4

I was taking the Iowa Gambling Task. I examined what happens when the same option is selected on every run. To my surprise I got a pretty good value when choosing options C and D (\$4500) in every run and a pretty bad one ($-500) with options A and B.

Is this intended by the test or is it a bug / feature of my downloaded program?

(I got $4700 when mixing some A's among plenty of C's)

Jeromy Anglim
  • 30,741
  • 11
  • 93
  • 221
draks ...
  • 1,818
  • 1
  • 16
  • 41

1 Answers1

4

In the standard version of the task there are "good" decks (overall positive payoff) and "bad" decks (overall negative payoff), so the optimal strategy is to figure out which ones are the good decks and stick with those. In principle, which deck is which should be randomized, but I don't know if your version implements that. There are also many variations that manipulate amount of gain/loss (e.g., rare big gains vs. consistent small gains) or change the payoffs in the middle, so the optimal strategy can vary.

Dan M.
  • 1,719
  • 11
  • 13
  • But what if I would know the good deck, call it C. Is there a reason for the $4500? – draks ... Nov 06 '12 at 11:22
  • 1
    I'm not sure I understand your question. The key aspect of the task is the relative values of the payoffs, not the absolute values. So (typically) the optimal strategy is to always choose the good deck, whether that gets you $4 or $40 or $4000 is an essentially arbitrary design decision. Depending on how you set up your study, you may want to use big numbers (make it more exciting?) or small numbers (actually pay participants their winnings?), the key piece is still the relative value of the "good" vs. "bad" decks. – Dan M. Nov 06 '12 at 15:13