2013 January 01 | Matteo Carandini and David J. Heeger
The article discusses the role of normalization as a canonical neural computation across various sensory systems and brain regions. Normalization involves dividing a neuron's response by a common factor, typically the summed activity of a pool of neurons. This computation was initially developed to explain responses in primary visual cortex but has since been found to operate in other sensory modalities and brain areas, including the olfactory system, retina, primary visual cortex (V1), higher visual areas, and non-visual cortical areas. Normalization helps explain phenomena such as response saturation, cross-orientation suppression, surround suppression, and winner-take-all competition. The authors suggest that normalization is a fundamental computational process that may underlie a wide range of neural functions, including sensory processing, attention, and multisensory integration. They also explore the underlying circuits and mechanisms of normalization, including the role of GABA-mediated inhibition and non-linearities in afferent inputs. Finally, they discuss the potential implications of normalization for understanding and treating neurological and psychiatric disorders.The article discusses the role of normalization as a canonical neural computation across various sensory systems and brain regions. Normalization involves dividing a neuron's response by a common factor, typically the summed activity of a pool of neurons. This computation was initially developed to explain responses in primary visual cortex but has since been found to operate in other sensory modalities and brain areas, including the olfactory system, retina, primary visual cortex (V1), higher visual areas, and non-visual cortical areas. Normalization helps explain phenomena such as response saturation, cross-orientation suppression, surround suppression, and winner-take-all competition. The authors suggest that normalization is a fundamental computational process that may underlie a wide range of neural functions, including sensory processing, attention, and multisensory integration. They also explore the underlying circuits and mechanisms of normalization, including the role of GABA-mediated inhibition and non-linearities in afferent inputs. Finally, they discuss the potential implications of normalization for understanding and treating neurological and psychiatric disorders.