Figure 5. Identifying the shape of the neural input to the pupil diameter system during decision formation.

(a) Schematic depicting the analysis approach employed to parse the contributions of distinct temporal components to the observed pupil dilation waveforms and identify the likely shape of the input to the pupil system during decision formation. Three temporal components (onset, sustained, response) were convolved with a canonical pupil impulse response function (top left) and regressed onto the observed pupil diameter time series. The shape of the sustained decisional component differed across 8 candidate models and their goodness-of-fits were quantitatively compared. (b,c) BIC scores for each of the 8 models fit to the FR (b) and DL (c) data, relative to the winning model in each case (model 1 for FR; model 2 for DL). (d,e) Effect sizes of parameter estimates for each of the temporal components from the winning FR (d) and DL (e) models. Error bars=s.e.m. ***=P<0.001, **=P<0.01, *=P<0.05.