What this measures
Four decks. Draw 24 cards, in any order you like, and finish with as many points as you can. Some decks pay well and punish hard; others pay less and stay steady. Nothing tells you which is which — you have to work it out from the outcomes.
This is the Iowa Gambling Task, developed by Bechara, Damasio and colleagues in 1994. It was built to study decision-making in patients with damage to the ventromedial prefrontal cortex, who scored normally on IQ tests yet made catastrophic real-life decisions — and the task caught what the IQ tests missed.
How the decks are built
Two decks are net positive over time and two are net negative, by roughly equal margins. The trap is that the bad decks give larger immediate rewards and hide their losses in occasional heavy penalties.
So the deck that feels best early is the one that ruins you, and detecting this requires integrating outcomes across many draws rather than reacting to the last one. That gap — between what feels rewarding and what actually pays — is the whole design.
What good play looks like
- Sample all four early. Spend your first several draws gathering information. Committing to a deck in the first five cards is the most common failure.
- Track net, not gross. A deck paying 100 a card while occasionally taking 350 is worse than one paying 50 with small penalties. Keep a rough running total per deck.
- Do not flee after one big loss. A single penalty says almost nothing. Two heavy penalties in a short span from the same deck is real evidence.
- Once you have evidence, commit. Continuing to "check" a known bad deck is expensive curiosity.
The finding that made this task famous
In the original studies, healthy participants began avoiding the bad decks before they could explain why — their skin conductance rose when reaching toward a bad deck while they still reported having no idea what was going on. The patients showed neither the physiological signal nor the behavioural shift, and kept choosing badly even after they could verbally state which decks were bad.
That result launched the somatic marker hypothesis — the idea that bodily signals guide decisions ahead of conscious reasoning. It has been argued about ever since, and the methodology has drawn substantial criticism, but the core observation is robust and worth knowing: you will probably start avoiding the bad decks before you can say why.
Frequently asked
Q. Is there a way to know which decks are good without losing anything?
A. No. The information cost is built in — that is the task.
Q. I ended negative. Is that meaningful?
A. In 24 draws, variance is large. One run tells you little; a consistent pattern across
several tells you more.
Q. Does this measure impulsivity?
A. Partly, but it also measures how well you integrate outcomes over time. Two people can
fail it for opposite reasons — one by chasing rewards, another by over-reacting to
individual losses.
For risk-taking without a learning component, try the balloon task.