Friday, January 23, 2004

Logical fallacies as clues to cognitive instincts

Robert Cialdini's book Influence argues that we fall for social manipulations like laugh tracks because they exploit psychological short cuts that are correct and useful most of the time. (The laugh track is an example of "social proof"; in most cases, the fact that other people like something is a good predictor that we'll like it.) For this very reason, such tricks are very hard to resist.

Logical fallacies are another category of behavior where a huge cautionary apparatus is required. The fact that we are so vulnerable to them suggests, following Cialdini, that logical fallacies represent another set of psychological short cuts. (This was an example of using "authority" - that of Cialdini, who is much more reputable than me in these matters - to get you to agree to my point of view.)

There are some good lists of fallacies on the web: Michael Labossiere's list on the Nizcor Project, Stephen Downes's Guide, and "Logic and Fallacies" on the Atheism Web.

Some of the fallacies tie directly to one of Cialdini's six principles: Appeal to Authority and Appeal to Force , to Authority, Appeal to Pity to Liking, Prejudicial Language to Social Proof. I'm more interested here in ones that would reveal something new, like inductive fallacies or syllogistic errors.

Take causal fallacies like Post Hoc, and the "non sequitur" errors of affirming or denying the consequent. They represent our tendency to assume that implications are reciprocal. For example: if A implies B, and B is true, then A is true; similary starting again from A => B, derive that A being false implies that B is false. Our vulnerability to this error suggests that our sense of causality contains a strong dose of correlation: if A implies B then they're "so close" that B would imply A. Since causality is a difficult phenomenon to pin down, the efficient short cut is correlation - if A and B are in proximity, there must be some tie between them. It doesn't matter which causes which; if you see an A or a B, a B or an A (respectively) can't be far behind.

Superstition is built on this phenomenon. Think back a few millennia: I got sick around the full new moon when we found that deer carcass just after I spat over my left shoulder while my mate's mother was angry with me. The best bet is to correlate all those phenomena with getting sick - one of them is bound to be the cause, but it would take too many experiments -- which are tough to fund while barely surviving on the savannah -- to figure out for sure which one.

The inductive fallacies (eg, hasty generalization, unrepresentative sample, and fallacy of exclusion) are cases where one jumps to conclusions. This is a very adaptive behavior where one has to act quickly in the absence of information - a typical situation when survival against the odds is at stake.

No comments: