Holyoak Pilot2 Specs

programmer: Grant

Categorization Task

Stimuli: Each stimulus is a 10X10 grid of red or black squares against white background. Should measure about 1 cm per square (total size 10 cm X 10 cm).

Categories: Stimuli fall into 3 categories, designated A, B, and C. For each category, at the very beginning separately generate a prototype (same prototype is used for every subject). To generate a prototype, simply assign each of the 100 squares to be either black or red (p=0.5, independently). Do this only one time when you make the program, rather than at run time. So now you have prototype A, prototype B, and prototype C and those are fixed across the whole experiment. (Save them because we may use them in followup experiments.)

Generating stimuli: All stimuli are generated by copying the prototype and then randomly perturbing the stimulus as follows:

1. Determine whether the stimulus should be A, B, or C (independent prob=1/3 for each).

2. Now create a stimulus as follows:

for x = 1 to 100;

with probability perturb_prob flip the square from its color in the prototype to the other value red->black or black->red;

end;

Thus, each square within the stimulus is the same as its prototype with a probability (1-perturb_prob). To start with, let’s set perturb_prob to .25. This is a pilot to determine if that number works well.

Procedure for an individual categorization trial:

1. select stimulus

2. present plus sign in center for 500 msec

3. present stimulus until response (response made by clicking on box A, B, or C).

4. present feedback (confirmation tone if right; error tone if wrong together with an asterisk over the correct answer).

5. Pause 1 second before next trial.

Trials.

This is a pilot experiment. We just want to have 6 subjects do this experiment for 300 trials with feedback and see how they do. **There were actually 299 trials according to the data files.

Data File

One row per trial:

<Trial#; What prototype was used, A, B, or C; Subject’s response; Correct vs. Wrong>