Distorted Cognitive Processes in Major Depression: A Predictive Processing Perspective

Review

Biological Psychiatry

Distorted Cognitive Processes in Major Depression: A Predictive Processing Perspective Tobias Kube, Rainer Schwarting, Liron Rozenkrantz, Julia Anna Glombiewski, and Winfried Rief

ABSTRACT The cognitive model of depression has significantly influenced the understanding of distorted cognitive processes in major depression; however, this model’s conception of cognition has recently been criticized as possibly too broad and unspecific. In this review, we connect insights from cognitive neuroscience and psychiatry to suggest that the traditional cognitive model may benefit from a reformulation that takes current Bayesian models of the brain into account. Appealing to a predictive processing account, we explain that healthy human learning is normally based on making predictions and experiencing discrepancies between predicted and actual events or experiences. We present evidence suggesting that this learning mechanism is distorted in depression: current research indicates that people with depression tend to negatively reappraise or disregard positive information that disconfirms negative expectations, thus resulting in sustained negative predictions and biased learning. We also review the neurophysiological correlates of such deficits in processing prediction errors in people with depression. Synthesizing these findings, we propose a novel mechanistic model of depression suggesting that people with depression have the tendency to predominantly expect negative events or experiences, which they subjectively feel confirmed due to reappraisal of disconfirming evidence, thus creating a self-reinforcing negative feedback loop. Computationally, we consider too much precision afforded to negative prior beliefs as the main candidate of pathology, accompanied by an attenuation of positive prediction errors. We conclude by outlining some directions for future research into the understanding of the behavioral and neurophysiological underpinnings of this model and point to clinical implications of it. Keywords: Belief update, Cognitive immunization, Depression, Expectation, Prediction error, Predictive processing https://doi.org/10.1016/j.biopsych.2019.07.017

TRADITIONAL COGNITIVE MODEL The cognitive model of depression described by Beck et al. (1–3) has provided a fruitful theoretical framework for understanding major depressive disorder (MDD), assuming that people with MDD tend to interpret environmental experiences in a negative fashion. It has been hypothesized that this maladaptive information processing is caused by dysfunctional cognitions (1), as illustrated in Supplemental Figure S1. Although this model has been deeply influential in research into depression for decades and has even inspired the development of cognitive behavioral treatment as the gold standard in the treatment of MDD (1), its conception of cognition has recently been criticized as possibly too broad and unspecific (4). Rief and Joormann (4) argued that the concept of expectations,1 representing the subgroup of cognitions that refer to future events or experiences (5,6), might more aptly describe certain phenomena with which people with MDD struggle (such as the expectation that [subjective] 1

The terms expectation and expectancy are often used interchangeably. However, expectation is more frequently used as a specific, verbalized construct, whereas expectancies may be present without full awareness (i.e., implicit expectancies). Here, we use only the term expectation.

ISSN: 0006-3223

loneliness will never end). In line with this notion, there has been a considerable increase of empirical studies revealing a significant impact of expectations on depression. In this article, we review findings from psychiatry and clinical psychology literature on the role of expectations in MDD and connect them with recent evidence from cognitive neuroscience on predictive processing. Bridging these bodies of literature, we will introduce an expectation-focused model of depression that may provide new insights into dysfunctional cognitive processes in MDD.

RELEVANCE OF EXPECTATIONS FOR THE DEVELOPMENT OF DEPRESSIVE SYMPTOMS Research has consistently revealed associations between depressive symptoms and different types of expectations, such as low self-efficacy expectancies (7–9) and negative global expectations about future events (10–12). Furthermore, a longitudinal study has shown that in youths seeking emergency psychiatric care, patients’ self-rated expectations of suicidal behavior predicted actual suicidal attempts over a period of 18 months (13). Recent studies have further specified how exactly expectations influence the development of depressive symptoms. For example, a study on high-risk adolescents found that a

ª 2019 Society of Biological Psychiatry. 1 Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

Biological Psychiatry

A Predictive Processing Approach to Depression

lack of positive expectations about future events, rather than excessively negative expectations, predicts depressive symptoms and suicidal behavior over a period of 2 to 4 years (14). Moreover, our group found that in people with MDD, the effects of negative global cognitions on depressive symptoms were mediated through dysfunctional situational expectations (15). That is, negative global cognitions may elicit negative situation-specific predictions, which then cause depressive symptoms. As will be explained in more detail below, there is reason to assume that dysfunctional situational expectations affect depressive symptoms because people with MDD typically feel that their negative predictions are confirmed, whereas positive predictions (if there are any) are not confirmed.

PREDICTIVE PROCESSING Parallel to the clinical literature, expectations have been studied in a very dynamic field of research in cognitive neuroscience, which, with some important exceptions (16,17), has rarely been connected with theoretical models of depression: predictive coding and error processing. According to this literature, the brain is neither passive nor stimulus driven. Rather, with reference to Bayesian models, which explain how the brain handles uncertainty (18), the brain actively generates top-down predictions about expected sensory input. Incoming sensory signals are constantly matched with prior predictions, referred to as predictive processing (19–22). In this terminology, a mismatch between predicted and actual sensory data is referred to as prediction error (PE). PEs thus provide corrective feedback, which normally results in an update of predictions, with the aim of optimizing the use of incoming signals and minimizing PEs (19–22). An important aspect related to updating predictions after PEs refers to the precision of predictions and perceptions (19,23,24). That is, the more precisely an incoming signal is perceived, the stronger is its impact on the adjustment of predictions after PEs. In other words, in the case of imprecise perceptions, prior expectations can be maintained despite the experience of PEs. Relatedly, prior predictions have more impact on perception if imprecise (noisy) input is expected. Precision is modulated by attending to sensory input versus attenuating it. We will draw on this concept when discussing the phenomenon of persistent negative expectations in MDD later in this article. Another concept that will be important in what follows is active inference. Active inference means that action and perception are closely linked, in the sense that people are inclined to construct an environment (and physiology) that matches their predictions to minimize PEs (23,25–27). Evidence supporting this predictive processing framework comes particularly from research into visual (28–32) and auditory (33,34) processing. Besides exteroceptive sensations, the predictive processing account has recently been applied to interoception (i.e., sensations from the inner milieu of the body) (23,26,35,36). Following this account, predictive processing has been used to provide a new understanding of autism (26,37), pain perception (38), placebo analgesia (39–41), and symptom perception in medically unexplained symptoms (42,43) and hysteria (44).

2

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

Neurobiology of Predictive Processing Predictive processing at the neurobiological level has been described in terms of connections between cerebral hierarchies: in higher levels of cortical hierarchies, predictions are generated (encoded by synaptic activity) and descend to lower levels, where top-down predictions are compared with (bottom-up) ascending sensory signals (in superficial pyramidal cells). The confidence in the descending predictions or the reliability of ascending sensory input (i.e., precision) is encoded by the weight given to these signals (19,45). For a detailed description of the neural architecture of predictive processing in interoception, see Barrett and Simmons (23). In terms of brain activity related to PE processing, research has indicated that, besides the insula and the fusiform gyrus, the ventral striatum (VS) is crucially involved (46,47). A recent metaanalysis has further specified that the VS is particularly activated in studies that encourage a behavioral adaptation after a PE compared with studies in which participants only passively experience PEs (48).

Implications for the Understanding of Depression An important implication of this concept of a predictive mind (49) for the understanding of depression is that “what we perceive is not the world as it actually is, but the brain’s best guess of it” (41). Applying this framework to depression, it has been argued that people with MDD frequently anticipate negative events or experiences, whereas they rarely anticipate positive events or experiences (4). According to Bayesian brain models, this skewed pattern of predictions may affect the way people with MDD perceive the world: based on their prior predictions, they are likely to perceive their environment as predominantly negative. In other words, in terms of a selffulfilling prophecy, people with MDD perceive their world as being predominantly negative because they expect it to be predominantly negative. A similar account of depression has been proposed by Barrett et al. (17). Drawing on the predictive processing framework in interoception, the authors argue that depression results from a locked-in brain, meaning that negative predictions continue to shape perception negatively, resulting in persistent negative affect. In line with this notion, the literature on affective forecasting (50) indicates that dysphoric people are biased in predicting future emotional states toward the overestimation of negative emotional reactions to future events (51,52). Moreover, Barrett et al. (17) postulate that reduced activity and increased withdrawal may further contribute to this locked-in brain through lack of exploration that usually involves encoding new information and processing PEs. Similarly, Clark et al. (16) have provided a predictive processing approach to mood disorders. In their account, the authors relate negative emotions to the prediction of an unpredictable (i.e., uncertain) situation, whereas positive emotions are assumed to be associated with events that resolve uncertainty, thus enabling the person to have a sense of control. The authors postulate that depression, as a form of chronically negative mood, is characterized by inappropriately high precision in the prediction of an unpredictable, uncontrollable world; the authors state: “[.] depression occurs when the brain is certain that it will encounter an uncertain

Biological Psychiatry

A Predictive Processing Approach to Depression

event” (p. 2278) (16) [for other computational perspectives on mood disorders, see (26,53,54)]. A novel aspect we want to add in this article is to use the predictive processing framework to account for distorted cognitive processes in depression, enabling us to explain why the experience of positive PEs in depression often does not lead to an update of negative expectations.

LACK OF EXPECTATION UPDATE IN DEPRESSION Behavioral Studies Several lines of research converge on the finding that people with MDD have difficulty updating negative expectations after unexpected positive experiences. First, research on interpretations biases has shown that people with MDD tend to interpret ambiguous situations often negatively and less often positively, especially if they contain self-referential stimuli (55). Moreover, it has been indicated that people with MDD maintain established negative interpretations of ambiguous information even after novel information that disconfirms the initial negative interpretation in favor of a more positive interpretation (56,57). Second, cognitive inflexibility, reflecting (the lack of) “human abilities to recognize and adapt to various situational demands” (58), has been discussed as a vulnerability factor of MDD (59). In particular, cognitive inflexibility has been shown to predict the first onset of a depressive episode (60) and suicidal ideation (61). In addition, depressive symptoms have been linked to inflexibility in processing unexpected positive information (56). Third, it has been shown that healthy people update their expectations of future events in an optimistically biased way even after receiving negative information (62–65). This optimism bias seems to be absent in MDD (66) [for the neural substrates of this effect, see (67)].

Cognitive Immunization. Consistent with this previous work, our group has found in a series of experimental studies that healthy people are able to adjust negative performancerelated expectations when receiving positive expectationdisconfirming performance feedback; in contrast, people with MDD maintain negative performance expectations despite positive feedback (68). To explain this phenomenon, the concept of cognitive immunization has been introduced. In the clinical psychology literature, cognitive immunization has been defined as a reappraisal of expectation-disconfirming experiences in such a way that the person’s previous expectations are maintained (69). People with MDD could, for instance, engage in cognitive immunization strategies by considering disconfirming evidence to be an exception rather than the rule or by questioning its credibility (70). For example, an unexpectedly positive social interaction could be negatively reappraised by thinking, “This person was just pretending to like me. In fact, she does not like me.” Such kind of reappraisal may make people with MDD immune to positive disconfirmatory evidence. This hypothesis has recently been confirmed: experimentally promoting the use of cognitive immunization strategies increased the likelihood of persisting negative expectations, whereas an inhibition of cognitive immunization strategies boosted expectation change in people reporting elevated

levels of depression (68). In a subsequent experiment, we compared 3 strategies aimed at inhibiting cognitive immunization and found that particularly a strategy aimed at increasing the value of disconfirming information facilitated the adjustment of negative expectations after positive experiences in people with MDD (71). Conversely, in another experiment, we examined whether the negative reappraisal of disconfirming positive information could be mimicked in healthy people by experimentally promoting the use of cognitive immunization strategies (T. Kube, Ph.D., et al., unpublished data, August 2019). We found that promoting the use of cognitive immunization strategies did not affect perceptions of healthy people of unexpectedly positive feedback; healthy people updated negative expectations after positive feedback regardless of immunization-promoting strategies. These findings suggest that the engagement in cognitive immunization strategies after unexpected positive information is specific to MDD, but not to healthy people (T. Kube, Ph.D., et al., unpublished data, August 2019).

Neurophysiological Studies Current neuroscientific research on predictive processing in depression is largely based on the study of reward PEs (discrepancies between expected and actual reward), built on the literature on reward insensitivity relating depression to hyposensitivity to rewarding and hypersensitivity to aversive (i.e., punishment) stimuli [for reviews, see (72,73)]. It has been suggested that striatal and prefrontal regions are crucially involved in reward processing, whereas the amygdala and insula are involved in processing aversive stimuli (74,75). Consistent with this view, some (76–79), albeit not all (80), studies found reduced activation in the (ventral) striatum for anticipated or delivered rewards in MDD. Relatedly, despite considerable heterogeneity in the literature, research into reinforcement learning has linked depression to hyposensitivity to rewards and hypersensitivity to punishments (81–84). More recently, researchers in this area appealed to the predictive processing framework and argued that reinforcement learning might not primarily be driven by the amount of rewards versus punishments, but rather by the corresponding PEs (85,86). Following this argument, we next review the results from 12 studies examining neural responses to reward PEs [11 using functional magnetic resonance imaging and 1 using electroencephalography (87)], which are described in more detail in Supplemental Table S2. At the outset, it should be noted that most studies did not find clear differences between people with MDD and healthy control (HC) subjects at the behavioral level (i.e., reaction time, accuracy, pleasantness), whereas the studies did find differences at the level of brain activity. As a region of particular interest, 5 studies reported reduced activation in the VS among individuals with MDD compared with HC subjects when processing reward PEs (88–92); 4 other studies, however, did not find such differences (85,93–95). Furthermore, the specificity of this effect is somewhat questioned by the fact that 2 studies that additionally included patients with schizophrenia found similar VS activation for reward PEs in MDD and schizophrenia (88,91). Some studies, albeit not all (94), indicated correlations between VS-related blunted reward PE processing and illness

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

3

Biological Psychiatry

A Predictive Processing Approach to Depression

Figure 1. The role of expectations in the develLearning and conditioning

opment of depressive symptoms. In this model, we distinguish between generalized expectations and situation-specific expectations. Whereas generalized expectations refer to a wide range of situations, situation-specific expectations are focused on a Dysfunctional Dysfunctional Social influences particular situation (6,69). Generalized expectations Depressive symptoms generalized situation-specific (e.g., peers, group norms, media) expectations expectations develop based on learning experiences and conditioning, social influences (e.g., peers, group norms, and media), and interindividual differences (including genetic vulnerabilities) (69). They can reflect both Inter-individual positive (“I’m always optimistic about my future”) differences (incl. genetic factors) and negative (“I hardly ever expect things to go my way”) predictions of future events or experiences. People with MDD are characterized by both a lack of positive generalized expectations and overly negative generalized expectations (10,14). When exposed to specific situations, negative generalized expectations can elicit negative situational expectations, which evoke depressive symptoms (15,122). Of note, the presence of dysfunctional expectations may not necessarily result in full-blown MDD, but it does increase the likelihood of the development of depressive symptoms.

severity or anhedonia severity (85,88,89). One study examining a large sample may have partially resolved the inconsistent findings regarding the role of the VS and anhedonia in reward PE processing (95): assessing both expected reward and reward PE, the authors demonstrated a negative correlation between reward expectancy-related and PE-related reactivity in the right VS in HC subjects, but not in subjects with MDD. Across all participants, this relationship was moderated by anhedonia severity. The authors concluded that there is no abnormal PE processing in the VS in people with MDD per se, but rather an aberrant relationship between reward expectancy and PE processing in the VS, depending on the level of anhedonia (95). Aside from the VS, one study examining a sample of elderly individuals with depression found disrupted encoding of reward PEs in several cortical sites and the thalamus compared with HC subjects (86). Another study found decreased activation for reward PEs in the right rostral anterior cingulate cortex in participants with MDD compared with HC subjects (96). For punishment PEs, the literature to date is similarly heterogeneous: whereas some studies found increased sensitivity to negative PEs (86,96), others did not (85,89,90,93). In sum, the evidence for abnormalities in processing PEs in depression is not fully consistent; this is probably due to factors such as choice of task, brain region, sample size, disease status, symptom severity, and medication. Given this empirical heterogeneity, simplified conclusions are undue.

SYNTHESIS OF EVIDENCE: AN EXPECTATIONFOCUSED MODEL OF DEPRESSION As a synthesis of the recent findings reviewed above, we propose a novel explanatory model for the development and maintenance of depression. The first part of the model refers to the role of expectations in the exacerbation of depression, as illustrated and explained in more detail in Figure 1. In brief, this model suggests that people with MDD hold negative generalized expectations, which, when exposed to particular situations, elicit negative situational predictions that evoke depressive symptoms. To explain the maintenance of depressive symptoms, we use the predictive processing framework and distinguish between positive versus negative expectations and positive

4

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

versus negative PEs (Figure 2A). If negative expectations are positively disconfirmed, healthy people change their expectations in a positive direction, whereas people with MDD maintain their negative expectations after unexpectedly positive experiences, as they tend to devalue disconfirming positive experiences through cognitive immunization strategies (68). Whereas the evidence concerning asymmetries in expectation update between healthy and depressed people is to date largely based on the disconfirmation of negative expectations, our model hypothesizes that a reverse pattern of differences can be found when positive expectations are negatively disconfirmed. We assume that people with MDD change positive expectations in a negative direction after negative experiences, whereas healthy people sustain positive expectations by disregarding disconfirming negative information. This hypothesis is supported by research on the optimism bias as described above (62–64,66,97) and is further supported by research on depressive realism suggesting that mildly depressed people show a more realistic evaluation of self than healthy people (98–103). Based on these findings, we suggest that both updating negative expectations in a positive direction after disconfirming positive experiences and engaging in cognitive immunization strategies after unexpectedly negative experiences are associated with mental health. Combining the two components of the model concerning the exacerbation and the maintenance of depressive symptoms (Figure 3), we propose that information processing in people with MDD resembles a self-reinforcing negative feedback loop: people with MDD predominantly predict negative experiences, which they often feel to be confirmed because potential positive information is discarded through cognitive immunization strategies, hence further stabilizing negative predictions. In addition, positive experiences are rarely expected, thus decreasing the likelihood of perceiving them. Connecting this account with active inference approaches to mood (16,53), it follows that the (subjective) confirmation of negative predictions (resulting in negative posterior predictions) leads to negative mood. This is likely to promote behaviors that confirm negative predictions [i.e., sickness behaviors (17,23,26,104)], which further contributes to sustained negative mood and amplifies the selffulfilling prophecy concerning expectation and perception as described above. From a clinical point of view, there is reason

Biological Psychiatry

A Predictive Processing Approach to Depression

A Negative Expectation

Positive Expectation

Healthy people

Generalization of positive information

Expectation update

Depressed people

Cognitive immunization

Expectation maintenance

Healthy people

Cognitive immunization

Expectation maintenance

Depressed people

Generalization of negative information

Expectation update

Positively valenced disconfirmatory information

Negatively valenced disconfirmatory information

B

Figure 2. Expectation update vs. maintenance in healthy vs. depressed people. (A) It is assumed that healthy people change negative expectations in a positive direction after disconfirmatory positive information because they are able to generalize this positive disconfirmatory evidence and update their predictions accordingly. By contrast, people with depression are assumed to engage in cognitive immunization strategies after unexpected positive information (e.g., by considering the disconfirmatory experience to be an atypical exception), thus maintaining negative expectations. A mirrored pattern is predicted for negative disconfirmatory information: whereas depressed people generalize negative information and update their expectations accordingly, healthy people use cognitive immunization strategies to positively reappraise unexpected negative information, resulting in sustained optimistic expectations. (B) Suggested neural mechanism of persisting expectations in depressed people vs. healthy people: suppression of prediction error processing in the reward system. Specifically, the authors assume that cognitive immunization is associated with the suppression of prediction error processing in the ventral striatum by the prefrontal cortex. Importantly, whereas it is postulated that this mechanism in depressed people applies to disregarding reward prediction errors, it may account for disregarding aversive prediction errors in healthy people.

to assume that this dysfunctional pattern of information processing is particularly pronounced in people with persistent depressive disorders, as discussed in the Supplement.

Neuronal and Computational Specification of the Model Research has provided evidence for both dysfunctional expectations per se in depression and lack of updating them. From a predictive processing perspective, this raises the question of whether the main candidate of pathology is the

abnormal prior beliefs or too much precision given to these beliefs. Appealing to the predictive processing accounts of mood (16,53,54), we suggest the latter. In particular, we propose that people with MDD afford too much precision to negative predictions [note that this corresponds well to the psychological literature on depressive predictive certainty (105–109)]; as a result, positive PEs are attenuated and hence given reduced weight. The psychological equivalent of this attenuation is assumed to be cognitive immunization. In other words, if (negative) prior expectations are afforded too much precision, they can become immune to belief updating and

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

5

Biological Psychiatry

A Predictive Processing Approach to Depression

Chronification

Disconfirming experiences

Persistence

Learning and conditioning

Cognitive immunization

Figure 3. Integrative expectation-focused model of depression. It is assumed that depressive symptoms are caused by dysfunctional expectations, which become increasingly immune against positive disconfirming experiences, thus maintaining symptoms of depression and contributing to its chronicity.

Dysfunctional expectations

Social influences

Generalized expectations

Situation-specific expectations

Depressive symptoms

Inter-individual differences

learning from new experience. On the other side of abnormal modulation of precision in depression, we suggest that an attention bias toward negative self-referential stimuli (110,111) increases the precision of confirmatory negative sensory input, thereby contributing to a stabilization of negative predictions. In computational terms, this model can be formalized using the framework of hierarchical reinforcement learning, which accounts for optimistically biased learning in healthy people (112). In Figure 4, we use this framework to illustrate different learning rates in healthy people versus depressed people depending on the valence of the PE (see the Supplement for the corresponding computational models). In terms of neuronal networks involved in the persistence of negative expectations in MDD, we suggest the involvement of the prefrontal cortex (PFC) in addition to the VS [for their interplay, see also (113,114)]. A recent study on persistent placebo hypoalgesia performing a connectivity analysis found negative coupling between the activation in the VS and the anterior PFC for PE processing (115). The authors interpreted these findings in such a way that the PFC can suppress the

processing of PEs in the VS. We postulate that in depression a similar suppression of positive PEs in the VS by the PFC might account for a lack of expectation updating (Figure 2B). This could also shed light into the partly inconsistent results on reward PE processing in depression because, with the exception of one study (95), most studies investigated VS activation alone and not in connection with other brain regions.

Novelty of This Model In comparison with the traditional cognitive model (1), our model does not consider cognitions that relate to the past or the present, but focuses only on expectations as predictions of future events or experiences. In doing so, it is more consistent with current neuroscientific views of the brain as a prediction machine and can aptly explain how expectations bias perception and become immune to disconfirmatory evidence. Though previous predictive processing approaches to depression nicely accounted for behavioral (i.e., reduced activity and withdrawal) (17,104) and mood-related (16,53) Figure 4. Illustration of different learning rates

among healthy people vs. depressed people for positive and negative prediction errors. (A) It is assumed that healthy people update their expectations quite quickly after a few positive prediction errors, whereas people with depression tend to maintain negative prior predictions longer before updating them. Further, the authors suggest that the degree of update after positive prediction errors is overall greater for healthy people than for people with depression. (B) For negative prediction errors, a mirrored pattern is postulated: people with depression update their expectations faster and to a greater extent than healthy people, which corresponds to a recent study suggesting that people with mood and anxiety disorders update their behavior in a reinforcement learning paradigm faster in response to negative outcomes (formalized by increased punishment learning rates) (123). Of note, we recognize that previous studies on reward learning have not consistently found such differences between healthy people and depressed people [e.g., (94,124)], although some others did (86,89–91). Outside the context of reward, the differences between healthy people and depressed people seem somewhat more unequivocal with respect to different responses to good news vs. bad news (66,67), so we assume that the different learning rates postulated here are particularly pertinent to prediction errors in the form of unexpected information about desired vs. unwanted life events, social feedback, or performance feedback.

6

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

Biological Psychiatry

A Predictive Processing Approach to Depression

Table 1. Testable Hypotheses for Future Research Hypothesis Behavioral Level

Number of prediction errors

Experimental Examination

Healthy people update negative expectations after a few disconfirmatory positive experiences (that is, PEs), whereas people with MDD need more positive experiences before updating negative expectations

Trial-by-trial design from reinforcement learning; consider different learning rates for people with MDD and HC subjects

For the update of positive expectations after novel negative information, the pattern is reversed: people with MDD update their expectations after few negative PEs; HC subjects maintain optimistic expectations longer Magnitude of prediction errors

For both people with MDD and HC subjects, there is an “optimal window” for expectation update depending on the magnitude of the PE (that is, the extent to which the actual experience deviates from the predicted experience), as illustrated in Supplemental Figure S3

Experimental variation of the magnitude of the PE

This optimal window can be formalized in terms of an inverse Ushaped relationship between the degree of expectation update and the magnitude of the PE For the update of negative expectations after positive PEs, this optimal window for people with MDD is narrower than for HC subjects, and the overall degree of expectation update is lower (for negative PEs, vice versa)

Neurobiological Level

Chronicity of depression

The persistence of negative expectations despite positive PEs is more pronounced in people with PDD than in people with nonchronic forms of depression (in terms of lower learning rate for positive PEs and a narrower window for expectation update in relation to PE magnitude)

Comparison of people with PDD, people with nonpersistent MDD, and HC subjects

Suppression of PE processing

Lack of expectation update is associated with negative coupling between activation in the VS and the PFC (that is, PE processing in the VS is suppressed by the PFC)

Connectivity analysis in fMRI

The psychological process that corresponds to this attenuation of PE processing is cognitive immunization Role of dopamine

For reward PEs, dopamine encodes the precision of PEs rather than PEs per se

fMRI, functional magnetic resonance imaging; HC, healthy control; MDD, major depressive disorder; PDD, persistent depressive disorder; PE, prediction error; PFC, prefrontal cortex; VS, ventral striatum.

aspects of depression, our model is the first to specifically focus on distorted cognitive processes. In particular, by linking the predictive processing framework with the literature on cognitive immunization, our model provides a coherent explanation as to why the experience of positive PEs in MDD often does not lead to an update of negative predictions. Moreover, beyond traditional cognitive and learning–oriented models of MDD, this model is well suited to derive novel psychotherapeutic interventions (i.e., inhibiting cognitive immunization) (116), as discussed in the Supplement. However, it should be kept in mind that depression is a highly heterogeneous disorder, so the provision of a unifying framework demands explanation. As with other predictive processing accounts of mental disorders (117,118), our model does not reduce depression to a single cause; rather, predictive processing offers a framework to elucidate how different underlying pathologies can result in overlapping phenomenologies.

FUTURE WORK A significant limitation of previous research is that researchers often distinguished between expectation confirmation versus disconfirmation as if they were binary concepts. In fact, disconfirming experiences can vary greatly in the extent to which

they contradict one’s expectations. Therefore, it may be important for future research to examine how healthy people versus people with MDD update their expectations depending on the magnitude of the PE. As illustrated in Supplemental Figure S3, we suggest an inverse U–shaped relationship of PE size and expectation update [for evidence of such boundary effects in pain, see also (119)]. To better understand the neurobiological underpinnings of the proposed cognitive dysfunctions in MDD, it may be important for future work to examine more broadly which PEs are pertinent in depression. To date, neurophysiological research in this area is largely based on reward PEs, revealing partly inconsistent findings. We believe it would be an important next step to expand the focus to modalities other than reward PEs. Given the inspiring theoretical accounts of active interoceptive inference in depression (16,17,23,53), we encourage researchers to empirically examine interoceptive PEs in MDD. One recent study investigating responses to glucose load has provided promising proof-of-concept evidence in this regard (36) [for abnormal interoceptive activation of the insula in MDD, see also (120)]. Relatedly, it may be important not only to examine motor aspects of behavior but also the autonomic actions that are encompassed by predictive processing. Furthermore, it may be worthwhile to gain a

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

7

Biological Psychiatry

A Predictive Processing Approach to Depression

more precise understanding of the self-reinforcing feedback loops in depression as proposed in this article. To do so, researchers may consider an intricate study on self-reinforcing feedback loops in pain perception (121). In Table 1, we present some specific hypotheses derived from our model that may be tested in future work.

5.

CONCLUSIONS

8.

This article aimed to connect disparate bodies of literature to provide a new framework for understanding distorted cognitive processes in depression. We proposed an explanatory model suggesting that patients with depression hold negative expectations about future experiences, which they subjectively feel confirmed owing to discounting disconfirmatory positive information. In computational terms, we suggest that the main candidate of pathology in MDD is too much precision afforded to prior predictions, which may lead to an attenuation of PEs and failure to update negative predictions. This has important implications for the treatment of MDD, pointing to the potential of modifying dysfunctional expectations and reducing engagement in cognitive immunization strategies.

6.

7.

9.

10.

11.

12. 13.

ACKNOWLEDGMENTS AND DISCLOSURES Although the present article received no funding, it was conducted in the context of the Research Training Group 2271 “Breaking Expectations: Expectation Maintenance vs. Change in the Context of Expectation Violations,” located at Philipps-University of Marburg. We thank all members of Research Training Group 2271, who inspired and supported the present work. Figure 2B was created with Motifolio drawing toolkits (www.motifolio.com). The authors report no biomedical financial interests or potential conflicts of interest.

14.

15.

16.

ARTICLE INFORMATION From the Department of Clinical Psychology and Psychotherapy (TK, JAG), Faculty of Psychology, University of Koblenz-Landau, Landau; Department of Clinical Psychology and Psychotherapy (TK, JAG, WR) and Department of Behavioral Neuroscience (RS), Faculty of Psychology, Philipps-University of Marburg, Marburg, Germany; Program in Placebo Studies (TK, LR), Harvard Medical School, Beth Israel Deaconess Medical Center, Boston; and Department of Brain and Cognitive Sciences (LR), Massachusetts Institute of Technology, Cambridge, Massachusetts. Address correspondence to Tobias Kube, Ph.D., Department of Clinical Psychology and Psychotherapy, Faculty of Psychology, University of Koblenz-Landau, Ostbahnstrasse 10, Landau 76829, Germany; E-mail: [email protected]. Received Apr 18, 2019; revised Jul 11, 2019; accepted Jul 18, 2019. Supplementary material cited in this article is available online at https:// doi.org/10.1016/j.biopsych.2019.07.017.

REFERENCES 1. 2. 3. 4.

8

Beck AT, Rush A, Shaw B, Emery G (1979): Cognitive Therapy of Depression. New York: Guilford Press. Beck AT (1963): Thinking and depression: I. Idiosyncratic content and cognitive distortions. Arch Gen Psychiatry 9:324–333. Beck AT (1964): Thinking and depression: II. Theory and therapy. Arch Gen Psychiatry 10:561–571. Rief W, Joormann J (2019): Revisiting the cognitive model of depression: The role of expectations. Clin Psychology in Europe 1: e32605.

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

17.

18. 19.

20. 21. 22.

23. 24.

25.

26. 27.

Olson JM, Roese NJ, Zanna MP (1996): Expectancies. In: Higgins ET, Kruglanski AW, editors. Social Psychology: A Handbook of Basic Principles. New York: Guilford Press. Laferton JA, Kube T, Salzmann S, Auer CJ, Shedden-Mora MC (2017): Patients’ expectations regarding medical treatment: A critical review of concepts and their assessment. Front Psychol 8:233. Gordon MS, Tonge B, Melvin GA (2011): Outcome of adolescent depression: 6 months after treatment. Aust N Z J Psychiatry 45: 232–239. Gopinath S, Katon WJ, Russo JE, Ludman EJ (2007): Clinical factors associated with relapse in primary care patients with chronic or recurrent depression. J Affect Disord 101:57–63. Ludman E, Katon W, Bush T, Rutter C, Lin E, Simon G, et al. (2003): Behavioural factors associated with symptom outcomes in a primary care-based depression prevention intervention trial. Psychol Med 33:1061–1070. Strunk DR, Lopez H, DeRubeis RJ (2006): Depressive symptoms are associated with unrealistic negative predictions of future life events. Behav Res Ther 44:861–882. Vilhauer JS, Young S, Kealoha C, Borrmann J, IsHak WW, Rapaport MH, et al. (2012): Treating major depression by creating positive expectations for the future: A pilot study for the effectiveness of future-directed therapy (FDT) on symptom severity and quality of life. CNS Neurosci Ther 18:102–109. Thimm J, Holte A, Brennen T, Wang C (2013): Hope and expectancies for future events in depression. Front Psychol 4:470–470. Czyz EK, Horwitz AG, King CA (2016): Self-rated expectations of suicidal behavior predict future suicide attempts among adolescent and young adult psychiatric emergency patients. Depress Anxiety 33:512–519. Horwitz AG, Berona J, Czyz EK, Yeguez CE, King CA (2017): Positive and negative expectations of hopelessness as longitudinal predictors of depression, suicidal ideation, and suicidal behavior in high-risk adolescents. Suicide Life Threat Behav 47:168–176. Kube T, Siebers VHA, Herzog P, Glombiewski JA, Doering BK, Rief W (2018): Integrating situation-specific dysfunctional expectations and dispositional optimism into the cognitive model of depression—a path-analytic approach. J Affect Disord 229: 199–205. Clark JE, Watson S, Friston KJ (2018): What is mood? A computational perspective. Psychol Med 48:2277–2284. Barrett LF, Quigley KS, Hamilton P (2016): An active inference theory of allostasis and interoception in depression. Philos Trans R Soc Lond B Biol Sci 371:20160011. O’Reilly JX, Jbabdi S, Behrens TE (2012): How can a Bayesian approach inform neuroscience? Eur J Neurosci 35:1169–1179. Kanai R, Komura Y, Shipp S, Friston K (2015): Cerebral hierarchies: Predictive processing, precision and the pulvinar. Philos Trans R Soc Lond B Biol Sci 370:20140169. Clark A (2013): Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behav Brain Sci 36:181–204. Huang Y, Rao RP (2011): Predictive coding. Wiley Interdiscip Rev Cogn Sci 2:580–593. Knill DC, Pouget A (2004): The Bayesian brain: The role of uncertainty in neural coding and computation. Trends Neurosci 27:712–719. Barrett LF, Simmons WK (2015): Interoceptive predictions in the brain. Nat Rev Neurosci 16:419–430. Kwisthout J, Bekkering H, van Rooij I (2017): To be precise, the details don’t matter: On predictive processing, precision, and level of detail of predictions. Brain Cogn 112:84–91. Friston K, Rigoli F, Ognibene D, Mathys C, Fitzgerald T, Pezzulo G (2015): Active inference and epistemic value. Cogn Neurosci 6: 187–214. Seth AK, Friston KJ (2016): Active interoceptive inference and the emotional brain. Philos Trans R Soc Lond B Biol Sci 371:20160007. Friston KJ, Stephan KE, Montague R, Dolan RJ (2014): Computational psychiatry: The brain as a phantastic organ. Lancet Psychiatry 1:148–158.

Biological Psychiatry

A Predictive Processing Approach to Depression

28. 29.

30.

31. 32. 33.

34.

35. 36.

37.

38. 39. 40. 41. 42.

43.

44. 45.

46.

47.

48.

49. 50. 51.

52.

53.

Hohwy J, Roepstorff A, Friston K (2008): Predictive coding explains binocular rivalry: An epistemological review. Cognition 108:687–701. Egner T, Monti JM, Summerfield C (2010): Expectation and surprise determine neural population responses in the ventral visual stream. J Neurosci 30:16601–16608. Summerfield C, Egner T, Greene M, Koechlin E, Mangels J, Hirsch J (2006): Predictive codes for forthcoming perception in the frontal cortex. Science 314:1311–1314. Sterzer P, Frith C, Petrovic P (2008): Believing is seeing: Expectations alter visual awareness. Curr Biol 18:R697–R698. Hosoya T, Baccus S, Meister M (2005): Dynamic predictive coding by the retina. Nature 436:71–77. Moran RJ, Campo P, Symmonds M, Stephan KE, Dolan RJ, Friston KJ (2013): Free energy, precision and learning: The role of cholinergic neuromodulation. J Neurosci 33:8227–8236. Todorovic A, van Ede F, Maris E, de Lange FP (2011): Prior expectation mediates neural adaptation to repeated sounds in the auditory cortex: An MEG study. J Neurosci 31:9118–9123. Seth AK, Suzuki K, Critchley HD (2012): An interoceptive predictive coding model of conscious presence. Front Psychol 2:395. Young HA, Gaylor CM, de Kerckhove D, Watkins H, Benton D (2019): Interoceptive accuracy moderates the response to a glucose load: A test of the predictive coding framework. Proc Biol Soc 286:20190244. Pellicano E, Burr D (2012): When the world becomes ‘too real’: A Bayesian explanation of autistic perception. Trends Cogn Sci 16:504–510. Wiech K (2016): Deconstructing the sensation of pain: The influence of cognitive processes on pain perception. Science 354:584–587. Buchel C, Geuter S, Sprenger C, Eippert F (2014): Placebo analgesia: A predictive coding perspective. Neuron 81:1223–1239. Kaptchuk TJ (2018): Open-label placebo: Reflections on a research agenda. Perspect Biol Med 61:311–334. Ongaro G, Kaptchuk TJ (2019): Symptom perception, placebo effects, and the Bayesian brain. Pain 160:1–4. Van den Bergh O, Witthöft M, Petersen S, Brown RJ (2017): Symptoms and the body: Taking the inferential leap. Neurosci Biobehav Rev 74:185–203. Henningsen P, Gündel H, Kop WJ, Löwe B, Martin A, Rief W, et al. (2018): Persistent physical symptoms as perceptual dysregulation: A neuropsychobehavioral model and its clinical implications. Psychosom Med 80:422–431. Edwards MJ, Adams RA, Brown H, Parees I, Friston KJ (2012): A Bayesian account of ‘hysteria’. Brain 135:3495–3512. Bastos AM, Usrey WM, Adams RA, Mangun GR, Fries P, Friston KJ (2012): Canonical microcircuits for predictive coding. Neuron 76:695– 711. Garrison J, Erdeniz B, Done J (2013): Prediction error in reinforcement learning: A meta-analysis of neuroimaging studies. Neurosci Biobehav Rev 37:1297–1310. O’Doherty JP, Dayan P, Friston K, Critchley H, Dolan RJ (2003): Temporal difference models and reward-related learning in the human brain. Neuron 38:329–337. D’Astolfo L, Rief W (2017): Learning about expectation violation from prediction error paradigms—a meta-analysis on brain processes following a prediction error. Front Psychol 8:1253. Hohwy J (2013): The Predictive Mind. Oxford: Oxford University Press. Wilson TD, Gilbert DT (2003): Affective forecasting. Adv Exp Soc Psychol 35:345–411. Hoerger M, Quirk SW, Chapman BP, Duberstein PR (2012): Affective forecasting and self-rated symptoms of depression, anxiety, and hypomania: Evidence for a dysphoric forecasting bias. Cogn Emot 26:1098–1106. Marroquín B, Nolen-Hoeksema S (2015): Event prediction and affective forecasting in depressive cognition: Using emotion as information about the future. J Soc Clin Psychol 34:117. Eldar E, Rutledge RB, Dolan RJ, Niv Y (2016): Mood as representation of momentum. Trends Cogn Sci 1:15–24.

54.

55.

56.

57.

58. 59.

60.

61.

62. 63.

64. 65.

66.

67.

68.

69.

70.

71.

72. 73.

74.

75.

Adams RA, Huys QJ, Roiser JP (2016): Computational psychiatry: Towards a mathematically informed understanding of mental illness. J Neurol Neurosurg Psychiatry 87:53–63. Everaert J, Podina IR, Koster EHW (2017): A comprehensive metaanalysis of interpretation biases in depression. Clin Psychol Rev 58:33–48. Everaert J, Bronstein MV, Cannon TD, Joormann J (2018): Looking through tinted glasses: Depression and social anxiety are related to both interpretation biases and inflexible negative interpretations. Clin Psychol Sci 6:517–528. Liknaitzky P, Smillie LD, Allen NB (2017): Out-of-the-blue: Depressive symptoms are associated with deficits in processing inferential expectancy-violations using a novel cognitive rigidity task. Cogn Ther Res 41:757–776. Kashdan TB, Rottenberg J (2010): Psychological flexibility as a fundamental aspect of health. Clin Psychol Rev 30:865–878. Stange JP, Alloy LB, Fresco DM (2017): Inflexibility as a vulnerability to depression: A systematic qualitative review. Clinical Psychol Sci Pract 24:245–276. Stange JP, Connolly SL, Burke TA, Hamilton JL, Hamlat EJ, Abramson LY, et al. (2016): Inflexible cognition predicts first onset of major depressive episodes in adolescence. Depress Anxiety 33:1005–1012. Miranda R, Gallagher M, Bauchner B, Vaysman R, Marroquín B (2012): Cognitive inflexibility as a prospective predictor of suicidal ideation among young adults with a suicide attempt history. Depress Anxiety 29:180–186. Sharot T (2011): The optimism bias. Curr Biol 21:R941–R945. Korn CW, Prehn K, Park SQ, Walter H, Heekeren HR (2012): Positively biased processing of self-relevant social feedback. J Neurosci 32:16832–16844. Sharot T, Korn CW, Dolan RJ (2011): How unrealistic optimism is maintained in the face of reality. Nat Neurosci 14:1475–1482. Garrett N, Sharot T (2017): Optimistic update bias holds firm: Three tests of robustness following Shah et al. Conscious Cogn 50:12–22. Korn CW, Sharot T, Walter H, Heekeren HR, Dolan RJ (2014): Depression is related to an absence of optimistically biased belief updating about future life events. Psychol Med 44:579–592. Garrett N, Sharot T, Faulkner P, Korn CW, Roiser JP, Dolan RJ (2014): Losing the rose tinted glasses: Neural substrates of unbiased belief updating in depression. Front Hum Neurosci 8:639. Kube T, Rief W, Gollwitzer M, Gärtner T, Glombiewski JA (2019): Why dysfunctional expectations in depression persist—results from two experimental studies investigating cognitive immunization. Psychol Med 49:1532–1544. Rief W, Glombiewski JA, Gollwitzer M, Schubo A, Schwarting R, Thorwart A (2015): Expectancies as core features of mental disorders. Curr Opin Psychiatry 28:378–385. Kube T, Rief W, Glombiewski JA (2017): On the maintenance of expectations in major depression—investigating a neglected phenomenon. Front Psychol 8:9. Kube T, Glombiewski JA, Gall J, Toussaint L, Gärtner T, Rief W (2019): How to modify persisting negative expectations in major depression? An experimental study comparing three strategies to inhibit cognitive immunization against novel positive experiences. J Affect Disord 250C:231–240. Eshel N, Roiser JP (2010): Reward and punishment processing in depression. Biol Psychiatry 68:118–124. Whitton AE, Treadway MT, Pizzagalli DA (2015): Reward processing dysfunction in major depression, bipolar disorder and schizophrenia. Curr Opin Psychiatry 28:7–12. Sescousse G, Caldú X, Segura B, Dreher J-C (2013): Processing of primary and secondary rewards: A quantitative meta-analysis and review of human functional neuroimaging studies. Neurosci Biobehav Rev 37:681–696. Wächter T, Lungu OV, Liu T, Willingham DT, Ashe J (2009): Differential effect of reward and punishment on procedural learning. J Neurosci 29:436–443.

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

9

Biological Psychiatry

A Predictive Processing Approach to Depression

76.

77. 78.

79.

80.

81. 82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

10

Smoski MJ, Felder J, Bizzell J, Green SR, Ernst M, Lynch TR, et al. (2009): fMRI of alterations in reward selection, anticipation, and feedback in major depressive disorder. J Affect Disord 118:69–78. Steele J, Kumar P, Ebmeier KP (2007): Blunted response to feedback information in depressive illness. Brain 130:2367–2374. Stoy M, Schlagenhauf F, Sterzer P, Bermpohl F, Hägele C, Suchotzki K, et al. (2012): Hyporeactivity of ventral striatum towards incentive stimuli in unmedicated depressed patients normalizes after treatment with escitalopram. J Psychopharmacol 26:677–688. Pizzagalli DA, Holmes AJ, Dillon DG, Goetz EL, Birk JL, Bogdan R, et al. (2009): Reduced caudate and nucleus accumbens response to rewards in unmedicated individuals with major depressive disorder. Am J Psychiatry 166:702–710. Knutson B, Bhanji JP, Cooney RE, Atlas LY, Gotlib IH (2008): Neural responses to monetary incentives in major depression. Biol Psychiatry 63:686–692. Henriques JB, Davidson RJ (2000): Decreased responsiveness to reward in depression. Cogn Emot 14:711–724. Vrieze E, Pizzagalli DA, Demyttenaere K, Hompes T, Sienaert P, de Boer P, et al. (2013): Reduced reward learning predicts outcome in major depressive disorder. Biol Psychiatry 73:639–645. Pizzagalli DA, Jahn AL, O’Shea JP (2005): Toward an objective characterization of an anhedonic phenotype: A signal-detection approach. Biol Psychiatry 57:319–327. Tavares JVT, Clark L, Furey ML, Williams GB, Sahakian BJ, Drevets WC (2008): Neural basis of abnormal response to negative feedback in unmedicated mood disorders. Neuroimage 42:1118– 1126. Rothkirch M, Tonn J, Köhler S, Sterzer P (2017): Neural mechanisms of reinforcement learning in unmedicated patients with major depressive disorder. Brain 140:1147–1157. Dombrovski AY, Szanto K, Clark L, Reynolds CF, Siegle GJ (2013): Reward signals, attempted suicide, and impulsivity in late-life depression. JAMA Psychiatry 70:1020–1030. Bakic J, Pourtois G, Jepma M, Duprat R, De Raedt R, Baeken C (2017): Spared internal but impaired external reward prediction error signals in major depressive disorder during reinforcement learning. Depress Anxiety 34:89–96. Gradin VB, Kumar P, Waiter G, Ahearn T, Stickle C, Milders M, et al. (2011): Expected value and prediction error abnormalities in depression and schizophrenia. Brain 134:1751–1764. Kumar P, Goer F, Murray L, Dillon DG, Beltzer ML, Cohen AL, et al. (2018): Impaired reward prediction error encoding and striatalmidbrain connectivity in depression. Neuropsychopharmacology 43:1581–1588. Robinson OJ, Cools R, Carlisi CO, Sahakian BJ, Drevets WC (2012): Ventral striatum response during reward and punishment reversal learning in unmedicated major depressive disorder. Am J Psychiatry 169:152–159. Segarra N, Metastasio A, Ziauddeen H, Spencer J, Reinders NR, Dudas RB, et al. (2016): Abnormal frontostriatal activity during unexpected reward receipt in depression and schizophrenia: Relationship to anhedonia. Neuropsychopharmacology 41:2001–2010. Kumar P, Waiter G, Ahearn T, Milders M, Reid I, Steele JD (2008): Abnormal temporal difference reward-learning signals in major depression. Brain 131:2084–2093. Chase HW, Nusslock R, Almeida JR, Forbes EE, LaBarbara EJ, Phillips ML (2013): Dissociable patterns of abnormal frontal cortical activation during anticipation of an uncertain reward or loss in bipolar versus major depression. Bipolar Disord 15:839–854. Rutledge RB, Moutoussis M, Smittenaar P, Zeidman P, Taylor T, Hrynkiewicz L, et al. (2017): Association of neural and emotional impacts of reward prediction errors with major depression. JAMA Psychiatry 74:790–797. Greenberg T, Chase HW, Almeida JR, Stiffler R, Zevallos CR, Aslam HA, et al. (2015): Moderation of the relationship between reward expectancy and prediction error-related ventral striatal

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

96.

97. 98.

99.

100.

101.

102. 103.

104.

105.

106.

107.

108.

109.

110.

111. 112.

113.

114. 115.

116.

117.

reactivity by anhedonia in unmedicated major depressive disorder: Findings from the EMBARC study. Am J Psychiatry 172:881–891. Ubl B, Kuehner C, Kirsch P, Ruttorf M, Diener C, Flor H (2015): Altered neural reward and loss processing and prediction error signalling in depression. Soc Cogn Affect Neurosci 10:1102– 1112. Sharot T, Riccardi AM, Raio CM, Phelps EA (2007): Neural mechanisms mediating optimism bias. Nature 450:102–105. Brookings JB, Serratelli AJ (2006): Positive illusions: Positively correlated with subjective well-being, negatively correlated with a measure of personal growth. Psychol Rep 98:407–413. Taylor SE, Brown JD (1988): Illusion and well-being: A social psychological perspective on mental health. Psychol Bull 103:193–210. Taylor SE, Kemeny ME, Reed GM, Bower JE, Gruenewald TL (2000): Psychological resources, positive illusions, and health. Am Psychol 55:99–109. Alloy LB, Abramson LY (1979): Judgment of contingency in depressed and nondepressed students: Sadder but wiser? J Exp Psychol Gen 108:441–485. Carson RC, Hollon SD, Shelton RC (2010): Depressive realism and clinical depression. Behav Res Ther 48:257–265. Dunning D, Story AL (1991): Depression, realism, and the overconfidence effect: Are the sadder wiser when predicting future actions and events? J Pers Soc Psychol 61:521–532. Badcock PB, Davey CG, Whittle S, Allen NB, Friston KJ (2017): The depressed brain: An evolutionary systems theory. Trends Cogn Sci 21:182–194. Andersen SM (1990): The inevitability of future suffering: The role of depressive predictive certainty in depression. Soc Cogn 8:203– 228. Andersen SM, Spielman LA, Bargh JA (1992): Future-event schemas and certainty about the future: Automaticity in depressives’ futureevent predictions. J Pers Soc Psychol 63:711–723. Miranda R, Mennin DS (2007): Depression, generalized anxiety disorder, and certainty in pessimistic predictions about the future. Cogn Ther Res 31:71–82. Miranda R, Fontes M, Marroquín B (2008): Cognitive contentspecificity in future expectancies: Role of hopelessness and intolerance of uncertainty in depression and GAD symptoms. Behav Res Ther 46:1151–1159. Miranda R, Weierich M, Khait V, Jurska J, Andersen SM (2017): Induced optimism as mental rehearsal to decrease depressive predictive certainty. Behav Res Ther 90:1–8. Eizenman M, Lawrence HY, Grupp L, Eizenman E, Ellenbogen M, Gemar M, et al. (2003): A naturalistic visual scanning approach to assess selective attention in major depressive disorder. Psychiatry Res 118:117–128. Joormann J, Gotlib IH (2007): Selective attention to emotional faces following recovery from depression. J Abnorm Psychol 116:80–85. Lefebvre G, Lebreton M, Meyniel F, Bourgeois-Gironde S, Palminteri S (2017): Behavioural and neural characterization of optimistic reinforcement learning. Nature Human Behaviour 1: 0067. Pfeifer JH, Allen NB (2012): Arrested development? Reconsidering dual-systems models of brain function in adolescence and disorders. Trends Cogn Sci 6:322–329. Price JL, Drevets WC (2012): Neural circuits underlying the pathophysiology of mood disorders. Trends Cogn Sci 16:61–71. Schenk LA, Sprenger C, Onat S, Colloca L, Büchel C (2017): Suppression of striatal prediction errors by the prefrontal cortex in placebo hypoalgesia. J Neurosci 37:9715–9723. Kube T, Glombiewski JA, Rief W (in press): Expectation-focused psychotherapeutic interventions for people with depressive symptoms. Verhaltenstherapie. Sterzer P, Adams RA, Fletcher P, Frith C, Lawrie SM, Muckli L, et al. (2018): The predictive coding account of psychosis. Biol Psychiatry 84:634–643.

Biological Psychiatry

A Predictive Processing Approach to Depression

118. 119.

120.

121.

Wilkinson S, Dodgson G, Meares K (2017): Predictive processing and the varieties of psychological trauma. Front Psychol 8:1840. Hird EJ, Charalambous C, El-Deredy W, Jones AK, Talmi D (2019): Boundary effects of expectation in human pain perception. Sci Rep 9:9443. Avery JA, Drevets WC, Moseman SE, Bodurka J, Barcalow JC, Simmons WK (2014): Major depressive disorder is associated with abnormal interoceptive activity and functional connectivity in the insula. Biol Psychiatry 76:258–266. Jepma M, Koban L, van Doorn J, Jones M, Wager T (2018): Behavioural and neural evidence for self-reinforcing expectancy effects on pain. Nature Human Behaviour 2:838–855.

122.

123.

124.

Kube T, Glombiewski JA, Rief W (2018): Situational expectations mediate the effect of global beliefs on depressive symptoms: A 1-year prospective study using a student sample. Curr Opin Psychiatry 31:409–416. Aylward J, Valton V, Ahn WY, Bond RL, Dayan P, Roiser JP, Robinson OJ (2019): Altered learning under uncertainty in unmedicated mood and anxiety disorders [published online ahead of print Jun 17]. Nature Human Behaviour. Moutoussis M, Rutledge RB, Prabhu G, Hrynkiewicz L, Lam J, Ousdal OT, et al. (2018): Neural activity and fundamental learning, motivated by monetary loss and reward, are intact in mild to moderate major depressive disorder. PloS One 13:e0201451.

Biological Psychiatry - -, 2019; -:-–- www.sobp.org/journal

11