The functional neuroanatomy of speech perception has been difficult to char
acterize. Part of the difficulty, we suggest, stems from the fact that the
neural systems supporting 'speech perception' vary as a function of the tas
k. Specifically, the set of cognitive and neural systems involved in perfor
ming traditional laboratory speech perception tasks, such as syllable discr
imination or identification, only partially overlap those involved in speec
h perception as it occurs during natural language comprehension. In this re
view, we argue that cortical fields in the posterior-superior temporal lobe
, bilaterally, constitute the primary substrate for constructing sound-base
d representations of speech, and that these sound-based representations int
erface with different supramodal systems in a task-dependent manner. Tasks
that require access to the mental lexicon (i.e. accessing meaning-based rep
resentations) rely on auditory-to-meaning interface systems in the cortex i
n the vicinity of the left temporal-parietal-occipital junction. Tasks that
require explicit access to speech segments rely on auditory-motor interfac
e systems in the left frontal and parietal lobes. this auditory-motor inter
face system also appears to be recruited in phonological working memory.