Elsevier

Schizophrenia Research

Volume 222, August 2020, Pages 185-194
Schizophrenia Research

Aberrant connectivity in auditory precision encoding in schizophrenia spectrum disorder and across the continuum of psychotic-like experiences

https://doi.org/10.1016/j.schres.2020.05.061Get rights and content

Abstract

Background

The ability to generate a precise internal model of statistical regularities is impaired in schizophrenia. Predictive coding accounts of schizophrenia suggest that psychotic symptoms may be explained by a failure to build precise beliefs or a model of the world. The precision of this model may vary with context. For example, in a noisy environment the model will be more imprecise compared to a model built in an environment with lower noise. However compelling, this idea has not yet been empirically studied in schizophrenia. Methods: In this study, 62 participants engaged in a stochastic mismatch negativity paradigm with high and low precision. We included inpatients with a schizophrenia spectrum disorder (N = 20), inpatients with a psychiatric disorder but without psychosis (N = 20), and healthy controls (N = 22), with comparable sex ratio and age distribution. Bayesian mapping and dynamic causal modelling were employed to investigate the underlying microcircuitry of precision encoding of auditory stimuli. Results: We found strong evidence (exceedance P > 0.99) for differences in the underlying connectivity associated with precision encoding between the three groups as well as on the continuum of psychotic-like experiences assessed across all participants. Critically, we show changes in interhemispheric connectivity between the two inpatient groups, with some connections further aligning on the continuum of psychotic-like experiences. Conclusions: While our results suggest continuity in backward connectivity alterations with psychotic-like experiences regardless of diagnosis, they also point to specificity for the schizophrenia spectrum disorder group in interhemispheric connectivity alterations.

Introduction

The ability to adapt to the ever changing environment and make inferences about future events is robustly reduced in people with schizophrenia (Adams et al., 2013; Fletcher and Frith, 2009; Kapur, 2003). This has been studied in a classical laboratory setting using auditory oddball paradigms where unpredictable (“oddball”) sounds are interspersed in a stream of frequently occurring standard sounds. Such paradigms elicit a mismatch (MMN) response (Garrido et al., 2009b; Näätänen, 1995) or sensory prediction error. According to the predictive coding framework and the model-adjustment hypothesis for MMN (Winkler et al., 1996), this error response is caused by a violation to the regularities, arising when sensory input does not match the prediction according to a learnt model (Friston, 2005). The precision of this model varies according to the precision of context itself, such that a noisier environment will lead to the formation of more unreliable predictions than those formed in a stable environment (Mathys et al., 2011).

The MMN is robustly reduced in schizophrenia (Erickson et al., 2016; Umbricht and Krljes, 2005), first episode psychosis (Haigh et al., 2016) and in individuals with high risk for schizophrenia (Atkinson et al., 2012; Perez et al., 2014), hence suggesting that the ability to generate a precise model of the environment is reduced in schizophrenia and to some extent in the continuum of psychosis (Randeniya et al., 2017; van Os et al., 2009). Recent predictive coding accounts put forward that the range of psychotic symptoms in schizophrenia can be explained by a failure to build precise beliefs or a model of the world (Adams et al., 2013). The underlying processes of MMN generation involve both adaptation (neural habituation due to repeated standard sound stimuli) and prediction formation (guessing what might come next) (Garrido et al., 2009b; Larsen et al., 2019) It is unclear whether the consistent reductions of MMN in schizophrenia are caused by a failure of either adaptation or prediction, or both (Michie and Malmierca, 2016). Adaptation and prediction formation are hard to disentangle with classical oddball paradigms because they typically evoke both processes simultaneously. Effective connectivity modelling attempts (i.e. inferring the dynamic effect one brain region has on another (Friston et al., 2003)), point to both adaptive and predictive processes being affected in schizophrenia (Dima et al., 2012). In that study, dynamic causal modelling (DCM) was used to show that patients with schizophrenia have altered connectivity within the top-down connection from the right inferior frontal gyrus to the right superior temporal gyrus, as well as in the intrinsic connection (self-connection) within right primary auditory cortex. These two types of connections (intrinsic and top-down) are believed to reflect adaptation and prediction processes respectively, indicating that both processes were affected in that sample. We have further shown evidence that the same connections are altered in non-psychotic individuals with a genetic high risk for developing schizophrenia (Larsen et al., 2018). Such connectivity disruptions might explain the failure to build a precise (top-down) belief or model of the world.

While the majority of previous studies on MMN in schizophrenia have been case-control studies assessing group differences, recent results indicate that a continuum perspective of psychosis is more powerful in explaining real-world functioning rather than a categorical approach as proposed in the Diagnostic and Statistical Manual of Mental Disorders (DSM) (Hanlon et al., 2019; Owen, 2014; Owen and O'Donovan, 2017). According to the continuum view, the major clinical symptoms reflect the degree of alterations in brain function resulting in functional abnormalities (Owen and O'Donovan, 2017). Hence, while the categorical approach is very useful for diagnostic purposes, the behavioural and brain alterations may not always follow this dichotomy and its aetiology might be best aligned on a continuum. An evident manifestation of this is precisely the MMN attenuation over the psychosis continuum (Randeniya et al., 2017).

Here, we set out to investigate prediction errors with precision of the prediction manipulated to be either high or low in inpatients with a schizophrenia spectrum disorder (schizoaffective disorder or schizophrenia), inpatients with a psychiatric disorder but without psychosis, and healthy controls. We will adopt two approaches, a group comparison and a continuum approach along the dimension of psychotic-like experiences across all participants (N = 62, patients and controls) and parameterised by psychotic like experiences. The group approach will enable inferences on the specificity of connectivity changes in the schizophrenia spectrum group given our inpatient control group with a psychiatric diagnosis but without psychosis. The continuum approach will allow us to make inferences about how precision encoding and brain connectivity varies with the degree of psychotic-like symptoms irrespective of group membership. We use a stochastic MMN paradigm previously validated in healthy controls (Garrido et al., 2013) that allows us to tap into predictive processes while mitigating adaptive processes, as well as to manipulating precision levels in the auditory environment. We hypothesise that the ability to encode the level of precision is reduced in the schizophrenia spectrum group compared to the two (non-psychosis and healthy) control groups. We expect this will be present both at the scalp level as well as in the effective connectivity. In addition, we hypothesise that the ability to encode precision decreases over the continuum of psychotic-like experiences across the whole sample.

Section snippets

Participants

62 participants took part in the study, with 20 inpatients with a schizophrenia spectrum disorder (SZS), 20 non-psychotic inpatients controls (NP), and 22 healthy controls (HC). Inpatients were recruited from the Monash Medical Centre, acute adult psychiatric inpatient facility. HC's were recruited through the Psychology Research Participation Scheme (SONA), using Gumtree, and flyers distributed around the Monash Medical Centre. Participants provided written informed consent before taking part

Results

There was no overall difference in the performance on the 1-back task when comparing d′ and reaction times across groups, see Fig. 2 in Supplementary material. This rather easy task was intentionally chosen to keep participants engaged during the experiment, while avoiding potential confounds of task performance across groups.

Discussion

Here we show that patients with schizophrenia spectrum disorder have alterations in the encoding of contextual precision that are associated with aberrant brain effective connectivity. Critically, we show that part of this connectivity alteration is specific to schizophrenia spectrum with the effect being present in the schizophrenia spectrum group but not the two control groups. We found not only group differences in the connectivity, but also connectivity changes across individuals on the

Role of funding body

The funding body had no role in conducting this study.

Contributors

MIG, SS, ID and OC contributed in the design of the study. ID, HD and HP collected the data. KML analysed the data and wrote the first draft of the manuscript. All authors contributed to the interpretation of the data, revised the manuscript and agreed with the final content of the manuscript.

Financial disclosures

All authors declare that there are no conflicts of interest in relation to this study.

Acknowledgements

We would like to thank all included participants for their time in this study as well as the nurses in the psychiatric ward at the Monash Medical Centre, Adult Inpatient Psychiatric Facility.

Funding

This work was funded by the Australian Research Council Centre of Excellence for Integrative Brain Function (ARC Centre Grant CE140100007), a University of Queensland Fellowship (2016000071), and a Women's Academic Fund Maternity funding from Queensland Government, to MIG. OC is supported by the Australian Research Council FFT40100807.

References (46)

  • S.J. Kiebel et al.

    Dynamic causal modelling of evoked responses: the role of intrinsic connections

    Neuroimage

    (2007)
  • K.M. Larsen et al.

    Individuals with 22q11.2 deletion syndrome show intact prediction but reduced adaptation in responses to repeated sounds: evidence from Bayesian mapping

    Neuro. Image Clin.

    (2019)
  • P.T. Michie et al.

    The neurobiology of MMN and implications for schizophrenia

    Biol. Psychol.

    (2016)
  • N. Mossaheb et al.

    The Community Assessment of Psychic Experience (CAPE) questionnaire as a screening-instrument in the detection of individuals at ultra-high risk for psychosis

    Schizophr. Res.

    (2012)
  • M.J. Owen

    New approaches to psychiatric diagnostic classification

    Neuron

    (2014)
  • W.D. Penny et al.

    Bayesian fMRI time series analysis with spatial priors

    Neuroimage

    (2005)
  • V.B. Perez et al.

    Automatic auditory processing deficits in schizophrenia and clinical high-risk patients: forecasting psychosis risk with mismatch negativity

    Biol. Psychiatry

    (2014)
  • M.J. Rosa et al.

    Bayesian model selection maps for group studies

    Neuroimage

    (2010)
  • D. Umbricht et al.

    Mismatch negativity in schizophrenia: a meta-analysis

    Schizophr. Res.

    (2005)
  • I. Winkler et al.

    Adaptive modeling of the unattended acoustic environment reflected in the mismatch negativity event-related potential

    Brain Res.

    (1996)
  • Y. Zhou et al.

    Altered intrinsic and extrinsic connectivity in schizophrenia

    NeuroImage. Clin.

    (2018)
  • R.A. Adams et al.

    The computational anatomy of psychosis

    Front. Psychiatry

    (2013)
  • X. Chang et al.

    Distinct inter-hemispheric dysconnectivity in schizophrenia patients with and without auditory verbal hallucinations

    Sci. Rep.

    (2015)
  • View full text