Wednesday, November 21, 2007

Heuristics, Gut Feelings, and Reason

Newsweek has an article Less Information is More that proposes that ‘gut feeling’ is better than reason when it comes to making important decisions. As is generally the case of articles in the popular press, this one uses language in a loose and imprecise way that blurs several concepts that should be left distinct.

Heuristics

One of the claims made in the article is that considering all information relevant to a decision often simply is not worth the investment. It is sufficient to consider only the most relevant factors, and simply ignore the rest.

This is a version of a claim that I have made against Harris and some of the ‘new atheists’ who claim that anybody can be condemned if they do not base their conclusion solely on the evidence. Against this view, I have argued that we simply do not have the time to hold every one of our beliefs up to the light of reason. Instead, we are better off resorting to quick rules which, though they are less accurate than reason, allow us to reach actionable decisions in a timely manner.

A good example – one that is alluded to in the article – is the decision on the part of some early human ancestor to run from a suspected threat or to sit and gather more evidence before deciding what to do. A quick heuristic that keeps the individual alive may serve him far better than spending a lot of time contemplating every available piece of evidence. We are the descendents of such creatures, and we have evolved quick but fallible ways of reaching conclusions other than reason, and for good reason.

Satisficing

Another method of decision making mentioned in the article, which the authors blurred with the concept of ‘heuristics’, is the concept of ‘satisficing’.

‘Satisficing’ theories of action are often contrasted with ‘maximizing’ theories in that ‘satisficing’ theories are content with what is ‘good enough’. On a satisficing theory it is possible for option B to be better than option A. However, if option A is ‘good enough’, then there is no reason to bring about option B, even though option B is better.

This is absurd.

Like all absurdities, it is well-intentioned. Act-utilitarian theories embraced the concept of ‘satisficing’ to deal with one of the common objections to their theory. In act utilitarianism, only the act that maximizes utility is ‘right’. All other acts are (to various degrees) ‘wrong’. The concept of ‘satisficing’ allows actions that fall short of maximizing utility to still be counted as ‘right’.

It is an arbitrary and ad-hoc solution used to try to rescue a theory that ultimately does not merit being rescued. The fact that act-utilitarianism requires something like ‘satisficing’ to make it work should be taken as good evidence that it does not work.

One way to see the flaws in a satisficing theory of action is to look at what a satisficing theory of physics might look like. It would be quite reasonable for a physicist, trying to calculate the motion of an object through space, to get an estimate of its motion by calculating only the most significant forces acting on it. He does not need to calculate the effect of every little force – because most of them will not have enough of an effect to change the outcome.

However, this rational heuristic approach becomes an absurdity when the physicist says that the lesser forces have no effect. A satisficing theory of physics says that, once an object is acted on by the larger and more significant forces, that this is ‘good enough’, and all lesser forces have no further bearing on the motion of that object. Somehow, it is possible for a force to exist and to act on this object without having an effect.

The satisficing moral theorist says that it is possible for a desire to exist, and yet have absolutely no impact on the quality of different options that an agent might choose. Once a certain outcome is ‘good enough’, a desire that says that option B would still be slightly better than option A has no motivational force at all.

The kindest thing to say about this theory is that these hypothetical entities – desires without motivational force – are so strange that we need extraordinary reasons to justify their existence. In the absence of this type of evidence, we have more reason to assert that they have the same status as God and ghosts – that they are an imaginary ad-hoc invention whose sole purpose is to try to hammer together pieces of a puzzle that simply do not fit.

Gut Feeling

A third concept that the authors of this article included in their essay was the concept of a ‘gut feeling’.

Notice that the physicist in the example above is not trusting to a ‘gut feeling’. He is still basing his conclusions based on measurements made on real-world phenomena. There is no connection – or at least no necessary connection – between a conclusion reached by ‘gut feeling’ and a conclusion reached by heuristics.

Nature may well have (and probably did) program into us a set of heuristics related to ‘gut feeling’. That is to say, the brain has some heuristic programming in it that leads to certain conclusions (e.g., this is safe, that is dangerous) that it then communicates to the conscious mind through its influence on the emotions. A heuristic that argues that there is something dangerous prepares the body for a fight or flight response. It does not tell the conscious mind directly, “This is dangerous.” Rather, the conscious mind picks up only the physiological symptoms of a fight or flight response and starts looking for a source of this anxiety.

However, ‘gut feelings’ are linked to what we like and dislike – to our own personal desires. Relying on gut feelings to decide what we should or should not do is dangerous, and not something that is to be too loosely defended.

I have no doubt that just about every atrocity committed in human history – if not every atrocity in fact – was committed by people who trusted to a ‘gut feeling’ that they were doing the right thing. It is, in fact, very tempting to use ‘gut feeling’ as a substitute for morality. Since ‘gut feeling’ is an indicator of what the agent does and does not like, using ‘gut feeling’ as a source of morality ultimately translate into, “Do what pleases you – whatever it pleases you to do is right.” This is an extremely easy moral system to live by, which I suspect is one of the things that leads to its popularity.

The problem comes when others who do what feels right to them decide that it feels right to kill, enslave, rape, or otherwise harm others. The question that comes up using the ‘gut feeling’ method is, “What should we do with gut feelings when the feelings are caused by a love of doing harm to others, or a love of something that produces harm to others as a side effect?”

Desire utilitarianism advocates a certain amount of heuristic thinking. Desire utilitarianism holds that a desire is good to the degree that it tends to fulfill other desires, and bad to the degree that it tends to thwart other desires. In order to determine whether a desire tends to fulfill or thwart other desires we do not need to consider its affect against every single other desire and to determine the answer with minute precision. It is sufficient to consider only the major effects, and to dismiss the minor effects as just so much noise that has a low chance of affecting the final result.

A love of honesty and a distaste for sophistry and deception does not require a complex calculation that pits them against every other factor that it might touch. It is sufficient to note that while people seek states that fulfill their desires, they act so as to fulfill their desires given their beliefs. From this, we can see that false beliefs pose a great risk of thwarting desires. Consequently, we can defend promoting a strong aversion to liars, sophists, and intellectually reckless individuals who promote false beliefs. Unless somebody can come up with a reason to belief that some of the other factors we have ignored might actually change the results, we can move on to the next issue.

Heuristics also tells us that, if there is anything wrong with studying architecture (for instance), a look at the major effects of a desire to study architecture give us no reason to condemn this particular interest. We also have no reason to promote it as a universal desire. It is a desire that we are best off allowing some people to acquire, while others acquire desires that would best fit them to different professions (e.g., teaching or medicine). A precise calculation of every single factor involved is simply unnecessary.

However, the calculation of the value of any given desire is not dependent on ‘gut feeling’. Like the physicist who makes his calculation based on a consideration of only the largest forces, the ethicist is looking at real-world relationships that are fully independent of ‘gut feeling’. He simply recognizes that to get a practical real-world actionable item, he does not need to precisely measure each and every real-world relationship.

Summary

This article, then, blurs the distinction between three forms of decision making. Heuristics makes sense – it allows an individual to make reliable (though fallible) decisions much more quickly than a careful evaluation of every piece of evidence. “Satisficing” theories, on the other hand, are metaphysical nonsense.

The fact that it is not efficient to weigh the relationships between a desire and every other desire that exists does not prove that those desires have no weight. In fact, the very concept of a desire lacking motivational force is so bizarre that we would need extraordinary evidence to suggest that such things really exist.

Finally, the fact that there is good reason to engage in ‘heuristic’ thinking in order to reach actionable conclusions in a timely manner is no proof that ‘gut feeling’ has any reliability at all. Gut feelings give us a measure of whether we do or do not like a particular conclusion. It does not tell us whether we should or should not like a particular conclusion. The history books (and prisons) are filled with examples of people who acted on ‘gut feelings’ in ways that they should not have.

3 comments:

Doug S. said...

I thought the concept of "satisficing" was generally justified based on the idea that once you find an "option A" that is "good enough," the cost of continuing to search for a better option often tends to be greater than the benefits of the search? (See also "rational ignorance".)

Alonzo Fyfe said...

doug s.

What you are talking about is a maximizing strategy in which one continues to pursue an effort until the marginal cost of an additional unit of input exceeds the marginal benefit. This, in fact, is the formula for maximization when we are facing a product that has diminishing marginal returns.

Doug S. said...

Yes, I guess you're right; I suppose there is a difference in principle between "Decision making is costly, so I'll choose the first 'acceptable' result I can think of" and "Any acceptable solution is as good as an optimal solution".