Introduction
In Kant's view of transcendental idealism, the world we perceive is shaped through an intrinsically generated process and not by external objects as they are. This implies an interaction between the observer and the observed, where the spatial and temporal properties of the perception are constrained by sensory faculties and therefore essentially subjective experiences. This fundamental subjectivity of perception is most vividly illustrated by the phenomenon of hallucinations. Hallucinations can occur in every sensory modality and often accompany a psychiatric, neurological, general medical condition or substance use. In the general population, hallucinations most often occur in the auditory domain and may take the form of auditory verbal hallucinations (AVH) (Beavan et al. Reference Beavan, Read and Cartwright2011). This type of hallucination, in disease and health, will be the focus of this review.
In psychiatry, AVH are classified as a psychotic symptom. Indeed, 70% of patients with schizophrenia report AVH (Larøi et al. Reference Larøi, Sommer, Blom, Fernyhough, Ffytche, Hugdahl, Johns, McCarthy-Jones, Preti, Raballo, Slotema, Stephane and Waters2012). However, AVH occur across a broad range of psychiatric diagnoses such as bipolar disorder (11–63%; Toh et al. Reference Toh, Thomas and Rossell2015), borderline personality disorder (20–50%; Schroeder et al. Reference Schroeder, Fisher and Schafera2013), autism (20%; Kyriakopoulos et al. Reference Kyriakopoulos, Stringaris, Manolesou, Radobuljac, Jacobs, Reichenberg, Stahl, Simonoff and Frangou2015) and post-traumatic stress disorder (50%; Anketell et al. Reference Anketell, Dorahy, Shannon, Elder, Hamilton, Corry, MacSherry, Curran and O'Rawe2010). In addition, AVH occur in people with hearing loss (Linszen et al. Reference Linszen, Brouwer, Heringa and Sommer2016) and in neurological disorders that excite the temporal cortex, such as epilepsy and migraine (Vreeburg et al. Reference Vreeburg, Leijten and Sommer2016). Thus, AVH are an important phenomenon in its own right, and can be said to represent an altered state of consciousness from a phenomenological point of view. As pointed out by Hugdahl (Reference Hugdahl2015) and exemplified above, AVH cross the border between different psychiatric disorders, and neurological diseases, since they are found also in Parkinson's disease, Alzheimer's diseases and epilepsy, to mention a few examples. Finally, they cross the border between disease and health, by being present also in a certain percentage of the general population (Sommer et al. Reference Sommer, Daalman, Rietkerk, Diederen, Bakker, Wijkstra and Books2010).
Several theories exist to explain AVH, including overactivation of the auditory cortex (Penfield, Reference Penfield1958), misattribution of internally generated speech (Frith & Done, Reference Frith and Done1988; McGuire et al. Reference McGuire, Silbersweig, Wright, Murray, David, Rrackowiak and Firth1995), errors in the corollary discharge for labelling internally generated events (Feinberg, Reference Feinberg1978; Allen et al. Reference Allen, Freeman, Johns and McGuire2006) and disturbances in episodic memory retrieval (Copolov et al. Reference Copolov, Seal, Maruff, Ulusoy, Wong, Tochon-Danguy and Egan2003). A growing body of neuroimaging research has provided support for a neurobiological basis of AVH (Hugdahl et al. Reference Hugdahl, Løberg and Nygård2009; Jardri et al. Reference Jardri, Pouchet, Pins and Thomas2011; Palaniyappan et al. Reference Palaniyappan, Balain, Radua and Liddle2012; Modinos et al. Reference Modinos, Costafreda, Van Tol, McGuire, Aleman and Allen2013) and also for the efficacy of various treatment strategies. Here, we provide an overview of structural, physiological and functional imaging literature in AVH. Potential underlying mechanisms are discussed with a particular interest for the interaction of available treatment options (pharmacotherapy, brain stimulation and cognitive–behavioural therapy; CBT) with the neurobiology of AVH.
Anatomy and AVH
Early magnetic resonance imaging (MRI) studies reported significant negative associations between superior temporal gyrus (STG) volume (especially Heschl's gyrus; HG) and AVH severity in patients with schizophrenia (Barta et al. Reference Barta, Powers, Richards, Pearison and Tune1990; Levitan et al. Reference Levitan, Ward and Catts1999). This association was also observed in bipolar disorder with AVH (Stanfield et al. Reference Stanfield, Moorhead, Job, McKirdy, Sussmann, Hall, Giles, Johnstone, Lawrie and Mcintosh2009). Later, voxel-based morphometry (VBM) studies complemented these findings by showing that volume reductions in the bilateral STG and the left insula are associated with AVH severity in schizophrenia (for meta-analyses, see Palaniyappan et al. Reference Palaniyappan, Balain, Radua and Liddle2012; Modinos et al. Reference Modinos, Costafreda, Van Tol, McGuire, Aleman and Allen2013).
In schizophrenia patients with AVH, cortical thickness is reduced in HG, Broca's area and Wernicke's area of the left hemisphere, while an increase is observed in the bilateral insula, cingulate gyrus, dorsal middle frontal gyrus and parietal lobe compared with non-hallucinators (Van Swam et al. Reference Van Swam, Federspiel, Hubl, Wiest, Boesch, Vermathen, Kreis, Strik and Dierks2012); others report a reduction in right HG (Chen et al. Reference Chen, Liang, Pu, Song, Mwansisya, Yang, Liu, Liu, Shan and Xue2015). L Mørch-Johnsen et al. (unpublished observations) found that cortical thickness, but not area, was significantly reduced in AVH schizophrenia patients, thus extending the VBM studies which only provide data on the tissue volume or density. Interestingly, the only significant reduction, after correcting for multiple comparisons, was in the left STG, and HG. Furthermore, there is evidence that non-clinical hallucinators show an intermediate phenotype compared with controls and schizophrenia patients with AVH. Especially, cortical thicknesses of the left paracentral lobule, left pars orbitalis, right fusiform gyrus and right inferior temporal gyrus were found to show this continuum effect (Van Lutterveld et al. Reference Van Lutterveld, van den Heuvel, Diederen, de Weijer, Begemann, Brouwer, Daalman, Blom, Kahn and Sommer2014).
Furthermore, the occurrence of AVH was classified with an accuracy of 83.6% in first-episode psychotic patients based upon the structural covariance of Broca's area, insula and intraparietal sulcus (Yun et al. Reference Yun, Nyun Kim, Young Lee, Chon and Soo Kwon2015), suggesting that morphological aspects of these regions are predictive of AVH in psychotic disorders.
Diffusion-weighted MRI studies show that schizophrenia patients with AVH have significantly lower white matter integrity, as defined by fractional anisotropy (FA), in the left arcuate fasciculus compared with controls and non-hallucinating schizophrenia patients (Geoffroy et al. Reference Geoffroy, Houenou, Duhamel, Amad, De Weijer, Ćurčić-Blake, Linden, Thomas and Jardri2014; McCarthy-Jones et al. Reference McCarthy-Jones, Oestreich and Whitford2015). In addition, magnetization transfer of the left arcuate fasciculus is increased in schizophrenia patients and non-clinical hallucinators, suggesting a higher free water concentration due to degradation of axons or glia cells (De Weijer et al. Reference De Weijer, Neggers, Diederen, Mandl, Kahn, Hulshoff Pol and Sommer2013). The arcuate fasciculus connects the STG with the inferior frontal gyrus (IFG) and is believed to be a key white matter tract in the language network (Catani et al. Reference Catani, Jones and Ffytche2005; Catani & de Schotten, Reference Catani and de Schotten2012).
Furthermore, AVH are associated with a distributed pattern of connectivity reductions, involving the uncinate fasciculus, corpus callosum, thalamic radiation and fronto-occipital fibers (Knöchel et al. Reference Knöchel, O'Dwyer, Alves, Reinke, Magerkurth, Rotarska-Jagiela, Prvulovic, Hampel, Linden and Oertel-Knöchel2012; Ćurčić-Blake et al. Reference Ćurčić-Blake, Nanetti, van der Meer, Cerliani, Renken, Pijnenborg and Aleman2013b ). Alterations in white matter integrity of fronto-temporal tracts have been observed in adolescents with psychotic experiences (mainly AVH), consisting of FA increases in the uncinate fasciculus, arcuate fasciculus and white matter near the striatum (O'Hanlon et al. Reference O'Hanlon, Leemans, Kelleher, Clarke, Roddy, Coughlan, Harley, Amico, Hoscheit, Tiedt, Tabish, McGettigan, Frodl and Cannon2015). The uncinate fasciculus connects temporal limbic regions such as the hippocampus and amygdala to the IFG and orbitofrontal gyrus (Kier et al. Reference Kier, Staib, Davis and Bronen2004).
The findings of increased white matter connectivity during adolescence and decreases associated with AVH in adulthood may indicate an aberrant developmental trajectory in which early maturation of connectivity in language and emotional brain networks could introduce a mismatch with the maturation of executive control also occurring during this period (Blakemore & Choudhury, Reference Blakemore and Choudhury2006), rendering the system more susceptible to errors of self-source monitoring and inhibition (Waters et al. Reference Waters, Allen, Aleman, Fernyhough, Woodward, Badcock, Barkus, Johns, Varese, Menon, Vercammen and Laroi2012). However, longitudinal investigations are much needed to support this.
Blood-flow physiology and AVH
Positron emission tomography (PET), single photon emission tomography (SPECT) and arterial spin labelling (ASL) techniques are used to study changes in blood flow and metabolism associated with AVH. Using fluoro-deoxy-glucose (FDG)-PET, both negative (Gur et al. Reference Gur, Mozley, Resnick, Mozley, Shtasel, Gallacher and Gur1995) and positive (Volkow et al. Reference Volkow, Wolf, van Gelder, Brodie, Overall, Cancro and Gomez-Mont1987; Liddle et al. Reference Liddle, Friston, Frith, Hirsch, Jones and Frackowiak1992) associations have been reported between hallucination severity and temporal lobe glucose metabolism. Using the same technique, several brain regions including the bilateral supplementary motor area, anterior cingulate cortex (ACC), medial and superior frontal gyri and cerebellum showed a higher glucose metabolism during periods of active AVH compared with silent periods (Parellada et al. Reference Parellada, Lomena, Font, Pareto, Gutierrez, Simo, Fernandez-Egea, Pavia, Ros and Bernardo2008). Furthermore, antipsychotic-naive patients with AVH showed significantly increased metabolism in the left middle and superior temporal gyrus, bilateral middle and superior frontal gyrus and left caudate nucleus compared with those without AVH, while decreased metabolism was observed in the hippocampus, cerebellum and parietal cortex (Horga et al. Reference Horga, Parellada, Lomena, Fernandez-Egea, Mane, Font, Falcon, Konova, Pavia, Ros and Bernardo2011).
Using a within-subject comparison design, a 15O-PET study revealed activation in the thalamus and right striatum, as well as the bilateral parahippocampal gyrus, right ACC and left orbitofrontal cortex during active hallucination (Silbersweig et al. Reference Silbersweig, Stern, Frith, Cahill, Holmes, Grootoonk, Seaward, McKenna, Chua, Schnorr, Jones and Frackowiak1995). Another symptom-capture study using this technique reports activation in the right medial prefrontal and medial temporal gyrus, left STG, left parahippocampus and left posterior cingulate gyrus (Copolov et al. Reference Copolov, Seal, Maruff, Ulusoy, Wong, Tochon-Danguy and Egan2003).
Sabri et al. (Reference Sabri, Erkwoh, Schreckenberger, Owega, Sass and Buell1997) observed a negative association between left thalamic blood flow and AVH severity in never-medicated schizophrenia patients which disappeared after treatment with antipsychotic medication, using SPECT. Furthermore, Gordon et al. (Reference Gordon, Barry, Anderson, Fawdry, Yong, Grunewald and Meares1994) found that blood flow in the left temporal lobe negatively correlated with hallucination severity. Within-subject comparison of SPECT data between periods of regular v. remitted AVH showed associations with activity in language-related areas such as Broca's region, the left temporal cortex and the ACC (Suzuki et al. Reference Suzuki, Yuasa, Minabe, Murata and Kurachi1993; McGuire et al. Reference McGuire, Silbersweig, Wright, Murray, David, Rrackowiak and Firth1995). Using ASL, increased perfusion of left HG was found to be associated with the experience of AVH (Homan et al. Reference Homan, Kindler, Hauf, Walther, Hubl and Dierks2013).
Brain physiology studies, thus, show that AVH are related to abnormal metabolism and blood flow in the left temporal cortex, hippocampal complex, thalamus, ACC and frontal cortex. Interestingly, the studies sensitive to state effects (Silbersweig et al. Reference Silbersweig, Stern, Frith, Cahill, Holmes, Grootoonk, Seaward, McKenna, Chua, Schnorr, Jones and Frackowiak1995; Copolov et al. Reference Copolov, Seal, Maruff, Ulusoy, Wong, Tochon-Danguy and Egan2003; Parellada et al. Reference Parellada, Lomena, Font, Pareto, Gutierrez, Simo, Fernandez-Egea, Pavia, Ros and Bernardo2008; Homan et al. Reference Homan, Kindler, Hauf, Walther, Hubl and Dierks2013) consistently show evidence of increased metabolism and blood flow during the experience of AVH v. inactive periods. However, studies more sensitive to trait aspects remain equivocal in their findings, possibly related to limited sample sizes (Volkow et al. Reference Volkow, Wolf, van Gelder, Brodie, Overall, Cancro and Gomez-Mont1987; Liddle et al. Reference Liddle, Friston, Frith, Hirsch, Jones and Frackowiak1992; Gordon et al. Reference Gordon, Barry, Anderson, Fawdry, Yong, Grunewald and Meares1994; Gur et al. Reference Gur, Mozley, Resnick, Mozley, Shtasel, Gallacher and Gur1995; Sabri et al. Reference Sabri, Erkwoh, Schreckenberger, Owega, Sass and Buell1997; Horga et al. Reference Horga, Parellada, Lomena, Fernandez-Egea, Mane, Font, Falcon, Konova, Pavia, Ros and Bernardo2011).
Trait-associated brain activation and AVH
Functional MRI (fMRI) studies measuring trait aspects of AVH show that regions including the bilateral ACC), the left superior and middle temporal gyrus and the left premotor cortex are hypoactive in patients with psychosis experiencing AVH during language-related tasks (for a meta-analysis, see Kühn & Gallinat, Reference Kühn and Gallinat2012). These findings are similar to what Hubl et al. (Reference Hubl, Koening, Strik, Melie Garcia and Dierks2007) and Woodruff et al. (Reference Woodruff, Wright, Bullmore, Brammer, Howard, Williams, Shapleske, Rossell, David, McGuire and Murray1997) found where there was reduced activation in the STG area, particularly on the left, when patients with active hallucinations were exposed to externally presented sounds. Both studies concluded that this phenomenon indicated a kind of ‘competition for neuronal resources’, between internally and externally generated neuronal activity, where the internal activity ‘wins’. See also Kompus et al. (Reference Kompus, Westerhausen and Hugdahl2011), who called this phenomenon a ‘paradoxical effect’. It could be argued that the reason for lack of increased activity in response to an external sound source is because the hyperactivation caused by the AVH satiates the fMRI-BOLD (blood oxygen level-dependent) response, a kind of ceiling effect that prohibits additional activation caused by the external source. If so, the ‘paradoxical effect’ suggested by Kompus et al. (Reference Kompus, Westerhausen and Hugdahl2011) would be a methodological artefact. This is, however, not a likely explanation, since Hubl et al. (Reference Hubl, Koening, Strik, Melie Garcia and Dierks2007) used electrophysiology (event-related potential) recordings, which are not dependent on haemodynamic changes. Therefore, the most likely explanation for a ‘race model’, and ‘paradoxical effect’, is that AVH cause spontaneous hyperactivity, resulting in increased metabolic demands, seen in the fMRI activation changes.
An important and largely unsolved issue (but see the Discussion), is what changes at the transmitter and receptor level of explanation can explain these changes in neuronal activity (event-related potential) and activation (fMRI) at the imaging level of explanation. To answer questions like these we need to get down to the neurochemistry of AVH, which is a new area for research (Homan et al. Reference Homan, Vermathen, Van Swam, Federspiel, Boesch, Strik, Dierks, Hubl and Kreis2014; Hugdahl et al. Reference Hugdahl2015). Homan et al. (Reference Homan, Vermathen, Van Swam, Federspiel, Boesch, Strik, Dierks, Hubl and Kreis2014) and Hugdahl et al. (Reference Hugdahl2015) used MR spectroscopy as a new avenue into revealing the neurochemistry of AVH, and Homan et al. (Reference Homan, Vermathen, Van Swam, Federspiel, Boesch, Strik, Dierks, Hubl and Kreis2014) reported increased levels of the metabolite N-acetylaspartate in healthy controls in frontal areas, while Hugdahl et al. (Reference Hugdahl2015) found increased levels of glutamate/glutamine in both frontal and temporal lobe areas in trait-AVH subjects. Findings like these may have important consequences for the development of new, symptom-targeted, drugs that point beyond the classic dopamine hypothesis.
Kang et al. (Reference Kang, Kim, Seok, Chun, Lee and Park2009) report reduced activation of the amygdala and hippocampus in patients with AVH compared with non-hallucinators and controls in response to a gender discrimination task involving negative auditory stimuli. The connectivity between Wernicke's and Broca's areas is found to be reduced in patients experiencing AVH compared with those without AVH, suggesting a reduced temporo-frontal flow of information during a semantic decision task (Ćurčić-Blake et al. Reference Ćurčić-Blake, Liemburg, Vercammen, Swart, Knegtering, Bruggeman and Aleman2013a ).
In contrast, Sanjuan et al. (Reference Sanjuan, Lull, Aguilar, Martí-Bonmatí, Moratal, Gonzalez, Robles and Keshavan2007) report increased activation of frontal, temporal and limbic regions in schizophrenia patients with AVH compared with healthy controls when confronted with emotional spoken words. Furthermore, patients showed increased activation of the parahippocampus and amygdala compared with patients without hallucinations and controls in response to emotional words (Escarti et al. Reference Escarti, de la Iglesia-Vaya, Marti-Bonmati, Robles, Carbonell, Lull, Garcia-Marti, Manjon, Aguilar, Aleman and Sanjuan2010). It has also been reported that schizophrenia patients experiencing AVH show temporo-frontal hypercoupling during a speech perception task (Lavigne et al. Reference Lavigne, Rapin, Metzak, Whitman, Jung, Dohen, Lœvenbruck and Woodward2015).
Although these findings seem contradictory, they could be explained by differences in experimental setup. Where studies involving a cognitive component to verbal processing (Kang et al. Reference Kang, Kim, Seok, Chun, Lee and Park2009; Ćurčić-Blake et al. Reference Ćurčić-Blake, Nanetti, van der Meer, Cerliani, Renken, Pijnenborg and Aleman2013b ) tend to show decreases in activity and connectivity, studies involving speech perception tend to report overactivation and overconnectivity (Sanjuan et al. Reference Sanjuan, Lull, Aguilar, Martí-Bonmatí, Moratal, Gonzalez, Robles and Keshavan2007; Escarti et al. Reference Escarti, de la Iglesia-Vaya, Marti-Bonmati, Robles, Carbonell, Lull, Garcia-Marti, Manjon, Aguilar, Aleman and Sanjuan2010; Lavigne et al. Reference Lavigne, Rapin, Metzak, Whitman, Jung, Dohen, Lœvenbruck and Woodward2015). These findings thus suggest that AVH are associated with a hypersensitivity to auditory stimuli, while also showing reduced processing of stimuli in association with cognitive demand (Kang et al. Reference Kang, Kim, Seok, Chun, Lee and Park2009; Ćurčić-Blake et al. Reference Ćurčić-Blake, Nanetti, van der Meer, Cerliani, Renken, Pijnenborg and Aleman2013b ). This latter reduced processing may be caused by competition for cerebral resources between the AVH and language-related tasks, possibly related to an attentional bias for internally driven events (Kompus et al. Reference Kompus, Westerhausen and Hugdahl2011). Therefore, these findings seem consistent with both the auditory overactivation hypothesis (Penfield, Reference Penfield1958), as well as the misattribution of internally generated speech hypothesis (Frith & Done, Reference Frith and Done1988; McGuire et al. Reference McGuire, Silbersweig, Wright, Murray, David, Rrackowiak and Firth1995).
State-associated brain activation and AVH
During the experience of AVH, Broca's area, insula, IFG, left middle and superior temporal gyrus, left inferior parietal lobule and left (para)hippocampal region are consistently reported to show overactivation (Jardri et al. Reference Jardri, Pouchet, Pins and Thomas2011; Kühn & Gallinat, Reference Kühn and Gallinat2012).
Activity in right homologues of language regions is associated with the occurrence of AVH, in contrast to production of inner speech, which is strongly associated with left language regions (Woodruff et al. Reference Woodruff, Wright, Bullmore, Brammer, Howard, Williams, Shapleske, Rossell, David, McGuire and Murray1997; Copolov et al. Reference Copolov, Seal, Maruff, Ulusoy, Wong, Tochon-Danguy and Egan2003; Sommer et al. Reference Sommer, Diederen, Blom, Willems, Kushan, Slotema, Boks, Daalman, Hoek, Neggers and Kahn2008). The same regions were observed to activate in non-clinical hallucinators, suggesting a common neural network underlying AVH in the psychosis continuum (Diederen et al. Reference Diederen, Daalman, De Weijer, Neggers, Van Gastel, Blom, Kahn and Sommer2012). Lastly, non-clinical hallucinators show increased connectivity between the bilateral STG, and between the left parahippocampal area and left IFG (Diederen et al. Reference Diederen, Neggers, de Weijer, van Lutterveld, Daalman, Eickhoff, Clos, Kahn and Sommer2013).
The period before the onset of AVH is associated with decreased activity in the left parahippocampal gyrus, and also in the left STG and medial frontal gyrus and right IFG (Diederen et al. Reference Diederen, Neggers, Daalman, Blom, Goekoop, Kahn and Sommer2010). Hoffman et al. (Reference Hoffman, Fernandez, Pittman and Hampson2011) report increased activation of the left IFG compared with its right homologue preceding AVH; the opposite pattern was observed in the left and right posterior temporal area. This activation could correspond to the triggering mechanism of AVH, which may have a component in memory.
Thus, contrary to trait-associated brain activation studies, the literature above clearly suggests that the phenomenon of AVH itself is accompanied by spontaneous over-activity of language-related brain regions and preceded by activity in memory-related areas. Therefore, this literature provides some evidence for the misattribution of internally generated speech hypothesis, the corollary discharge hypothesis (Feinberg, Reference Feinberg1978; Allen et al. Reference Allen, Freeman, Johns and McGuire2006) and episodic memory retrieval hypothesis (Copolov et al. Reference Copolov, Seal, Maruff, Ulusoy, Wong, Tochon-Danguy and Egan2003). The reviewed studies also cast light on an alternative model, provided by Hugdahl (Reference Hugdahl2009), who suggested that AVH arise because of spontaneous bottom-up hyperactivation, which are not inhibited because of top-down hypoactivation. It was further suggested that bottom-up and top-down processes may be differentially mediated through glutamate overactivity, being the major excitatory transmitter, and γ-amino-butyric-acid (GABA) underactivity, being the major inhibitory transmitter (Hugdahl et al. Reference Hugdahl, Craven, Nygård, Løberg, Berle, Johnsen, Kroken, Specht, Andreassen and Ersland2015).
Treatment strategies
Pharmacotherapy
Two studies have specifically investigated the efficacy of antipsychotic medication for AVH in patients with schizophrenia by using the P3 item of the Positive and Negative Syndrome Scale as the outcome measure (Sommer et al. Reference Sommer, Slotema, Daskalakis, Derks, Blom and Van Der Gaag2012; Johnsen et al. Reference Johnsen, Jørgensen, Kroken and Løberg2013). Both studies point out a significant and clinically relevant treatment effect of several first- and second-generation antipsychotics.
Striatal dopamine synthesis and release are elevated in most patients with a psychotic disorder, yet this is not the case in every individual patient (Howes et al. Reference Howes, Montgomery, Asselin, Murray, Grasby and McGuire2007; Bonoldi & Howes, Reference Bonoldi and Howes2014). Antipsychotic medication blocks the D2/3 receptor, thereby acting downstream of the dopaminergic abnormalities (Howes et al. Reference Howes, Kambeitz, Kim, Stahl, Slifstein, Abi-Dargham and Kapur2012). The efficacy of antipsychotic medication against AVH may thus be related to normalization of dopaminergic signalling. However, one in three patients with psychosis reports antipsychotic-resistant AVH (Johnsen et al. Reference Johnsen, Jørgensen, Kroken and Løberg2013), suggesting that AVH are not related to disrupted dopaminergic signalling in these patients. Instead, disrupted glutamine signalling could also be underlying AVH (Hugdahl et al. Reference Hugdahl, Craven, Nygård, Løberg, Berle, Johnsen, Kroken, Specht, Andreassen and Ersland2015).
Treatment with atypical antipsychotics is associated with a normalization of ACC activation during cognitive demand (Snitz et al. Reference Snitz, MacDonald and Carter2006), increased activity in the IFG during a verbal fluency task (Fusar-Poli et al. Reference Fusar-Poli, Broome, Matthiasson, Williams, Brammer and McGuire2007), decreased functional connectivity in frontal, temporal and subcortical regions including the IFG, STG and parahippocampal gyrus during rest (Lui et al. Reference Lui, Li, Deng, Jiang, Wu, Tang, Yue, Huang, Chan, Collier, Meda, Pearlson, Mechelli, Sweeney and Gong2010) and an increase in functional connectivity during rest between the striatum and the ACC, dorsolateral prefrontal cortex (DLPFC), hippocampus and anterior insula (Sarpal et al. Reference Sarpal, Robinson, Lencz, Argyelan, Ikuta, Karlsgodt and Malhotra2015). Although none of these studies specifically investigated AVH as an outcome measure, it is interesting that many of the regions implicated in AVH such as the IFG, STG, hippocampus and ACC also show a normalization of functional responses after treatment. These studies thus provide circumstantial evidence that a normalization of brain function induced by antipsychotic medication may underlie efficacy on AVH in responders. Direct investigation of this putative effect, however, is much needed.
Brain stimulation
Transcranial magnetic stimulation (TMS) is a widely investigated technique for treatment of AVH. Efficacy of 1 Hz-TMS is supported by meta-analysis with an effect size of 0.44 (Slotema et al. Reference Slotema, Blom, Van Lutterveld, Hoek and Sommer2014), reducing frequency and severity of AVH. A large trial with theta-burst TMS reported negative findings (Koops et al. Reference Koops, van Dellen, Schutte, Nieuwdorp, Neggers and Sommer2016).
Usually, the area of stimulation is the left temporo-parietal junction as this is found to be most effective (Hoffman et al. Reference Hoffman, Hampson, Wu, Anderson, Gore, Buchanan, Constable, Hawkins, Sahay and Krystal2007; Slotema et al. Reference Slotema, Blom, Van Lutterveld, Hoek and Sommer2014). Responders to TMS show increased cerebral blood flow in the left STG preceding 4 weeks of treatment compared with non-responders (Homan et al. Reference Homan, Kindler, Hauf, Walther, Hubl and Dierks2013). Also, decreases in cerebral blood flow of the auditory cortex, left Broca's area and cingulate gyrus after TMS treatment compared with sham are reported (Kindler et al. Reference Kindler, Homan, Jann, Federspiel, Flury, Hauf, Strik, Dierks and Hubl2013). These findings suggest that the clinical effect of TMS may come from a normalization of the hyperperfusion of the language regions implicated in AVH.
Using FDG-PET, it was shown that metabolism in the left temporal gyrus decreases after stimulation on this area, correlating with AVH severity. Also, a metabolic increase was observed in the right temporal gyrus and the frontal cortex. Furthermore, better treatment response was predicted by higher pretreatment metabolism in the left temporal and parahippocampal regions (Horacek et al. Reference Horacek, Brunovsky, Novak, Skrdlantova, Klirova, Bubenikova-Valesova, Krajca, Tislerova, Kopecek, Spaniel, Mohr and Heschl2007). Recently, Klirova et al. (Reference Klirova, Horacek, Novak, Cermak, Spaniel, Skrdlantova and Höschl2013) reported that metabolism-guided TMS stimulation on the temporoparietal cortex results in stronger treatment effects.
The clinical effect of 1 Hz-TMS is no longer observed after 1 month of follow-up (Slotema et al. Reference Slotema, Daalman, Blom, Diederen, Hoek and Sommer2012), indicating that repetitive TMS treatment may need to be applied periodically in order to remain its efficacy. The lack of a long-lasting clinical effect suggests that the observed altered neural dynamics discussed above return to their pathological state at follow-up, but this needs to be investigated. Until now, TMS is mainly used to treat medication-resistant AVH. Given its low risk for side effects it could also be used in earlier stages, even before a first antipsychotic attempted, especially when delusions are absent.
Another emerging treatment for AVH is transcranial direct current stimulation (tDCS), which is based on increasing cortical excitability near the anode (placed over the temporo-parital cortex) and decreasing excitability near the cathode (placed over the prefrontal cortex). tDCS was found to be effective against AVH in medication-resistant schizophrenia patients (Brunelin et al. Reference Brunelin, Mondino, Gassab, Haesebaert, Gaha, Suaud-Chagny, Saoud, Mechri and Poulet2012). Mondino et al. (Reference Mondino, Jardri, Suaud-Chagny, Saoud, Poulet and Brunelin2016) showed that active tDCS treatment is associated with a reduction in AVH and a reduction in resting-state functional connectivity between the left temporo-parietal cortex and right IFG. Also, they observed increased resting-state connectivity between the left temporo-parietal cortex and the left angular gyrus, left DLPFC and left precuneus.
Psychological interventions
CBT is the most investigated psychological intervention for AVH, with an average effect size of 0.44 (Van der Gaag et al. Reference Van der Gaag, Valmaggia and Smit2014). The primary aim of CBT is not to reduce the presence of AVH but to accomplish a reappraisal of the meaning of AVH, thereby reducing the emotional reaction and stress in response to them. The neural response to CBT has not yet been studied for AVH specifically, but there are some reports on the effects in relation to positive psychotic symptoms.
Kumari et al. (Reference Kumari, Fannon, Peters, Ffytche, Sumich, Premkumar, Anilkumar, Andrew, Phillips, Williams and Kuipers2011) report on the neural response to fearful stimuli before and after CBT for psychosis. Patients receiving CBT showed an attenuated neural response co-occurring with positive symptom reductions after receiving CBT. The regions showing this treatment response comprised the IFG, insula, putamen, thalamus and occipital areas. Mason et al. (Reference Mason, Peters, Dima, Williams and Kumari2016) report that patients receiving CBT for psychosis show a normalization of functional connectivity of the amygdala with the insula and visual areas compared with controls. Also a greater increase in connectivity of the amygdala with the DLPFC and inferior parietal lobule was observed compared with patients receiving treatment as usual. Lastly, the DLPFC showed greater connectivity with other prefrontal regions after treatment with CBT compared with treatment as usual.
In relation to the earlier discussed altered neural response in the amygdala and parahippocampal complex in response to emotional auditory stimuli in AVH (Kang et al. Reference Kang, Kim, Seok, Chun, Lee and Park2009; Escarti et al. Reference Escarti, de la Iglesia-Vaya, Marti-Bonmati, Robles, Carbonell, Lull, Garcia-Marti, Manjon, Aguilar, Aleman and Sanjuan2010), these studies show that CBT can attenuate the neural response to emotional stimuli. Especially the findings by Mason et al. (Reference Mason, Peters, Dima, Williams and Kumari2016) of the attenuation of amygdala connectivity after CBT treatment are in line with the observation of aberrant amygdala response to emotional auditory stimuli, possibly suggesting how CBT for AVH may be efficacious at the neural level.
Discussion
Kant has taught us the subjectivity of perception due to constrains in the sensory faculties, enabling the existence of AVH as a veridical perception. In line with this philosophical framework, the neurobiological correlates of AVH overlap with brain regions associated with the perception of auditory verbal stimuli. Indeed, structural imaging studies converge on the finding of grey matter reductions in the STG, insula and IFG as well as abnormalities in the connecting white matter between these regions. Some findings suggest an atypical developmental trajectory, as well as a dose–response effect between brain structure abnormalities and the psychosis continuum. This suggests that subtle structural brain abnormalities are associated with hallucination proneness, also in the absence of a psychotic disorder.
The metabolic and physiological processes associated with AVH show a strong overlap with the areas that are anatomically affected. However, the evidence regarding these processes is equivocal in terms of the direction of effect, with some studies pointing out increased blood flow and glucose consumption and others finding the opposite effect.
Functional brain imaging studies unequivocally show that language regions are implicated in AVH. However, other regions such as right hemisphere homologues of language regions and the parahippocampal cortex also show abnormal activation during AVH, suggesting that activity outside of primary language regions is a contributor to the occurrence of AVH.
The question remains open as to how the reductions in grey and white matter are related to the disturbances in brain metabolism and activation associated with AVH. Prospective longitudinal studies are needed to further clarify whether the observed structural abnormalities constitute a risk factor or a pathological process post-AVH onset. As argued above, for this question to be resolved it is necessary to move beyond current imaging protocols and search for mechanisms underlying the changes in functional electrophysiology and haemodynamic changes, i.e. looking for alterations and abnormalities at the neurochemistry level. As also suggested above, the use of MR spectroscopy may be a novel and largely untested pathway into the neurochemistry of AVH. Although promising, it is important to also have a critical approach to MR spectroscopy, since there are several unsolved measurement issues, when it comes to both data acquisition, and data analysis, not the least current technical measurement issues with regard to how to best obtain valid data on glutamate and GABA.
Antipsychotic medication is an effective treatment for AVH in psychotic disorders (Sommer et al. Reference Sommer, Slotema, Daskalakis, Derks, Blom and Van Der Gaag2012; Johnsen et al. Reference Johnsen, Jørgensen, Kroken and Løberg2013). Imaging evidence suggests that this efficacy may come from a normalization of brain activity in areas typically implicated in AVH. However, a considerable minority of the psychosis patients shows no treatment effect of antipsychotics. Also, antipsychotic medication may not be a favourable treatment strategy for AVH in other diagnoses where a dysregulation of the dopaminergic system is not evident. This suggests that while aberrant dopaminergic signalling may induce AVH in psychotic patients with dysregulated dopaminergic signalling, it is not necessary for AVH to emerge.
There are considerable individual differences in the phenomenology of AVH considering frequency, (negative) emotional content, imperativeness, repetitiveness and omnipotence (Daalman et al. Reference Daalman, Boks, Diederen, de Weijer, Blom, Kahn and Sommer2011; McCarthy-Jones et al. Reference McCarthy-Jones, Oestreich and Whitford2015). Brain stimulation and CBT are both effective treatments for AVH, yet in different ways. Both treatments appear to have a normalizing effect on activity in the IFG and insula. Selective effects for TMS include a normalized response of the auditory cortex and ACC, whereas tDCS is associated with a normalization of temporofrontal functional connectivity. CBT on the other hand is reported to have normalizing effects on amygdala and DLPFC activation. Interestingly, this shows that interventions effective at the level of reducing AVH occurrence and severity mainly interact with the language network while interventions effective at the level of appraisal may be more involved with the limbic system. This may indicate that while occurrence of AVH is associated with aberrant functioning of the language network, the appraisal and emotional reaction to AVH are mediated by the limbic system. Intervention studies using a combined design with CBT and TMS may be able to elucidate this further.
Although AVH are prevalent in a broad range of disorders and also occur in the general population, current evidence for neurobiological correlates of AVH relies heavily on research in schizophrenia spectrum disorders. A common neurobiology for AVH could be expected considering the overlap in phenomenology over different diagnoses (Slotema et al. Reference Slotema, Daalman, Blom, Diederen, Hoek and Sommer2012; McCarthy-Jones & Longden, Reference McCarthy-Jones and Longden2015) and indications that non-clinical hallucinators also show brain abnormalities that lie on a continuum between controls and patients with psychosis (De Weijer et al. Reference De Weijer, Neggers, Diederen, Mandl, Kahn, Hulshoff Pol and Sommer2013; Diederen et al. Reference Diederen, Neggers, de Weijer, van Lutterveld, Daalman, Eickhoff, Clos, Kahn and Sommer2013; Van Lutterveld et al. Reference Van Lutterveld, van den Heuvel, Diederen, de Weijer, Begemann, Brouwer, Daalman, Blom, Kahn and Sommer2014). However, the finding of unaffected dopamine synthesis in non-clinical hallucinators (Howes et al. Reference Howes, Shotbolt, Bloomfield, Daalman, Demjaha, Diederen, Ibrahim, Kim, McGuire, Kahn and Sommer2013) suggests that differences also exist. A transdiagnostic approach in neuroimaging research is much needed to tease apart the neurobiology of AVH from the clinical context in which they occur.
Acknowledgements
None.
Declaration of Interest
None.
