Wednesday, February 16, 2011

The Wason Selection Task and the Limits of Human Philosophical Cognition

The famous Wason Selection Task runs like this: You are to imagine four cards, each with a number on one side and a letter on the other side. Of two cards, you see only the letter side and you don't know what's on the number side. Of the other two cards, you see only the number side and don't know what's on the letter side. Imagine the four cards laid out as follows:

(image from: http://www.psypress.com/groome/figures/)

Here's the question: What card or cards do you need to turn over to test the rule "If there is a K on one side there is a 2 on the other", to see if it is violated? The large majority of undergraduates (and of almost every group tested) gets this question wrong. One common answer: the K and the 2. The correct answer is the K and the 7.

That K and 7 is the correct answer is much more intuitively evident when we put the task in a context of social cognition or cheating detection rather than in an abstract context. Imagine that instead of numbers and letters the cards featured beverages and ages and the rule in question was "If there is an alcoholic beverage on one side, there is an age of 21 or greater on the other". Then the cards would look like this: sprite, gin, 32 years old, 17 years old. The large majority of people will correctly say that to check for violations of the rule you need to turn over the gin card and the 17-year-old card. But that's exactly the same logical structure as the much more difficult letter-number task.

Psychologists have discussed at length the cognitive mechanisms and what makes some versions of the task easier than others, but the lesson I want to draw is metaphilosophical: We humans stink at abstract thought. Logically speaking, the Wason selection task is incredibly simple. But still it's hard, even for well educated people -- and even logic teachers familiar with the task need to stop and think a minute, slow down to get it right; they don't find it intuitive. Compare this to how intuitively we negotiate the immense complexities of language and visual perception.

Nor is it just the Wason selection task we are horribly bad at, but all kinds of abstract reasoning. I used to write questions for the Law School Admissions Test. One formula I had for a difficult question was just a question with several options expressed in ordinary language with a conditional and a negation and no intuitive support. The poor victim of one of my questions would first read a paragraph about the economics of Peru (for example). Then I would ask which of the following must be true if everything in the paragraph is true: (a.) Unless crop subsidies are increased, the exchange rate will decline, (b.) The exchange rate will rise only if crop subsidies are not decreased, (c.) If crop subsidies are decreased, the exchange rate will not rise, etc. My own mind would melt trying to figure out which of the options is correct (or whether, even, some might be equivalent), and all we're talking about is a very simple conditional and negation! So yes, you have me to blame for your less-than-perfect LSAT score if you took the exam in the late 1990s. Sorry!

Okay, now consider... Hegel. Or even something as structurally simple as the Sleeping Beauty Problem.

Most of our philosophical ambitions are far beyond ordinary human cognitive capacity. Our philosophical opinions are thus much more likely to be driven by sociological and psychological factors that have little to do with the real merit of the arguments for and against. This partly explains why we make so little philosophical progress over the centuries. If we are succeeded by a species (biological or robotic) with greater cognitive capacities, a species that finds the Wason selection task and abstractly combining conditionals and negations to be intuitively simple, they will laugh at our struggles with Sleeping Beauty, with Kant's Transcendental Deduction, with paradoxes of self-reference, etc., in the same way that we, once we are fully briefed and armed on the Wason Selection Task, are tempted to (but shouldn't) laugh at all the undergraduates who fail it.

16 comments:

Anonymous said...

Don't you need to specify that those are the only two letters and the only two numbers that could be on the cards?

shane said...

ha I got it right. Of course I just started teaching conditionals in my critical thinking class. I might not have got it a week ago.

Eric Schwitzgebel said...

Anon Feb 16: Nope!

Anonymous said...

Philosophical problems are problems about what it is to be human. If you weren't human (or human-like), you wouldn't be faced with the problems and thus wouldn't need any solution.

The Transcendental Deduction, for instance, deals with what we are to make of the world we perceive. As humans, we perceive the world a certain way. What does that say about us and how does the way we see the world relate to the way others see the world. These are important questions for humans, and humans naturally grapple with them.

Eric Schwitzgebel said...

Agreed! Do you think I say anything that conflicts with that?

Neil said...

Or consider the meta-induction from our cognitive frailties to the conclusion that our views on complex problems are more likely to be driven by sociological and psychological factors than good argument. Clearly we ought to be sceptical about that view, on the grounds advanced here. If not, why not?

Eric Schwitzgebel said...

I agree that the view applies to itself too, Neil. So there I am, stuck. It would be more comfortable not to have such a meta-view of one's own philosophy.

G. Randolph Mayes said...

I wonder what we mean, though, when we say that we're bad at abstract thought. After all,is there anything out there that does it better? I'm guessing the reason it feels so right to describe it that way is that when the correct answers to these questions are revealed to us, we feel stupid, because the solution is now so obvious. But if we really do get the answer so quickly after it is explained, can we really be that bad at the kind of thinking it requires? What if, by contrast, we had to grind on it for hours before the solution made any sense? Then we wouldn't say we're bad at abstract thought, we'd just say that is a really friggin' hard problem! So maybe we're actually really great at abstract thought, though it's not what we do best.

Eric Schwitzgebel said...

Interesting thought, Randy! But I guess, too, it seems so structurally simple. That isn't just a matter of perceived effort, is it? Of course, simplicity is notoriously difficult to define.

G. Randolph Mayes said...

I guess you're right about that. I guess we're just saying that there is a remarkable competence/performance difference here. I don't have a sense for how remarkable that's allowed to be before it becomes suspicious.

Anonymous said...

why are we bad at this task? or why do we find it so difficult to do?

chris dyre said...

hi i am a little confused, i did this in class only two days ago, i got this example wrong, and said the obvious two cards, when i should have stopped and think about,only when the lecturer used concrete example, of things we could relate to i could get it, of course the logic was the same and didn't change when concrete words were used, in fact the same cards were used for all examples we were shown in class, only the words written on the cards, however i wonder if any one can please help me understand why i found / we find this so difficult to do, of course i wouldn't make the same mistake again but i understand the task now. can any one please help me understand why this is so. thank you!!!!

Eric Schwitzgebel said...

I don't think there's much consensus on this! You might search for "Wason selection" in the PsycInfo database to see recent research articles on the issue, if you're interested.

Karl said...

I don't understand this question at all. It is true that those two cards are possible to falsify the rule given... but just flipping over those two cannot always prove whether the rule is right or wrong. It is easy to create a scenario where you have flipped all four cards over, and you still cannot ascertain whether the rule is valid or not. Obviously, given the way the question is worded, an even number can have a consonant on the other side - that was never an issue; HOWEVER, the way the question is worded misled me into thinking "Which cards must you flip over to determine whether the rule is right or wrong?" - in this case, the answer would be all four cards plus two hidden ones.

Let's say you word the question this way: "Which card(s) must be turned over to test the idea that..." (Wikipedia), then you could still say all four cards... or every single card, because even if you flip over "irrelevant" cards you are narrowing down the possibilities, making it more likely for the rule to be right/wrong. That, for me, is not much different from flipping over cards that can POSSIBLY falsify the rule.

Eric Schwitzgebel said...

Karl: What if the relevant universe of cards is just those four, so that "always" right or wrong is true if and only if it holds for all four of the cards?

Unknown said...

Not sure if you still check this but wanted your thoughts on this Wason test from facebook

http://bigthink.com/praxis/a-three-question-quiz-to-test-your-rationality?utm_campaign=Echobox&utm_medium=Social&utm_source=Facebook

I contend that by saying the cards represent the deck but not having access to the deck this is not a Wason test but something entirely different and flawed.

The rule of the deck is unknown,so what if the actual rule of the entire deck was "If a vowel is printed on one side of the card, then an even number is printed on the other side, except the letter A" If this was the rule then the cards showing would still represent the deck but the rule they list could seem to be true even though it wasn't.