Modulatory feedback determines attentional object segmentation in a model of the ventral stream

Abstract

AbstractStudies in neuroscience inspired progress in the design of artificial neural networks (ANNs), and, vice versa, ANNs provide new insights into the functioning of brain circuits. So far, the focus has been on how ANNs can help to explain the tuning of neurons at various stages of the visual cortical hierarchy. However, the role of modulatory feedback connections, which play a role in attention and perceptual organization, has not been resolved yet. The present study presents a biologically plausible neural network that performs scene segmentation and can shift attention using modulatory feedback connections from higher to lower brain areas. The model replicates several neurophysiological signatures of recurrent processing. Specifically, figural regions elicit more activity in model units than background regions. The modulation of activity by figure and ground occurs at a delay after the first feedforward response, because it depends on a loop through the higher model areas. Importantly, the figural response enhancement is enhanced by object-based attention, which stays focused on the figural regions and does not spill over to the adjacent background, just as is observed in the visual cortex. Our results indicate how progress in artificial intelligence can be used to garner insight into the recurrent cortical processing for scene segmentation and object-based attention.Author SummaryRecent feedforward networks in artificial intelligence provide unmatched models of tuning of neurons in the visual cortex. However, these feedforward models do not explain the influences of object-based attention and image segmentation on neuronal responses, which rely on feedback interactions between cortical regions that are not included in the feedforward networks. In particular, the role of feedback connections from higher brain regions that modulate neural activity in lower cortical regions has not yet been studied extensively so that we still lack anin silicomodel of the role of these connections. Here, we present a biologically plausible neural network that successfully performs image segmentation and can shift object-based attention using modulatory feedback connections. The model evolved representations that mirror the properties of neurons in the visual cortex, including orientation tuning, shape-selectivity, surround suppression and a sensitivity to figure-ground organization, while trained only on a segmentation task. The new model provides insight into how the perception of coherent objects can emerge from the interaction between lower and higher visual cortical areas.