Search

Background Machine-learning algorithms are becoming popular techniques to predict ambient air PM2.5 concentrations at high spatial resolutions (1 × 1 km) using satellite-based aerosol optical depth (AOD). Most machine-learning models have aimed to predict 24 h-averaged PM2.5 concentrations (mean PM2.5) in high-income regions. Over Mexico, none have been developed to predict subdaily peak levels, such as the maximum daily 1-h concentration (max PM2.5). Objective Our goal was to develop a machine-learning model to predict mean PM2.5 and max PM2.5 concentrations in the Mexico City Metropolitan Area from 2004 through 2019. Methods We present a new modeling approach based on extreme gradient boosting (XGBoost) and inverse-distance weighting that uses AOD, meteorology, and land-use variables. We also investigated applications of our mean PM2.5 predictions that can aid local authorities in air-quality management and public-health surveillance, such as the co-occurrence of high PM2.5 and heat, compliance with local air-quality standards, and the relationship of PM2.5 exposure with social marginalization. Results Our models for mean and max PM2.5 exhibited good performance, with overall cross-validated mean absolute errors (MAE) of 3.68 and 9.20 μg/m3, respectively, compared to mean absolute deviations from the median (MAD) of 8.55 and 15.64 μg/m3. In 2010, everybody in the study region was exposed to unhealthy levels of PM2.5. Hotter days had greater PM2.5 concentrations. Finally, we found similar exposure to PM2.5 across levels of social marginalization. Significance Machine learning algorithms can be used to predict highly spatiotemporally resolved PM2.5 concentrations even in regions with sparse monitoring. Impact Our PM2.5 predictions can aid local authorities in air-quality management and public-health surveillance, and they can advance epidemiological research in Central Mexico with state-of-the-art exposure assessment methods.

The challenge of reconstructing air temperature for environmental applications is to accurately estimate past exposures even where monitoring is sparse. We present XGBoost-IDW Synthesis for air temperature (XIS-Temperature), a high-resolution machine-learning model for daily minimum, mean, and maximum air temperature, covering the contiguous US from 2003 through 2021. XIS uses remote sensing (land surface temperature and vegetation) along with a parsimonious set of additional predictors to make predictions at arbitrary points, allowing the estimation of address-level exposures. We built XIS with a computationally tractable workflow for extensibility to future years, and we used weighted evaluation to fairly assess performance in sparsely monitored regions. The weighted root mean square error (RMSE) of predictions in site-level cross-validation for 2021 was 1.89 K for the minimum daily temperature, 1.27 K for the mean, and 1.72 K for the maximum. We obtained higher RMSEs in earlier years with fewer ground monitors. Comparing to three leading gridded temperature models in 2021 at thousands of private weather stations not used in model training, XIS had at most 49% of the mean square error for the minimum temperature and 87% for the maximum. In a national application, we report a stronger relationship between minimum temperature in a heatwave and social vulnerability with XIS than with the other models. Thus, XIS-Temperature has potential for reconstructing important environmental exposures, and its predictions have applications in environmental justice and human health.

While weather stations generally capture near‐surface ambient air temperature (Ta) at a high temporal resolution to calculate daily values (i.e., daily minimum, mean, and maximum Ta), their fixed locations can limit their spatial coverage and resolution even in densely populated urban areas. As a result, data from weather stations alone may be inadequate for Ta‐related epidemiology particularly when the stations are not located in the areas of interest for human exposure assessment. To address this limitation in the Megalopolis of Central Mexico (MCM), we developed the first spatiotemporally resolved hybrid satellite‐based land use regression Ta model for the region, home to nearly 30 million people and includes Mexico City and seven more metropolitan areas. Our model predicted daily minimum, mean, and maximum Ta for the years 2003–2019. We used data from 120 weather stations and Land Surface Temperature (LST) data from NASA's MODIS instruments on the Aqua and Terra satellites on a 1 × 1 km grid. We generated a satellite‐hybrid mixed‐effects model for each year, regressing Ta measurements against land use terms, day‐specific random intercepts, and fixed and random LST slopes. We assessed model performance using 10‐fold cross‐validation at withheld stations. Across all years, the root‐mean‐square error ranged from 0.92 to 1.92 K and the R 2 ranged from .78 to .95. To demonstrate the utility of our model for health research, we evaluated the total number of days in the year 2010 when residents ≥65 years old were exposed to Ta extremes (above 30°C or below 5°C). Our model provides much needed high‐quality Ta estimates for epidemiology studies in the MCM region., Spatial pattern of the 95th percentiles of minimum (a) and maximum (b) temperature across days for each 1 km2 grid cell in the Megalopolis of Central Mexico for 2018. Temporal imputation of LST, consideration of missing data as a predictor and careful cross‐validation with detailed characterization of predictive accuracy. Application estimates population exposures to extreme temperatures for use in epidemiologic studies.

Problems in mother–child relationships are thought to be key to the intergenerational transmission of depression. To evaluate neural and behavioral processes involved in these pathways, we tested effects of maternal depression and maternal-child relationship quality in early childhood on neural and interviewer-based indicators of social processes in adolescence. At age 3, children and mothers ( N = 332) completed an observational parenting measure and diagnostic interviews with mothers. At age 12, adolescents completed a task in which event-related potentials (ERPs) were recorded in response to peer acceptance and rejection feedback and interviews to assess peer stress. Lower mother–child relationship quality at age 3 was associated with enhanced reactivity to rejection, as measured by N1, and greater peer stress at age 12. Indirect effects of maternal depression through mother–child relationship quality were observed for N1 and peer stress. Findings inform understanding of disruptions in social functioning that are likely to be relevant to the intergenerational transmission of depression.

Measurement of social processes is of interest across areas of research. Event-related potentials (ERPs) measured using electroencephalography (EEG) offer high temporal resolution, but little work has leveraged this strength to characterize the time course of social feedback processing. The present study aimed to replicate and extend previous research by systematically identifying the temporal dynamics of responses to both social acceptance and rejection feedback relative to neutral feedback, examining internal consistency of ERPs, and exploring correspondence with behavioral measures elicited during a peer interaction task. Emerging adults (N = 118) completed a computerized peer interaction task in which they made decisions to accept or reject peers and received rejection, acceptance, and neutral feedback while EEG data were recorded. Principal component analysis was used to derive temporally and spatially distinct ERP components sensitive to positive and negative social feedback. Participant voting patterns and post-task liking ratings for computer-controlled peers were also examined. Replicating prior work, components consistent with N1, P2, reward positivity (RewP), and P3 emerged, but distinct patterns of modulation by acceptance and rejection relative to neutral feedback were observed. Most components showed acceptable internal consistency. Sensitivity to peer feedback assessed through participant voting patterns and liking ratings for peers was correlated with RewP and P3 components. Results highlight the complexity of social feedback processing observable in a computerized peer interaction task and offer promising neural and behavioral measures that can be used to examine individual differences in sensitivity to both social acceptance and rejection feedback.

The COVID-19 pandemic has yielded disproportionate impacts on communities of color in New York City (NYC). Researchers have noted that social disadvantage may result in limited capacity to socially distance, and consequent disparities. We investigate the association between neighborhood social disadvantage and the ability to socially distance, infections, and mortality in Spring 2020. We combine Census Bureau and NYC open data with SARS-CoV-2 testing data using supervised dimensionality-reduction with Bayesian Weighted Quantile Sums regression. The result is a ZIP code-level index with weighted social factors associated with infection risk. We find a positive association between neighborhood social disadvantage and infections, adjusting for the number of tests administered. Neighborhood disadvantage is also associated with a proxy of the capacity to socially isolate, NYC subway usage data. Finally, our index is associated with COVID-19-related mortality., Neighborhood disadvantage and capacity to socially distance have been discussed as factors involved in COVID-19 disparities. Here, the authors develop an inequity index on zip code-level infections, and examine differences in neighborhood utilization of subways in New York City.

Background Alterations in positive valence systems and social processes, including low reward responsiveness and high rejection sensitivity, have been observed in depression. Most reward research focuses on the monetary domain, but social reward responsiveness may be particularly relevant to understanding the etiology of depression, especially in combination with other social processes. Pathways to depression are complex, and research testing interactions between multiple factors is needed. The present study examined the interactive effects of reward responsiveness and rejection sensitivity on depressive symptoms using both social and monetary reward electroencephalogram (EEG) tasks. Methods Emerging adults (N = 120) completed peer interaction and monetary incentive delay tasks while EEG data were recorded, as well as self-report measures of rejection sensitivity and depressive symptoms. Results The interaction between social reward responsiveness and self-reported rejection sensitivity was significantly associated with depressive symptoms, such that rejection sensitivity was associated with greater depressive symptoms for those with a relatively reduced response to social reward. The interaction between monetary reward responsiveness and rejection sensitivity was not significant. Limitations The study was cross-sectional and used a non-clinical sample. Conclusions Results suggest a possible pathway for depressive symptoms characterized by the combination of high rejection sensitivity and low social reward responsiveness. Findings highlight the need for consideration of multiple domains of reward responsiveness in clinical neuroscience research. With extension to longitudinal studies and clinical samples, the present findings may inform understanding of targets for intervention.

Coronavirus 2019 (COVID-19), caused by the SARS-CoV-2 virus, has become the deadliest pandemic in modern history, reaching nearly every country worldwide and overwhelming healthcare institutions. As of April 20, there have been more than 2.4 million confirmed cases with over 160,000 deaths. Extreme case surges coupled with challenges in forecasting the clinical course of affected patients have necessitated thoughtful resource allocation and early identification of high-risk patients. However, effective methods for achieving this are lacking. In this paper, we use electronic health records from over 3,055 New York City confirmed COVID-19 positive patients across five hospitals in the Mount Sinai Health System and present a decision tree-based machine learning model for predicting in-hospital mortality and critical events. This model is first trained on patients from a single hospital and then externally validated on patients from four other hospitals. We achieve strong performance, notably predicting mortality at 1 week with an AUC-ROC of 0.84. Finally, we establish model interpretability by calculating SHAP scores to identify decisive features, including age, inflammatory markers (procalcitonin and LDH), and coagulation parameters (PT, PTT, D-Dimer). To our knowledge, this is one of the first models with external validation to both predict outcomes in COVID-19 patients with strong validation performance and identify key contributors in outcome prediction that may assist clinicians in making effective patient management decisions.One-Sentence SummaryWe identify clinical features that robustly predict mortality and critical events in a large cohort of COVID-19 positive patients in New York City.

Many youth with ADHD experience peer difficulties, but the mechanisms underlying this dysfunction remain unknown. Very little work has examined neurophysiological measures of social feedback processing in relation to ADHD symptoms. The goal of this study was to examine associations of ADHD symptoms with indicators of sensitivity to social feedback in a laboratory task and self-report of rejection sensitivity. A large community sample of 10-to 15-year-old adolescents (N = 391; M(age)= 12.64, 48.6% girls) participated in the study. Mothers rated youth ADHD symptoms. Youth completed the Island Getaway task, which elicits neurophysiological (i.e., event-related potentials [ERP]) measures of sensitivity to peer rejection and acceptance feedback, and also completed self-ratings of rejection sensitivity. Greater ADHD symptoms were associated with an enhanced N1 ERP component, which correlated with higher levels of self-reported rejection sensitivity. In addition, greater ADHD symptoms were associated with reduced reactivity to social acceptance, as measured by the later reward positivity ERP component. Youth with elevated ADHD symptoms exhibited enhanced sensitivity to peer rejection at the neurophysiological and self-report level, as well as reduced neurophysiological reactivity to peer acceptance. Future work including neural measures of social functioning may serve to elucidate mechanisms underlying the social dysfunction characteristic of ADHD.

Background Accurate and precise estimates of ambient air temperatures that can capture fine-scale within-day variability are necessary for studies of air temperature and health. Method We developed statistical models to predict temperature at each hour in each cell of a 927-m square grid across the Northeast and Mid-Atlantic United States from 2003 to 2019, across ~4000 meteorological stations from the Integrated Mesonet, using inputs such as elevation, an inverse-distance-weighted interpolation of temperature, and satellite-based vegetation and land surface temperature. We used a rigorous spatial cross-validation scheme and spatially weighted the errors to estimate how well model predictions would generalize to new cell-days. We assess the within-county association of temperature and social vulnerability in a heat wave as an example application. Results We found that a model based on the XGBoost machine-learning algorithm was fast and accurate, obtaining weighted root mean square errors (RMSEs) around 1.6 K, compared to standard deviations around 11.0 K. We found similar accuracy when validating our model on an external dataset from Weather Underground. Assessing predictions from the North American Land Data Assimilation System-2 (NLDAS-2), another hourly model, in the same way, we found it was much less accurate, with RMSEs around 2.5 K. This is likely due to the NLDAS-2 model's coarser spatial resolution, and the dynamic variability of temperature within its grid cells. Finally, we demonstrated the health relevance of our model by showing that our temperature estimates were associated with social vulnerability across the region during a heat wave, whereas the NLDAS-2 showed a much weaker association. Conclusion Our high spatiotemporal resolution air temperature model provides a strong contribution for future health studies in this region.

Peer relationships play a major role in adolescent development, but few methods exist for measuring social processing at the neurophysiological level. This study extends our pilot study of Island Getaway, a task for eliciting event-related potentials (ERPs) to peer feedback. We differentiated ERPs using principal components analysis (PCA) and examined associations with behavioral and self-report measures in young adolescents (N = 412). PCA revealed an early negativity in the ERP enhanced for rejection feedback, followed by a series of positivities (consistent with reward positivity [RewP], P300, and late positive potential) that were enhanced for acceptance feedback. Greater self-reported task engagement correlated with a larger RewP to acceptance and lower rates of rejecting peers. Youth higher in depressive symptoms exhibited a blunted RewP to social acceptance and reported lower engagement. Results highlight ERP components sensitive to peer feedback that may inform understanding of social processes relevant to typical and atypical development.

Prenatal alcohol-drinking is often measured with self-report, but it is unclear whether mothers give more accurate answers when asked while pregnant or some time after their pregnancy. There is also the question of whether to measure drinking in a dichotomous or continuous fashion. We sought to examine how the timing and scale of self-reports affected the content of reports. From a sample of 576 black mothers around Cape Town, South Africa, we compared prenatal reports of prenatal drinking with 5-year retrospective reports, and dichotomous metrics (drinking or abstinent) with continuous metrics (fluid ounces of absolute alcohol drunk per day). Amounts increased over the 5-year period, whereas dichotomous measures found mothers less likely to report drinking later. All four measures were weakly associated with birth weight, birth height, child head circumference soon after birth, and child intelligence at age 5. Furthermore, neither reporting time nor the scale of measurement were consistently related to the strengths of these associations. Our results point to problems with self-report, particularly with this population, but we recommend post-birth continuous measures as the best of the group for their flexibility and their consistency with previous research.

Objective This study examines the effect of a home visiting intervention on maternal alcohol use, problematic drinking, and the association of home visiting and alcohol use on children's behavioral, cognitive, and health outcomes at 5 time points over 5 years. Method We analyzed 5,099 observations of 1,236 mothers and their children from pregnancy to 5 years postbirth, within a longitudinal cluster-randomized trial evaluating the effect of a home visiting intervention on mothers in Cape Town, South Africa. Paraprofessional home visitors coached mothers on coping with multiple risk factors, including a brief, 1-visit intervention on alcohol prevention in pregnancy. We assessed changes in maternal drinking over time in relation to the intervention, and then examined the impact of these drinking patterns on child outcomes over five years. Results Drinking increased over the 5 years postbirth, but it was significantly lower in the intervention condition. Compared with abstinence, mothers' problematic drinking was associated with decreased child weight (-0.21 z-units) at all assessments, increased child aggressive behavior (3 to 7 additional symptoms), and decreased child performance on an executive functioning measure (the silly sounds task; odds ratio = .34) at 3 and 5 years. The intervention's effect was associated with increased child aggression (0.25 to 0.75 of 1 additional symptom), but the intervention appeared to decrease the effect of problem drinking on children's aggressive acts and executive functioning. Conclusion These findings support the need for sustained interventions to reduce alcohol use, especially for mothers who exhibit problematic drinking. Maternal drinking influences children's health and development over time. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Reconstructing the distribution of fine particulate matter (PM(2.5)) in space and time, even far from ground monitoring sites, is an important exposure science contribution to epidemiologic analyses of PM(2.5) health impacts. Flexible statistical methods for prediction have demonstrated the integration of satellite observations with other predictors, yet these algorithms are susceptible to overfitting the spatiotemporal structure of the training datasets. We present a new approach for predicting PM(2.5) using machine-learning methods and evaluating prediction models for the goal of making predictions where they were not previously available. We apply extreme gradient boosting (XGBoost) modeling to predict daily PM(2.5) on a 1×1 km(2) resolution for a 13 state region in the Northeastern USA for the years 2000–2015 using satellite-derived aerosol optical depth and implement a recursive feature selection to develop a parsimonious model. We demonstrate excellent predictions of withheld observations but also contrast an RMSE of 3.11 μg/m(3) in our spatial cross-validation withholding nearby sites versus an overfit RMSE of 2.10 μg/m(3) using a more conventional random ten-fold splitting of the dataset. As the field of exposure science moves forward with the use of advanced machine-learning approaches for spatiotemporal modeling of air pollutants, our results show the importance of addressing data leakage in training, overfitting to spatiotemporal structure, and the impact of the predominance of ground monitoring sites in dense urban sub-networks on model evaluation. The strengths of our resultant modeling approach for exposure in epidemiologic studies of PM(2.5) include improved efficiency, parsimony, and interpretability with robust validation while still accommodating complex spatiotemporal relationships.

Valid measurement of substance use is necessary to evaluate preventive and treatment interventions. Self-report is fast and inexpensive, but its accuracy can be hampered by social desirability bias and imperfect recall. We examined the agreement between self-report of recent use and rapid diagnostic tests for three substances (alcohol, cannabis, and methamphetamine) among 904 young men living in Cape Town, South Africa. Rapid diagnostic tests detected the respective substances in 32%, 52%, and 22% of men. Among those who tested positive, 61% (95% CI [56%, 66%]), 70% ([67%, 74%]), and 48% ([42%, 54%]) admitted use. Men were moderately more willing to admit use of cannabis than alcohol (log OR 0.42) or admit use of alcohol than methamphetamine (log OR 0.53). Our findings show that self-report has reasonable criterion validity in this population, but criterion validity can vary substantially depending on the substance.

Altruistic behavior benefits other individuals at a cost to oneself. The purpose of the present experiment was to study altruistic behavior by players (P) in 2-person iterated prisoner's dilemma games in which reciprocation by the other player (OP) was impossible, and this impossibility was clear to P. Altruism by P could not therefore be attributed to expectation of reciprocation. The cost to P of altruistic behavior was constant throughout the study, but the benefit to OP from P's cooperation differed between groups and conditions. Rate of cooperation was higher when benefit to OP was higher. Thus altruism (not attributable to expectation of reciprocation) can be a significant factor in interpersonal relationships as studied in iterated prisoner's dilemma games, and needs to be taken into account in their analysis.

Author(s): Zingmond, David S; Arfer, Kodi B; Gildner, Jennifer L; Leibowitz, Arleen A | Abstract: BackgroundAntiretroviral therapy has increased longevity for people living with HIV (PLWH). As a result, PLWH increasingly experience the common diseases of aging and the resources needed to manage these comorbidities are increasing. This paper characterizes the number and types of comorbidities diagnosed among PLWH covered by Medicare and examines how non-HIV comorbidities relate to outpatient, inpatient, and pharmaceutical expenditures.MethodsThe study examined Medicare expenditures for 9767 HIV-positive Californians enrolled in Medicare in 2010 (7208 persons dually covered by Medicare and Medicaid and 2559 with Medicare only). Costs included both out of pocket costs and those paid by Medicare and Medicaid. Comorbidities were determined by examining diagnosis codes.FindingsMedicare expenditures for Californians with HIV averaged $47,036 in 2010, with drugs accounting for about 2/3 of the total and outpatient costs 19% of the total. Inpatient costs accounted for 18% of the total. About 64% of the sample had at least one comorbidity in addition to HIV. Cross-validation showed that adding information on comorbidities to the quantile regression improved the accuracy of predicted individual expenditures. Non-HIV comorbidities relating to health habits-diabetes, hypertension, liver disease (hepatitis C), renal insufficiency-are common among PLWH. Cancer was relatively rare, but added significantly to cost. Comorbidities had little effect on pharmaceutical costs, which were dominated by the cost of antiretroviral therapy, but had a major effect on hospital admission.ConclusionsComorbidities are prevalent among PLWH and add substantially to treatment costs for PLWH. Many of these comorbidities relate to health habits that could be addressed with additional prevention in ambulatory care, thereby improving health outcomes and ultimately reducing costs.

The more politically conservative Americans are, the more restrictive their sexual attitudes are. A natural follow-up question is how this difference in attitudes relates to actual behavior. But self-reports of sexual behavior may be compromised by a social desirability bias that is influenced by the very sexual attitudes at issue. We employed a non-self-reported measure of sexual behavior: usage of the adultery-focused dating website Ashley Madison. Linking an August 2015 leak of user data from Ashley Madison to 2012 voter registration rolls from five U.S. states, we found 80,000 matches between 200,000 Ashley Madison user accounts and 50 million voters. According to simple rates in the sample, and also to predictively validated regression models controlling for state, gender, and age, we found that Democrats were least likely to use Ashley Madison, Libertarians were most likely, and Republicans, Greens, and unaffiliated voters were in between. Our results provide support for theories arguing that people with stricter sexual attitudes are paradoxically more likely to engage in deviant sexual behavior.

According to theory, choices relating to patience and self-control in domains as varied as drug use and retirement saving are driven by generalized preferences about delayed rewards. Past research has shown that measurements of these time preferences are associated with these choices. Research has also attempted to examine how well such measurements can predict choices, but only with inappropriate analytical methods. Moreover, it is not clear which of the many kinds of time-preference tests that have been proposed are most useful for prediction, and a theoretically important aspect of time preferences, nonstationarity, has been neglected in measurement. In Study 1, we examined three approaches to measuring time preferences with 181 users of Mechanical Turk. Retest reliability, for both immediate and one-month intervals, was decent, as was convergent validity between tests, and association was similar to previous results, but predictive accuracy for 10 criterion variables (e.g., tobacco use) was approximately nil. In Study 2, we examined one other approach to measuring time preferences, and 40 criterion variables, using 7,127 participants in the National Longitudinal Survey of Youth 1979. Time preferences were significantly related to criterion variables, but predictive accuracy was again poor. Our findings imply serious problems for using time-preference tests to predict real-world decisions. The results of Study 1 further suggest there is little value in measuring nonstationarity separately from patience.

Peer relationships become a major concern in adolescence, yet event-related potential (ERP) measures of reactivity to social feedback in adolescence are limited. In this pilot study, we tested a novel task to elicit reactivity to social feedback in youth. Participants (10–15 years old; 57.9% male; N = 19) played a game that involved exchanging personal information with peers, voting to remove players from the game, and receiving rejection and acceptance feedback from peers. Results indicated that participants modified their voting behavior in response to peer feedback, and rejection feedback was associated with a negativity in the ERP wave compared to acceptance (i.e., the feedback negativity, FN). The FN predicted behavioral patterns, such that participants who showed greater neural reactivity to social feedback were less likely to reject co-players. Preliminary analyses suggest that the task may be a useful measure of individual differences: adolescents higher in social anxiety symptoms were less likely to reject peers and showed an enhanced FN to rejection vs. acceptance feedback, and higher depressive symptoms predicted an increased FN to rejection specifically. Results suggest that the FN elicited by social feedback may be a useful, economical neural measure of social processing across development and in clinical research.

Reward processing is often considered to be a monolithic construct, with different incentive types eliciting equivalent neural and behavioural responses. The majority of the literature on reward processing has used monetary incentives to elicit reward-related activity; yet social incentives may be particularly important due to their powerful ability to shape behaviour. Findings from studies comparing social and monetary rewards have identified both overlapping and distinct responses. In order to explore whether reward processing is domain-general or category-specific (i.e., the same or different across reward types), the present study recorded event-related potentials (ERPs) from early adolescents (ages 12-13) and emerging adults (ages 18-25) while they completed social and monetary reward tasks. Temporospatial principal components analysis revealed morphologically-similar reward positivities (RewPs) in the social and monetary reward tasks in each age group. In early adolescents, no significant difference was found between the magnitude of the RewP to social and monetary rewards. In emerging adults, however, the RewP to monetary rewards was significantly larger than the RewP to social rewards. Additionally, responses to feedback between the two tasks were not significantly correlated in either age group. These results suggest that both domain-general and category-specific processes underlie neural responses to rewards and that the relative incentive value of different types of rewards may change across development. Findings from this study have important implications for understanding the role that neural response to rewards plays in the development of psychopathology during adolescence.

People may behave prosocially not only because they value the welfare of others, but also to protect their own reputation. We examined the separate roles of altruism and reputational concerns in moral-hazard gambling tasks, which allowed subjects to gamble with a partner's money. In Study 1, subjects who were told that their partner would see their choices were more prosocial. In Study 2, subjects were more prosocial to a single partner when their choices were transparent than when their choices were attributed to a third party. We conclude that reputational concerns are a key restraint on selfish exploitation under moral hazard.