key: cord-0699124-5x98zroi authors: Suthaharan, Praveen; Reed, Erin; Leptourgos, Pantelis; Kenney, Joshua; Uddenberg, Stefan; Mathys, Christoph; Litman, Leib; Robinson, Jonathan; Moss, Aaron; Taylor, Jane; Groman, Stephanie; Corlett, Philip title: Paranoia and belief updating during a crisis date: 2021-01-18 journal: Res Sq DOI: 10.21203/rs.3.rs-145987/v1 sha: bd0a09488d827c37d14ee1c29dafc550d6ad4d50 doc_id: 699124 cord_uid: 5x98zroi The 2019 coronavirus (COVID-19) pandemic has made the world seem unpredictable. During such crises we can experience concerns that others might be against us, culminating perhaps in paranoid conspiracy theories. Here, we investigate paranoia and belief updating in an online sample (N=1,010) in the United States of America (U.S.A). We demonstrate the pandemic increased individuals’ self-rated paranoia and rendered their task-based belief updating more erratic. Local lockdown and reopening policies, as well as culture more broadly, markedly influenced participants’ belief-updating: an early and sustained lockdown rendered people’s belief updating less capricious. Masks are clearly an effective public health measure against COVID-19. However, state-mandated mask wearing increased paranoia and induced more erratic behaviour. Remarkably, this was most evident in those states where adherence to mask wearing rules was poor but where rule following is typically more common. This paranoia may explain the lack of compliance with this simple and effective countermeasure. Computational analyses of participant behaviour suggested that people with higher paranoia expected the task to be more unstable, but at the same time predicted more rewards. In a follow-up study we found people who were more paranoid endorsed conspiracies about mask-wearing and potential vaccines – again, mask attitude and conspiratorial beliefs were associated with erratic task behaviour and changed priors. Future public health responses to the pandemic might leverage these observations, mollifying paranoia and increasing adherence by tempering people’s expectations of other’s behaviour, and the environment more broadly, and reinforcing compliance. (N=1,010) in the United States of America (U.S.A). We demonstrate the pandemic increased 23 individuals' self-rated paranoia and rendered their task-based belief updating more erratic. Local 24 lockdown and reopening policies, as well as culture more broadly, markedly influenced participants' 25 belief-updating: an early and sustained lockdown rendered people's belief updating less capricious. Masks are clearly an effective public health measure against COVID-19. However, state-mandated 27 mask wearing increased paranoia and induced more erratic behaviour. Remarkably, this was most 28 evident in those states where adherence to mask wearing rules was poor but where rule following is 29 typically more common. This paranoia may explain the lack of compliance with this simple and effective 30 countermeasure. Computational analyses of participant behaviour suggested that people with higher 31 paranoia expected the task to be more unstable, but at the same time predicted more rewards. In a 32 follow-up study we found people who were more paranoid endorsed conspiracies about mask-wearing 33 and potential vaccines -again, mask attitude and conspiratorial beliefs were associated with erratic 34 task behaviour and changed priors. Future public health responses to the pandemic might leverage 35 these observations, mollifying paranoia and increasing adherence by tempering people's expectations Introduction zMETA=4.035, pMETA=5.45E-5). However, 2 was lower in high paranoia, indicating that tonic task 120 changes were less impactful on their choices ( Fig. 1a ; social task, F(1, 128)=5.091, p=0.026, ηp 2 =0.038; 121 non-social task, F(1, 70)=8.681, p=0.004, ηp 2 =0.11). Across social and non-social contexts, high paranoia The impact of an evolving pandemic on paranoia and belief updating 128 After the pandemic was declared we continued to acquire data on both tasks (3/19/2020-7/17/2020). We found an interaction between paranoia and pandemic period for win-switching (F(2, 593)=9.075, Lose-stay rates also exhibited a period by paranoia interaction (F(2, 593)=6.51, p=0.002, ηp 2 =0.021, 147 Figure 2b ). During reopening, high paranoia participants were less likely than participants with low 148 paranoia to persist after negative feedback. Lose-stay rates declined in high paranoia participants on 149 reopening. In parallel, we observed an increase in their contingency prior ( 2 0 ) after reopening (F (2, 150 593)=8.996, p=0.0001, ηp 2 =0.029, Figure 2c ). Across the three pandemic periods, 2 0 correlated 151 negatively with lose-stay behavior (r=-0.69, p=1.3E-7). These findings suggest that paranoid subjects 152 had higher expectations of reward during reopening and were less likely to tolerate negative feedback. Specifically, low paranoia appeared to temper reward expectations. Tonic belief updating parameters 154 showed a paranoia group effect ( 3: F(1, 593)=19.31, p=1.32E-5, ηp 2 =0.032), and a significant block by 155 paranoia interaction ( 2: F(1, 593)=5.446, p=0.02, ηp 2 =0.009). High paranoia subjects were slower to 156 update their volatility and reinforcement beliefs. We asked participants in the social task to rate whether or not they believed that the avatars had 159 deliberately sabotaged them. Win-switch rate (r=0.259, p=1.2E-5, n=280), 2 0 (r=0.124, p=0.038), and We recorded a significant increase in paranoia when Americans were emerging from lockdown ( Figure 179 2A Mandated mask wearing was associated with an estimated 48% increase in paranoia (gDID = 0.48, p = 201 0.018), relative to states in which mask wearing was recommended but not required (Figure 4a ). This 202 increase in paranoia was mirrored as significantly higher win-switch rates in participant task We examined whether any other features might illuminate this variation in paranoia by local mask 209 policy 17 . There are state-level cultural differences -measured by the Cultural Tightness and Looseness 210 (CTL) index 17 -with regards to rule following and tolerance for deviance. Tighter states have more 211 rules and tolerate less deviance, whereas looser states have few strongly enforced rules and greater 212 tolerance for deviance 17 . We also tried to assess whether people were following the mask rules. We 213 acquired independent survey data gathered in the U.S.A. from 250,000 respondents who, between July 214 2 and July 14, were asked: How often do you wear a mask in public when you expect to be within six 215 feet of another person? 18 These data were used to compute an estimated frequency of mask wearing in 216 each state during the reopening period ( Figure 4c ). We found that in culturally tighter states where mask wearing was mandated, mask wearing was lowest 219 (mloose=0.787, mtight=0.760, t32=2.87, p=0.007). Furthermore, even in states where mask wearing was recommended, mask wearing was lowest in culturally tighter states (mloose=0.674, mtight=0.629, 221 t107=2.46, p=0.016). Through backward linear regression with removal, we fit a series of models attempting to predict 224 individuals' self-rated paranoia (N=172) from the features of their environment, including whether they 225 were subject to a mask mandate or not, the cultural tightness of their state, state-level mask-wearing, and Coronavirus cases in their state. In the best fitting model (F(11,160)=1.91,p=0.04) there was a 227 significant three way interaction between mandate, state tightness and perceived mask wearing (t24=- Whilst mask-mandate and mask-recommend states were matched at baseline, it is possible that 258 increases in cases and deaths at reopening explain the increase in paranoia, rather than the mask 259 mandate. Our data militate against this explanation. There were no significant differences in cases (t=-1. 79 lockdown, reopening, Figure 5 ). Furthermore, given that the effects we describe depend on 277 geographical location, we confirm that the proportions of participants recruited from each state did not 278 differ across our study periods (χ 2 =6.63, d.f.=6, p=0.34, Figure 6 ). Finally, in order to assuage concerns lockdown, r = 0.78 p = 5.8E-9; reopening, r = 0.81 p = 8.5E-10 ( Figure 6 ). Thus, we did not, by chance, recruit more participants from mask-mandating states or tighter states, for example. Furthermore, 287 focusing on the data that went into the DiD, there were no demographic differences pre-versus post-288 reopening for mask-mandate versus mask-recommended states (age, p=0.45, gender, p=0.73, race, 289 p=0.17, Figure 7 ). Taken together with our task and self-report results, these control analyses increase 290 our confidence that during reopening, people were most paranoid in the presence of rules and 291 perceived rule breaking, particularly in states where people usually tend to follow the rules. The lockdown rendered participants in less proactive states more susceptible to paranoia in terms of 311 their expectations about volatility. However, we also found that people who were less paranoid during 312 lockdown and reopening were more forgiving of collaborators, returning to those characters even after 313 they have delivered losses in the social task. The increase in paranoia that we observed appeared to coincide with reopening from lockdown and to 316 be particularly pronounced in states that mandated that their residents wear masks when in public. We Perhaps a more vigorous lockdown provided fewer opportunities to misinterpret social interactions, whereas reopening provided more opportunities to encounter others and thence for paranoia. Abiding by lockdown is a personal choice whose effectiveness depends on ones' own choice (to stay 354 home and avoid others). Choosing to wear a mask also offers personal protection. However, mask-355 wearing also protects others from the wearer; it is something one does for others. Table 1 for further 630 information. We recruited 130 (20 high paranoia) participants who completed the social task. Similarly, 631 of the 231 (see Table 2 for details), we recruited 119 (27 high paranoia) and 112 (23 high paranoia) 632 participants who completed the non-social and social tasks, respectively. Lastly, of the 172, we 633 recruited 93 (35 high paranoia) and 79 (35 high paranoia) participants who completed the non-social 634 and social tasks, respectively (See Table 3 for details). In addition to CloudResearch's safeguard from bot submissions, we implemented the same study advertisement, submission review, approval and 636 bonusing as described in our previous study 5 . We excluded a total of 163 submissions -18 from pre- partner. We instructed participants to select an avatar (or partner) to work with to gain as many points 659 towards their group project. Like the non-social, they were instructed that the best partner could 660 change. For both tasks, the contingencies began as 90% reward, 50% reward, and 10% reward with 661 the allocation across deck/partner switching after 9 out of 10 consecutive rewards. At the end of the 662 second block, unbeknownst to the participants, the underlying contingencies transition to 80% reward, 663 40% reward, and 20% reward -making it more difficult to discern whether a loss of points was due to 664 normal variations (probabilistic noise) or whether the best option has changed. Questionnaires. Following task completion, questionnaires were administered via Qualtrics, we For the replication study, we adopted a survey 43 that investigated beliefs on mask usage of individual 677 US consumers and a survey 44 of COVID-19. The 9-item mask questionnaire was used for our study to (1) The coronavirus vaccine will contain microchips to control the people. (2) Coronavirus was created to force everyone to get vaccinated. (3) The vaccine will be used to carry out mass sterilization. (4) The coronavirus is bait to scare the whole globe into accepting a vaccine that will introduce the 'real' 687 deadly virus. (5) The WHO already has a vaccine and are withholding it. We adopted a 7-point scale: strongly disagree (1), disagree (2), somewhat disagree (3), neutral (4), 691 somewhat agree (5), agree (6) and strongly agree (7). A higher score indicates greater endorsement of 692 a question. Additional features. Along with the task and questionnaire data, we examined state-level 695 unemployment rate 45 , confirmed COVID-19 cases 46 , and mask usage 18 in the USA. Unemployment. The Carsey School of Public Policy reported unemployment rates for the months of February, April, May and June in 2020. We utilized the rates in April and June as our markers for measuring the 698 difference in unemployment between the pre-pandemic period and pandemic period, respectively. Protests. We accessed the publicly available data from the armed conflict location and event data 719 project (ACLED, https://acleddata.com/special-projects/us-crisis-monitor/), which has been recording We also defined a proactivity metric (or score) to measure how adequately or inadequately a state 731 reacted to COVID-19 47 . This score was calculated based on two features: ∶ number of days from baseline to introduce the stay-at-home order (i.e., baseline date -introduced date). ∶ number of days before the order was lifted (i.e., expiration date -introduced date). where baseline date is defined as the date at which the first stay-at-home order was implemented. California was the first to enforce the order on March 19 th , 2020 (i.e., baseline date = 0). States where 738 stay-at-home orders were not implemented had 'N/A' values and were set to 0 in our calculation. Moreover, states that had an indefinite time frame for the orders were set to 100 in our calculation (i.e., 740 expiration date = 100). To compute the proactivity score, we perform the following sum: This metric -ranging from 0 (inadequate) to 100 (adequate) -offers a reasonable approach for 747 measuring proactive state interventions in response to the pandemic. Causal inference. To measure attribution of mask-wearing policy on paranoia, we adopt a differences- 768 769 h = + * X lmnon + * X qrstlu + * X lws + * X ymnz + * X lmnon * qrstlu + * X lmnon * lws 770 + * X qrstlu * lsw + * X lmnon * ymnz + * X lws * ymnz + * X lws * ymnz + 771 * X lmnon * qrstlu * lws + * X lmnon * qrstlu * ymnz + * X lmnon * lws * ymnz + 772 * X qrstlu * lws * ymnz + * •‚ƒ"ƒ * … † ‡ˆ•‰ * •Š ‡ * ‹‚ƒOE + e 773 774 Reduced model: h = + * X lmnon + * X qrstlu + * X lws + * X ymnz + * X lmnon * qrstlu + * X lmnon * lws 777 + * X qrstlu * lws + * X qrstlu * ymnz + * X lws * ymnz + * X lmnon * qrstlu * lws + 778 * X qrstlu * lws * ymnz + e 779 780 781 See We estimated perceptual parameters individually for the first and second halves of the task (i.e., blocks 787 1 and 2). Each participant's choices (i.e., deck 1, 2, or 3) and outcomes (win or loss) were entered as 788 separate column vectors with rows corresponding to trials. Wins were encoded as '1', losses as '0', and 789 choices as '1', '2', or '3'. We selected the autoregressive 3-level HGF multi-arm bandit configuration for 790 our perceptual model and paired it with the softmax-mu03 decision model. Table 4 To conduct meta-analyses of effect replication across experiments, we fit random effects models in the Conspiracy theories as part of history: The role of societal 827 crisis situations Whom do You Trust? Doubt and Conspiracy Theories in the 2009 Influenza 829 Pandemic An evolutionary perspective on paranoia Uncertainty Through Basic and Computational Neuroscience Paranoia as a deficit in non-social belief updating The revised Green et al., Paranoid Thoughts Scale (R-GPTS): psychometric 836 properties, severity ranges, and clinical cut-offs Adaptive learning under expected and unexpected uncertainty Evaluation of the SCID-II personality disorder traits for 840 DSM-IV: coherence, discrimination, relations with general personality traits, and functional 841 impairment A bayesian foundation for individual 843 learning under uncertainty Uncertainty in perception and the Hierarchical Gaussian Filter. Frontiers in 845 human neuroscience Computational Psychiatry: A Rosetta Stone linking the brain to 847 mental illness Acting on persecutory delusions: the importance of safety seeking Quasi-experimental causality in neuroscience and 851 behavioural research Mostly Harmless Econometrics Using Difference-in-Differences to Identify Causal Effects of 855 COVID-19 Policies The Pursuit of the Millenium Tightness-looseness across the 50 united states Estimates from The New York Times, based on roughly 250,000 interviews 860 conducted by Dynata from Assessment of obsessive-compulsive symptom dimensions: 862 development and evaluation of the Dimensional Obsessive-Compulsive Scale Despite COVID-19 com: A versatile crowdsourcing data 867 acquisition platform for the behavioral sciences How paranoid are conspiracy believers? Toward a more fine-grained 869 understanding of the connect and disconnect between paranoia and belief in conspiracy theories Two neurocomputational building blocks of social norm compliance Disrupted prediction-error signal in psychosis: evidence for an associative 874 account of delusions Toward a neurobiology of 876 delusions Midbrain activation during Pavlovian conditioning and delusional symptoms 878 in schizophrenia Social norms and human cooperation Altruistic punishment in humans Some US police resist enforcing coronavirus mask mandates The mask police will not be patrolling': How Indiana is enforcing mask mandate Punishment and counter-punishment in public good games: can we really govern 886 ourselves Norms and expectations in social decision-making Social image concerns promote cooperation more 890 than altruistic punishment Both Rewards and Moral Praise Can Increase the Prosocial 892 Decisions: Revealed in a Modified Ultimatum Game Task Mistrust: An ethnographic theory Collective Action and the Evolution of Social Norms Not-so-social learning strategies Prevalence and correlates of self-reported psychotic symptoms in the British 899 population. The British journal of psychiatry : the journal of mental science 185 Concomitants of paranoia in the general population Differences between tight and loose cultures: a 33-nation study An inventory for measuring clinical anxiety: 905 psychometric properties An inventory for measuring 907 depression Consumers 909 and COVID-19: Survey Results on Mask-Wearing Behaviors and Beliefs. Economic Commentary Coronavirus conspiracy beliefs, mistrust, and compliance with government 911 guidelines in England An ongoing repository of data on coronavirus cases and deaths in the n/a n/a n/a n/a n/a n/a Green et al. , 73 SCID Paranoid Personality a n/a n/a n/a n/a n/a n/a Green et al.