How We Know What Isn't So: The Fallibility of Reason in Everyday Life
by Thomas Gilovich
Reviewed by Steven Rubio
Thursday, July 30 1998, 1:45 PM
The title of this 1991 book, while not misleading, nonetheless suggests a different book than actually exists between the covers. While Gilovich is concerned with the fallibility of human reason, his book is not anti-reason. Far from it. Gilovich is firmly rooted in rationalism. What he describes here is the danger inherent in our flawed use of reason in combination with the good old common sense which leads us to believe things that aren't true. Gilovich methodically walks the reader through the various ways human judgment fails us, not in order to dismiss reason, but so that, recognizing our mistakes, we might work to conquer them in our daily lives.
An example might help, from a section titled "Misunderstanding Instances of Statistical Regression." (It's worth noting at this point that the book is very readable and nowhere near as dry as it might sound.) To simplify, hopefully without doing too much damage to the statistical concepts, the regression effect is one of collapsing towards the center. For instance, "very tall parents tend to have tall children, but not as tall (on average) as they are themselves." Gilovich moves on to discuss "regression fallacy ... the tendency to fail to recognize statistical regression when it occurs, and instead to 'explain' the observed phenomena with superfluous and often complicated causal theories."
OK, so far we're still in statland. But Gilovich takes this statistical knowledge into the realm of everyday life, using as an example the "effectiveness of reward and punishment in producing desired behavior." While rewarding desired behavior is more productive than punishing "bad" behavior, many people persist in believing that if you spare the rod you spoil the child. Gilovich argues that this is the regression fallacy at work. A child does something exceptionally well, and the parent rewards them with something nice. Regression suggests that the child's next behavior is likely to be less exceptional; the parent will then associate reward with a falloff in the child's behavior. When the child does something exceptionally bad, the parent will punish the child. As will be expected if we consider regression, the child's next behavior is likely to be less exceptionally bad; the parent will associate punishment with improved behavior. The parent's perception will lead to a belief in the usefulness of punishment. They will know something that isn't so, and it will "make sense" to them. And a statistically consistent, random blip, which marks a surprising rise in "crime", will lead to a call for increased "law enforcement"; when the expected regression takes effect, "crime" will drop. People will believe the increased enforcement was the "reason."
The above is only a small portion of what Gilovich discusses in his book. He also takes on specific "questionable and erroneous beliefs" in chapters on "alternative" health practices and belief in ESP. More to the point, he concludes with a hopeful chapter on what social scientists and laypersons can offer a society that knows what isn't so, arguing that we must all participate in the battle to dispute dubious beliefs that lead us to know what isn't so. His book is highly recommended.
Free Press, 1991