« Rarity Anomalies Remain | Main | Future Incompetence »

April 20, 2009


Great post.

I'm reminded of this interview of F. A. Hayek, where (around the 13 minute mark) he talks about three different levels of moral beliefs: genetic influences (which he believes is the source of desires for social justice; I'm not sure what they thought about signaling back then), traditions (some of which he believes are property rights and the family) and logical reasoning (e.g., Objectivism or any other deduced set of ethics). Though I admittedly haven't thought about it very much, I'm inclined to agree with Hayek: Genetic intuitions gained in an environment different from our own probably aren't useful and may need to be discarded. Intuitions gained via tradition and cultural group-selection processes may be "smarter" than we are, and so should be trusted unless we have good reason not to. Logical deduction of morals will probably contain many errors and should be greeted with great skepticism.

I think this can apply to non-moral intuitions as well. We obviously evolved (or inherited?) intuitions for good reasons, so it seems prudent to utilize this cognitive tool as we might a calculator or spreadsheet. To me the question is not whether to rely more or less on intuition, but specifically when to do so? Some examples are easy, e.g. using intuition to read another's mood and predict her future actions. That intuition becomes a lot less accurate if the other person is a Cylon sleeper agent, so the choice in this instance is easy.

how do we know when a string needs repair. The more I think about it, the more it seems that Caplan's position is that the edge cases aren't always as obvious as a trillion nazi's. The harder it is to identify when a rule may or may not apply the riskier following that rule is.

"We can often test those reasons through criticism, increasing confidence when criticism is less effective than expected, and decreasing confidence when criticism is more effective than expected."

This causes problems when your expectation of the efficacy of criticism is miscalibrated. I have a very hard time calibrating my expectations on arguments for theism; I find it hard to accept on a gut level that such smart people will proffer such bad arguments.

Grant, yes the question more precisely is when to rely more or less on intuition.

Paul, yes it can be hard to form expectations of criticism effectiveness.

nazgul, I wasn't talking about rules in this post.

Genetic intuitions gained in an environment different from our own probably aren't useful and may need to be discarded. Intuitions gained via tradition and cultural group-selection processes may be "smarter" than we are, and so should be trusted unless we have good reason not to.

Aren't many traditions founded in "an environment different from our own"?

How do you feel about situations where we can't use introspection very well? When you look out the window and can tell which shapes are shadows and which are real objects, or recognize a face, I doubt you can introspect into the chain of reasoning that makes that work. There are chains of reasoning there, but they seem to be extremely messy, so much that they are only practical when done very fast in specialized neural hardware, and so much that even if you could use introspection to inspect what was going on, it would take you a very, very long time to describe the logic to someone else.

Looking at the experience of mathematicians and computer scientists working with formal proof systems (pre-computer ones like Principia Mathematica, and modern computerized ones like HOL and Coq) it seems as though it's pretty normal for even rigorous mathematical reasoning to be grounded in messy tangles --- tangles which are so messy that it's not very practical for humans to communicate them explicitly, or to reason about them explicitly with any process which runs at about the same speed as speech. My experience talking with Chess and Go players about their tactical analyses suggests to me that something similar is going on there as well.

You may know this already, but you are recapitulating parts of Blaise Pascal's discussion, in his Pensees, of the difference between what he called geometric and intuitive modes of reasoning. By geometric, he means what we'd probably call mathematical/deductive/axiomatic. Pascal's reflection is fairly short but it is well-known and has probably nucleated further discussion that would interest you.

Here's the first English translation google provides (which translates geometric as "mathematical"):

My (limited) exposure to this kind of argument is from Rawls' discussions of reflective equilibrium, which seems similar to what Bryan is striving for. What possibly bothered me more about what Bryan said wasn't simply that he used an intuitive process (and I'm sure the usual cognitive pratfalls associated with this are a worthy method of poking at this approach), but that he claimed to ascribe to a strand of moral realism.. given more time to follow up, I probably would have tried to press him more on how he comes to perceive moral facts in the universe in ways that others (those who disagree with him) do not.

Susan Haack discusses the problem in "Evidence and Inquiry"; she tries to develop a middle way between "foundationalism" and "coherentism". One that has the web of mutually and interdependently supporting facts/ideas but is supported by a few solid foundational facts/ideas.

Of course, the main idea in your post is the need for a better understanding of intuitions, which I agree with. "Intuition" as a term has collected a lot of baggage from New Agers and others, so I mostly try to avoid using it, I mostly use "insight" instead, which seems to capture the epistemic characteristics of intuition. Whatever you call it, it seems to mostly be the result of an unconscious search and judgement system (all too often with unconscious goals as well).

Robin, et. al.,

a number of top philosophers have been turning with heated interest to precisely this issue in the past recent years. For an excellent introduction to this current discussions, here is perhaps the best place to start:


All our reasoning reduces itself to yielding to feeling. (Pascal's Thoughts, 274)

"I'm not even sure we have much in the way of empirical data on belief accuracy of folks who rely more versus less on intuition."

most successful high-stakes no-limit hold ém professionals rely almost entirely on intuition when making decisions at the table. the number of relevant variables and the complexity of their interactions is such that deliberative reasoning is impossible. professional NLHE players can enumerate the variables they're considering, e.g., recent history, the rhythm or timing of their opponent's actions, the impact the present hand/decision node will have on future interactions, etc., but they very rarely point to several specific things and say, 'this, this, and this, therefore that, and thus i did x'. the interactions between the variables are so subtle that they can only be apprehended after hundreds of thousands of hands. 'this, this, this, this, and this, and so i feel like x would be good here'.

the high-stakes community is made up predominantly of individuals who rely more on intuitive judgment than 'deliberative' judgment. analytical types are few and far between above a certain threshold of competence. those who ''need'' or ''look for'' explanations with well-defined terms and relations between them seem to do well but not exceptionally well (they flounder in low or midstakes) on average.

the above is true for many, many games where accurate beliefs about your opponent's meta-strategy are required for success at the highest level (from rock-paper-scissors to any well-balanced real-time strategy game). we can discuss this in more depth if you're interested.

pascal essay is excellent.

This is my intuition. There are many like it, but this one is mine.

@ Doug S. re: This is my intuition. There are many like it, but this one is mine.

Doug, while I was forced to recite this mantra repeatedly during my training for the Tet Offensive, I doubt that many OB readers will recognize your reference (unless it's from a movie).

Isn't relying on intuition the same as relying on one's priors?

Robin: since when is morality not rules?

Unit, priors are consciously held, while intuitions result from at least partially unconscious processes.

@Retired: it is in a movie. http://www.youtube.com/watch?v=Ar0_um--LDQ

@ Michael: "it is in a movie"

That was the point. It was a movie about Viet Nam, an old topic to most OB readers, but quite real to me.

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30