Saccade Cookbook

From biophysics
Jump to navigation Jump to search

A tutorial, for students and coworkers: a recipe to calibrate, detect, and analyze saccades


Easily calibrate, detect, and analyze saccades with PandA. For a more detailed description, visit the other broken link sections, starting with the broken link.

Calibration

First you have to calibrate (see also saccade [[1]]) by training an artificial neural network to learn the relationship between measured voltages and head orientation:

pa_calibrate

(choose the calibration dat-file in the pop-up menu, usually ending in 0000). Then you can calibrate (and low-pass filter) the data-files, by using the button in the pa_calibrate interface (this will calibrate ALL dat-files with the .net-file you got from the pa_calibrate-procedure).

Detection

Then you have to detect (see also saccade [[2]]) with:

pa_sacdet

(choose an experimental hv-file, usually ending with numbers 0001 and up).

Parameters

Obtain all movement parameters, such as end-point location, reaction time, peak velocity.

pa_sac2mat

(these will be saved in a mat-file)

Analysis

Finally, you can start some high-level analysis (see also saccade [[3]])

After all these obligatory steps, you can actually start analyzing the data. This usually involves custom-made analysis functions, suited for your experiment. First, you have to load the data:

load MW-RG-2011-03-02-0001

You will now have a Sac- and a Stim-matrix. The Sac-matrix contains all the movement parameters, the Stim-matrix all the stimulus parameters. You can combine them into a single matrix, containing in each column a relevant parameter, and each row containing a single saccade.

SupSac = pa_supersac(Sac,Stim,XX,YY);

Note that the XX and YY should be numbers describing the type of stimulus, and the number of that type of stimulus in the trial, e.g.

SupSac = pa_supersac(Sac,Stim,1,2);

for the second (YY=2) stimulus of type 1 (XX=1, usually an LED) in a trial, or

SupSac = pa_supersac(Sac,Stim,2,1);

for the first (YY=1) stimulus of type 2 (XX=2, usually a sound) in the trial (you have to check what number is assigned to which stimulus type, by typing:

help pa_readcsv

in the trial information of the LOG-matrix, line 5).

To plot target location versus response location:

plot(SupSac(:,23),SupSac(:,8),'k.');

or

pa_plotloc(SupSac);