To the Editor: Ralph E. Hoffman, M.D., and colleagues (1) reported impaired performance of hallucinating schizophrenic patients on a narrative speech perception task and a sentence-repetition task relative to nonhallucinating patients and normal comparison subjects. However, the hallucinating and nonhallucinating patients did not differ on a measure of continuous performance, which indicated that the differences in speech perception were not due to deficits in sustained attention. Dr. Hoffman et al. concluded, "Results support the hypothesis that hallucinated voices in schizophrenia arise from disrupted speech perception and verbal working memory systems rather than from nonlanguage cognitive or attentional deficits" (p. 393). We think that the first part of the conclusion is adequate (i.e., it is supported by the data), but the latter part may be premature. Specifically, the two nonlanguage cognitive functions that have figured most prominently in the cognitive theories of hallucination—namely, reality monitoring and imagery vividness—were not assessed by the authors. Previous studies have revealed differences between hallucinating and nonhallucinating patients on tasks measuring these functions (e.g., references 2–4). Although, as Dr. Hoffman et al. pointed out, one PET study of hallucinations failed to find an activation of Broca’s area, which challenges the reality-monitoring account that hallucinations are misidentified thoughts. All studies of functional brain imaging during auditory hallucinations (e.g., reference 5) have demonstrated activation of the temporal areas, consistent with speech perception as imagery accounts (auditory imagery has been shown to activate temporal auditory association areas ). In addition, the speech perception tasks used by Dr. Hoffman et al. may involve mental imagery to a significant extent. The process of filling in the blanks during speech perception to reduce acoustically ambiguous speech is clearly dependent on imagery. Indeed, Kosslyn and Sussman (7) have argued that "immaculate perception" does not exist: in most circumstances, perception and imagery are intimately intertwined. This underscores the importance of a thorough investigation of the differences between patients with and without hallucinations on multiple tasks aimed at measuring language-specific (e.g., speech perception) as well as non-language-specific (e.g., reality monitoring and mental imagery) cognitive functions that may be involved in the cognitive basis of hallucinations. It is too early yet to discard the possible contribution of disruptions in nonlanguage cognitive functions to the occurrence of hallucinations in schizophrenia.