ADAPTATION IS AUTOMATIC

Two experiments were used to test whether selective adaptation for spe ech occurs automatically or instead requires attentional resources. A control condition demonstrated the usual large identification shifts c aused by repeatedly presenting an adapting sound (/wa/, with listeners identifying members of a /ba/-/wa/ test series). Two types of distrac ter tasks were used: (1) Subjects did a rapid series of arithmetic pro blems during the adaptation periods (Experiments 1 and 2), or (2) they made a series of rhyming judgments, requiring phonetic coding (Experi ment 2). A control experiment (Ex periment 3) demonstrated that these tasks normally impose a heavy attentional cost on phonetic processing. Despite this, for both experimental conditions, the observed adaptati on effect was just as large as in the control condition. This result i ndicates that adaptation is automatic, operating at an early, preatten tive level. The implications of these results for current models of sp eech perception are discussed.