Delusions and prediction error: re-examining the behavioural evidence for disrupted error signalling in delusion formation

Introduction. There is now significant evidence that prediction error signalling is mediated by dopamine in the midbrain, and that dopamine dysfunction is implicated in people experiencing psychotic symptoms, including delusions. There has also been significant theorizing and experimentation concerning the remaining link in this triad, namely that deviant prediction error signalling produces or maintains psychotic symptoms.

Methods. The research supporting the link between prediction error signalling and delusional symptoms was reviewed. Numerous studies indirectly support this link, but only one set of studies claim to directly test this hypothesis by combining three crucial elements: a patient sample, a manipulation of prediction error and neuroimaging. This particular set of studies were examined in detail.

Results. Important methodological limitations in these studies were observed, and a reinterpretation of their data was offered.

Conclusions. Methodological inconsistencies significantly weaken the claims made by these studies, but their data are consistent with current theorizing and they are instructive for future lines of inquiry in this field.     

Reward prediction error signals associated with a modified time estimation task

The feedback error-related negativity (fERN) is a component of the human event-related brain potential (ERP) elicited by feedback stimuli. A recent theory holds that the fERN indexes a reward prediction error signal associated with the adaptive modification of behavior. Here we present behavioral and ERP data recorded from participants engaged in a modified time estimation task. As predicted by the theory, our results indicate that fERN amplitude reflects a reward prediction error signal and that the size of this error signal is correlated across participants with changes in task performance.     

Delusions and prediction error: clarifying the roles of behavioural and brain responses

Griffiths and colleagues provided a clear and thoughtful review of the prediction error model of delusion formation [Cognitive Neuropsychiatry, 2014 April 4 (Epub ahead of print)]. As well as reviewing the central ideas and concluding that the existing evidence base is broadly supportive of the model, they provide a detailed critique of some of the experiments that we have performed to study it. Though they conclude that the shortcomings that they identify in these experiments do not fundamentally challenge the prediction error model, we nevertheless respond to these criticisms. We begin by providing a more detailed outline of the model itself as there are certain important aspects of it that were not covered in their review. We then respond to their specific criticisms of the empirical evidence. We defend the neuroimaging contrasts that we used to explore this model of psychosis arguing that, while any single contrast entails some ambiguity, our assumptions have been justified by our extensive background work before and since.     

Pupil dilation indicates the coding of past prediction errors: Evidence for attentional learning theory

The attentional learning theory of Pearce and Hall ( 1980 ) predicts more attention to uncertain cues that have caused a high prediction error in the past. We examined how the cue‐elicited pupil dilation during associative learning was linked to such error‐driven attentional processes. In three experiments, participants were trained to acquire associations between different cues and their appetitive (Experiment 1), motor (Experiment 2), or aversive (Experiment 3) outcomes. All experiments were designed to examine differences in the processing of continuously reinforced cues (consistently followed by the outcome) versus partially reinforced, uncertain cues (randomly followed by the outcome). We measured the pupil dilation elicited by the cues in anticipation of the outcome and analyzed how this conditioned pupil response changed over the course of learning. In all experiments, changes in pupil size complied with the same basic pattern: During early learning, consistently reinforced cues elicited greater pupil dilation than uncertain, randomly reinforced cues, but this effect gradually reversed to yield a greater pupil dilation for uncertain cues toward the end of learning. The pattern of data accords with the changes in prediction error and error‐driven attention formalized by the Pearce‐Hall theory.     

Manipulation of feedback expectancy and valence induces negative and positive reward prediction error signals manifest in event-related brain potentials

The feedback-related negativity (FRN) has been hypothesized to be most sensitive to unexpected negative feedback. The present study investigated feedback expectancy and valence using a probabilistic gambling paradigm where subjects encountered expected or unexpected positive and negative feedback outcomes. In line with previous studies, FRN amplitude reflected a negative reward prediction error, but to a minor extent also a positive reward prediction error. Moreover, the P300 amplitude was largest after unexpected feedback, irrespective of valence. We propose to interpret the FRN in terms of a reinforcement learning signal which is detecting mismatch between internal and external representations indexed by the ACC to extract motivationally salient outcomes.     

Quantification of error between the heartbeat intervals measured form photoplethysmogram and electrocardiogram by synchronisation

Abstract

Currently, heartbeat intervals required for the analysis of heart rate variability (HRV) are derived from electrocardiogram (ECG). Many investigators have explored the possibility of using photoplethysmography (PPG), for the analysis of HRV. However, all these studies are based on statistical approach and have used the correlation coefficient for the comparison of HRV data obtained using ECG and PPG, which is inappropriate as the causal relationship between the R-peaks in ECG and the systolic peaks in PPG is well known in physiology. In this study, the heart beat intervals measured from PPG, are compared, beat by beat, with the corresponding beat intervals of same cardiac cycle obtained from the synchronously recorded ECG and the differences between them are taken as errors. These errors are verified to exactly match with the variations in the pulse transit times (PTTs), beat by beat. The error in the measurement of heartbeat intervals using PPG is quantified by obtaining the root mean square of the errors associated with each beat interval for a subject. The rms error, which is found to vary between 0.17 and 1.81% across the study group of 42 subjects, can be treated as insignificant, considering the nonstationary character of physiological signals. The errors are compared and interpreted with the variations in PTT. In view of these findings, PPG can be considered as a low cost, safe and more convenient alternative to ECG, as a wearable sensor outside hospital environment, for the analysis of HRV, without compromising on accuracy.     

Developmental differences in learning and error processing: Evidence from ERPs

This study examined developmental differences in the ERP correlates of internal and external error processing (ERN and FRN) during learning. A probabilistic learning task was applied in which feedback validity was manipulated. The behavioral data showed similar accuracy for children and adults when feedback was valid, whereas age differences were obtained when it was partially invalid. We found no reduction of the ERN for children compared to adults when performance levels were equated. Yet, contrary to adults, children did not differentiate between responses when feedback was partially invalid, indicating that they are less able to represent the correctness of a response when there is interference during learning. Moreover, we found a larger FRN and reduced ERP learning effects for positive feedback for children, suggesting that they are more sensitive to external error feedback and less able to disengage from positive feedback during learning.     

When the outcome is different than expected: Subjective expectancy shapes reward prediction error at the FRN level

Converging evidence in human electrophysiology suggests that evaluative feedback provided during performance monitoring (PM) elicits two distinctive and successive ERP components: the feedback‐related negativity (FRN) and the P3b. Whereas the FRN has previously been linked to reward prediction error (RPE), the P3b has been conceived as reflecting motivational or attentional processes following the early processing of the RPE, including action value updating. However, it remains unclear whether these two consecutive neurophysiological effects depend on the direction of the unexpectedness (better‐ or worse‐than‐expected outcomes; signed RPE) or instead only on the degree of unexpectedness irrespective of direction (i.e., unsigned RPE). To address this question, we devised an experiment in which we manipulated the objective reward probability and the subjective reward expectancy (via instructions) in a factorial within‐subject design and explored amplitude changes of the FRN and the P3b. A 64‐channel EEG was recorded while 32 participants performed a speeded go/no‐go task in which evaluative feedback based on the reward probability either violated expectancy (thereby creating a RPE) or did not. This violation corresponded either to better‐ or worse‐than‐expected events. Results showed that the FRN was larger when RPE occurred than when it did not, but irrespective of the direction of this violation. Interestingly, in these two conditions, action value was updated for the positive feedback selectively, as shown by the P3b amplitude. These results obey a two‐stage model of PM assuming that unsigned RPE is first rapidly detected (FRN level) before the positive feedback’s value is updated selectively (P3b effect).     

Temporal dissociation of salience and prediction error responses to appetitive and aversive taste

The feedback‐related negativity (FRN), a frontocentral ERP occurring 200–350 ms after emotionally valued outcomes, has been posited as the neural correlate of reward prediction error, a key component of associative learning. Recent evidence challenged this interpretation and has led to the suggestion that this ERP expresses salience instead. Here, we distinguish between utility prediction error and salience by delivering or withholding hedonistically matched appetitive and aversive tastes, and measure ERPs to cues signaling each taste. We observed a typical FRN (computed as the loss‐minus‐gain difference wave) to appetitive taste, but a reverse FRN to aversive taste. When tested axiomatically, frontocentral ERPs showed a salience response across tastes, with a particularly early response to outcome delivery, supporting recent propositions of a fast, unsigned, and unspecific response to salient stimuli. ERPs also expressed aversive prediction error peaking at 285 ms, which conformed to the logic of an axiomatic model of prediction error. With stimuli that most resemble those used in animal models, we did not detect any frontocentral ERP signal for utility prediction error, in contrast with dominant views of the functional role of the FRN ERP. We link the animal and human literature and present a challenge for current perspectives on associative learning research using ERPs.     

Individual differences in decisiveness: ERP correlates of response inhibition and error monitoring

The aim of the current study was to examine whether and how self‐reported decisiveness is associated with response inhibition and performance monitoring. We hypothesized that these two cognitive control mechanisms, both of which are often associated with decision making, would differ in individuals varying in decisiveness. We focused on ERP correlates and behavioral measures of inhibition and error processing in the stop‐signal task. We expected a negative relationship between decisiveness and behavioral measures of inhibitory control. We also hypothesized that stop‐signal‐locked N1 and P3 components and response‐locked error‐related negativity (ERN) would be less pronounced when participants self‐reported higher levels of decisiveness. Correlation analysis identified an association between high decisiveness, long stop‐signal reaction time, and low inhibition rate. Analysis with mixed‐effects linear models revealed that stop signals evoked less pronounced N1 and P3 in individuals scoring higher on decisiveness in both successfully and unsuccessfully inhibited trials. Additionally, high decisiveness was linked to reduced error monitoring, as indicated by decreased ERNs. Importantly, we also found positive association between P3 onset latency and decisiveness, suggesting that individuals scoring higher on this measure have relatively less ability to rapidly engage the stopping process. Thus, our findings primarily indicate that decisiveness is negatively associated with the efficiency of both response inhibition and error monitoring. They also suggest that highly decisive people may share some characteristics of diminished cognitive control with impulsive individuals.     

Investigating the relationship between negative symptoms and metacognitive functioning in psychosis: An individual participant data meta-analysis

Purpose

Negative symptoms are a persistent, yet under-explored problem in psychosis. Disturbances in metacognition are a potential causal factor in negative symptom development and maintenance. This meta-analysis uses individual participant data (IPD) from existing research to assess the relationship between negative symptoms and metacognition treated as summed scores and domains.

Methods

Data sets containing individuals with negative symptoms and metacognition data, aged 16+ with psychosis, were identified according to pre-specific parameters. IPD integrity and completeness were checked and data were synthesized in two-stage meta-analyses of each negative symptoms cluster compared with metacognition in seemingly unrelated regression using restricted maximum likelihood estimation. Planned and exploratory sensitivity analyses were also conducted.

Results

Thirty-three eligible data sets were identified with 21 with sufficient similarity and availability to be included in meta-analyses, corresponding to 1301 participants. The strongest relationships observed were between summed scores of negative symptoms and metacognition. Metacognitive domains of self-reflectivity and understanding others' minds, and expressive negative symptoms emerged as significant in some meta-analyses. The uncertainty of several effect estimates increased significantly when controlling for covariates.

Conclusions

This robust meta-analysis highlights the impact of using summed versus domain-specific scores of metacognition and negative symptoms, and relationships are not as clear-cut as once believed. Findings support arguments for further differentiation of negative symptom profiles and continued granular exploration of the relationship between metacognition and negative symptoms.     

Relationship between prepulse inhibition of acoustic startle response and schizotypy in healthy Japanese subjects

Prepulse inhibition (PPI) of the acoustic startle reflex (ASR) is the most common psychophysiological index of sensorimotor gating. Several studies have investigated the relationship of PPI of ASR to schizotypy in Caucasians. However, little has been reported on this relationship in Asians. We investigated a possible relationship between PPI of ASR and schizotypy in 79 healthy Japanese subjects. Schizotypy was assessed by the Schizotypal personality Questionnaire (SPQ). PPI was evaluated at signal‐to‐noise ratios (SnRs: difference between background noise intensity and prepulse intensity) of +12, +16, and +20 dB. The total SPQ score, cognitive/perceptual score, and interpersonal score correlated negatively with PPI at SnR of +16 and +20 dB. We conclude that PPI is associated with the trait of schizotypy in healthy Asian subjects.     

An exploration of the relationship between use of safety‐seeking behaviours and psychosis: A systematic review and meta‐analysis

Safety‐seeking behaviours are responses employed to protect against perceived threat. In relation to anxiety disorders, safety‐seeking behaviours have been implicated in both the formation and maintenance of distress. Several studies have highlighted similar findings in relation to psychosis; however, this literature has not yet been synthesized. This review is, therefore, being conducted in order to synthesize the literature on safety seeking in people with psychosis to increase the understanding of this relationship. A systematic search identified and included 43 studies comprising 2,592 participants, published between 1995 and 2015. The results indicated that people experiencing psychosis commonly respond to their experiences with behavioural and cognitive strategies intended to manage their difficulties. In relation to safety seeking, avoidance, and resistance, there was a pattern that these responses are associated with increased distress and appraisals of threat. The results relating to engagement response styles showed the opposite pattern. These results provide support for cognitive models of safety seeking and psychosis with many of the meta‐analyses reported here showing a clear pattern of association between behavioural responses and distress. However, the results reported within individual studies are mixed. This appears to be particularly true with the response style of distraction, with our analyses unable to clarify this relationship. It is possible that the mixed results could reflect the complexities in defining safety seeking and distinguishing it from coping in this population. The clinical implications of this are discussed.     

Guidance of spatial attention during associative learning: Contributions of predictability and intention to learn

Expectations of an event can facilitate its neural processing. One of the ways we build expectations is through associative learning. Interestingly, the learning of contingencies between events can also occur without intention. Here, we study feature‐based attention during associative learning, by asking how a learned association between a cue and a target outcome impacts the attention allocated to this outcome. Moreover, we investigate attention in learning depending on the intention to learn the association. We used an associative learning paradigm where we manipulated outcome predictability and intention to learn an association within streams of cue‐target outcome visual stimuli, while stimulus characteristics and probability were held constant. In order to measure the event‐related component N2pc, widely recognized to reflect allocation of spatial attention, every outcome was embedded among distractors. Importantly, the location of the target outcome could not be anticipated. We found that predictable target outcomes showed an increased spatial attention as indexed by a greater N2pc component. A later component, the P300, was sensitive to the intention to learn the association between the cue and the target outcome. The current study confirms the remarkable ability of the brain to extract and update predictive information, in accordance with a predictive‐coding model of brain function. Associative learning can guide a visual search and shape covert attentional selection in our rich environments.     

Lessons in diagnostic virology: expected and unexpected sources of error

Viral diagnostics have shown continued innovation, with serological and molecular diagnostic assays pushing the limits of sensitivity. Technology has provided new automated shared diagnostic platforms that reduce hands‐on time, while with globalisation of the diagnostic market, commercial assays are applied across epidemiologically diverse settings on different patient and viral populations. However, with these novel developments, new and often unexpected sources of diagnostic error emerge. In this review we will reflect on case studies that highlight these often underappreciated or unexpected diagnostic errors spanning pre‐analytical, analytical, and post‐analytic processes. We will also suggest approaches that could help identify error and reduce the impact on patient management.     

Clinical characteristics of hospitalized children with community-acquired pneumonia and respiratory infections: Using machine learning approaches to support pathogen prediction at admission

BackgroundAcute respiratory infections (ARIs) are common in children. We developed machine learning models to predict pediatric ARI pathogens at admission.MethodsWe included hospitalized children with respiratory infections between 2010 and 2018. Clinical features were collected within 24 h of admission to construct models. The outcome of interest was the prediction of 6 common respiratory pathogens, including adenovirus, influenza virus types A and B, parainfluenza virus (PIV), respiratory syncytial virus (RSV), and Mycoplasma pneumoniae (MP). Model performance was estimated using area under the receiver operating characteristic curve (AUROC). Feature importance was measured using Shapley Additive exPlanation (SHAP) values.ResultsA total of 12,694 admissions were included. Models trained with 9 features (age, event pattern, fever, C-reactive protein, white blood cell count, platelet count, lymphocyte ratio, peak temperature, peak heart rate) achieved the best performance (AUROC: MP 0.87, 95% CI 0.83–0.90; RSV 0.84, 95% CI 0.82–0.86; adenovirus 0.81, 95% CI 0.77–0.84; influenza A 0.77, 95% CI 0.73–0.80; influenza B 0.70, 95% CI 0.65–0.75; PIV 0.73, 95% CI 0.69–0.77). Age was the most important feature to predict MP, RSV and PIV infections. Event patterns were useful for influenza virus prediction, and C-reactive protein had the highest SHAP value for adenovirus infections.ConclusionWe demonstrate how artificial intelligence can assist clinicians identify potential pathogens associated with pediatric ARIs upon admission. Our models provide explainable results that could help optimize the use of diagnostic testing. Integrating our models into clinical workflows may lead to improved patient outcomes and reduce unnecessary medical costs.     

The power of frontal midline theta and post‐error slowing to predict performance recovery: Evidence for compensatory mechanisms

Past studies utilizing cognitive control tasks have noted that trials following errors are characterized by slowed reaction time. Despite the assumption long held by researchers that this slowing is compensatory (in the service of post‐error performance recovery), studies consistently show that post‐error trials are no more accurate than post‐correct trials. As a result, it has recently been proposed that post‐error slowing (PES) is merely part of an orienting response that serves no task‐relevant cognitive control purpose. Frontal midline theta (FMθ) oscillations represent another potential compensatory mechanism serving cognitive control processes, yet past studies relying on ERPs have failed to find an association between FMθ and post‐error accuracy. The present study investigated the potentially adaptive role of PES and FMθ oscillations during a flanker task using trial‐by‐trial comparisons. Results indicated that error‐related FMθ oscillations signal the need for enhanced top‐down cognitive control and that PES supports cognitive control by providing the added time needed to achieve greater confidence in judgment. Overall, findings provide convergent evidence that both error‐related FMθ and PES predict performance recovery following errors.     

Human eye movement response to z-axis linear acceleration: the effect of varying the phase relationships between visual and vestibular inputs

We investigated the effect of systematically varying the phase relationship between 0.5-Hz sinusoidal z-axis optokinetic (OKN) and linear acceleration stimuli upon the resulting vertical eye movement responses of five humans. Subjects lay supine on a linear sled which accelerated them sinusoidally along their z-axis at 0.4 g peak acceleration (peak velocity 1.25 m/s). A high-contrast, striped z-axis OKN stimulus moving sinusoidally at 0.5 Hz, 70°/s peak velocity was presented either concurrently or with the acceleration stimulus or alone. Subjects' vertical eye movements were recorded using scleral search coils. When stimuli were paired in the naturally occurring relationship (e.g., visual stripes moving upward paired with downward physical acceleration), the response was enhanced over the response to the visual stimulus presented alone. When the stimuli were opposed (e.g., visual stripes moving upward during upward physical acceleration, a combination that does not occur naturally), the response was not significantly different from the response to the visual stimulus presented alone. Enhancement was maximized when the velocities of the visual and motion stimuli were in their normal phase relationship, while the response took intermediate values for other phase relationships. The phase of the response depended upon the phase difference between the two inputs. We suggest that linear self-motion processing looks at agreement between the two stimuli — a sensory conflict model.