A number of PET and SPECT studies have directly compared brain activity between schizophrenia patients when they are experiencing hallucinations and when they are not (Table 20.2). In general, patients in these studies were symptomatic and using antipsychotic medication at the time of scanning. In a series of SPECT studies, the group of Musalek and colleagues compared cerebral blood flow between 17 schizophrenia patients with chronic, treatment-resistant AVH and healthy controls. Auditory hallucinations were associated with increased blood flow in the anterior basal ganglia and the medial temporal region bilaterally, as well as reductions in blood flow in frontal cortical areas when compared to 28 non-hallucinating control subjects (Musalek et al. 1988, 1989). A comparison between the same group of schizophrenia patients with AVH and a control group who reported musical hallucinations only while hypnotised (n = 10) demonstrated increased activity in the medial temporal region and decreased activity in the frontal lobes and thalamus bilaterally of patients (Walter et al. 1990). However, the substantial age difference between patients and control groups could be a serious confounder in the interpretation of the results of these studies as cerebral blood flow is known to decline with age (e.g. Melamed et al. 1980). Cleghorn and colleagues evaluated auditory hallucinations using FDG-PET in two different studies. The first study found no differences in regional glucose uptake between nine schizophrenia patients with AVH, ten schizophrenia patients who had recovered from AVH, and ten control subjects. Interestingly, auditory hallucinations were associated with a highly correlated pattern of activity in specific language centres, including Broca’s area and its right hemisphere homologue (involved in speech production), anterior cingulate, and left superior temporal gyrus (Cleghorn et al. 1990). In a second study, using the same imaging technique, a group of 12 medication-free schizophrenia patients who were experiencing auditory hallucinations during scanning demonstrated lower relative metabolism in the bilateral posterior superior temporal gyrus when compared to ten medication-free schizophrenia patients who did not hallucinate during imaging. The intensity of reported hallucinations was significantly correlated with glucose metabolism in the striatum and anterior cingulate cortex (involved in attentional processes) (Cleghorn et al. 1992). McGuire et al. (1993) scanned 13 schizophrenia patients using SPECT during an episode of their illness in which they regularly experienced auditory hallucinations. They were scanned again on a second occasion, during a remission period, when the hallucinations were absent. When the scans were compared, brain activity associated with hallucinations was observed in language-related regions, especially Broca’s area. To a lesser extent, activity was also found in the anterior cingulate and left temporal cortex (McGuire et al. 1993). Using a similar approach, comparable results were reported by Suzuki et al. (1993), who demonstrated with SPECT an increase in cerebral blood flow in the left temporal lobe (auditory association cortex) and anterior cingulate cortex in five hallucinating patients. Blood flow patterns were normalised after clinical improvement (Suzuki et al. 1993). Silbersweig and colleagues (1995) again used patients as their own control subjects, but measured cerebral blood flow during hallucinations and quiescent periods in the same scanning session with H₂ ¹⁵O-PET. Five patients pushed a button to indicate the duration of the auditory verbal hallucination while they were scanned. Activation during hallucinations was shown in subcortical structures (bilateral thalamus, right putamen, and caudate), bilateral parahippocampal gyrus, right anterior cingulate, and left orbitofrontal cortex (areas involved in emotion regulation). The authors suggest that activity in deep brain structures might generate or modulate hallucinations, whereas cortical activity may affect the specific perceptual content of the hallucinations (Silbersweig et al. 1995). An H₂ ¹⁵O-PET study of auditory hallucinations performed by Szechtman et al. (1998) did not include schizophrenia patients but compared brain activity in highly hypnotisable volunteers during different experimental conditions: hearing, imagining, and hallucinating. Eight of these volunteers were able to hallucinate under hypnosis (these were termed hallucinators), whereas six lacked this ability (control group). A region in the right anterior cingulate cortex was activated in the group of hallucinators when they heard an auditory stimulus and when they hallucinated hearing it, but not when they merely imagined hearing it. The same experimental conditions did not yield such activation in the control group. The investigators suggest that the anterior cingulate activation ‘tags’ an auditory event as originating from the external world. Thus, in hallucinations, such activation may reflect a mismatch between externally directed attention and internally generated events. They propose that the involvement of rostral anterior cingulate cortex (which has been implicated in modulating affect) may imply that the attention of hallucinators is more affect-laden than that of non-hallucinators, and speculate that when attention is more affect-laden, self-generation of the expected auditory event is more likely to occur (Szechtman et al. 1998). A major problem, however, is that no anterior cingulate activation was observed in the hearing versus baseline condition for the non-hallucinators. If anterior cingulate activation tags an auditory event as originating from the external world, one would expect such activation also in the control group. Similar to Silbersweig et al. (1995), eight schizophrenia patients scanned with H₂ ¹⁵O-PET by Copolov and colleagues (2003) indicated their AVH episodes by a button press. During hallucinations, a network of cortical regions was activated, including right medial frontal and prefrontal regions, right medial temporal gyrus, left superior temporal gyrus, left parahippocampal gyrus, and left posterior cingulate (Copolov et al. 2003). Parellada and colleagues (2008) performed three consecutive FDG-PET scans of nine first-episode schizophrenia patients with prominent AVH. Scans were obtained during hallucinations, after medication-induced remission, and in remission during a linguistic activation task, respectively. Patients demonstrated significantly increased activity in bilateral supplementary motor area, anterior cingulate cortex, medial and superior frontal lobes, and cerebellum during auditory verbal hallucinations compared to the remission scan. Activation was also observed in the right superior temporal pole and right orbitofrontal cortex. During linguistic activation, higher activity was observed in bilateral middle and superior temporal cortex and left parahippocampus compared to the remission scan. The authors suggest that the different patterns of glucose metabolism between hallucinations and physiological auditory activation indicate that frontal cortical regions implicated in the generation of inner speech rather than auditory-linguistic pathways may be involved in auditory verbal hallucinations of acute schizophrenia patients (Parellada et al. 2008). Horga et al. (2011) investigated differences in glucose metabolism with FDG-PET between antipsychotic-naïve first-episode schizophrenia patients with (n = 9) and without (n = 7) AVH during scanning. Hallucinators exhibited significantly increased activity in the left superior and middle temporal region, bilateral superior medial frontal cortex, and left caudate and decreased activity in the hippocampal complex, cerebellum, and parietal cortex compared to patients without hallucinations during the scan. The severity of hallucinations showed a positive correlation with metabolic rate in the right superior temporal cortex and cerebellum (Horga et al. 2011).

Altogether, results of these studies convincingly suggest a role for the temporal lobe in AVH, as seven of the 11 studies showed significantly increased temporal lobe activity during hallucinations (three bilateral and four left lateralised). In addition, several studies report the involvement of cortical and subcortical areas in the experience of hallucinations, such as anterior cingulate cortex, Broca’s area, and basal ganglia. Although the exact role of these regions is not yet clear, a possible hypothesis is that activity in subcortical areas and modality-specific association cortices account for the conscious perceptual experience of hallucinations. Inappropriate anterior cingulate activation may reflect impairments in the monitoring of speech and erroneously tags internally generated imagery as originating from an external source. An alternative explanation could be that increased temporal activation drives cingulate activation, which then results in greater attention being directed towards the sensory cortex (Hunter et al. 2006). What is clear however is that studies have consistently observed activity in either language-production areas or in the primary auditory cortex during auditory hallucinations. This strongly implicates the temporal lobe, more specifically the middle or superior temporal gyri.

Neuroimaging studies have also been performed with schizophrenia patients who experienced hallucinations in other sensory modalities and with other groups of patients with hallucinations. For example, one SPECT study examined cerebral blood flow of schizophrenia patients while they were having tactile hallucinations (feeling imagined bugs). A comparison between 11 patients with only tactile hallucinations and a group of 28 non-hallucinating control patients with schizophrenia revealed a significant reduction in blood flow in the inferior temporal regions (Musalek et al. 1989). Furthermore, although visual hallucinations are being reported more often than auditory hallucinations in patients with Parkinson’s disease, Matsui and colleagues (2007) examined brain perfusion with SPECT in patients with Parkinson’s disease during the occurrence of AVH. They found that those with AVH (n = 4) showed significantly higher perfusion of the right thalamus during AVH than patients without these symptoms (n = 77) (Matsui et al. 2007). In a SPECT study with five patients with Charles Bonnet syndrome (CBS), a condition in which psychologically healthy people with significant visual loss experience recurrent complex visual hallucinations, Adachi et al. (2000) showed increased perfusion in the left thalamus of all individual patients. However, the authors did not perform a group analysis in which they examined brain regions showing significantly different activity levels in their patient group.

Several interesting case studies in which brain function was assessed while patients experienced hallucinations have been reported. The first study used SPECT to evaluate a 45-year-old alcohol-dependent male with AVH. Images obtained while the subject was actively hallucinating showed the highest degree of radioactive accumulation in the left superior temporal lobe, an area corresponding to primary and secondary auditory cortex (Matsuda et al. 1988). Unfortunately, the patient was not imaged again while not hallucinating. Notardonato and colleagues scanned a 41-year-old female schizophrenia patient with a 4-year history of auditory hallucinations with SPECT before and after pharmacological treatment. Hallucinations were associated with increased activity in the basal ganglia and right temporal lobe. Abnormalities in brain activity resolved with clinical improvement after treatment (Notardonato et al. 1989). In a study of a 70-year-old female with a 1-month history of ‘seeing ghosts’ following a left hemispheric stroke, increased cerebral blood flow was demonstrated with SPECT in the bilateral parietooccipital lobes. Once again, resolution of the hallucinatory phenomenon correlated with a normalisation of parietooccipital activity (Kim et al. 1993). A 23-year-old male, drug-naïve schizophrenia patient who experienced both visual and auditory verbal hallucinations, was scanned with H₂ ¹⁵O-PET while he indicated the presence of hallucinations with a button press. Hallucinations were associated with activity in visual areas (lingual, fusiform, and occipital gyri) and in the superior and middle temporal cortex (Silbersweig et al. 1995). Kasai and colleagues (1999) examined cerebral blood flow with SPECT in a cognitively intact 88-year-old woman during musical hallucinations and in their absence and compared both states with a group of 18 elderly controls. Activity was increased in the right superior temporal and inferior frontal gyri during the occurrence of hallucinations (Kasai et al. 1999). Izumi et al. (2002) obtained three SPECT scans of a 51-year-old male during musical hallucinations, during verbal hallucinations, and in the absence of hallucinations, respectively. They found evidence of differing patterns of regional cerebral blood flow during musical hallucinations and verbal hallucinations, with increased activity in the bilateral lower frontal area and basal ganglia during musical hallucinations, and increased activity in the left lower temporal area, right lower frontal area, and left basal ganglia during verbal hallucinations (Izumi et al. 2002). Using SPECT, Mori et al. (2006) observed increased regional cerebral blood flow in left temporal regions and left angular gyrus in a patient with Alzheimer’s disease who experienced musical hallucinations when compared to nine patients without hallucinations (Mori et al. 2006). Finally, Godani and colleagues (2012) examined glucose metabolism in a 67-year-old woman with auditory hallucinations after right parietal haemorrhagic stroke and compared it to a normal control database. They demonstrated significantly increased metabolism in the middle and posterior areas of corpus callosum as well as in the precuneus and cuneus cortex (Godani et al. 2012).