Normalization is a widespread neural computation that operates across various sensory systems and brain regions. It involves dividing neuronal responses by a common factor, often the summed activity of a neuronal pool. This computation is thought to underlie processes such as odor representation, visual attention modulation, value encoding, and multisensory integration. Evidence suggests that normalization is a canonical neural computation, present in diverse systems from invertebrates to mammals. In the invertebrate olfactory system, normalization explains how mask odorants suppress responses to test odorants. In the retina, normalization adjusts sensitivity to light intensity, enabling responses that represent contrast rather than absolute intensity. In primary visual cortex (V1), normalization explains saturation of responses with increasing stimulus contrast and cross-orientation suppression. It also underlies surround suppression and winner-take-all competition in response to multiple stimuli. Normalization is also observed in other cortical areas, such as MT, where it helps in motion selectivity. In the ventral visual pathway, normalization plays a role in object recognition through non-linear pooling. In auditory systems, normalization is evident in primary auditory cortex (A1), where responses show logarithmic dependence on sound intensity and are affected by background noise. In multisensory integration, normalization explains how neurons weigh different sensory modalities based on their strength. Normalization is also involved in encoding the value of actions, such as in the lateral intraparietal area (LIP) of the macaque monkey, where responses depend on the relative value of alternatives. Attention modulates normalization by enhancing stimulus drive before normalization, leading to stronger responses to attended stimuli. The mechanisms underlying normalization vary across systems. In the fruitfly, it may involve GABA-mediated inhibition, while in mammalian V1, it is less dependent on GABA inhibition. Other mechanisms include synaptic depression and fluctuations in membrane potential. Normalization can also be achieved through feedback circuits, with different timing and spatial characteristics depending on the region. Behavioral studies support normalization, showing effects similar to those observed in neural systems. Deficits in normalization are linked to disorders such as amblyopia, epilepsy, depression, and schizophrenia. Understanding normalization as a canonical computation could provide insights into brain function and disorders. It is a modular computation that repeats across brain systems and is essential for various neural processes.Normalization is a widespread neural computation that operates across various sensory systems and brain regions. It involves dividing neuronal responses by a common factor, often the summed activity of a neuronal pool. This computation is thought to underlie processes such as odor representation, visual attention modulation, value encoding, and multisensory integration. Evidence suggests that normalization is a canonical neural computation, present in diverse systems from invertebrates to mammals. In the invertebrate olfactory system, normalization explains how mask odorants suppress responses to test odorants. In the retina, normalization adjusts sensitivity to light intensity, enabling responses that represent contrast rather than absolute intensity. In primary visual cortex (V1), normalization explains saturation of responses with increasing stimulus contrast and cross-orientation suppression. It also underlies surround suppression and winner-take-all competition in response to multiple stimuli. Normalization is also observed in other cortical areas, such as MT, where it helps in motion selectivity. In the ventral visual pathway, normalization plays a role in object recognition through non-linear pooling. In auditory systems, normalization is evident in primary auditory cortex (A1), where responses show logarithmic dependence on sound intensity and are affected by background noise. In multisensory integration, normalization explains how neurons weigh different sensory modalities based on their strength. Normalization is also involved in encoding the value of actions, such as in the lateral intraparietal area (LIP) of the macaque monkey, where responses depend on the relative value of alternatives. Attention modulates normalization by enhancing stimulus drive before normalization, leading to stronger responses to attended stimuli. The mechanisms underlying normalization vary across systems. In the fruitfly, it may involve GABA-mediated inhibition, while in mammalian V1, it is less dependent on GABA inhibition. Other mechanisms include synaptic depression and fluctuations in membrane potential. Normalization can also be achieved through feedback circuits, with different timing and spatial characteristics depending on the region. Behavioral studies support normalization, showing effects similar to those observed in neural systems. Deficits in normalization are linked to disorders such as amblyopia, epilepsy, depression, and schizophrenia. Understanding normalization as a canonical computation could provide insights into brain function and disorders. It is a modular computation that repeats across brain systems and is essential for various neural processes.