Saturday, June 9, 2012

Rationality and Thinking Fast vs. Slow

I just listened to an interview with Daniel Kahneman, the author of Thinking Fast and Thinking Slow.  He made a remark about an interesting case.  Suppose that a person is informed that a loved one needs immediate surgery in order to save her life, and that this surgery has a 99% success rate (i.e., only 1% of patients do not survive the surgery).  Kahneman suggests that this is one of many cases in which people behave irrationally.  Most people, he suggests, would be very nervous and fearful that the person they love would die.  He points out that this is irrational because there is but a very small percentage of people that do not survive the surgery.  This verdict doesn't seem right to me.  To call such people irrational or to say that such emotions aren't apt for the situation strikes me as a mistake.  It seems to me that because the stakes or so high (a person's life is on the line) the 1% chance is rather significant.  If the situation involved something with lower stakes, then I'd be prepared to say of such a person that she were irrational or not appreciating the statistical facts in forming her beliefs, emotions, etc.  Consider a different case:  Suppose that I have a cold and my physician says that such-and-such a treatment has a 99% chance of substantially reducing my symptoms and a 1% chance of exacerbating them. In this sort of case, it seems right to say that I would be irrational in believing that my condition would get worse upon taking the medication.  It would be strange for me to feel afraid of taking it.  It seems to me that the statistics matter and must inform our beliefs, guide our emotions, etc.  But it also seems understandable that the stakes/severity of the possible consequences matter and can help explain why a person might think or feel certain things; indeed, such considerations might even justify what would otherwise be a bad inference/ill-formed reaction. 

What do you think of all this? 

(I should confess that I haven't read his book yet.  Maybe he offers some replies to my worry.  It's on my list for summer reading and hope to get to it soon.)


Jeremy Stangroom said...

I wonder if Kahneman would get on an aircraft if he was told it has a 1% chance of crashing...

Of course, he'd probably get on it if it were the only way to save his own life, but... I'd bet he'd be nervous during the flight!

Moti Mizrahi said...

This is a very interesting question.

I think there are a couple of things one could say about why it is irrational to fear death from a procedure that has 99% success rate.

(1) Driving a car carries a risk of death of 1 in 100. But people keep driving cars. Very few people, I suppose, fear death before they drive a car or ride in a car, a bus, etc. So, one could say that people are irrational insofar as they fear death from a minor procedure but they don’t fear death from car accidents, given that the two carry more or less the same risk of death.

(2) When we say that people behave irrationally, we judge their behavior from a purely logical perspective. So, in the case of fearing death from a minor procedure, it is irrational to fear death insofar as the odds of dying from such a procedure are very low. However, people are not purely rational creatures. So, although their behavior may be irrational from a purely logical point of view, their behavior might still seem *reasonable*, once we abandon the purely logical perspective and consider the fact that people are creatures of emotions as well. Once we take that into consideration, it seems *unreasonable* to expect people to consider the fact that a loved one is about to undergo surgery from a purely logical point of view.

Jesse Steinberg said...

Thanks for the comments.

Maybe some distinctions are in order. First, we could ask whether it's rational to believe that the loved one is going to die. (Answer: Of course not) Second, we could ask whether the agent's fear or dread that she experiences is an emotion that is ill-suited given the evidence. (Answer: I'm not sure about this one) And we could continue along these lines with more questions.

I agree that (typically) assessments of rationality ought to involve purely logical grounds. And so I think it's right to say that one is irrational if one believes that her loved one will die as a result of the operation. One has no good reason for believing this and has great reasons for believing that she will survive. But a perfectly rational person might reply as follows:

"I realize the chances are slim that she'll die, but I really love her and I would be devastated if she were not to make it out of the operation. I am thus extremely nervous and fearful of this (albeit unlikely) possibility. I do not believe that she will die. But I am not sure that she will survive either. That would really set my mind at ease."

It seems to me that this kind of reaction is appropriate and we ought not to judge such a person negatively. What is required of her (in terms of her being rational) is simply that she do the best she can in keeping her beliefs consistent. And I don't see any inconsistency in the above.

A quick google check indicated that (according to the World Health Organization) the mortality rate per 100,000 inhabitants in the U.S. due to car accidents is 12.6. This is much lower than one percent. I think this is partly what explains why it would be irrational to believe that one is likely going to die due to an accident/one would be seen as odd if she were extremely fearful of riding in a car. But even such a meager percentage warrants wearing seat belts, driving carefully, etc. And I sometimes do feel a sense of nervousness when traveling fast on an unlit highway if the weather is bad. There are, of course, reasons for these things. And this is presumably what makes driving carefully, etc. not irrational.

My point is simply that even if it's very unlikely that a certain behavior is going to result in dire consequences, the fact that these unlikely consequences are so dire makes them more salient than if they weren't as dire. It seems to me that the chances can be amplified by the severity of what might happen. And this should inform how we judge a person's reaction to a given situation.

Jesse Steinberg said...

If I'm remembering it correctly, there's a case described by Paul Griffiths in his *What Emotions Really Are* that might be interesting to discuss. Moti's second point made me think of it. (I'm very sympathetic to Moti's remarks.) The case is quite different from the others and has more to do with emotions than beliefs. I haven't given it much thought and I wonder how it relates to the stuff we've been discussing...

Suppose I offer you a beer but before I hand it to you, I spit into it. You understand that it's very unlikely that you'll suffer any negative consequences from drinking it (other than having a certain emotional reaction of disgust). You won't be able to taste the difference and you won't get ill from ingesting it. I imagine that you wouldn't drink it. What if a worker at the brewery spit into a full vat of beer such that the beer I purchased contained a very minute fraction of his saliva. Would you drink it then? Would not drinking the beer (in either case) be irrational? Certainly believing that the beer is harmful to you would be irrational. But, it seems that there's something not wholly inappropriate in refusing to drink saliva-laden beer--and this is true even when it's a very small amount of saliva. But, here, the consequences aren't as dire. You'd simply have a strong emotional reaction. Now imagine I redescribe this case in terms of chances, and I tell you that the beer I've handed you has a 1% chance of containing saliva. Would you drink it? Even with the very low chance, I'd opt for the wine.

Anonymous said...

The case, as you initially described it, is that the person needs the surgery to save his/her life. If this means that without the surgery the person would die, then it would clearly be irrational to refuse the surgery on the grounds that there is a small chance that the surgery would not be successful.

Jesse Steinberg said...

Thanks for commenting, Anon.

You're right that the person who is need of surgery would be irrational to refuse to have surgery. But that's not the issue with which Kahneman is concerned. It's about the LOVED ONE'S rationality when she feels very nervous about the outcome of the surgery.

The other cases discussed in the comments section are not analogous in some important ways. Maybe this is what you're trying to get at. If so, I think you're right. And it seems to me that we have to think about a myriad of cases before we can confidently make proclamations about what counts as rational/irrational.

Anonymous said...

If the person will die without the procedure, then it would be irrational to refuse the procedure on the grounds that the person would have a 1% chance of dying as a result of the procedure. That is, there is a 100% chance of dying without the procedure and a 1% chance of dying from the procedure.

Jesse Steinberg said...

Thanks for commenting, Anon. As I say in my last comment, the question is not whether it would be rational for the patient to seek surgery--of course it would be rational--rather, it's whether it would be rational for a loved one of the patient to, say, feel extremely nervous about the surgery.

Anonymous said...

I think I agree.

A fascinating issue is whether one can apply the term ‘rational’ to feelings or emotions, as we do to choices and behavior. Choices are rational or irrational based on the extent to which decision-making procedures stem from justified premises and proceed by logical steps (or something like this). Behaviors are rational or irrational (if we are talking about goal-directed behavior) to the extent that they are reasonable for bringing about the desired end or result.

Now, what does it mean to say of a feeling or emotion that it is rational or irrational? Well, we talk about a feeling that doesn’t “fit” a situation or circumstance. We say that a feeling is inappropriate to a situation; or that a feeling is out of line with what’s happening; or that a feeling is not warranted or justified. Perhaps, in this sense, we say that a feeling is irrational. Do these constitute different senses of ‘rational’?

Is it irrational to feel anxious about the surgery, when the person would die without it; and there is a 99% chance of success with only a 1% chance of death?

Let’s be clearer about what the loved one is anxious about. Wouldn’t it be appropriate for the loved one to feel anxious that the 1% chance of death might occur? Here, as others have noted, the consequence (death) is great although the probability is small.

Let’s say I’m playing poker and I have a hand that is 99% a sure winner. If I’ve bet 50 cents on the hand, then being anxious about the result would be odd. But if I have bet my life on it, or the life of a loved one, then I don’t think anyone would hold that anxiety is irrational or inappropriate. The upshot seems to be that whether it is appropriate to feel anxiety doesn’t only relate to probability but also to consequence.