Robust representation of natural images by sparse and variable population of active neurons in visual cortex

Natural scenes sparsely activate neurons in the primary visual cortex (V1). However, how sparsely active neurons robustly represent natural images and how the information is optimally decoded from the representation have not been revealed. We reconstructed natural images from V1 activity in anaesthetized and awake mice. A single natural image was linearly decodable from a surprisingly small number of highly responsive neurons, and an additional use of remaining neurons even degraded the decoding. This representation was achieved by diverse receptive fields (RFs) of the small number of highly responsive neurons. Furthermore, these neurons reliably represented the image across trials, regardless of trial-to-trial response variability. The reliable representation was supported by multiple neurons with overlapping RFs. Based on our results, the diverse, partially overlapping RFs ensure sparse and reliable representation. We propose a new representation scheme in which information is reliably represented while the representing neuronal patterns change across trials and that collecting only the activity of highly responsive neurons is an optimal decoding strategy for the downstream neurons