Re: [Pebl-list] WCST

[Shane M. <smu...@ob...>]

Check out the latest version in CVS.  That should fix the trial number
error and the perseveration error:
http://pebl.cvs.sourceforge.net/*checkout*/pebl/pebl/battery/bcst/cardsort.pbl?revision=1.7

> Dear Shane,
>
> I have a couple of questions regarding the Wisconsin Card Sorting test.
> I have noticed that that the three rules (shape, color and number) are
> presented a different number of times each time the task is run. Could
> the program be made to present each of them the same number of times
> (i.e., 3) in an experiment (in randomized order, of course)?

I believe this was designed as described in Berg, 1954.  The first rule is
sampled randomly.  After that, the rule is sampled according to the
function
 rule <- SampleNewRule(rule)
on line 199, which makes sure that the rule does in fact change, but does
not require a set distribution of rules.

It is fairly simple to modify to produce the behavior you want.  First,
you would set up the rule-change sequence, using the
ShuffleWithoutAdjacents function, somewhere at the beginning of the
script:

rules <- ShuffleWithoutAdjacents([[1,1,1],[2,2,2],[3,3,3])

Then, instead of sampling randomly (i.e., rule <- RandomDiscrete(3)), 
before the beginning of the loop at line 145,  do:

rule <- First(rules)
rules <- Rotate(rules,1)


and similarly, around line 197 replace

rule <- SampleNewRule(rule)
with:
rule <- First(rules)
rules <- Rotate(rules,1)

If you change the experiment to go longer than 9 runs, it will just repeat
the same sequence because we rotated the deck.  You can re-shuffle if need
be.

These changes are simple enough that I think I'll make them and update CVS.

> I have also noticed a couple of problems in the report file:
>
> - the number of trials appearing in the report file is always 1 unit
> greater than the actual number of trials presented (according to the
> output file).
This should be fixed.


> - the number of perseverations is always "0" no matter how many
> responses one gives according to a no longer valid rule.
This should be fixed, although the proper way to define perseveration is
quite complicated, and you should do it yourself using the raw output
data.