Evaluative conditioning (EC) is an important variant of Pavlovian condition
ing in which the outcome is a change in affective response to the condition
ed stimulus (CS). It is the best extant account, with evidence, for affecti
ve change in humans. Good laboratory models are available. This paper revie
ws a set of findings which suggest that the actual occurrence of EC, and it
s magnitude, varies widely in both real world and laboratory situations. At
tention to known parameters of Pavlovian conditioning may account for some,
but not all, of the failures and successes. Six empirical studies on human
s are described; two document frequent failures of EC to occur in real worl
d situations where the Pavlovian conditions for development of animal phobi
as or taste aversions are present, two are real-world experiments in which
no evidence for EC is obtained, and two are laboratory failures to produce
EC, by pairing neutral odor CSs with a variety of unconditioned stimuli (US
s). We suggest that there are important, not understood factors, that modul
ate the appearance of EC, and that for both theoretical and applied reasons
, these factors should be identified. (C) 1998 Academic Press.