Auditory verbal hallucinations (AVH) in schizophrenia patients assumingly result from a state inadequate activation of the primary auditory system. We tested brain responsiveness to auditory stimulation in healthy controls (n = 26), and in schizophrenia patients that frequently (n = 18) or never (n = 11) experienced AVH. Responsiveness was assessed by driving the EEG with click-tones at 20, 30 and 40 Hz. We compared stimulus induced EEG changes between groups using spectral amplitude maps and a global measure of phase-locking (GFS). As expected, the 40 Hz stimulation elicited the strongest changes. However, while controls and non-hallucinators increased 40 Hz EEG activity during stimulation, a left-lateralized decrease was observed in the hallucinators. These differences were significant (p = .02). As expected, GFS increased during stimulation in controls (p = .08) and non-hallucinating patients (p = .06), which was significant when combining the two groups (p = .01). In contrast, GFS decreased with stimulation in hallucinating patients (p = 0.13), resulting in a significantly different GFS response when comparing subjects with and without AVH (p < .01). Our data suggests that normally, 40 Hz stimulation leads to the activation of a synchronized network representing the sensory input, but in hallucinating patients, the same stimulation partly disrupts ongoing activity in this network.
About 70% of patients with schizophrenia patients have auditory verbal hallucinations (AVH); these voices often comment on inner aspects of the patients' live. In the absence of sensory input, AVH therefore are likely to involve top-down frontal functions (producing context related language), sensory areas (accounting for the compelling realness) and automatic perceptual processes making the voices immediately recognizable. The currently most accepted hypothesis on the biological mechanisms of AVH is thus that during inner speech, there is an abnormal co-activation of sensory and perceptual systems leading to a faulty misattribution of inner-psychic processes to external sources (Allen et al., 2008).
Research on the neurobiology of AVH has therefore emphasized a network perspective, trying to understand the unitary quality of AVH across multiple perceptual and cognitive domains as dysfunctional interactions among the involved functions and regions. AVHs correlate with reduced cortical gray matter in the left temporal lobe (e.g. Flaum et al., 1995 and Gaser et al., 2004), and Heschl's gyrus (e.g. Gaser et al., 2004, Sumich et al., 2005 and Hubl et al., 2010), whereas neuronal activity in similar regions seemed to be increased during the acute experience of hallucinations (e.g. Dierks et al., 1999 and Hubl et al., 2007). Accordingly, inhibitory transcranial magnetic stimulation of the left temporal cortex diminished AVH (Jandl et al., 2006 and Horacek et al., 2007). Recent studies investigating white-matter structural connectivity showed increased connectivity between left frontal language-related areas and temporal regions associated with auditory perception (Hubl et al., 2007 and Shergill et al., 2007). Allen (Allen et al., 2008) therefore suggested a model in which an imbalance of bottom-up and top-down processes terminates in these erroneous perceptions.
Independent of schizophrenia, the issue of how information distributed across multiple cognitive–perceptual modules is transiently assembled into a unitary mental representation (the so called binding problem) has received widespread attention. The current hypothesis is that binding occurs when the different modules synchronize in the gamma (~ 40 Hz) frequency range (Gray and Singer, 1989 and Singer and Gray, 1995). Additionally, there is a long-known link between gamma-band activity and auditory processing. When subjects hear sounds that repeat at a certain frequency, EEG spectral amplitude increases at that frequency (Galambos et al., 1981), which is called auditory steady-state response (ASSR). Interestingly, in healthy subjects, the ASSR is strongest in the gamma band around 40 Hz (Galambos et al., 1981) and has been localized to the auditory cortices (Pantev et al., 1991), and is strongly reduced when consciousness is lost (Plourde et al., 2008).
The relation of gamma-band activity, auditory processing, and AVH in schizophrenia is thus interesting. However, only a few studies have specifically related gamma band activity to AVH. Spencer (Spencer et al., 2009) reported that the phase-locking of gamma-band steady-state responses in the left superior temporal lobe correlated positively with AVH, and Mulert et al., 2011, found that during 40 Hz stimulation, lagged coherence between left and right primary auditory cortices increased with the severity of AVH. Ford (Ford and Mathalon, 2005) however failed to find an association of gamma-band coherence during auditory–verbal integration and AVH.
The current study further investigated the relation of the 40 Hz steady state response and AVH. The novelty of the present study lays in the method to quantify synchronization. In previous studies, synchronization has either been defined as the stability of the relation between stimulus onset and EEG phase (e.g. Spencer et al., 2009 and Uhlhaas and Singer, 2010) or as coherence with a 90° phase lag, avoiding problems of volume conduction (Mulert et al., 2011). Conceptually, binding is however assumed to occur thru simultaneous, non-lagged oscillations of different regions (Gray and Singer, 1989, Singer and Gray, 1995 and Koenig et al., 2005b). This is neither assessed when investigating phase-locking to a stimulus, nor by lagged coherence, which explicitly discards simultaneous oscillations. We have thus proposed a method called global field synchronization (GFS) that estimates the amount of phase-locking among all active regions at a given frequency.
Under the hypotheses that AVH arise from an excessive coactivation of internal mental representations with sensory systems, and that co-activation of remote, but functionally related brain regions is mediated by phase-locking of gamma-band activity, we expected that gamma-band GFS during auditory driving would show deviations that are selective to patients prone to AVH in contrast to patients that do not experience AVH and to healthy controls. To provide links to the existing literature, we also analyzed EEG scalp spectral amplitude maps and estimates of spectral amplitudes in regions of interest.
