Tuesday, December 30, 2008

Which is Better: Conscious or Unconscious Decision Making?

Which is Better: Conscious or Unconscious Decision Making ?


The snippet above from Science Daily reports on the superiority of unconscious decision making in the case of estimating likelihoods.

My thoughts ...

There's a fairly sizable literature on decision making, and typically the headlines hide the facts in this field. I'd estimate that about 70% of the recent experiments went the way of this article with their conclusion (and of course one of Malcolm Gladwell's popular books) and about 30% went the opposite way, saying that the work 'proves' that complex decisions should not be left to unconscious proceses. It's like the nature/nurture question, we feel compelled to slide toward the extremes.

My impression is that what the literature (collectively) has shown so far is:

1. There are perceptual tracking mechanisms and very powerful computational mechanisms that operate completely outside of awareness for the most part, yet are important in decision making (rapid cognition, etc., the "human intuition is so amazing" school)

2. Our conscious reasoning processes are built very recently (phylogenetically) on a substrate of other things, and so they have a lot of biases and blind spots and rely heavily on our limited and easily distrated working memory capacity (Kahnemann, Tversky, etc., the "human reasoning is so fallible" school)

3. The unconscious mechanisms don't work by magic, i.e. with the exception of some basic category templates for recognizing and sorting things, they still rely on an acquired store of experience built onto those templates (Kant plus Darwin)

4. Good decision making for complex problems is good largely because because it tends to use the unconscious processes effectively, and does this by focusing attention on the relevant aspects of the problem (knowing what is relevant and what is not is sometimes the meat of the problem, and sometimes a given)

5. Focusing attention on the relevant aspects of a complex problem generally requires us to draw from the skills that rely on working memory and are generally conscious and build on structural rules, sequencing, and explicit algorithms (the virtual von Neumann machine that many of us we believe we possess in our head)

So while the above is a very simplistic account of a big topic, I think it helps illustrate that "conscious" vs. "unconscious" is not really the important issue in decision making, although it is certainly useful to distinguish automatic processes. The more compelling issue is how to structure problems so we can use the powerful resources of the mind, and how to learn the skills for making best use of those abilities, and the skills for recognizing our own biases and blind spots when we structure problems.

Many natural mechanisms work best on imperfect information but fail when compared to systematic reasoning with good information. However most real and meaningful human problems in nature require responses with limited resources, limited time, and very limited information, so a lot of the research actually demonstrates that while we can do better than intuition in theory, we can't do better than intuition in practice.

That is, unless we have some way of "cheating" by having unusually good information or knowing the answer ahead of time. Our species dependence on social imitation probably reflects exactly that, a way of "cheating" the limitations of individual problem solving by copying what other people are doing. For every problem we find that legitimately shows that people screwed up by trusting systematic reasoning too far, we can find one where someone trusted an intuition that had catastrophic consequences. In my opinion, the key skill set is in learning clues for recognizing when to trust each tool. That is, rather than espousing either blind trust or mistrust in powerful automatic processes.

No comments:

Post a Comment