Unstable Belief Formation and Slowed Decision-making: Evidence That the Jumping-to-Conclusions Bias in Schizophrenia Is Not Linked to Impulsive Decision-making

Abstract

Background

Jumping-to-conclusions (JTC) is a prominent reasoning bias in schizophrenia (SCZ). While it has been linked to not only psychopathological abnormalities (delusions and impulsive decision-making) but also unstable belief formation, its origin remains unclear. We here directly test to which extend JTC is associated with delusional ideation, impulsive decision-making, and unstable belief formation.

Methods

In total, 45 SCZ patients were compared with matched samples of 45 patients with major depressive disorder (MDD) and 45 healthy controls (HC) as delusions and JTC also occur in other mental disorders and the general population. Participants performed a probabilistic beads task. To test the association of JTC with measures of delusions (Positive and Negative Syndrome Scale [PANSS]_positive, PANSS_{positive-factor}, and Peter Delusions Inventory [PDI]), Bayesian linear regressions were computed. For the link between JTC and impulsive decision-making and unstable beliefs, we conducted between-group comparisons of “draws to decision” (DTD), “decision times” (DT), and “disconfirmatory evidence scores” (DES).

Results

Bayesian regression obtained no robust relationship between PDI and DTD (all |R²_adj| ≤ .057, all P ≥ .022, all Bayes Factors [BF₀₁] ≤ 0.046; α _adj = .00833). Compared with MDD and HC, patients with SCZ needed more time to decide (significantly higher DT in ambiguous trials: all P ≤ .005, r² ≥ .216; numerically higher DT in other trials). Further, SCZ had unstable beliefs about the correct source jar whenever unexpected changes in bead sequences (disconfirmatory evidence) occurred (compared with MDD: all P ≤ .004 and all r² ≥ .232; compared with HC: numerically higher DES). No significant correlation was observed between DT and DTD (all P ≥ .050).

Conclusions

Our findings point toward a relationship of JTC with unstable belief formation and do not support the assumption that JTC is associated with impulsive decision-making.

Introduction

People with schizophrenia (SCZ) often gather less information before arriving at a conclusion.^1–6 This jumping-to-conclusions (JTC) bias is most commonly assessed with the beads task,⁷ which requires participants to sample random sequences of colored beads and to infer from which of 2 potential source jars the beads were taken. Essentially, the prediction of the beads task is that patients with SCZ will view fewer beads before deciding and commonly do so with greater confidence than controls. Two variants of the beads task have commonly been used to assess JTC in SCZ. In the so-called draws-to-decision (DTD) version,^5,7,8 participants are allowed to sample any desired number of up to 20 beads and following every new bead either continue sampling or decide from which of the 2 source jars the beads were drawn—as soon as they feel sufficiently certain to make this decision. In the graded-estimates (GE) version,^9,10 participants are shown a fixed number of beads and asked on each new bead view to rate the probability (on a Likert scale ranging from 0% to 100%) that the currently viewed bead sequence is drawn from one of the two source jars.

Patients with SCZ consistently show JTC^1,4,11 and overestimate the probability that the beads are drawn from a specific jar.^9,10,12–14 However, the reasons for this bias remain unclear.^{1,4,11,15–18} On the one hand, JTC has been associated with delusions^{1–4,11,19–23} and impulsive decision-making.^1,2,22,24 On the other hand, a growing number of studies have shown that patients with SCZ display JTC even without the presence of delusions and other authors have attributed JTC to impaired probabilistic reasoning.^{10,11,16,18,25–27} Specifically, patients seem to overweight unexpected recent bead occurrences—leading to unstable beliefs about the correct source jar.^{9,12,15,17,28} Resolving these disparate views is of relevance as the JTC bias has become important for evaluating behavioral treatment interventions aimed at improving impulsive decision-making and adhering to inordinately fixed beliefs or aberrant reasoning.^29,30

Against this background, we here directly assess the relationship of JTC with delusions, impulsive decision-making, and belief formation. For this purpose, we applied a novel probabilistic beads task that allows systematic assessments of data gathering and explicit probability estimates (PEs) from varying sensory and cognitive information (via different trial types) and in different probabilistic contexts (via different source jar distributions). We hypothesized that (1) JTC is more pronounced in SCZ patients compared with matched clinical and healthy controls (HC). Furthermore, we reasoned that (2) JTC is not strongly associated with delusions or (3) impulsive decision-making. Finally, we hypothesized that (4) JTC would be associated with specific alterations of probabilistic reasoning, namely unstable belief formation.

Methods

Study Population

Inpatients currently treated at the Department of Psychiatry of the University Hospital Munich (Germany) diagnosed with SCZ (n = 45) or major depression disorder (MDD, n = 45) participated in this study. Diagnoses were based on nonstructured clinical interviews following international classification of diseases version 10 (ICD-10) criteria and were confirmed by 2 independent clinical interviewers. Additionally, n = 45 HC participants were tested. Both groups were matched to SCZ participants (see supplement “Methods”) with respect to age, gender, and intelligence quotient (IQ) as these factors have been identified to impact beads task performance.^20,31Furthermore, clinical samples were matched according to disease duration. We included a clinical control group of MDD patients because dysfunctions in decision-making are considered a central psychopathological feature of depression and to remain comparable with previous studies on MDD patients.^1,4,31,32 Our sample size met the requirements of a corresponding power analysis (G * Power),³³ assuming a small expected effect size of f = 0.25 an alpha error probability of α = .05, a power of (1 − β err prob) = 0.9, and a correlation among repeated measures of 0.5 (see supplement “Methods”). Inclusion and exclusion criteria and procedures for obtaining clinical and neuropsychological characteristics are detailed in supplement “Methods.” All participants provided informed consent in accordance with the standards of the Declaration of Helsinki, and the study protocol was approved by the local Ethics Committee. Monetary compensation amounted to 20€ per testing hour. In the case of SCZ and HC, available data sets³⁴ were extended with newly recruited participants from n = 32 to 45 each. Symptom severity and global functioning were assessed via Clinical Global Impression (CGI)³⁵ and Global Assessment of Functioning (GAF).³⁶ For SCZ, we further conducted the Positive and Negative Syndrome Scale (PANSS)³⁷ and computed Wallworks’ PANSS_{positive-factor} (consisting of items P1, P3, P5, and G9).³⁸ In the case of MDD, we surveyed Beck Depression Inventory (BDI)³⁹ and Hamilton Major Depression Rating Scale (H-MDRS).⁴⁰ All participants further underwent Peter Delusions Inventory (PDI)⁴¹ and neuropsychological assessments of attention,⁴² executive functioning,⁴³ and premorbid intelligence.⁴⁴

Behavioral Tasks

Upon inclusion, all participants performed the same set of behavioral reach decision tasks that were computerized using Matlab (Mathworks, version R2017b) and the Cogent Graphics toolbox for Windows⁴⁵: first, the Choice Reaction-time Task (CRT), followed by our computational instantiation of the beads task^7,10,46 (see figure 1).

Choice Reaction-time Task

The CRT was used to obtain individual reaction times (RT_CRT), which are considered⁴⁷ aggregates of the delays attributable to individual differences in attention span and sensory processing, muscle response initiation, and cognitive slowing. As in previous work,⁴⁷ we corrected the response times in the beads task (RT_Beadstask) for these aggregates of delays on an individual subject basis (correction = RT_Beadstask − RT_CRT) to estimate individual “decision times” (DT), as a proxy for the amount of time participants spent to cognitively process bead sequences and make a decision.

Beads Task

The complete task design is detailed in the supplement “Methods.” For the beads task, we applied 2 DTD conditions (2 × 18 trials), which differed regarding the prior information that participants were given about the distributions in each jar (P_80/20: 80:20%, blue:green; P_60/40: 60:40%, violet:orange). Here, participants could stop viewing additional beads whenever they felt sufficiently confident to make a decision about a source jar. Additionally, participants undertook 12 trials of the GE condition, which also used a P_80/20 distribution of beads. In the GE, participants could not terminate a trial after viewing a bead but instead were required to view 10 beads successively and to report PEs on a Likert scale (ranging from 0% to 100% probability in 2 directions, with both source jars presented at the extremes). These PEs reflected how likely participants estimated a bead sequence originated from a source jar that they selected (participants could decide for either of the 2 potential source jars, but not for both). As in previous work,^12,34 disconfirmatory evidence scores (DES) were further assessed that quantified the cumulative amount by which participants changed their PEs following changes in bead color after viewing ≥2 preceding beads of the same color. Participants were instructed to view and rate random bead sequences; however, prespecified trial sequences were applied^34,46: (I) easy trials with a likelihood of 80% for one predominant bead color, (II) difficult trials with a likelihood of 60%, and (III) ambiguous trials with a likelihood of 50%. All sequences were counterbalanced and the order of trials was randomized to control for sequence effects. In all 3 parts of the task, no feedback was provided about the correctness of responses. Bead sequences and illustrations of the source jars were kept displayed on the computer screen throughout the whole task to reduce working memory load, which has been demonstrated to bias beads task findings.^16,19,48

Statistical Analyses

All statistical analyses were conducted using IBM SPSS 27. Level of significance was set to alpha = .05 for group-level comparisons of baseline characteristics using one-way analysis of variance (ANOVA) and Chi-square tests. We defined DTD, JTC frequencies, DT, PE, and DES as main outcome variables. As the assumption of normal distribution was violated for 52 of the 53 metric main outcome variables (Shapiro-Wilk tests: fifth PE rating in easy trials: W₍₁₃₅₎ = 0.982, P = .074; all other W₍₁₃₅₎ ≤ 0.980, all P ≤ .048) and several methods of transformation did not achieve normal distribution, between-group comparisons (SCZ/MDD/HC) were computed using nonparametric Kruskal-Wallis tests (KWT) and—where appropriate—corrected Mann-Whitney U (MWU) tests, instead of repeated measures ANOVA assumed for our power analysis (see supplement “Methods”). For hypothesis (1), JTC frequencies were compared with Freeman-Halton tests⁴⁹ (in analogy of Chi-Square tests as contingency tables measured 2 × 3 and some expected cell counts were <5). Data gathering was investigated comparing DTD between groups using KWT and MWU. Hypothesis (2) was assessed with Bayesian linear regression between DTD and measures of delusions (PDI, PANSS_positive, and PANSS_{positive-factor}³⁸), and age, gender, IQ, and disease severity included as co-factors. For hypothesis (3), we compared DT between groups using KWT and MWU. To assess hypothesis (4), KWT and MWU were computed comparing PE and DES between groups. Effect sizes were estimated using $Φc=(√χ2/√n)$ for Freeman-Halton tests, $η2=(H−k+1)/(n−k)$ for KWT, r² = (Z²/n) for MWU, and R² for Bayesian regression.⁵⁰To correct for the number of tests per main outcome analysis, we adjusted the significance level to α _JTC = .00833, α _DTD = .00333, α_{Bayesian-Regression} = .00833, α _DT = .00833, α _PE = .00125, and α _DES = .01667 (see supplement “Methods,” section 1).

Results

Baseline Characteristics Across Groups

Participants were well matched regarding group distributions of age, gender, IQ, and attention load capacity (see supplement “Results,” section 1 and table 1). SCZ and MDD displayed gradually slower response times in the CRT test compared with HC. Clinical characteristics further categorized symptom burden in both patient groups as moderate to severe^51,52 and psychosocial level of function as higher in MDD compared with SCZ (see table 1). Post hoc analyses of individual responses on a trial-by-trial basis obtained no indicators of noncomprehension or reduced motivation (eg, uniform responses). All participants were able to adequately differentiate between trial types and conditions (see supplement “Results,” section 1 and tables 1–6).

JTC Frequencies and Data Gathering

To assess our hypothesis (1) whether SCZ patients showed increased rates of JTC, we first classified JTC as present according to established criteria,^1,3,4,53 that is, if participants decided after <3 bead views in the DTD conditions (P_80/20/P_60/40). For the P_80/20 condition, we observed JTC frequencies of 44.4% (n = 20) in easy trials, 40.0% (n = 18) in difficult trials, and 44.4% (n = 20) in ambiguous trials for SCZ patients. By comparison, JTC was less frequently displayed by MDD (easy trials: 8.9%, n = 4; difficult trials: 6.7%, n = 3; ambiguous trials: 4.4%, n = 2) and by HC (easy trials: 13.3%, n = 6; difficult trials: 13.3%, n = 6; ambiguous trials: 2.2%, n = 1). Freeman-Halton tests comparing these distributions for each trial type confirmed the significance of these observed differences (overall analysis: all X²₍₂₎ ≥ 16.4, all P ≤ .001, ϕ _c ≥ 0.360) and obtained higher JTC rates in SCZ compared with MDD (all X²₍₁₎ ≥ 13.98, all P < .001, all ϕ _c ≥ 0.394) and compared with HC (all X²₍₁₎ ≥ 8.18, all P ≤ .008, all ϕ _c ≥ 0.302), whereas MDD and HC showed no differences (all X²₍₁₎ ≤ 1.11, all P ≥ .242). In contrast, SCZ displayed JTC less frequently in the P_60/40 condition (easy trials: 15.6%, n = 7; difficult trials: 11.1%, n = 5; ambiguous trials: 6.7%, n=3), and we observed no significant group differences (all P ≥ .507; see supplement “Results,” tables 7 and 8).

We next compared DTD in each trial type between groups since the definition of JTC (at <3 bead views) does not consider the composition of the first 2 bead views in a sequence. This can largely impact the PEs, as the first 2 beads can be of the same color (and point into the direction of one specific source jar) or not. For the P_80/20 condition, respective KWT obtained significant group differences in easy (H₍₂₎ = 18.43, P < .001, η ²_H = 0.391), difficult (H₍₂₎ = 15.87, P < .001, η ²_H = 0.330), and ambiguous trials (H₍₂₎ = 19.04, P < .001, η ²_H = 0.406). Subsequent Sidak-corrected MWU explained the observed effects through significantly fewer DTD of SCZ patients compared with MDD patients (all U ≤ 541.5, all P < .001, all r² ≥ .321) and numerical differences compared with HC participants (all U ≤ 685.0, all P ≤ .024, all r² ≥ .155, α _adj = .0033), while no group differences were observed between MDD and HC (all U ≥ 796.5, all P ≥ .240). By contrast, no significant differences were observed for the P_60/40 condition (all H₍₂₎≤5.14, all P ≥ .077) (see figure 2A and supplement “Results,” tables 5 and 6).

Delusions and JTC

Regarding our second hypothesis (2) whether JTC is linked to delusions, we computed Bayesian linear regression between PDI scores and DTD and included age, gender, IQ, and disease severity as co-factors into our model. Since the PDI is considered transdiagnostic to assess delusional thinking across a continuum,^41,54 we first conducted a cross-sectional analysis on all participants (SCZ, MDD, and HC combined). This analysis obtained no significant correlations for all trial types of the P_80/20 condition (all |R²_adj| ≤ .057, all F_(4,128) ≤ 2.98, all P ≥ .022, all BF₀₁ ≤ 0.046; α _adj = .00833) and for the P_60/40 condition (all |R²_adj| ≤ .007, all F_(4,128) ≤ 1.23, all P ≥ .302, all BF₀₁ ≤ 0.002). Next, the same analysis was repeated separately for SCZ patients (only with PANSS_positive instead of PDI), which showed no significant correlations (all |R²_adj| ≤ 0.057, all F_(4,40) ≤ 0.609, all P ≥ .658, all BF₀₁ ≤ 0.004). Similarly, no significant correlations were observed between Wallworks’ PANSS_{positive-factor}³⁸ and DTD (all |R²_adj| ≤ .057, all F_(4,40) ≤ 0.602, all P ≥ .663, all BF₀₁ ≤ 0.004). However, our sample size was only sufficiently powered to detect PDI differences across groups, as secondary analyses for PDI differences within groups using G*Power³³ obtained larger necessary sample sizes (see supplement “Results,” section 2).

Decision Times and JTC

To investigate our hypothesis (3) whether JTC is associated with impulsive decision-making, we compared individual DT derived from reaction times of the CRT offset against beads task reaction times (as detailed in Methods). Respective KWT obtained significant group differences for ambiguous trials of both conditions (P_80/20 condition: H₍₂₎ = 19.7, P < .001, η ²_H = 0.421; P_60/40 condition: H₍₂₎ = 11.5, P = .003, η ²_H = 0.227), while only numerically slower DT were observed for easy and difficult trials in both conditions (all H₍₂₎ ≥ 6.87, all P ≤ .032, all η ²_H ≤ 0.148, α _adj = .00833). To further explore the direction of these differences, we further computed (in part exploratory) Sidak-corrected MWU for the P_80/20 condition. This analysis obtained no differences between MDD and HC (all U ≥ 813.0, all P ≥ .317), while slower DT were observed in ambiguous trials comparing SCZ patients with MDD (U = 488.0, P < .001, r² = .398) and with HC (U = 626.0, P = .005, r² = .216) (see figure 2B). For the P_60/40 condition, we observed significantly slower DT in ambiguous trials of SCZ compared with MDD (U = 606.0, P = .003, r² = .239). Interestingly, SCZ patients also showed a pattern of numerically slower DT in easy and difficult trials of both conditions (P_80/20 and P_60/40) compared with MDD and HC (supplement “Results,” section 3, tables 9 and 10).

Probability Ratings and Unstable Beliefs

In the case of the GE version of the beads task, group comparisons using KWT and Sidak-corrected MWU obtained a pattern of significantly and numerically higher PE of SCZ patients in easy and difficult trials compared with MDD and HC, while no such differences were observed between MDD and HC (see figure 3A and supplement “Results,” section 4 and tables 11–13). Further exploratory analyses obtained significant Spearman correlations between DES and DTD in easy (all |r_S| ≥ .303, all P < .001) and difficult trials (all |r_S| ≥ .242, all P ≤ .005; α _adj = .006), but not for ambiguous trials (all P ≥ .014).

Further, with respect to our hypothesis (4) whether JTC is associated with unstable belief formation, we computed DES, which quantify changes of probability ratings following switches of bead color compared with ≥2 preceding beads.¹² This approach was based on previous studies observing that patients with SCZ display unstable beliefs following unexpected changes in bead sequences.^{12,34,55–59} Respective KWT between SCZ, MDD, and HC obtained significant group differences for easy (H₍₂₎ = 17.87, P < .001, η ²_H = 0.237), difficult (H₍₂₎ = 22.19, P < .001, η ²_H = 0.233), and ambiguous trials (H₍₂₎ = 11.81, P = .003, η ²_H = 0.142). Subsequent Sidak-corrected MWU explained the observed effects through significantly higher DES of SCZ patients compared with MDD patients in easy and difficult trials (all U ≤ 612.0, all P ≤ .004, all r² ≥ .232; ambiguous trials: U = 746.0, P = .092), while only numerically higher DES were observed in difficult and ambiguous trails of SCZ patients compared with HC (all U ≤ 678.5, all P ≥ .020; easy trials: U = 748.0, P = .097). Furthermore, no group differences were observed between MDD and HC (all U ≥ 874.0, all P ≥ .778) (see figure 3B). Finally, an exploratory analysis did not obtain a significant correlation between DES and DT (all |r_S| ≤ .169, all P ≥ .050, all BF₀₁ ≥ 1.736).

Discussion

As main findings, we here observed significantly higher rates of JTC and reduced DTD in patients with SCZ compared with MDD and HC and that SCZ patients displayed unstable belief formation following unexpected changes in bead sequences. Our results also show for the first time that SCZ patients needed more time to view less beads before a decision. The lower DTD and correspondingly increased rates of JTC in SCZ patients compared with MDD and HC (see figure 2A) side with recent meta-analyses.^1–4 By contrast, we did not find support for the hypothesis that these findings are also associated with delusions. This is in line with previous reports^26,27 that challenge the view that JTC may reflect delusion-specific alterations. Of note in this regard, subsequent analyses showed that our sample size was too small to detect small effect sizes (as reported in the study of Ross et al⁶⁰) for associations of DTD with PDI; additionally, our sample of SCZ patients showed similar levels of delusional ideation, which probably precluded us from obtaining correlations of PANSS_positive or PANSS_{positive-factor} with DTD in our Bayesian linear regression. Due to these limitations, the association between JTC and delusions remains unclear.

However, with respect to the proposed association of JTC with impulsive decision-making, group-wise comparisons of DT (calculated specifically for this purpose) showed differences in ambiguous trials and numerical differences in easy and difficult trials but in an opposite way as previously assumed: although SCZ patients tended to decide after fewer bead views they showed a pattern of needing more time than MDD and HC to make their decisions. While we are aware that our approach to correct for contributors to slower reaction times in patients with SCZ (see the Methods section and reference ⁴⁷) poses only a general approximation, our findings are not compatible with an assumption of impulsive decision-making and are in line with previous reports that did not correct for reaction time delays.¹⁰ This finding suggests that impulsiveness does not necessarily govern decisions in SCZ. We speculate that the decisions often occurred after just a few bead views because of impairments in stable belief formation. Further research is needed to reconfirm our finding and to disentangle the underlying processes of increased time expenditure and decision-making based on fewer bead views.

Finally, SCZ patients tended to overestimate the probabilities conveyed by the bead sequences presented in easy and difficult trials of the GE version (see figure 3A), which is in line with previous reports.^{12,13,27,55,61} Of note, these findings appeared not to be due to a global cognitive impairment, as SCZ patients were able to adapt their responses from easy to difficult and to ambiguous trials. In addition to assigning increased probabilities to viewed bead sequences, patients with SCZ showed increased DES, whenever there was an unexpected change in bead color in a bead sequence after ≥2 beads of the same color (figure 3B). This indicates that SCZ patients assign increased levels of significance to unexpected changes in bead color and struggle to form stable beliefs about the source jar as suggested by previous findings.^{9,10,12,55,57} Of note, we employed the same trial sequences in both the DTD and the GE versions of the task, respectively. Hence, JTC was only measured when patients could limit data gathering by making a decision and stopping the current trial (DTD version of the task). However, when SCZ patients were presented with more bead views from the same trial sequence in the GE version, they did not stick to their “JTC choice.” Rather, each time unexpected changes in bead sequences occurred (eg, a green bead after the 3 preceding beads were blue) they switched their probability ratings toward the opposite source jar (eg, switching from rating a high probability for the predominantly blue source jar after the first 3 blue beads to a medium probability for the source jar containing more green beads). We also observed changes in probability ratings following changes in bead sequences in MDD patients and HC participants. However, we only observed significantly increased DES and more pronounced changes in patients with SCZ. In sum, the GE version of the beads task, therefore, appears to be more suitable for detecting unstable belief formation in patients with SCZ in addition to the JTC bias that can be obtained from the DTD version. Of interest in this regard, reduced DTD and increased JTC in SCZ were only observed in the P_80/20 condition, but not in the more difficult P_60/40 condition, although the very same trial sequences and trial difficulty levels were applied (see supplement tables 5 and 6). SCZ patients, therefore, appeared as able as MDD and HC to adapt their performance to more difficult trial types and task conditions. However, future research is needed to clarify the role of probabilistic reasoning on beads task performance⁶¹ and to rule out the contribution of sequence effects to this finding as our two conditions (P_80/20/P_60/40) were presented in a fixed order.

Limitations and Conclusions

As we had adopted a group-comparison design, participants within each group had been thoroughly matched with respect to their sociodemographic, neuropsychological, and clinical characteristics. Additionally, PANSS scores suggest that SCZ patients were mildly paranoid, and we did not investigate acutely ill patients nor did we assess JTC longitudinally across the disease course. Furthermore, we investigated whether JTC was associated with current severity of delusions, not with their occurrence. Finally, our sample size was only sufficiently powered to detect PDI differences across all groups and our sample was not representative of the psychoses continuum in the general population.⁶⁰ Future studies could usefully address these issues by investigating larger samples regarding PDI associations and test both delusional and non-delusional patients with SCZ longitudinally in naturalistic designs.

Specific advantages of our study are that we could demonstrate for the first time the important contributory roles of unstable beliefs and individual DT to JTC, thereby challenging the notion that JTC reflects impulsive decision-making. As current psychotherapeutic and psychosocial interventions for patients with SCZ aim to modify impulsive decision-making, these findings have implications to inform further developments in psychotherapeutic treatment such as meta-cognitive training.^29,30 Importantly, our findings further imply a discussion about the construct validity of different versions of the beads task. By employing a novel beads task design that further evaluates explicit PE and disconfirmatory evidence responses, we were able to observe that SCZ patients tended to form unstable beliefs. Additionally, SCZ patients needed more time than MDD or HC to make their decisions. As the JTC bias has been demonstrated to be modifiable,^29,62 this contribution to our understanding of its components offers the potential to inform the further development of related treatment options.

Acknowledgments

We would like to thank all participants for so generously giving their time. W.S. has received a speaker’s honorarium from Mag&More GmbH. P.F. was honorary speaker for Janssen-Cilag, AstraZeneca, Eli Lilly, Bristol–Myers–Squibb, Lundbeck, Pfizer, Bayer Vital, SmithKline Beecham, Wyeth, and Essex. During the last 5 years, but not presently, he was a member of the advisory boards of Janssen-Cilag, AstraZeneca, Eli Lilly, and Lundbeck. A.H. has received a paid speakership from Janssen-Cilag, Otsuka, and Lundbeck. He was a member of an advisory board of Roche, Janssen-Cilag, Otsuka, and Lundbeck. S.B. and all other authors, C.C., M.U., Ö.Ö., T.S.-A., and L.M., declare no competing financial interests.

Funding

This work was supported by a research grant of the Deutsche Forschungsgemeinschaft (STR 1472/1-1), by a research grant of the Friedrich-Baur-Stiftung (FBS 59-17), and by a research grant of the Medical Faculty, University of Augsburg (89000101) (all to W.S). The funders had no role in the design and conduct of the study, the collection and management of the data as well as their respective analysis and the preparation or approval of the manuscript.

Author Contributions

Conceptualization: W.S., S.B., and A.H; formal analysis: W.S., S.B., A.H., and T.S.-A; funding acquisition: W.S. and S.B; investigation: W.S., C.C., and M.U; methodology: W.S., S.B., A.H., and T.S.-A; project administration: W.S. and A.H; software: W.S., S.B., and L.M; supervision: W.S., S.B., and A.H; visualization: W.S., S.B., and L.M; writing—original draft: W.S., A.H., and S.B; writing—review and editing: all authors.

Open Science Statement

The Matlab code for our experimental beads task instantiation can be downloaded at https://github.com/wstrube/Beadstask.

Study Registration Details

Trial name: “Behavioral investigation of the influence of impaired neurotransmission on perceptual and decision-making processes using computer-assisted mathematical model systems in people with schizophrenia and depression.” URL: https://www.drks.de/drks_web/navigate.do?navigationId=trial.HTML&TRIAL_ID=DRKS00019012 (Accessed July 5, 2021); https://www.drks.de/drks_web/setLocale_EN.do (Accessed July 5, 2021); Registration number: DRKS00019012.

References

Author notes