Linear models have limitations for modeling observed neural data. In this chapter, we will examine a simple nonlinear encoding model which can model discrete, non-negative data obtained by counting the number of spikes occurring in a given time bin. Luckily, the nonlinear encoding model introduced here can be fit using a built-in function available in MATLAB®.
exponential function; Poisson distribution; log-linear models; peri-event time histogram
Previously, you used a simple linear encoding model to predict a neuron’s firing rate, obtained by averaging over many trials for a range of stimuli. However, linear models have limitations for modeling ...