We report evidence that auditory verbal hallucinations (AVH) in schizophrenia patients are perceptual distortions lateralized to the left hemisphere. We used a dichotic listening task with repeated presentations of consonant–vowel syllables, a different syllable in the right and left ear. This task produces more correct reports for the right ear syllable in healthy individuals, indicative of left hemisphere speech processing focus. If AVHs are lateralized to the left hemisphere language receptive areas, then this should interfere with correct right ear reports in the dichotic task, which would result in significant negative correlations with severity of AVHs. We correlated the right and left ear correct reports with the PANSS hallucination symptom, and a randomly selected negative symptom, in addition to the sum total of the positive and negative symptoms, in 160 patients with schizophrenia. The results confirmed the predictions with significant negative correlations for the right ear scores with the PANSS hallucination item, and for the sum total of positive symptoms, while all other correlations were close to zero. The results are unambiguous evidence for AVHs as aberrant speech perceptions originating in the left hemisphere.
Auditory verbal hallucinations (AVHs) are a key symptom in schizophrenia (Wing et al., 1974 and David, 1999) which directs attention inwards towards the "voice", with consequences for social interaction and reality orientation. It is estimated that about 70% of diagnosed patients experience auditory hallucinations (Wing et al., 1974 and Shergill et al., 1998). Auditory verbal hallucinations are the subjective experience and conviction of someone speaking to the patient despite the absence of an acoustic signal (Hugdahl et al., 2009). The "voices” are commonly characterized as having an out-of-head origin, often having a negative and commanding emotional valence, and not possible to cognitively control and to avoid for the patient. Patients also often struggle with complying with the demands and requirements imposed by the "voices" (cf. Daalman et al., 2011). There is an ongoing discussion in the literature regarding the underlying mechanisms for AVH, and whether these can best be described as inner speech (e.g. Blakemore et al., 1998 and Allen et al., 2007), traumatic memory (e.g. Badcock et al., 2005 and Waters et al., 2006), or as aberrant speech perceptions (e.g. Fiszdon et al., 2005 and Spencer et al., 2009, see also Jones, 2010 for a thorough review and discussion of existing theoretical models for AVH). What is lacking, however, are hypothesis-driven empirical tests of the different models. In this study we aimed at an empirical test of a speech perception model, with the hypothesis that if AVHs are aberrant speech perceptions, then they should implicate the left hemisphere receptive language areas (Binder et al., 1996 and van den Noort et al., 2008), and interfere with the processing of an external speech sound, that is processed in the same brain areas. Such a hypothesis is derived from previous functional neuroimaging studies showing that the language regions in the brain are in a hyper-excited state during AVHs (Spencer et al., 2009), and that neuronal activation in the left speech perception area is correlated negatively with frequency and severity of AVHs (Plaze et al., 2006). Such a prediction would also follow from the findings by Woodruff et al. (1997) that patients with severe hallucinations showed reduced left temporal cortex response sensitivity to external speech sounds. Adding to this, Aleman and Vercammen (2012) concluded after a review of the existing activation literature that "if auditory hallucinations share a processing system with auditory sense perception, one would not expect an increase in activity upon external auditory stimulation in the auditory areas of patients actively experiencing AVH" (p. 272). A disadvantage with imaging data is however that it is not clear whether an implicated brain region is necessary and sufficient for the function in question, this would require experimental behavioral data. A behavioral task that is a valid indicator of lateralization of speech perception, and has been shown to unequivocally probe left hemisphere function is the consonant–vowel (CV) syllables dichotic listening (DL) task (Studdert-Kennedy and Shankweiler, 1970, Hugdahl and Andersson, 1984, Bryden, 1988 and van den Noort et al., 2008). The CV-syllables DL task produces a very robust right-ear advantage (REA) in the general population, with about 85–90% reliability (Hugdahl and Hammar, 1997), meaning that subjects correctly report the right ear syllable of the dichotic pair more frequently than the left ear syllable. Since the right ear report is a measure of lateralization of speech perception to the left hemisphere, a negative correlation with severity of AVHs, and a non-significant correlation for the left ear stimulus, will be direct evidence in support of a speech perceptual model. The reasoning is that the more frequent and severe AVHs are, the more should such experiences interfere with perception of an external speech sound, revealed as significant negative correlations between AVH scores and DL scores for the right, but not the left, ear stimulus. To test a perceptual hypothesis with a dichotic listening experimental paradigm, we examined a large sample of patients with schizophrenia diagnosis, and calculated the correlation between the patients’ hallucination score on the Positive and Negative Symptom Scale (PANSS) (Kay et al., 1987), and correct reports for the right and left ear stimulus in the DL task, respectively. To examine whether the relationship between DL performance and PANSS score is specific for hallucinations, we also correlated the right and left ear DL correct reports with a negative symptom, randomly chosen among the PANSS negative symptoms. The hypothesis was that a negative symptom should not correlate with DL reports, neither right nor left ear reports, since they should be unrelated to any perceptual disturbances. We have previously reported on the overall structure of relationships between DL performance and the full range of symptom scores in a non-hypothesis driven approach (Hugdahl et al., 2008), we therefore now chose to focus on key symptoms where we would have a priori directed hypotheses.
