The conventional theory of investor behavior was called "efficient markets." Investors "knew" everything worth knowing about markets or were supposed to. (In the "strong" form, this applied to private, inside information; in the semi-strong form, only to public information).
If this was the case, why do markets sometimes behave so violently? After all, it's "impossible" to introduce loads of new information that people can't quickly digest, right.
A new theory of markets, hypothesized by people like MIT's Andrew Lo called adaptive expectations seems to explain things better. Moreover, it correlated better with classical models put forth by nineteenth century economist, Leon Walras, called the tatonnement, theory in the original French ("groping") in English. That is, people knew approximately what they were doing in the marketplace, didn't know exactly, and were always searching for answers.
Such a theory explained a lot more. This meant that markets would be stable most of the time, like a lighted pathway. But, occasionally, the lights would go out, and there would be chaos, just like in the real world. Investors would fail to see the occasional pothole, and therefore panic when they saw it. And that would explain why crashes occasionally occurred..