Cheating on the Brain

By Carl Zimmer | May 2, 2005 5:43 pm

Evolutionary psychologists argue that we can understand the workings of the human mind by investigating how it evolved. Much of their research focuses on the past two million years of hominid evolution, during which our ancestors lived in small bands, eating meat they either scavenged or hunted as well as tubers and other plants they gathered. Living for so long in this arrangement, certain ways of thinking may have been favored by natural selection. Evolutionary psychologists believe that a lot of puzzling features of the human mind make sense if we keep our heritage in mind.

The classic example of these puzzles is known as the Wason Selection Task. People tend to do well on this task if it is presented in one way, and terribly if it is presented another way. You can try it out for yourself.

Version 1:

You are given four cards. Each card has a number on one side and a letter on the other. Indicate only the card or cards you need to turn over to see whether any of these cards violate the following rule: if a card has a D on one side, it has a 3 on the other side.

Wason cards-numbers.001.jpg


Version 2:

Now you’re a bouncer at a bar. You must enforce the rule that if a person is drinking beer, then he must be over 21 years old. The four cards below each represent one customer in your bar. One side shows what the person is drinking, and the other side shows the drinker’s age. Pick only the cards you definitely need to turn over to see if any of these people are breaking the law and need to be thrown out.

Wason cards-drinks.001.jpg


The answer to version one is D and 5. The answer to version two is beer and 17.

If you took these tests, chances are you bombed on version one and got version two right. Studies consistently show that in tests of the first sort, about 25% of people choose the right answer. But 65% of people get test number two right.

This is actually a very weird result. Both tests involve precisely the same logic: If P, then Q. Yet putting this statement in terms of social rules makes it far easier for people to solve than if it is purely descriptive.

Leda Cosmides and John Tooby of the University of California at Santa Barbara have argued that the difference reveals some of our evolutionary history. Small bands of hominids could only hold together if their members obeyed social rules. If people started cheating on one another–taking other people’s gifts of food, for example, without giving gifts of their own–the band might well fall apart. Under these conditions, natural selection produced a cheating detection system in the brain. On the other hand, our hominid ancestors did not live or die based on their performance on abstract logic tests. Rather than being a general-purpose problem-solver, the human brain became adapted to solving the problems that our ancestors regularly faced in life.

The Wason Selection Task has become the center of the debate over evolutionary psychology. Some critics, such as the French psychologist Dan Sperber, claim that Cosmides and Tooby can’t make such strong statements about human reasoning from the Wason Selection Task. Others claim that the brain can’t be sliced up into modules so nicely.

The controversy has taken a very interesting turn now, thanks to brain imaging. A team of Italian psychologists had people lie in an MRI scanner and work their way through a set of puzzles that followed the same line of logic as the ones I presented above. They then compared how the brain responded to the challenges to see if indeed the brain works differently when it is solving problems in terms of social exchange than when the problem is more abstract.

The psychologists didn’t use a conventional Wason Selection Task like the ones above, because they wanted to make the problems as similar as possible, except that one dealt with social exchanges. Brain imaging requires this sort of strict experimental design, because it’s very easy to see differences in brain activity that aren’t actually relevant to the question a scientist wants to answer. For example, if one puzzle just so happens to involve picturing an object, some of the brain’s visual processing may become active. So the researchers told their subjects that the puzzles would involve a hypothetical tribe. A purely descriptive puzzle might require subjects to consider the rule, "If a person cracks walnut shells, then he drinks pond water." The subjects might then see a set of cards that read, "He didn’t drink pond water," "He didn’t crack walnut shells," He cracked walnut shells," and "He drank pond water." The researchers also had their subjects solve puzzles that involved social exchanges. The rule in these cases might be, "If you give me sunflower-seeds, then I give you poppy petals."

The psychologists report the results of the test in a paper in press at the journal Human Brain Mapping (click the html link to get the whole paper for free). The results are fascinating–although the researchers don’t claim to have settled the debate over the cheater module. Both the social exchange and descriptive version of the puzzle activated the same network of regions on the left side of the brain. One region (the angular gyrus) is considered important for semantic tasks. A second region is located near the left temple (the dorsolateral prefrontal cortex). It’s essential for considering many different pieces of information at once. The third region, the medial prefrontal cortex, becomes active when people need to bear in mind a larger goal while they solve the many small problems it poses. Previous studies have shown that the left side of the brain plays a much more important role than the right in reasoning and coming up with explanations for how the world works in general.

Now here’s the kicker: the social exchange version of the problem doesn’t just activate this left-brain network. It also activates the same regions in the right side of the brain. Many studies in which people have thought about social situations have tended to turn on the right side of the brain more than the left, and so in one sense this result isn’t too surprising. But it is surprising when you consider that the descriptive version of the puzzle that only switch on parts of the left side of the brain involved thinking about other people and their actions. You might think that that would be social enough to engage any parts of the brain specializing in social thinking. Apparently not. Only when the puzzle involved rules for social exchanges did the right-brain network come on line.

Is this the cheater module? It’s conceivable that the Italian psychologists tapped into some social brain circuit that isn’t specifically adapted for enforcing social rules, but for some somewhat broader group of social problems. It would be interesting if a test other than the Wason Selection Task could trigger the same left versus left-right patterns. The precise evolutionary forces that shaped this feature of the mind may not be clear yet. But this experiment is an important step towards working out the biology between the strange results of the Wason test. Clearly, our brains throw a lot more neurons at logic problems when they concern our social lives instead of abstractions. Analytic philosophers are made, you could say, but political philosophers are born.

Update: 7:15 pm– I decided to change the first version of the test to avoid ambiguity.

Update: Tuesday, 8:15 am– Some commenters have argued that people do better with the bar version of the puzzle because people have more experience with it than with abstract logic. Actually, many variations of the puzzle have been tested out, and the same results emerge. Notice, for example, that the Italian scientists who did the most recent study put the puzzles in terms of a hypothetical tribe, with which the subjects had no experience at all. Despite this different format, almost precisely the same fraction of the subjects got the different versions write as in more familiar versions of the test, such as the bartending example.

Thanks also to the sharp readers who pointed out that the puzzles need to be If-Then propositions.


Comments (20)

  1. Horse

    This may not be true of Wason tests in general, but in the example test shown above, I’d argue the two questions are not logically equivalent.

    I got the first wrong and the second right, as expected. But I got the first question wrong because I picked up on “and you have to make sure that student files for a class have been properly labeled” and inferred that for a student marking system the grade-number correlations must be exclusive – so for the files to be “properly labelled” only the D grade should have a 3 and a 3 should only be applied to a D grade – so you’d have to check all 4 files.

    Now the final question doesn’t support this inference, saying just “Pick only the card or cards that you have to turn over to see if any of the files violates the rule.” But I’d argue this is in direct conflict with the first statement explaining the task.

    So I don’t know how much the given example demonstrates the supposed evolutionary difference – I think the two problems are not specified identically apart from the social context in one.

  2. Brent M Krupp

    I completely agree with Horse. That first question is horribly specified without any guidance on what we can assume or not (and if we’re to assume nothing, well, I don’t want to assume that). The second question involves concrete real-world details of ages and drinking so we know exactly what we can assume about the situation.

    I’m hoping this is just a poorly explicated Wason Test and not that the Wason Test is crap. =)

  3. Brent M Krupp

    Woah — delete my last comment (or ignore it). Carl updated the post while I was commenting. I like the new wording — so what I said is now moot.

  4. Horse

    Heh, yep – the first question has been changed. I’ll take that as vindication of my original criticism. :)

  5. Perhaps it’s because of finals week that I can’t figure out how you got the answer for the first one. Does not having a “5” on one side automatically mean it doesn’t follow the rule, so you wouldn’t need to turn it over? How is it not “D” (where 3 could or could not be on the other side) and “3” (where D could or could not be on the other side)? 5 outright breaks the rule, does it not?

  6. No, 5 doesn’t break the rule IF it has a letter on the other side that is NOT D (so you have to check if it is D).

  7. phanste

    What if the other side of F has a number that is not 3? Would that not too follow the rule?

  8. Mike Hopkins

    Well I got here after Carl reworded:

    “a card with a D on one side has a 3 on the other side.”

    When I did it I did assume that we were supposed to make sure cards with a 3 had as well, i.e. looking for a 3-D card. Rereading the rule after I read the solution I see that I misinterpreted the instructions by making an unwarrented assumption. (You know about assumptions: they make a you know what of you and me.)

    With the social example, there is nothing to interpret. I know the rules: an underaged person may not legally have ethanol but can have soda. Anyone twenty-one years or older can legally have the booze but does not have to.

    I think it is really clear that this is not a logic problem per se since the actually logic is trivially obvious. It is a communication problem. It is easier to to communicate a problem when people already know the rules. This should not be surprising at all. And this leads to a prediction. If you in subtle manner reworded the second problem so that the adult had to have a beer, people who get the problem wrong in droves.

  9. Jonathan Vause

    I agree with Mike, the Wason test seems completely meaningless. People solve the second version of the test from memory, and get the first one wrong because actually they have to think, abstractly, about it – tricky – and don’t absorb all the information before they answer. The real puzzle is why evolutionary psychologists get so excited about it.

  10. Robert Karls Stonjek

    There is an error in the form of the first question.

    The logic being checked is a simple “If p then q”, but Carl has omitted the IF from the first question.

    The CORRECT answer to the first question is “All four”.

    If the question is asked as Michael Gazzaniga does in “The Mind’s Past” P.168:
    “Determine whether the following rule has any exceptions. If a card has a D on one side, it has a 3 on the other side.” He also stipulates that all cards have a number on one side and a letter on the other side.

    Here is the logic: you only need to check the IF part of the statement ie only if there is an ‘F’ on the letter side of the card do you have to check to see if there is three on the other. The ‘letter’ side of the card may be obscured. A ‘3’ may be coupled with any letter.

    IF you omit to stipulate that all cards have a letter on one side and a number on the other, then any of the four cards may have an ‘F’ on one side, but you don’t need to check the ‘3’ card if the ‘IF’ stipulation is made (ie you must check three of the cards).

    If you omit the ‘IF’ part of the statement then one must check both the ‘3’ and the ‘F’ to make sure they are coupled correctly with the complimentary ‘F’ or ‘3’ and the other cards to make sure that they are not erroneously coupled with the ‘3’ or ‘F’ ie you must check all four cards.

    Carl omitted the ‘IF’ so all four cards must be checked. Unfortunately, Carl has been caught in his own net and has shown that he had not properly understood the underlying logic, therefore the error.

    But this is not a too terrible reflection on Carl as Cosmides and Tooby made a similar error – they omitted the explicit stipulation that all cards must have a number on one side and a letter on the other, so the correct answer to the original question as asked (by them) “all but the ‘3’ card”. See their book ‘The Adapted Mind”, Page 182. The roundabout way that the question is asked does not explicitly state that a number must appear on one side and a letter on the other, so F may appear on the other side of the card showing ‘D’.

    In general, it is important to ask these questions correctly – there just might be someone with greater insight and intellect than yourself watching. An error in the question will result in a correct answer to the erroneous question but not to the classic problem – highly encephalated individuals don’t like their moderately equipped colleagues pointing the finger at their expense!

    Kind regards,
    Robert Karl Stonjek

  11. Chris Ackerman

    Normally I’m a fan of this blog, but Evo Psych and fMRI make for serious danger areas for bad science.

    The Wason card test results might be compelling if they were found in people who were raised in isolation, ie, people who would have had as little practical experience with social exchange as most of us would with arbitrary logic puzzles. Then any differences would be attributable to genetically specified brain wiring. The happy rareness of such individuals means the tests must be conducted with normal people, who have had to solve reciprocal social exchange problems numerous times per day, every day of their lives. What’s more, these problems will vary widely in their specifics, so a generalized representation will be easy to form, so as to map quickly to new problems. Nor is it required that this social exchange knowledge is represented in the mind in logical form, as the nativists seem to assume. Research studies (and informal observation) show that untrained people are quite bad at logic, so perhaps the knowledge is more naturally represented as learned associative pairings (I give-I don’t get-danger!) where the activation of one part activates the others, drawing attention to the potential conflict.

    The brain study adds nothing to all this nothingness; if there are differences in cognitive performance as displayed in the behavioral tests, how else might this be manifested but in differences in activity in some brain areas? And if you hold everything constant but the social exchange aspect, where else but in areas of the brain already known to be involved in representing social interactions would you expect to find this differentiated activity?

    BTW, I got both problems right; maybe that’s because I have the logic puzzle-solving gene, or maybe it’s because I’ve seen these types of puzzles before.

  12. Robert Karls Stonjek

    Thanks to Carl Zimmer for updating the question with an ‘IF’ as I recommended and for acknowledging the minor error in an Email to me :)


  13. This is a classic case of logic – that is: A -> B does not imply B -> A. I knew that, and still bombed on the first test. Had I given it thought, I’m sure I could’ve solved it correctly, but the whole idea is to trick you into thinking you know it.

    I did fine on the second test, and I believe that, yes indeed, the reason people perform the way they do on the two tests is because we bring a scripted form of reasoning to the second one based on real world experience. In a way, it follows what Roger Schank has proposed in his book Dynamic Memory. He claims that we mostly solve problems through being reminded of previous similar circumstances, as opposed to truly working out all the gory details from first principles everytime. And when we encounter new situations, we remeber the exceptions of those new situations as compared to what we were most reminded of. Truly, we get through the day by applying our past experiences, rather than thinking too hard.

  14. Lawrie Hunter

    Hold on a second here: In version 1, ALL the cards need to be investigated: i.e. if the F card has a 3 on the back, it violates the rule. If the 5 card has a D on the back, it violates the rule.

    em tasol


  15. Nic

    Close, Lawrie, but not quite “write”. You need to leave F out of that since the first assumption is not asked to be questioned. There IS a letter on one side and not the other.

    F cannot possibly have a D on the other side.


  16. Nic

    D’oh! Answer was correct as given. For the 3,it doesn’t matter what’s on the other side. Any letter will fit the rules, as you all probably already knew.

  17. In order to properly explain the premise of Test-1, you have to spend a fair amount of time explaining the assumptions. Even someone comfortable with logic and “letter” -> “number” notation has to read the problem and think about it. The symbols just don’t differentiate themselves.

    In Test-2, the real-world context bypasses all the need to explain the assumptions and the symbols. With no thought whatsoever, the symbols “drinks” and “ages” are easily differentiated.

    Test-2 has the benefit of capitalizing on a pre-existing understanding that makes it easy to instantly absorb the task at hand. Test-1 does not. When the rules of the game map directly to preexisting experience, no time is needed to explain them.

    It’s not obvious to me that that’s evolution speaking.

  18. 1) Anyone with a grounding in logic would recognize that “if D, then 3” is logically equivalent to “if not 3, then not D”. This immediately suggests that you turn over all letter cards with a D on them as well as all number cards without a 3 on them.

    2) As pointed out by Bryan, many people think that if A implies B, then B implies A. (They may also confuse “if” with “only if”, “implies” with “implied by”, and “necessary condition” with “sufficient condition”.) I had to disabuse a graduate student in physics of this misconception not long ago; I had expected better of someone holding a physics degree.

    3) I, too, must agree with Mike Hopkins. Version 2 of the test concords with common practice in the US, and I suspect that at least some of those who answer this question correctly are merely falling back on their own experience. If the rule for version 2 was modified somewhat so that it deviated from common experience—say, “if a person is drinking soda, then he or she must be under 21 years of age”—I suspect that a smaller percentage would get it right (see previous point).

    4) Rather than just having the respondents give their answers, wouldn’t it be more instructive to ask them to explain how they arrived at their answers? (Multiple choice questions are such poor gauges of a person’s line of reasoning.) Of course, not asking the respondents to explain their reasoning gives the testers greater latitude to arrive at spurious and fallacious conclusions.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

The Loom

A blog about life, past and future. Written by DISCOVER contributing editor and columnist Carl Zimmer.

About Carl Zimmer

Carl Zimmer writes about science regularly for The New York Times and magazines such as DISCOVER, which also hosts his blog, The LoomHe is the author of 12 books, the most recent of which is Science Ink: Tattoos of the Science Obsessed.


See More

Collapse bottom bar