The psychology and policy of overcoming economic inequality.

Recent arguments claim that behavioral science has focused - to its detriment - on the individual over the system when construing behavioral interventions. In this commentary, we argue that tackling economic inequality using both framings in tandem is invaluable. By studying individuals who have overcome inequality, "positive deviants," and the system limitations they navigate, we offer potentially greater policy solutions.

The Inner Loop of Collective Human-Machine Intelligence.

With the rise of artificial intelligence (AI) and the desire to ensure that such machines work well with humans, it is essential for AI systems to actively model their human teammates, a capability referred to as Machine Theory of Mind (MToM). In this paper, we introduce the inner loop of human-machine teaming expressed as communication with MToM capability. We present three different approaches to MToM: (1) constructing models of human inference with well-validated psychological theories and empirical measurements; (2) modeling human as a copy of the AI; and (3) incorporating well-documented domain knowledge about human behavior into the above two approaches. We offer a formal language for machine communication and MToM, where each term has a clear mechanistic interpretation. We exemplify the overarching formalism and the specific approaches in two concrete example scenarios. Related work that demonstrates these approaches is highlighted along the way. The formalism, examples, and empirical support provide a holistic picture of the inner loop of human-machine teaming as a foundational building block of collective human-machine intelligence.

An Evaluation of Digital Partner Notification Tool Engagement and Impact for Patients Diagnosed With Gonorrhea and Syphilis.

BACKGROUND: Partner Notification (PN) is a key public health service to alert the partners of patients diagnosed with a sexually transmitted infection and then support these individuals to test. Partner Notification is a challenge to deliver because of the personal nature of this communication and the time it requires to deliver effectively. Digital tools have been developed to support patients to inform their partners; unfortunately, it has not been possible to measure engagement with-and the impact of-these solutions. A digital PN tool (dPNt) was developed for use by the health care provider to (1) deliver anonymous PN, (2) support the partner to locate and test at an appropriate local service, and (3) track real-time outcome metrics. To assess the usefulness of this tool we analyzed dPNt adoption and performance in the delivery of PN services. METHOD: A dPNt engagement for 5715 patients diagnosed by 23 health care providers with either gonorrhea or syphilis in 2019 was analyzed to determine what factors were associated with the use of this tool. The PN performance and a sub analysis of the patients who used dPNt was compared with the adjusted aggregate PN performance reported by Public Health England. RESULTS: Overall, dPNt engagement was 21% and 27% for patients with gonorrhea and syphilis, respectively. Male gender patients with gonorrhea and younger patients with either diagnosis were more likely to engage with dPNt. Nonengagement with dPNt was associated with a significantly higher number of partners reported as already seen and tested at the time that PN was initiated. The overall number of partners seen and tested per index patient increased in the clinics using dPNt from 0.43 to 0.84 for gonorrhea and 0.71 to 0.94 for syphilis, relative to Public Health England baseline results. Half of all prospective partner testing verification was done automatically by dPNt and no interaction was demonstrated between dPNt and standard methods. CONCLUSION: Digital PN tool engagement increased when more partners needed to be informed of their risk of infection. Future work is planned to improve the use of and engagement with dPNt to alert and test more partners or their risk and improve public health.

Humans actively sample evidence to support prior beliefs.

No one likes to be wrong. Previous research has shown that participants may underweight information incompatible with previous choices, a phenomenon called confirmation bias. In this paper, we argue that a similar bias exists in the way information is actively sought. We investigate how choice influences information gathering using a perceptual choice task and find that participants sample more information from a previously chosen alternative. Furthermore, the higher the confidence in the initial choice, the more biased information sampling becomes. As a consequence, when faced with the possibility of revising an earlier decision, participants are more likely to stick with their original choice, even when incorrect. Critically, we show that agency controls this phenomenon. The effect disappears in a fixed sampling condition where presentation of evidence is controlled by the experimenter, suggesting that the way in which confirmatory evidence is acquired critically impacts the decision process. These results suggest active information acquisition plays a critical role in the propagation of strongly held beliefs over time.

Unstandard Deviation: The Untapped Value of Positive Deviance for Reducing Inequalities.

Behavioral science is increasingly used in public policy to understand and address various manifestations of inequalities. Yet evidence from effective population-level interventions is limited. One framework, known as positive deviance, emphasizes individuals from disadvantaged circumstances who have significantly better outcomes than are typical for their group. Studying their behaviors and outcomes helps to understand what might explain their overall success. These insights could also be used to help others from these circumstances experience positive outcomes. Because positive deviance has been markedly understudied, we present a framework for doing so specifically within behavioral science for public policies aimed at reducing inequalities. Using examples from real-world and experimental insights on choices and outcomes of positive deviants, we encourage further study of their choices and trajectories over time to produce valuable insights. We propose that leveraging those findings would inform public policy by introducing interventions that are more ecologically sound and population-relevant and thus have a better chance at benefiting those who start off under adverse circumstances.

Mitigating belief projection in explainable artificial intelligence via Bayesian teaching.

State-of-the-art deep-learning systems use decision rules that are challenging for humans to model. Explainable AI (XAI) attempts to improve human understanding but rarely accounts for how people typically reason about unfamiliar agents. We propose explicitly modelling the human explainee via Bayesian teaching, which evaluates explanations by how much they shift explainees' inferences toward a desired goal. We assess Bayesian teaching in a binary image classification task across a variety of contexts. Absent intervention, participants predict that the AI's classifications will match their own, but explanations generated by Bayesian teaching improve their ability to predict the AI's judgements by moving them away from this prior belief. Bayesian teaching further allows each case to be broken down into sub-examples (here saliency maps). These sub-examples complement whole examples by improving error detection for familiar categories, whereas whole examples help predict correct AI judgements of unfamiliar cases.

Global mental health: an improved measure of well-being in multiple languages.

BACKGROUND: An increasing number of international organisations and national governments have committed to well-being promotion. Unfortunately, important questions regarding how to assess well-being are still unresolved, making policy implementation and evaluation difficult. METHODS: This research expanded on Huppert and So's (Soc Indic Res. 110, 837-861 2013) multidimensional subjective well-being framework by investigating the replicability of the model in two non-European regions (South America, represented by Brazil and Colombia, and Eastern Africa, represented by Uganda), and the United Kingdom. Additionally, previous limitations of the framework were also addressed. ESS Round Six items were crucially improved in terms of temporal and response scale consistency. Bayesian approximate measurement invariance was applied on a sample of 381 young adult participants to test for consistency across countries. RESULTS: The Huppert & So (Soc Indic Res. 110, 837-861 2013) model was found to fairly replicate across non-European regions, where meaningful differences in well-being patterns across regions were observed. Additionally, estimated well-being was related with other well-being measures (Five Ways): Learn and Connect were the strongest predictors of general well-being, with Take Notice and Give being associated with specific aspects of it. CONCLUSIONS: Based on this narrow sample of young adults, it appears the ten-item measure proposed by Huppert & So (Soc Indic Res. 110, 837-861 2013) is useful for assessing population mental health outside of Europe. This is only an initial attempt to assess qualities, so further testing should be done before applying at scale for identifying policy opportunities to address well-being of populations.

Replicating patterns of prospect theory for decision under risk.

Prospect theory is among the most influential frameworks in behavioural science, specifically in research on decision-making under risk. Kahneman and Tversky's 1979 study tested financial choices under risk, concluding that such judgements deviate significantly from the assumptions of expected utility theory, which had remarkable impacts on science, policy and industry. Though substantial evidence supports prospect theory, many presumed canonical theories have drawn scrutiny for recent replication failures. In response, we directly test the original methods in a multinational study (n = 4,098 participants, 19 countries, 13 languages), adjusting only for current and local currencies while requiring all participants to respond to all items. The results replicated for 94% of items, with some attenuation. Twelve of 13 theoretical contrasts replicated, with 100% replication in some countries. Heterogeneity between countries and intra-individual variation highlight meaningful avenues for future theorizing and applications. We conclude that the empirical foundations for prospect theory replicate beyond any reasonable thresholds.

Nudging New York: adaptive models and the limits of behavioral interventions to reduce no-shows and health inequalities.

BACKGROUND: Missed healthcare appointments (no-shows) are costly and operationally inefficient for health systems. No-show rates are particularly high for vulnerable populations, even though these populations often require additional care. Few studies on no-show behavior or potential interventions exist specifically for Federally Qualified Health Centers (FQHCs), which care for over 24 million disadvantaged individuals in the United States. The purpose of this study is to identify predictors of no-show behavior and to analyze the effects of a reminder intervention in urban FQHCs in order to design effective policy solutions to a protracted issue in healthcare. METHODS: This is a retrospective observational study using electronic medical record data from 11 facilities belonging to a New York City-based FQHC network between June 2017 to April 2018. This data includes 53,149 visits for 41,495 unique patients. Seven hierarchical generalized linear models and generalized additive models were used to predict no-shows, and multiple regression models evaluated the effectiveness of a reminder. All analyses were conducted in R. RESULTS: The strongest predictor of no-show rates in FQHCs is whether or not patients are assigned to empaneled providers (z = - 91.45, p < 10- 10), followed by lead time for appointments (z = 23.87, p < 10- 10). These effects were fairly stable across facilities. The reminder had minimal effects on no-show rates overall (No show rate before: 41.6%, after: 42.1%). For individuals with appointments before and after the reminder, there was a small decrease in no-shows of 2%. CONCLUSIONS: The limited effects of the reminder intervention suggest the need for more personalized behavioral interventions to reduce no-shows. We recommend that these begin with increasing the use of empaneled providers for preventive care appointments and reducing the lag time between setting the appointment and the actual date of the appointment, at least for individuals with a high rate of no-show. By complementing these with low-intensity, low-cost behavioral interventions, we would expect greater impacts for improved access to care, contributing to the well-being of vulnerable populations.

A bilingual disadvantage in metacognitive processing.

Recent research indicating that bilingualism is associated with enhanced executive function suggests that this enhancement may operate within a broader spectrum of cognitive abilities than previously thought (e.g., Stocco & Prat, 2014). In this study, we focus on metacognition or the ability to evaluate one's own cognitive performance (Flavell, 1979). Over the course of two experiments, we presented young healthy adult monolinguals and bilinguals with a perceptual two-alternative-forced-choice task followed by confidence judgements. Results from both experiments indicated that bilingual participants showed a disadvantage in metacognitive efficiency, determined through the calculation of Mratio (Maniscalco & Lau, 2014). Our findings provide novel insight into the potential differences in bilingual and monolingual cognition, which may indicate a bilingual disadvantage. Results are discussed with reference to the balance of advantages versus disadvantages associated with multilanguage learning.