English

Four-Deck Card Game

Four decks, hidden rules. Some decks win big and lose more; others win less and lose less. Can you tell them apart? About 5 minutes, free, no sign-up. We save scores, with your IP address and country, to build comparisons.

By the Helibrain teamReviewed by Robert, biophysics PhD candidatePublished Last reviewed

Pick cards, win points

You start with 2,000 points. Pick a card from any of the four decks. Every card wins points, and some cards also take points away.

Some decks are better than others. Find out which, and finish with as many points as you can. The game ends after a set number of picks.

How to read your result

  • Net score: picks from the two good decks minus picks from the two bad decks, over all 100 picks. It runs from −100 to +100; above 0 means you favoured the decks that pay off over time. This is the usual score of the Iowa Gambling Task [2].
  • Learning curve: the same score for each block of 20 picks. Rising bars mean you worked out which decks pay off as you went.
  • What the decks were: after the game, the result shows which deck was which. Their places on screen are shuffled for every game, so this page can't give the answer away.
  • Compared with other players: once at least 30 people have played, the result shows what share of players you beat.

What is a good score on the Iowa Gambling Task?

A net score above 0 means you picked the good decks more often than the bad ones. Many healthy adults score lower than you might think: in 330 healthy adults from 4 studies of this version of the task, the average net score over 100 picks was 2.5, and about half ended below 0 [2][3].

Quick facts

  • What it measures: decisions under uncertainty: learning from wins and losses which choices pay off in the long run, when the rules are never explained [1].
  • Where it comes from: Antoine Bechara and colleagues at the University of Iowa published it in 1994. People with damage to part of the frontal lobe (the ventromedial prefrontal cortex) kept choosing the bad decks, while most healthy people learned to avoid them [1].
  • The trap: decks with big wins also have the big losses. Decks with rare losses are popular even when they lose money: across 602 healthy adults in an open data pool, deck B (big wins, rare huge losses) was picked most often [2][3].
  • Learning is slow and varied: in the same data, the average net score rose from about −3 in the first 20 picks to about +4 in the last 20, and people differed a lot [3].
Bar chart of what each kind of deck pays per 10 cards. Decks A and B win 1,000 points and lose 1,250 (A in five losses, B in one): net minus 250. Decks C and D win 500 and lose 250 (C in five losses, D in one): net plus 250.
Figure 1. Wins and losses per 10 cards of each kind of deck in this game, the structure of the 1994 task [1]. The kinds are shuffled across the four places on screen in every game.

How the game works

  1. You start with 2,000 points and four decks, numbered 1 to 4.
  2. Pick a card from any deck. Every card wins points; some cards also take points away. The card shows both at once.
  3. Keep picking. The game ends after a set number of picks (100). Finish with as many points as you can.

Each deck has 40 cards; a deck that runs out can't be picked again, as in the original task. Two kinds of deck win 100 points a card but lose 1,250 per 10 cards; the other two win 50 a card and lose 250 per 10 cards (Figure 1). Within every 10 cards, our server shuffles where the losses fall, and it shuffles which kind of deck sits in which place for every game.

This is our own open version with the same structure as the published task, not the licensed Iowa Gambling Task™ sold by its publisher. Its cards, losses and order are our own, drawn fresh for each game.

What affects your score

  • How much you explore. Trying every deck early helps you learn, but each bad card costs points.
  • What you notice. Big, rare losses are easy to forget between them; many people keep returning to deck B for that reason [2].
  • Thinking in the long run. The task rewards the decks that win over many cards, not on the next card.
  • Practice. Once you know the trick, a second game is easier. Compare first games with first games.

What this game can and can't tell you

Your net score shows how quickly you found the better decks in one game. It isn't a measure of intelligence and can't tell you anything about your brain health: healthy adults vary so much on this task that a low score is common [2][3]. It is not a diagnosis.

Our version keeps the published structure, but the scores aren't strictly comparable with lab studies: they differed in how they paid people and in the exact card order [3]. The reference numbers on this page come from our own calculation with the open data of the 330 healthy adults who played versions with the same structure as ours [3].

Frequently asked questions

What is the Iowa Gambling Task?

The Iowa Gambling Task is a card game for studying decisions under uncertainty, published by Bechara and colleagues in 1994. Players pick 100 cards from four decks; two decks give big wins and bigger losses, two give smaller wins and smaller losses. The score is how much someone favours the good decks [1]. The Helibrain four-deck card game is an open version with the same structure.

Which decks are the good ones?

In the Helibrain four-deck card game, the good decks change places every game, so there is no fixed answer. The good decks win less per card (50 points instead of 100) but lose much less, so they come out ahead over every 10 cards. The result screen tells you which deck was which.

Why do so many people score below zero?

Big wins are tempting, and a deck with rare huge losses (like deck B) feels good most of the time. In open data from healthy adults, deck B was the most popular deck and many people never settled on the good decks [2][3]. A negative score is common and says nothing about your health.

Is this the real Iowa Gambling Task?

No. The Iowa Gambling Task is a licensed test sold by its publisher. The Helibrain four-deck card game is our own version with the same published structure: the same wins and losses per 10 cards, but our own card order, shuffled for each game [1]. So its scores are close to, but not the same as, scores from the licensed test.

Is the Helibrain card game really free?

Yes. The Helibrain four-deck card game and your full result are free: no sign-up, no email, no payment. We save results with your IP address and country to build fair comparisons, as our privacy policy explains. Helibrain makes money from its app, which has a free plan and paid plans; you don't need the app to play.

Related tests

See all 37 free brain tests

Want to work on this daily? The Helibrain app has a free plan and paid plans.

Sources

  1. Bechara A, Damasio AR, Damasio H, Anderson SW (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition 50(1–3):7–15. doi.org/10.1016/0010-0277(94)90018-3
  2. 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. doi.org/10.1037/a0029929
  3. Steingroever H, Fridberg DJ, Horstmann A, et al. (2015). Data from 617 healthy participants performing the Iowa Gambling Task: a "many labs" collaboration. Journal of Open Psychology Data 3:e5. doi.org/10.5334/jopd.ak (data: osf.io/8t7rm, CC BY 4.0; the per-block numbers on this page are our own calculation from these data)

How we made this page: written by the Helibrain team from the studies listed above, and reviewed by Robert (biophysics PhD candidate) on 4 November 2026. We update it when the sources change.