Properties of Input-Output Transformation in Sensory Systems
Neurons and neuronal systems in sensory systems can encode information with either a shallow function that extends through the entire possible range of inputs or one that is limited to only a subset of these inputs. If the input-output function extends through the entire range, its slope will necessarily be shallow, and therefore the neurons will be less sensitive to the input. However, if the function is limited to a particular range of inputs, then its slope will be steep and the neuron's firing rate will very sensitive to small changes in the input (thus giving rise to better signal detection). The role of adaptation is to move this steep function to the range in which it is needed.
In the visual system, the firing rate of cortical neurons is steeply determined by the level of contrast. Contrast adaptation moves this steep input-output relation to center around the range of average contrast, thereby making visual encoding more efficient.