R. Ratcliff and G. McKoon (1997) reported a set of findings they claim is i
nconsistent with all prior accounts of long-term priming, including (a) a p
attern of benefits and costs in an identification task suggestive of a bias
interpretation of priming and (b) a restriction on priming such that benef
its and costs are obtained only when the alternatives in the forced-choice
task are similar. On the basis of these and related findings, Ratcliff and
McKoon developed a bias theory of visual word priming that is implemented i
n a mathematical model. However, the present article shows that their empir
ical findings are ambiguous and can be explained more parsimoniously within
more traditional frameworks. Furthermore, 8 studies are reported that dire
ctly contradict their model. On the basis of these and related findings, it
is argued that priming is best understood as a by-product of learning with
in perceptual systems whose main function is to categorize inputs.