Experiments
You can work with a MATLAB script, spreadsheet tables, or a combination of the two...
Coding method
In the coding method of making an experiment we use a MATLAB script. That said, it's not really code in the usual sense, more like a list of statements setting parameters. Define each trial by making objects of various types—e.g. picture, sound, keyPress, or more specialized types like linearDotMask, bmlWalker, etc. Any object works anywhere in any experiment. Set properties you want to change from default using dot syntax. You only need to define each distinct trial once, and in any order, so you can use for loops through conditions to do it easily (or multiple for loops in more complex experiments). Finally set a list of distinct trials to run through, adding any repetition and ordering there. PsychBench includes little tools for various kinds of randomization and counterbalancing.
For example, the script below defines three distinct trials. It runs two repetitions of each, all in random order. In each trial a picture of a different shape shows at a height of 8 degrees visual angle and stops showing when the subject responds by any key press. Experiment results report picture file name and response latency for each trial.
<cdcm>% Mark start of experiment script<cdcm>
newExperiment
<cdcm>% Define 3 distinct trials<cdcm>
<cdkm>for<cdkm> fileName = [<cdsm>"sphere.png" "cube.png" "hourglass.png"<cdsm>]
picture = pictureObject;
picture.fileName = fileName;
picture.height = 8;
picture.end.response = true;
picture.report = <cdsm>"fileName"<cdsm>;
recorder = keyPressObject;
recorder.report = <cdsm>"responseLatency"<cdsm>;
addTrial(picture, recorder);
<cdkm>end<cdkm>
<cdcm>% Want 2 repetitions of each distinct trial in random order<cdcm>
nn = randomOrder(rep(1:3, 2));
setTrialList(nn)
<cdcm>% Run experiment<cdcm>
results = runExperiment;
This is just a toy example. Many dozens of properties are available for timing, visual options, response options, options specific to picture and keyPress objects, and more. We can implement any number of conditions of any kind, blocks, intros, breaks, outros, syncs, any kind of ordering, randomization, counterbalancing, etc. Methods like staircases, adjustment, scanners, online experiments, and more are available. For complex experiments you can lay out values in spreadsheet tables and load them into the script. See below for feature highlights.
For more realistic examples, see stroopDemo.m. Or for something more complex wisconsinDemo.m. Or see a list of all demos.
(Without PsychBench)
Just for comparison, here is some compact Psychtoolbox code that would run the same experiment as above, with the same timing precision and error handling. The difference in length and complexity is noticeable here and balloons up for more realistic experiments.
<cdkm>try<cdkm>
<cdcm>% Parameters<cdcm>
fileNames = {<cdsm>'sphere.png' 'cube.png' 'hourglass.png'<cdsm>};
pictureHeight_deg = 8;
screenHeight_cm = 17.9;
screenDistance_cm = 55;
iti = 0.75;
<cdcm>% Open on-screen window<cdcm>
[n_window, windowRect] = Screen(<cdsm>'OpenWindow'<cdsm>, 0, [0 0 0]);
flipInterval = Screen(<cdsm>'GetFlipInterval'<cdsm>, n_window);
<cdcm>% Pause keyboard output to MATLAB command line<cdcm>
ListenChar(2)
<cdcm>% Make 3 textures from pictures<cdcm>
<cdkm>for<cdkm> n_trialDef = 1:3
data = imread(fileNames{n_trialDef});
nn_textures(n_trialDef) = Screen(<cdsm>'MakeTexture'<cdsm>, n_window, data);
sourceRect = RectOfMatrix(data);
sourceHeight = RectHeight(sourceRect);
targetHeight = 2*screenDistance_cm*tand(pictureHeight_deg/2)/screenHeight_cm*RectHeight(windowRect);
r = targetHeight/sourceHeight;
targetRect = CenterRect(ScaleRect(sourceRect, r, r), windowRect);
targetRects(n_trialDef,:) = targetRect; <cdcm>%#ok<*SAGROW><cdcm>
<cdkm>end<cdkm>
<cdcm>% Run 2x3 trials in random order<cdcm>
nn_trialDefs = Shuffle(repmat(1:3, 1, 2));
<cdkm>for<cdkm> n_trial = 1:6
n_trialDef = nn_trialDefs(n_trial);
fileName = fileNames{n_trialDef};
n_texture = nn_textures(n_trialDef);
targetRect = targetRects(n_trialDef,:);
Screen(<cdsm>'DrawTexture'<cdsm>, n_window, n_texture, [], targetRect)
<cdkm>if<cdkm> n_trial == 1
t0 = [];
<cdkm>else<cdkm>
t0 = tf+iti-0.5*flipInterval;
<cdkm>end<cdkm>
t0 = Screen(<cdsm>'Flip'<cdsm>, n_window, t0);
[keyIsDown, t] = KbCheck;
<cdkm>while<cdkm> ~keyIsDown
Screen(<cdsm>'DrawTexture'<cdsm>, n_window, n_texture, [], targetRect)
Screen(<cdsm>'Flip'<cdsm>, n_window);
[keyIsDown, t] = KbCheck;
<cdkm>end<cdkm>
tf = Screen(<cdsm>'Flip'<cdsm>, n_window);
latency = t-t0;
results(n_trial,:) = {fileName latency};
<cdkm>end<cdkm>
<cdcm>%Resume keyboard output to MATLAB command line<cdcm>
ListenChar(0)
<cdcm>% Close on-screen window and textures<cdcm>
Screen(<cdsm>'CloseAll'<cdsm>)
<cdkm>catch<cdkm> X
<cdcm>% Close on-screen window and textures on error<cdcm>
ListenChar(0)
Screen(<cdsm>'CloseAll'<cdsm>)
rethrow(X)
<cdkm>end<cdkm>
Visual method
The full visual method is an option for experiments of low to moderate complexity. The concepts are the same as in the coding method but we lay everything out in tables. You can use any spreadsheet app for this, but Google Sheets allows dropdown menus, quick docs, and starting from examples. When done, save/export to an Excel file, use the command loadExperiment in MATLAB to load it, and runExperiment to run (same as the coding method from that point). For example, the two tables below would make the experiment from above.
