When I was a young child in early elementary school, I performed poorly on IQ tests, so my teachers and I had low expectations for me. In particular, my teachers thought I was stupid, and so I thought I was stupid; my teachers were happy that I acted stupid, and I was happy that they were happy. So everyone was pretty happy. It was not until I was in fourth grade, when I had Mrs. Alexa as a teacher, that I realized that I could do better and became an “A” student.

Things just chugged merrily along until I took introductory psychology as a freshman at Yale. I was determined to major in psychology because I wanted to understand why I had seemed so stupid as a child. But that was all in the past – or so I thought. The first test we took, I received a score of three out of ten. My professor, handing me my test paper, commented that there was a famous Sternberg (Saul) in psychology, and it looked like there wouldn't be another one. I ended up with a “C” in the course, which the professor referred to as a “gift.” Seven years later, that same professor was chair of the search committee that hired me back to Yale as an assistant professor of psychology. How times change!

Experiences such as these led me to believe – or at least to hope – that there is something more to intelligence than IQ (my fourth-grade experience) and being able to memorize books (my freshman-year experience). By my last year as a graduate student in psychology at Stanford, I was convinced that the “something more” was an understanding of the mental processes (or, as I called them, components) underlying intelligence. So I started a research program looking at the mental processes used to solve IQ-test-like problems, such as analogies and series completions. The idea was that I would be able to detect, for example, if someone was solving a verbal analogy incorrectly because she did not reason well or, instead, simply did not know the meanings of the words constituting the analogy.

Eventually, though, I realized that my “componential” theory of intelligence was incomplete. It specified processes, sure enough, but only those involved in solving IQ-test-like problems.