The neurophysiology of auditory hallucinations - a historical and contemporary review

From Sound Perception to Automatic Detection of Schizophrenia: An EEG-Based Deep Learning Approach

Deep learning techniques have been applied to electroencephalogram (EEG) signals, with promising applications in the field of psychiatry. Schizophrenia is one of the most disabling neuropsychiatric disorders, often characterized by the presence of auditory hallucinations. Auditory processing impairments have been studied using EEG-derived event-related potentials and have been associated with clinical symptoms and cognitive dysfunction in schizophrenia. Due to consistent changes in the amplitude of ERP components, such as the auditory N100, some have been proposed as biomarkers of schizophrenia. In this paper, we examine altered patterns in electrical brain activity during auditory processing and their potential to discriminate schizophrenia and healthy subjects. Using deep convolutional neural networks, we propose an architecture to perform the classification based on multi-channels auditory-related EEG single-trials, recorded during a passive listening task. We analyzed the effect of the number of electrodes used, as well as the laterality and distribution of the electrical activity over the scalp. Results show that the proposed model is able to classify schizophrenia and healthy subjects with an average accuracy of 78% using only 5 midline channels (Fz, FCz, Cz, CPz, and Pz). The present study shows the potential of deep learning methods in the study of impaired auditory processing in schizophrenia with implications for diagnosis. The proposed design can provide a base model for future developments in schizophrenia research.

A comparison and classification of oscillatory characteristics in speech perception and covert speech

Covert speech, the mental imagery of speaking, has been studied increasingly to understand and decode thoughts in the context of brain-computer interfaces. In studies of speech comprehension, neural oscillations are thought to play a key role in the temporal encoding of speech. However, little is known about the role of oscillations in covert speech. In this study, we investigated the oscillatory involvements in covert speech and speech perception. Data were collected from 10 participants with 64 channel EEG. Participants heard the words, 'blue' and 'orange', and subsequently mentally rehearsed them. First, continuous wavelet transform was performed on epoched signals and subsequently two-tailed t-tests between two classes were conducted to determine statistical differences in frequency and time (t-CWT). Features were also extracted using t-CWT and subsequently classified using a support vector machine. θ and γ phase amplitude coupling (PAC) was also assessed within and between tasks. All binary classifications produced accuracies significantly greater (80-90%) than chance level, supporting the use of t-CWT in determining relative oscillatory involvements. While the perception task dynamically invoked all frequencies with more prominent θ and α activity, the covert task favoured higher frequencies with significantly higher γ activity than perception. Moreover, the perception condition produced significant θ-γ PAC, corroborating a reported linkage between syllabic and phonemic sampling. Although this coupling was found to be suppressed in the covert condition, we found significant cross-task coupling between perception θ and covert speech γ. Covert speech processing appears to be largely associated with higher frequencies of EEG. Importantly, the significant cross-task coupling between speech perception and covert speech, in the absence of within-task covert speech PAC, supports the notion that the γ- and θ-bands subserve, respectively, shared and unique encoding processes across tasks.

Complex auditory musical hallucinations with ambivalent feelings

A 78-year-old woman with hypertension, diabetes mellitus type 2 and bilateral sensorineural hearing loss was referenced to geriatric psychiatry consultation. She presented cognitive dysfunction, erotomanic delusion and complex musical hallucinations (MH), described as hearing her neighbour singing a familiar church song along with bells in the background, making comments and talking to her. A computed tomography (CT) of the brain detected small right nucleocapsular and bilateral external capsules hypodensities of presumed vascular aetiology during hospitalisation. MH are a rare phenomenon with heterogeneous aetiology. Most frequently, the cause is hearing impairment; other causes include social isolation, cognitive dysfunction, vascular risk factors and medication. Studies suggest that some brain areas related to musical memory circuitry might be related and not fully mapped. Auditory verbal hallucinations with a voice that either comments, talks or sings to the patient have never been described in the literature, making this clinical case attractive.

Electrophysiological correlates of focused attention on low- and high-distressed tinnitus

OBJECTIVES

The study aimed at determining the EEG correlates of concentration on either low or high-distressed tinnitus.

METHODS

Sixty-seven patients (36 women, mean age = 50.34 ± 12.94 years) with chronic tinnitus were assigned to either a high (HD) or low (LD) tinnitus-related distress group based on THI results. All participants took part in the EEG study comprising two 3-4 min blocks of focusing on either tinnitus (Tinnitus Focus Condition, TFC) or the sensations from one's own body (Body Focus Condition, BFC). The absolute power and current density of 8 frequency bands in 7 clusters were compared between conditions and groups.

RESULTS

The most pronounced differences were found in the HD patients in the TFC, relative to the BFC, i.e. reduced power of frontally distributed low alpha (8-10 Hz) and posterior high alpha (10-12 Hz) as well as lower current density of 8-10 Hz rhythm over the right frontal/anterior cingulate cortex and higher middle beta (15-18 Hz) density in the precuneus. The HD, relative to LD patients, in both conditions, exhibited increased low beta (12-15 Hz) power over the left middle area and greater higher beta (15-25 Hz) power in the left posterior region.

CONCLUSIONS

The present study contrasted bioelectrical activity, acquired when concentrating on tinnitus with EEG data collected whilst patients focused on their body. Decreased alpha power and current density in the frontal/cingulate cortex when listening to bothersome tinnitus might reflect greater cortical arousal whereas increased beta power and density in the precuneus/posterior cingulate activity in this condition could be indicative for elevated tension or augmented cognitive/emotional processing of tinnitus sound. Enhanced beta rhythm in patients with high versus low tinnitus distress, observed independently of the study condition, may be due to greater self-focused attention or more active processing of sensations derived from the own body.

Altered parahippocampal gyrus activation and its connectivity with resting-state network areas in schizophrenia: An EEG study

Synchronized and coherent activity in resting-networks during normal brain functioning could be altered in disconnection syndrome like schizophrenia. Study of neural oscillations as assessed by EEG appears to be a promising proposition to understand the pathophysiology of schizophrenia in patients and their first-degree relatives, where disturbances in neural oscillations point towards genetic predisposition. Therefore, present study aims at establishing EEG based biomarkers for early detection and management strategies. Thirty-two patients with schizophrenia, 28 first-degree relatives and 31 healthy controls (HC) participated in the study. Resting brain activity was recorded using 128-channel electroencephalography. After pre-processing and independent component analysis (ICA), an equivalent current dipole was estimated for each IC. Total of 1551 independent and localizable EEG components across all groups were used in subsequent analysis. Power spectral density and source coherence between IC clusters were computed. Patients and first-degree relatives displayed significantly higher power spectral density (PSD) than HC for all frequency bands in left parahippocampal gyrus (PHG) (-7, -26, 8; BA 27). Another region within left deep PHG (-4, -28, 1), however, distinguished patients from first-degree relatives and HC in terms of significantly lower PSD in higher frequency bands. Functional connectivity (FC) was found to be lower in patients and higher in relatives compared to HC between different resting-state network areas. In patients, connectivity was lower compared to first-degree relatives. Altered activity within left PHG and FC of primarily this with other areas in resting-state network can serve as state and trait markers of schizophrenia.

The Predictive Coding Account of Psychosis

Fueled by developments in computational neuroscience, there has been increasing interest in the underlying neurocomputational mechanisms of psychosis. One successful approach involves predictive coding and Bayesian inference. Here, inferences regarding the current state of the world are made by combining prior beliefs with incoming sensory signals. Mismatches between prior beliefs and incoming signals constitute prediction errors that drive new learning. Psychosis has been suggested to result from a decreased precision in the encoding of prior beliefs relative to the sensory data, thereby garnering maladaptive inferences. Here, we review the current evidence for aberrant predictive coding and discuss challenges for this canonical predictive coding account of psychosis. For example, hallucinations and delusions may relate to distinct alterations in predictive coding, despite their common co-occurrence. More broadly, some studies implicate weakened prior beliefs in psychosis, and others find stronger priors. These challenges might be answered with a more nuanced view of predictive coding. Different priors may be specified for different sensory modalities and their integration, and deficits in each modality need not be uniform. Furthermore, hierarchical organization may be critical. Altered processes at lower levels of a hierarchy need not be linearly related to processes at higher levels (and vice versa). Finally, canonical theories do not highlight active inference-the process through which the effects of our actions on our sensations are anticipated and minimized. It is possible that conflicting findings might be reconciled by considering these complexities, portending a framework for psychosis more equipped to deal with its many manifestations.

Emotional self-other voice processing in schizophrenia and its relationship with hallucinations: ERP evidence

Abnormalities in self-other voice processing have been observed in schizophrenia, and may underlie the experience of hallucinations. More recent studies demonstrated that these impairments are enhanced for speech stimuli with negative content. Nonetheless, few studies probed the temporal dynamics of self versus nonself speech processing in schizophrenia and, particularly, the impact of semantic valence on self-other voice discrimination. In the current study, we examined these questions, and additionally probed whether impairments in these processes are associated with the experience of hallucinations. Fifteen schizophrenia patients and 16 healthy controls listened to 420 prerecorded adjectives differing in voice identity (self-generated [SGS] versus nonself speech [NSS]) and semantic valence (neutral, positive, and negative), while EEG data were recorded. The N1, P2, and late positive potential (LPP) ERP components were analyzed. ERP results revealed group differences in the interaction between voice identity and valence in the P2 and LPP components. Specifically, LPP amplitude was reduced in patients compared with healthy subjects for SGS and NSS with negative content. Further, auditory hallucinations severity was significantly predicted by LPP amplitude: the higher the SAPS "voices conversing" score, the larger the difference in LPP amplitude between negative and positive NSS. The absence of group differences in the N1 suggests that self-other voice processing abnormalities in schizophrenia are not primarily driven by disrupted sensory processing of voice acoustic information. The association between LPP amplitude and hallucination severity suggests that auditory hallucinations are associated with enhanced sustained attention to negative cues conveyed by a nonself voice.

Immune involvement in the pathogenesis of schizophrenia: a meta-analysis on postmortem brain studies

Although the precise pathogenesis of schizophrenia is unknown, genetic, biomarker and imaging studies suggest involvement of the immune system. In this study, we performed a systematic review and meta-analysis of studies investigating factors related to the immune system in postmortem brains of schizophrenia patients and healthy controls. Forty-one studies were included, reporting on 783 patients and 762 controls. We divided these studies into those investigating histological alterations of cellular composition and those assessing molecular parameters; meta-analyses were performed on both categories. Our pooled estimate on cellular level showed a significant increase in the density of microglia (P=0.0028) in the brains of schizophrenia patients compared with controls, albeit with substantial heterogeneity between studies. Meta-regression on brain regions demonstrated this increase was most consistently observed in the temporal cortex. Densities of macroglia (astrocytes and oligodendrocytes) did not differ significantly between schizophrenia patients and healthy controls. The results of postmortem histology are paralleled on the molecular level, where we observed an overall increase in expression of proinflammatory genes on transcript and protein level (P=0.0052) in patients, while anti-inflammatory gene expression levels were not different between schizophrenia and controls. The results of this meta-analysis strengthen the hypothesis that components of the immune system are involved in the pathogenesis of schizophrenia.

High-order interactions observed in multi-task intrinsic networks are dominant indicators of aberrant brain function in schizophrenia

Identifying the complex activity relationships present in rich, modern neuroimaging data sets remains a key challenge for neuroscience. The problem is hard because (a) the underlying spatial and temporal networks may be nonlinear and multivariate and (b) the observed data may be driven by numerous latent factors. Further, modern experiments often produce data sets containing multiple stimulus contexts or tasks processed by the same subjects. Fusing such multi-session data sets may reveal additional structure, but raises further statistical challenges. We present a novel analysis method for extracting complex activity networks from such multifaceted imaging data sets. Compared to previous methods, we choose a new point in the trade-off space, sacrificing detailed generative probability models and explicit latent variable inference in order to achieve robust estimation of multivariate, nonlinear group factors ("network clusters"). We apply our method to identify relationships of task-specific intrinsic networks in schizophrenia patients and control subjects from a large fMRI study. After identifying network-clusters characterized by within- and between-task interactions, we find significant differences between patient and control groups in interaction strength among networks. Our results are consistent with known findings of brain regions exhibiting deviations in schizophrenic patients. However, we also find high-order, nonlinear interactions that discriminate groups but that are not detected by linear, pairwise methods. We additionally identify high-order relationships that provide new insights into schizophrenia but that have not been found by traditional univariate or second-order methods. Overall, our approach can identify key relationships that are missed by existing analysis methods, without losing the ability to find relationships that are known to be important.

The role of the primary auditory cortex in the neural mechanism of auditory verbal hallucinations

Auditory verbal hallucinations (AVHs) are a subjective experience of "hearing voices" in the absence of corresponding physical stimulation in the environment. The most remarkable feature of AVHs is their perceptual quality, that is, the experience is subjectively often as vivid as hearing an actual voice, as opposed to mental imagery or auditory memories. This has lead to propositions that dysregulation of the primary auditory cortex (PAC) is a crucial component of the neural mechanism of AVHs. One possible mechanism by which the PAC could give rise to the experience of hallucinations is aberrant patterns of neuronal activity whereby the PAC is overly sensitive to activation arising from internal processing, while being less responsive to external stimulation. In this paper, we review recent research relevant to the role of the PAC in the generation of AVHs. We present new data from a functional magnetic resonance imaging (fMRI) study, examining the responsivity of the left and right PAC to parametrical modulation of the intensity of auditory verbal stimulation, and corresponding attentional top-down control in non-clinical participants with AVHs, and non-clinical participants with no AVHs. Non-clinical hallucinators showed reduced activation to speech sounds but intact attentional modulation in the right PAC. Additionally, we present data from a group of schizophrenia patients with AVHs, who do not show attentional modulation of left or right PAC. The context-appropriate modulation of the PAC may be a protective factor in non-clinical hallucinations.

The role of the primary auditory cortex in the neural mechanism of auditory verbal hallucinations

Auditory verbal hallucinations (AVHs) are a subjective experience of “hearing voices” in the absence of corresponding physical stimulation in the environment. The most remarkable feature of AVHs is their perceptual quality, that is, the experience is subjectively often as vivid as hearing an actual voice, as opposed to mental imagery or auditory memories. This has lead to propositions that dysregulation of the primary auditory cortex (PAC) is a crucial component of the neural mechanism of AVHs. One possible mechanism by which the PAC could give rise to the experience of hallucinations is aberrant patterns of neuronal activity whereby the PAC is overly sensitive to activation arising from internal processing, while being less responsive to external stimulation. In this paper, we review recent research relevant to the role of the PAC in the generation of AVHs. We present new data from a functional magnetic resonance imaging (fMRI) study, examining the responsivity of the left and right PAC to parametrical modulation of the intensity of auditory verbal stimulation, and corresponding attentional top-down control in non-clinical participants with AVHs, and non-clinical participants with no AVHs. Non-clinical hallucinators showed reduced activation to speech sounds but intact attentional modulation in the right PAC. Additionally, we present data from a group of schizophrenia patients with AVHs, who do not show attentional modulation of left or right PAC. The context-appropriate modulation of the PAC may be a protective factor in non-clinical hallucinations.

Network analysis of auditory hallucinations in nonpsychotic individuals

BACKGROUND

Auditory verbal hallucinations (AVH) are a cardinal feature of schizophrenia and can severely disrupt behavior and decrease quality of life. Identification of areas with high functional connectivity (so-called hub regions) that are associated with the predisposition to hallucinate may provide potential targets for neuromodulation in the treatment of AVH.

METHODS

Resting-state fMRI scans during which no hallucinations had occurred were acquired from 29 nonpsychotic individuals with AVH and 29 matched controls. These nonpsychotic individuals with AVH provide the opportunity to study AVH without several confounds associated with schizophrenia, such as antipsychotic medication use and other symptoms related to the illness. Hub regions were identified by assessing weighted connectivity strength and betweenness centrality across groups using a permutation analysis.

RESULTS

Nonpsychotic individuals with AVH exhibited increased functioning as hub regions in the temporal cortices and the posterior cingulate/precuneus, which is an important area in the default mode network (DMN), compared to the nonhallucinating controls. In addition, the right inferior temporal gyrus, left paracentral lobule and right amygdala were less important as a hub region in the AVH group.

CONCLUSIONS

These results suggest that the predisposition to hallucinate may be related to aberrant functioning of the DMN and the auditory cortices.

Neural correlates of the lombard effect in primate auditory cortex

Speaking is a sensory-motor process that involves constant self-monitoring to ensure accurate vocal production. Self-monitoring of vocal feedback allows rapid adjustment to correct perceived differences between intended and produced vocalizations. One important behavior in vocal feedback control is a compensatory increase in vocal intensity in response to noise masking during vocal production, commonly referred to as the Lombard effect. This behavior requires mechanisms for continuously monitoring auditory feedback during speaking. However, the underlying neural mechanisms are poorly understood. Here we show that when marmoset monkeys vocalize in the presence of masking noise that disrupts vocal feedback, the compensatory increase in vocal intensity is accompanied by a shift in auditory cortex activity toward neural response patterns seen during vocalizations under normal feedback condition. Furthermore, we show that neural activity in auditory cortex during a vocalization phrase predicts vocal intensity compensation in subsequent phrases. These observations demonstrate that the auditory cortex participates in self-monitoring during the Lombard effect, and may play a role in the compensation of noise masking during feedback-mediated vocal control.

Poor supplementary motor area activation differentiates auditory verbal hallucination from imagining the hallucination

Neuronal underpinnings of auditory verbal hallucination remain poorly understood. One suggested mechanism is brain activation that is similar to verbal imagery but occurs without the proper activation of the neuronal systems that are required to tag the origins of verbal imagery in one's mind. Such neuronal systems involve the supplementary motor area. The supplementary motor area has been associated with awareness of intention to make a hand movement, but whether this region is related to the sense of ownership of one's verbal thought remains poorly known. We hypothesized that the supplementary motor area is related to the distinction between one's own mental processing (auditory verbal imagery) and similar processing that is attributed to non-self author (auditory verbal hallucination). To test this hypothesis, we asked patients to signal the onset and offset of their auditory verbal hallucinations during functional magnetic resonance imaging. During non-hallucination periods, we asked the same patients to imagine the hallucination they had previously experienced. In addition, healthy control subjects signaled the onset and offset of self-paced imagery of similar voices. Both hallucinations and the imagery of hallucinations were associated with similar activation strengths of the fronto-temporal language-related circuitries, but the supplementary motor area was activated more strongly during the imagery than during hallucination. These findings suggest that auditory verbal hallucination resembles verbal imagery in language processing, but without the involvement of the supplementary motor area, which may subserve the sense of ownership of one's own verbal imagery.

Oscillatory cortical network involved in auditory verbal hallucinations in schizophrenia

Background

Auditory verbal hallucinations (AVH), a prominent symptom of schizophrenia, are often highly distressing for patients. Better understanding of the pathogenesis of hallucinations could increase therapeutic options. Magnetoencephalography (MEG) provides direct measures of neuronal activity and has an excellent temporal resolution, offering a unique opportunity to study AVH pathophysiology.

Methods

Twelve patients (10 paranoid schizophrenia, 2 psychosis not otherwise specified) indicated the presence of AVH by button-press while lying in a MEG scanner. As a control condition, patients performed a self-paced button-press task. AVH-state and non-AVH state were contrasted in a region-of-interest (ROI) approach. In addition, the two seconds before AVH onset were contrasted with the two seconds after AVH onset to elucidate a possible triggering mechanism.

Results

AVH correlated with a decrease in beta-band power in the left temporal cortex. A decrease in alpha-band power was observed in the right inferior frontal gyrus. AVH onset was related to a decrease in theta-band power in the right hippocampus.

Conclusions

These results suggest that AVH are triggered by a short aberration in the theta band in a memory-related structure, followed by activity in language areas accompanying the experience of AVH itself.

A neurophysiological deficit in early visual processing in schizophrenia patients with auditory hallucinations

Existing 67-channel event-related potentials, obtained during recognition and working memory paradigms with words or faces, were used to examine early visual processing in schizophrenia patients prone to auditory hallucinations (AH, n = 26) or not (NH, n = 49) and healthy controls (HC, n = 46). Current source density (CSD) transforms revealed distinct, strongly left- (words) or right-lateralized (faces; N170) inferior-temporal N1 sinks (150 ms) in each group. N1 was quantified by temporal PCA of peak-adjusted CSDs. For words and faces in both paradigms, N1 was substantially reduced in AH compared with NH and HC, who did not differ from each other. The difference in N1 between AH and NH was not due to overall symptom severity or performance accuracy, with both groups showing comparable memory deficits. Our findings extend prior reports of reduced auditory N1 in AH, suggesting a broader early perceptual integration deficit that is not limited to the auditory modality.

Neurophysiological studies of auditory verbal hallucinations

We discuss 3 neurophysiological approaches to study auditory verbal hallucinations (AVH). First, we describe "state" (or symptom capture) studies where periods with and without hallucinations are compared "within" a patient. These studies take 2 forms: passive studies, where brain activity during these states is compared, and probe studies, where brain responses to sounds during these states are compared. EEG (electroencephalography) and MEG (magnetoencephalography) data point to frontal and temporal lobe activity, the latter resulting in competition with external sounds for auditory resources. Second, we discuss "trait" studies where EEG and MEG responses to sounds are recorded from patients who hallucinate and those who do not. They suggest a tendency to hallucinate is associated with competition for auditory processing resources. Third, we discuss studies addressing possible mechanisms of AVH, including spontaneous neural activity, abnormal self-monitoring, and dysfunctional interregional communication. While most studies show differences in EEG and MEG responses between patients and controls, far fewer show symptom relationships. We conclude that efforts to understand the pathophysiology of AVH using EEG and MEG have been hindered by poor anatomical resolution of the EEG and MEG measures, poor assessment of symptoms, poor understanding of the phenomenon, poor models of the phenomenon, decoupling of the symptoms from the neurophysiology due to medications and comorbidites, and the possibility that the schizophrenia diagnosis breeds truer than the symptoms it comprises. These problems are common to studies of other psychiatric symptoms and should be considered when attempting to understand the basic neural mechanisms responsible for them.