Significance testing of the electrode contributions to the electrical receptive fields (ERFs) and model summary.

<p>A) Three stimulus frames (<i>t</i>₋₂, <i>t</i>₋₁, and <i>t</i>₀) of the true excitatory filter for a cell (black) are compared to a distribution of randomly time-shifted filters (grey). Each stimulus frame contains 20 amplitudes representing the amplitudes of the 20 stimulating electrodes. A 95% confidence limit is calculated from the distribution (dark grey) to which the actual values can be compared. In this example, three electrodes (denoted by stars) from <i>t</i>₀ significantly affected the response of the neuron, and none from <i>t</i>₋₁ or <i>t</i>₋₂. B) A block diagram illustrating the steps of the GQM procedure. An input stimulus, , is linearly filtered and squared (for the case of to ). The outputs are then added or subtracted depending on whether the linear filter is excitatory or suppressive, respectively. An estimate of the response is produced by transforming the result with a sigmoidal nonlinearity.