Anyone who has taken a course in introductory psychology has heard the story of how the behaviorist John B. Watson produced "conditioned fear" of a white rat — or was it a white rabbit? — in an unfortunate infant called Little Albert, and how Albert "generalized" that fear to other white, furry things (including, in some accounts, his mother's coat). It was a vividly convincing story and, like my fellow students, I saw no reason to doubt it. Nor did I see any reason, until many years later, to read Watson's original account of the experiment, published in 1920. What a mess! You could find better methodology at a high school science fair. Not surprisingly — at least it doesn't surprise me now — Watson's experiment has not stood up well to attempts to replicate it. But the failures to replicate are seldom mentioned in the introductory textbooks.
The idea of generalization is a very basic one in psychology. Psychologists of every stripe take it for granted that learned responses — behaviors, emotions, expectations, and so on — generalize readily and automatically to other stimuli of the same general type. It is assumed, for example, that once the baby has learned that his mother is dependable and his brother is aggressive, he will expect other adults to be dependable and other children to be aggressive.
I now believe that generalization is the exception, not the rule. Careful research has shown that babies arrive in the world with a bias against generalizing. This is true for learned motor skills and it is also true for expectations about people. Babies are born with the desire to learn about the beings who populate their world and the ability to store information about each individual separately. They do not expect all adults to behave like their mother or all children to behave like their siblings. Children who quarrel incessantly with their brothers and sisters generally get along much better with their peers. A firstborn who is accustomed to dominating his younger siblings at home is no more likely than a laterborn to try to dominate his schoolmates on the playground. A boy's relationship with his father does not form the template for his later relationship with his boss.
I am not, of course, the only one in the world who has given up the belief in ubiquitous generalization, but if we formed a club, we could probably hold meetings in my kitchen. Confirmation bias — the tendency to notice things that support one's assumptions and to ignore or explain away anything that doesn't fit — keeps most people faithful to what they learned in intro psych. They observe that the child who is agreeable or timid or conscientious at home tends, to a certain extent, to behave in a similar manner outside the home, and they interpret this correlation as evidence that the child learns patterns of behavior at home which she then carries along with her to other situations.
The mistake they are making is to ignore the effects of genes. Studies using advanced methods of data analysis have shown that the similarities in behavior from one context to another are due chiefly to genetic influences. Our inborn predispositions to behave in certain ways go with us wherever we go, but learned behaviors are tailored to the situation. The fact that genetic predispositions tend to show up early is the reason why some psychologists also make the mistake of attributing too much importance to early experiences.
What changed my mind about these things was the realization that if I tossed out the assumption about generalization, some hitherto puzzling findings about human behavior suddenly made more sense. I was 56 years old at the time but fairly new to the field of child development, and I had no stake in maintaining the status quo. It is a luxury to have the freedom to change one's mind.