Vaughan is so spot on over at mindhacks.com that I’m going to excerpt him here:
For example, an experiment might find that fear is associated with amygdala activation. But it’s impossible to say the reverse, that every time the amygdala is activated, the person is fearful.
Here’s an analogy. On average, people from New York may be more impatient than people from other cities.
If you predicted that all people from New York were impatient on the basis of this, you’d be grossly mistaken so many times that it would make your prediction invalid.
In fact, taking the average attributes of populations and applying them to individuals is stereotyping, and we avoid it because it is so often wrong as to cause us to misjudge people.
Alternatively, if you met an impatient person and therefore concluded that they must live in New York, you’d be equally inaccurate.