This toolbox contains the code necessary to perform the method described in the paper: "Time-domain analysis for extracting fast-paced pupil responses", currently under revision in Scientific Report. It also includes other functions that are useful for pupil data analysis.

loadWithPupil.m can be used to load data from experiments using either the eyelink or the pupil headset from PupilLabs as eyetrackers, and COSYgraphics as a toolbox for stimulus presentation and synchronization of the behavioural paradigm with the eyetrackers.

pupilARX.m implements the main method of applying an ARX system identification model to the pupil size data.

averageImpulseResponse.m takes a series of ARX models and computes their averaged impulse response while taking into account their individual variability.

pupilTrialResponses.m extracts pupillary responses from individual trials on the basis of previously fitted ARX model.

fastSmooth.m smooths a vector by applying sliding-window averaging. Runs faster than the smooth builtin function.

downsampleVector.m downsamples a vector by an integer ratio by taking the mean of the successive time bins.

loadData.m loads data acquired with Eyelink and psychophysics toolbox on Matlab and synchronizes pupil and behavioural data.

filterPupil.m applies high-pass filters to pupillary data. You should choose a cut-off freq that's around 2 to 3 times slower than the trial frequency.