An adequate theory of intelligence must provide an account for individ
ual variation in specific cognitive abilities but also for the common
variance between these that results in a general factor. However, it m
ust also be capable for providing an account of savant syndrome: rare
individuals who display high levels of skill beyond the accomplishment
s of most people, yet who have low IQs. On the basis of recent researc
h, two characteristics of savant performance are identified; the first
is soundly functioning long-term memory that is narrowly focused and
the second is a specific aptitude; that is, memory and cognitive proce
sses dedicated to a specific ability. It is concluded that savant skil
ls are not intelligent and that Gardner's (1983) theory of multiple in
telligences fails by overstating the relevance of savant skills and be
cause it disregards clear psychometric evidence for a general factor.
Vernon's (1971) hierarchical model fails because it cannot accommodate
high levels of cognitive skill, independent from general intelligence
. Evaluation of the Horn-Cattell model (Horn, 1987), Anderson's (1992)
process theory, and Detterman's (1982, 1986) system theory finds that
each can accommodate some aspects of savant syndrome but cannot provi
de a full account. An integration of the Anderson and Detterman models
is suggested, which overcomes the reliance of the former on a unitary
construct to explain a general factor and extends the latter to inclu
de specific aptitudes.