The experience of "hearing voices", i. e. auditory hallucinations in the absence of an external acoustic input is a perplexing phenomenon. In addition to being a defining characteristic of schizophrenia, experiences of "hearing voices" may be more common in the general population than what we normally think, which poses a theoretical challenge from a neuropsychological point of view. The overall goal is to track auditory hallucinations from the cognitive (phenomenological) to the neuronal (brain systems and synaptic) levels of explanation, by drawing on my previous research on hemispheric asymmetry and attention-modulation of dichotic listening and functional neuroimaging. I now suggest a new model for explaining "hearing voices" in patients and in healthy individuals. From the phenomenology of what patients and healthy individuals "hearing voices" actually report led me to question current models and theories that auditory hallucinations are "inner speech" or "traumatic memories". Since both patients and healthy individuals "hearing voices" subjectively report experiencing someone "speaking to them" it seems that a perceptual model would better fit the actual phenomenology. A perceptual model can however not explain why patients and healthy individuals differ in the way they cope with and interpret the "voice". An expanded model is therefore advanced that sees auditory hallucinations as a break-down of the dynamic interplay between bottom-up (perceptual) and top-down (inhibitory control) cognitive processes. It is suggested that while both groups show deficient perceptual processing, the patients in addition have impaired inhibitory control functions which prevents them from interpreting the "voices" as coming from inner thought processes. A series of experiments are proposed to test the model.