Auditory Hallucinations: Does a Continuum of Severity Entail Continuity in Mechanism?

The continuum view holds that psychotic symptoms vary along dimensions such as distress, vividness, and duration in clinical and nonclinical groups. At one extreme, symptoms are so severe and disabling that they require treatment and sometimes hospitalization. At the other end of the spectrum, individuals display similar symptoms without significant distress, disability, or dysfunction.

Recently, the continuum model has come under pressure from different directions.¹ Its postulation of gradual quantitative transitions across a continuum of symptom expression lends itself well to empirical evaluation. In support of the model, research efforts have identified similarities in psychosis-like experiences between healthy and psychiatric groups, from which common underlying mechanisms are sometimes inferred. This search for commonality might follow from an assumption that different levels of explanation map neatly onto each other. While similarities between groups in features of symptom expression might align with commonalities at the neurobiological level, they equally may not. Complex systems lend themselves to analysis at multiple levels of organization, and we should be wary of inferring from a continuum at one level that there must also be a continuum at another.

A case in point is the example of auditory hallucinations. There are few discernible differences in the descriptive features of experience between psychotic and nonpsychotic hallucinations.² Some nonclinical voice-hearers report vivid and frequent hallucinations, third-person hallucinations, and personification, as well as some negative contents, resembling the hallucinations of people diagnosed with a psychotic disorder. In the absence of differential phenomenological markers, a common approach in making diagnostic and treatment decisions is to focus on the presence of distress, or impairments in social, occupational, or other important areas of functioning, as a threshold for clinically significant symptoms.³ An over-emphasis on distress and functioning, however, may obscure important continuities and discontinuities across different population groups.

One example includes evidence of categorical differences within people presenting with psychosis-like experiences in the general population. At least 2 different continua have been proposed: a continuum of symptom expression, distributed across clinical and nonclinical groups, and a (dis)continuum of risk whereby only a subset of individuals are vulnerable to developing psychosis.^4,5 Even with a continuum of symptom expression, it seems likely that important discontinuities will exist. Studies are now investigating distinct nonclinical subgroups, such as professional psychics, voice-hearers with no psychiatric diagnosis, people at high risk or in the prodrome phase of psychosis, as well as individuals who score highly on measures of psychosis proneness. The extent to which these form a heterogeneous group within the category of nonclinical hallucinations is still unclear, and future research will be helpful in determining whether such groups differ on variables relating to etiology, phenomenology, subjective appraisals, and underlying mechanisms.

Neurobiological studies also reveal a complex pattern of continuity and discontinuity between pathological and nonpathological experiences. First, results of imaging studies show that brain changes in people with medical conditions who have hallucinations (acquired deafness, narcolepsy, etc.) are localized and specific,^6,7 which contrasts with the widespread changes observed in hallucinating individuals with schizophrenia.⁸ Second, studies of neurobiological processes involving direct comparisons of clinical and nonclinical samples show similarities as well as important differences. A functional magnetic resonance imaging (fMRI) study comparing a healthy and psychiatric group with frequent hallucinations revealed continuity in activation of the superior temporal gyrus and bilateral inferior frontal gyri,⁹ while a structural MRI study showed comparable changes in the insular region.¹⁰

Other evidence involving direct group comparisons points to discontinuity in mechanisms. Garrison et al (this issue¹¹) examined the structure of the paracingulate sulcus (PCS), a region in the anterior medial prefrontal cortex implicated in reality monitoring. Structural MRI scans revealed shorter PCS length in the psychotic group with frequent hallucinations, which differentiated them from the nonpsychiatric comparison groups (with and without hallucinations).

A limitation of all 3 studies (each conducted with the same large nonclinical sample from the Netherlands) is the absence of a psychosis comparison group without hallucinations, without which specificity to hallucinations and the contribution of other symptom mechanisms cannot be assessed. Nonetheless, the finding of typical PCS lengths in the nonclinical group with hallucinations is in line with a recent cognitive study which failed to find reality-monitoring deficits in nonclinical individuals scoring high on measures of hallucination proneness.¹² In further support for discontinuity in process, the cognitive literature in nonclinical samples with hallucinations shows largely mixed results,^5,12 although traditional difficulties in publishing negative findings likely act to bias the literature.

Further evidence of discontinuity in mechanisms includes pharmacological evidence that striatal dopaminergic overactivity may be specific to symptoms in psychosis. Positron emission tomography (PET) imaging studies suggest that elevated striatal dopamine production (which characterizes psychosis) is not a feature of nonclinical groups with hallucinations.¹³ Another interpretation of this finding is that hallucinations should not be considered a core feature of psychosis (or at least not as central a feature as striatal dopamine dysregulation), because they occur in both psychotic and nonpsychotic groups, as well as other psychiatric and medical conditions with distinct underlying mechanisms. In support, classical texts refer to hallucinations as a secondary (nonintegral) feature of schizophrenia.¹⁴ In any case, because antipsychotic medication seems only to benefit patients with increased striatal dopamine,¹⁵ a clinical implication of this finding is a caution about the introduction of antipsychotic medication until alternative medical causes for distressing hallucinations have been excluded.

An emerging theme from this recent literature is that there may be a mix of continuity and discontinuity in hallucination-related processes across the spectrum from wellness to disorder. One candidate for a continuous process is an alteration in spontaneous auditory cortex activation, which may feature in both clinical and nonclinical hallucinations.¹¹ By contrast, candidates for discontinuous mechanisms include prefrontal reality-monitoring processes and striatal dopaminergic dysregulation, which appear more specific to hallucinations in psychosis. These points of difference might explain intact reality-monitoring and preserved insight in the nonclinical groups.

Further research into the neural and psychological processes underlying nonclinical hallucinations will likely clarify this picture. Methodologically robust designs should include appropriate psychiatric and nonpsychiatric comparison groups, and a greater focus on potential differences between groups in medication, lifestyle, and clinical risk factors. Greater attention should also be paid to co-occurring experiences (eg, delusional thinking, negative symptoms) which are often neglected in nonclinical groups.

In conclusion, there are dangers in inferring continuity in mechanism from continuity in phenomenology, including a risk of incorrect treatment. On the other hand, discontinuity of mechanism between clinical and nonclinical groups presents opportunities for research. These include the development of sensitive measures to chart heterogeneity across the clinical divide and enhance the detection of individuals at risk of psychosis. Another challenge is to understand the therapeutic needs of people with different hallucination subtypes so that individuals can access support depending on symptom-specific needs. In the meantime, a graded approach to intervention (holding in mind the possibility of phenomenological continuity) is desirable, as is recognition of possible discontinuity in underlying mechanisms, with interventions tailored accordingly.¹⁶

Funding

C.F. was supported by Wellcome Trust (WT108720).