Figure 4.

Linear-Nonlinear Models and Response Predictions for IC Neurons Derived from the Baseline Stimulus Distribution

(A and E) The filter describes the stimulus feature that excites a given neuron, whereas the nonlinear input-output function (B and F) describes the sensitivity of the neuron to that feature. Most neurons exhibited largely monophasic filter shapes, such as these two examples, meaning that they were excited by negative deflections from the stimulus mean. (C and G) Recorded (averaged over 90 repeats) and predicted responses for these two neurons to the stimulus sequence shown in (D). A strong correspondence between recorded (thin dark line) and predicted (thick gray line) responses indicates that the linear-nonlinear model can successfully describe the relationship between stimulus and response. (H) Histogram of correlation coefficients between recorded and predicted responses for the whole sample of neurons in our study.