Russ Poldrack: UCLA Psychology

It has become common practice amongst neuroimaging researchers to infer the presence of mental
processes from activation in particular parts of the brain. The validity of this practice, which I refer to as
“reverse inference”, depends upon how selectively specific brain regions are associated with specific mental
processes. I will present evidence suggesting that these associations are often weak, providing little
evidence in favor of the reverse inference. I will also discuss general strategies by which one can use
neuroimaging to inform theories of mental processes.