Object vision articles from across Nature Portfolio

The precise role that inferotemporal cortex plays in object recognition remains poorly understood. Here, the authors combine high-throughput behavioral optogenetics in non-human primates with machine learning to graphically capture perceptual events evoked by local stimulation in the high-level visual cortex.

The phenomenon of blurry or degraded visual input in humans has been overlooked in the training of convolutional neural networks (CNNs). Here, the authors show that blur-trained CNNs outperform standard CNNs in predicting neural responses to objects and show improved correspondence with human perception.