses are discussed.