Decision & reward Bechara, Damasio, Damasio & Anderson, 1994

Iowa Gambling Task

Pick cards from four decks to win play money. Two decks pay more per card but lose in the long run, and most people work out which ones without being told.

Measures: Learning to choose under uncertainty
Template length: about 6 minutes (60 picks)

You start with $2,000 of play money and choose cards from four decks, A to D. Every card wins something, and some also lose something. Nobody tells you which decks are better: you have to learn it from the outcomes.

Decks A and B pay more per card but lose more over time; C and D pay less but come out ahead. The task was designed to show that patients with damage to the ventromedial prefrontal cortex keep choosing the risky decks, even after they can say which are better.

Try it yourself

A shortened run of the same template researchers use. Nothing is recorded and you do not need an account.

How the task works

1Click one of the four decks, A to D, to draw a card.
2The win, any loss and your new total are shown for 1.2 seconds.
3Keep drawing until the task ends. Try to finish with as much money as you can.

What people usually find

Over the course of the task most healthy participants shift towards the good decks, C and D, but learning is slower and more variable than often assumed. Many people keep choosing deck B, which in the original task gives large losses only rarely.

How to score it

Net score: (C + D) − (A + B) picks, computed per block of 20 picks to show learning.
Final balance is intuitive but noisy; the net score per block is the standard measure.
Computational models (such as expectancy-valence or prospect-valence learning) separate learning rate from sensitivity to losses.

Before you run it

The template keeps the original expected values (A and B lose $25 per card on average, C and D gain $25) but uses a simpler loss schedule: A and B are identical, as are C and D. The original contrasts frequent small losses with rare large ones; ask for the original schedule if that matters.
The original task uses 100 picks. The template uses 60, which may be too few for slower learners.

Run it as a study

Use the template as it is

The demo above is the same code as the Iowa Gambling Task template in the gallery. Add it to a project, change the settings in config.js, publish it and share the link. Reaction times and responses come back to your dashboard.

Create a free account to use it

Or build a variant: use the original deck schedule

Paste this into the AI panel in the experiment editor. Read the code it writes before you collect anything.

Build the Iowa Gambling Task with 100 picks and four decks labelled A to D, chosen by clicking a deck button, starting with $2,000. Deck A wins $100 per card with frequent losses (5 in every 10 cards, totalling $1,250); deck B wins $100 with one $1,250 loss in every 10 cards; deck C wins $50 with 5 losses totalling $250 in every 10 cards; deck D wins $50 with one $250 loss in every 10 cards. Use a fixed, pre-shuffled loss sequence for each deck so every participant sees the same schedule. Show the win, the loss and the running total after each pick, and record deck, win, loss and total.

Common questions

What does the Iowa Gambling Task measure?

Decision making under uncertainty: whether people learn from wins and losses to prefer options that pay off over time, rather than ones with bigger immediate rewards.

How is the Iowa Gambling Task scored?

Usually by the net score, good-deck picks (C + D) minus bad-deck picks (A + B), in blocks of 20, which shows whether choices improve over the task.

Is real money used?

In most studies the money is fictional. Some studies pay a small bonus tied to the final balance to increase motivation, which you can set up with study payments.

References

Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition, 50(1–3), 7–15.
Steingroever, H., Wetzels, R., Horstmann, A., Neumann, J., & Wagenmakers, E.-J. (2013). Performance of healthy participants on the Iowa Gambling Task. Psychological Assessment, 25(1), 180–193.