Theory of Knowledge
Fall 2006
Evidence of Our Irrationality?

Answer questions on the “Reader Survey” (or “Rationality Test”). Do questions 39, 1-3, 6-7, 12-13, 20, 23, 25-26, 31, 38

I. Introduction 

One question that has attracted considerable attention lately concerns the extent of human rationality. How good a job do people do when forming beliefs? Are people irrational? If so, how irrational? A negative answer could cast some doubt on the standard view. It might not suggest widespread skepticism, but it could seriously diminish the extent of our knowledge.

It’s important to see that answering any such question involves information of two different sorts. You need empirical information about what people actually do and also information about what the standards of rational or justified belief are. Here's a question to ask yourself: suppose we knew all there was to know about why people form beliefs as they do. What would we know about what constitutes good reasoning? Analogous question: suppose we knew everything there was to know about why people behave as they do. What would we able to infer about ethics? My answer, in both cases: nothing.

II. Grim Assessments of Human Rationality

Lots of people over the past 20 years have asserted that people are systematically irrational. Some assertions of this, from L.J. Cohen, "Can Human Irrationality Be Experimentally Demonstrated?"

Recent psychological studies have "bleak implications for human rationality." "For anyone who would wish to view man a reasonable intuitive statistician, [these] results are discouraging." "People systematically violate principles of rational decision making..."

III. Examples of Alleged Irrationality

The following examples are drawn from the “Rationality Tests” given in class.

A. The Selection Task

See text for discussion.

Commonly given answers: see "The Psychology of Judgment and Decision Making" p. 231.

E and 8: ~45%

E: 33%

E and 7: ~5%

Notably, people usually do much better when it’s realistic. This suggests one issue about these tests: how hard do people try.

(Some versions of the rationality test had a more “realistic” question of the same logical structure. The rule to be tested is “If a person is drinking beer, then the person is at least 21 years old.” The cards stated their age on one side and what they were drinking on the other. In general, people tend to do better on this version of the test.)

B. The Conjunction Problem

#1 on the test.

See text.

Here are some other questions, not on the test.

C. The Monty Hall Puzzle

There’s an old U.S. TV show Let’s Make A Deal. The emcee, Monty Hall, invited audience members to “come on down” to the stage and play the game. On the stage there were three curtains, labeled A, B, and C. The player is told that behind two of the curtains there are nearly worthless prizes (e.g., a goat). Behind the other is something truly wonderful and expensive. The contestant then gets to pick one of the curtains. Suppose it is A. After the contestant picks a curtain, Monty Hall then opens one of the other curtains - B or C. Monty always reveals a worthless prize. Let’s say he opens C. The contestant is then given the chance to switch to B or stick with A. Which is the most reasonable option:

      ___ Stick with A ___ Switch to B ____ It doesn’t matter - they are equally likely to win

Answer: Switch to C.

Explanation: Initially, there is a 1/3 chance it's A. 2/3 chance it is B or C. But he'll only uncover the dud. If C is the winner, he uncovers B. If B is the winner, he uncovers C. So, in those 2/3 of the cases, switching will get you the winner. After B has been uncovered, the chance that it's C is 2/3.

D. Selective Memory Problems

See text.

We could add to the mix other errors people make. For example, eyewitness reports are often terrible. Eg., experiment in which someone runs into lecture hall and steals the lecturer’s briefcase. Students asked to describe the person. Wildly different reports are given

The People are Irrational Argument - see text.

There’s some looseness in this argument - “frequently” and “significantly”. Let’s try not to worry about that. The argument is valid. Replies must be to one or the other of the premises. You can argue that the answers people give aren’t so bad. You can argue that there’s something special about these particular questions and making mistakes here is not indicative of any general irrationality. (The second response may end up turning how just what counts as “significantly” irrational.)

.

Assessing the Results: Notice first that to assess the results here fits with the general point made earlier - you need a general account of rationality and a description of what goes on in these cases in order to get any result. The negative assessments seem to be based on the idea that to be rational a person must conform to the rules of probability theory and logic.

IV. General Theoretical Defenses

Some people have argued that whatever these results show, they don’t show any general irrationality. The idea is that it couldn’t be that people are irrational. I’ll briefly describe a few of the reasons people have given for thinking this.

A. Irrationality Can’t Be Shown - The Irrationality Thesis is “Self-Refuting”

Anyone who argues that we are irrational, in the sense that we don’t reason well, is himself reasoning. So, he’s presupposing that he himself reasons well. So, his argument can’t succeed.

This is hopeless. 1) You can distinguish contexts or domains over which we make errors. 2) Anyway, this doesn’t tell us what to make of the results.

B. Evolutionary Arguments

It is sometimes said that this argument can’t be right. If we were systematically irrational, then we wouldn’t have survived. But we have. So the conclusion must be false.

Evolutionary arguments will at most show that people reason in ways that are evolutionarily advantageous. But that’s not to say that they must be epistemically rational. Hard to see how anything along these lines could work. Moreover, this does not tell us which premise of the argument is false.

C. The Analogy to Language

The main idea of this sort of response is that to reason properly is to reason in the way people actually do reason. Consider correct grammar. In some sense, it can’t be that everyone speaks incorrectly. Correct grammar must be a function of what people say. The rules of English don’t come from some other source. Analogously, correct reasoning is reasoning of the sort people regularly do. So (1) must be wrong.

Two problems with this: 1) Suppose it is true that grammatical correctness is somehow determined by the ways people speak. It is consistent with this that people usually don’t speak correctly. It may be some subset of overall linguistic tendencies that determines correct speech, while some larger set of tendencies determines how people usually speak. So, there could in fact be widespread and systematic errors. 2) The analogy to language doesn’t really seem very good. Language is simply a social practice. It is socially determined. We could in principle have hard wired tendencies to make gross logical errors. It’s very hard to see what that would contribute to rationality. That is, the idea that people routinely speak incorrectly does seem incoherent. But the idea that we reason incorrectly does not. 3) Even if the reply to the examples is right, it still follows that we make lots of errors. They are just “performance errors” rather than “competence errors.” This is hardly satisfying. We are still highly irrational in practice (provided the kinds of errors brought forth by the examples are in fact widespread.)

V. Rethinking the Examples- Details of the Experiments

Some of us find in ourselves a strong tendency to find the examples less than entirely convincing as evidence of irrationality. One can look at them one at a time and find various things to question in each case. I will do that with some of them. And then I will turn to some more general comments. See text.

A. Language Problems

One question about some of the examples concerns whether people understand the words in the same way as the experimenters intend. It is easy to see that people don’t typically take conditionals to express the material conditional of logic. It is far from clear what they take probability to be.

Take the conjunction rule violation. Here’s another question people could take themselves to be answering: “How likely a story is this?” See text.

B. Lack of Information

There is a difference between irrationality and ignorance. Psychologists Nisbett and Ross, describe people as giving "biased" estimates as a result of their use of "the availability heuristic" in some cases. They describe three cases: unemployed people overestimate the unemployment rate, and v.v.; Indiana people tend to overestimate the number of prominent people from Indiana (and same for other states); and the example we did about letters. Thus, it looks like people tend to use a rule like this:

 I can think of more As than Bs, so there probably are more As than Bs.

Or,

About x% of the As I can think of are B, so about x% of all As are B.

In other words, they judge on the basis of the available information about As. but this can be wrong: unemployed people tend to know unemployed people, people hear about, know and take notice of prominent folks from their state; people can more easily recall words by first letter than third.

What have they done wrong? Notice that the inference they make seems to be exactly the right one, given the premise they have. It is roughly, an inference from a fact about the observed (or remembered) As to all As. And that is in general ok. That is, suppose all the evidence people had was what they used as their premise. Then, their conclusion would be reasonable, even if wrong.

So, maybe the complaint is that they should have realized that they had additional evidence that was relevant, evidence that would tend to discount (or "defeat") the evidence they thought of - they *should* have thought of the factors they mention. Relate this to evidentialism: what is the evidence that the people "have" in the relevant sense of "have" in these cases? Do they just have as evidence the probabilistic and frequency information they think of? If so, then they believe what their evidence supports. Do they also have as evidence the other facts, the facts that show that their recalled data are likely to be biased?

Very hard to say. A restrictive notion of evidence would imply that people are rational in these cases. A more generous notion does not.

Conclusion: a plausible response to these alleged errors is that they are not cases of unjustified belief or bad inference. Instead, they are cases in which people fail to bring to mind some information that, in some sense, they could have considered. In any case, merely getting the wrong answer does not show irrationality. And not knowing that one’s memory works in a way that is apt to make what one remembers not representative does not show irrationality.

C. Using Different Rules

See text.

Consider again the conjunction case. Suppose that people are estimating probabilities (rather than interpreting the question differently). What rule do people use? See the rules mentioned in the text.

Conclusion: a plausible account of this case is that people are using a different, and pretty sensible, rule. It’s very hard to decide whether that makes for irrationality. It seems as if it’s a bias of some logicians and academics to think that using the rules they know of are constitutive of being rational. But not knowing a rule of logic, probability, statistics, or mathematics is not constitutive of rationality. Cf., not knowing a rule of baseball.

D. Ignorance.

Consider the Monty Hall case. There’s an inference you can make that will help you get a piece of information relevant to your choice. This is the reasoning that I went through above. You have to realize that: in the 2/3 of the cases in which the selected curtain is not “it”, Monty will open the only remaining curtain that is not it, thereby leaving closed the one that is it. You have to realize that he will never open the winner and will never open the one that you choose. All of that is obvious on reflection.

Again - this may be a case of not identifying all the evidence that others find in a case. It’s not that people so clearly make a bad inference.

V. Another Basis for Charges of Irrationality

Wishful thinking, fearful thinking, ignoring evidence.

VI. Conclusion

A. I don’t think that any general argument showing that we couldn’t be irrational is sound. So the three arguments - considered in section IV - designed to show that The Irrationality Argument must be mistaken don’t succeed. Note: those arguments didn’t say very clearly what was wrong with the argument anyway.

B. The various responses, - discussed in section V - dealing with different examples in different ways, seems better. But these response are far from conclusive. In part, they end up highlighting complexities and obscurities in the concept of rationality. Chief among these is a question about the role of logic and probability. The idea that any violation of a rule of logic or probability constitutes a case of irrationality strikes me as a major mistake.

C. I think that a stronger case for widespread and systematic irrationality comes from the fact that not all our beliefs are guided by our evidence. Sometimes we are guided by emotions. Sometimes we ignore part of our evidence. Sometimes we make mistakes about what our evidence supports.