Steps towards neural decoding of colors. (December 2019)

• Record Type: Journal Article
• Authors: Zaidi, Qasim; Conway, Bevil
• Is Part Of: Current opinion in behavioral sciences. Volume 30(2019)
• Page Start: 169
• Page End: 177
• Publication Date: 2019-12
• Subjects: Psychology -- Periodicals; 150.5
• DOI: 10.1016/j.cobeha.2019.10.011
• Languages: English
• ISSNs: 2352-1546

Abstract

Highlights: We show that colors can be decoded successfully from responses of narrowly tuned Inferior Temporal Cortex (IT) neurons. We show how to efficiently measure narrow 3-D color tuning curves peaking at any point in color space for IT cells. We propose neural circuits that generate narrow-tuned responses around scattered peaks as signals move through the cortex. We show how decoding with a winner-take-all rule from narrow-tuned neurons approximates optimal Bayesian decoding. Abstract : How we see colors is a great mystery, but also a route to understanding how we experience any quale, because color does not exist in the world outside our brains, and is undetectable by other senses. From photoreception to primary visual cortex, the neural encoding and transmission of color signals is well understood, providing a foundation for understanding cortical computations of color appearance. We describe how probabilistic models coupled with fMRI-guided microelectrode recordings from inferior-temporal macaque cortex (IT) could help us understand color decoding: i.e. how appearance is extracted from the neuronal responses evoked by a stimulus. Neurons in IT respond to a narrow range of colors with their peak responses scattered around the color circle. We discuss how intra-cellular processes and cortical circuits could generate such tuning curves, and how they approximate optimal Bayesian decoders in winner-take-all schemes.