Spatial frequency adaptation and contrast gain control

Spatial frequency adaptation is shown to cause a decrease in the exponent of the power law describing contrast increment thresholds. Next we show that spatial frequency adaptation produces almost no threshold elevation at a 30 msec test duration but a normal threshold elevation at 500 msec. Control experiments rule out an explanation in terms of sustained and transient mechanisms. Together these results show that spatial frequency adaptation cannot result from neuronal fatigue. The data can be explained by a contrast gain control network in which unit responses provide feedback signals that divide input contrast. Adaptation is hypothesized to produce an increase in the strength of this divisive feedback via temporary synaptic modification. A simple network model fits the data and also predicts both the magnitude and temporal dependence of the tilt aftereffect.