Post-January 6th deplatforming reduced the reach of misinformation on Twitter.

The social media platforms of the twenty-first century have an enormous role in regulating speech in the USA and worldwide1. However, there has been little research on platform-wide interventions on speech2,3. Here we evaluate the effect of the decision by Twitter to suddenly deplatform 70,000 misinformation traffickers in response to the violence at the US Capitol on 6 January 2021 (a series of events commonly known as and referred to here as 'January 6th'). Using a panel of more than 500,000 active Twitter users4,5 and natural experimental designs6,7, we evaluate the effects of this intervention on the circulation of misinformation on Twitter. We show that the intervention reduced circulation of misinformation by the deplatformed users as well as by those who followed the deplatformed users, though we cannot identify the magnitude of the causal estimates owing to the co-occurrence of the deplatforming intervention with the events surrounding January 6th. We also find that many of the misinformation traffickers who were not deplatformed left Twitter following the intervention. The results inform the historical record surrounding the insurrection, a momentous event in US history, and indicate the capacity of social media platforms to control the circulation of misinformation, and more generally to regulate public discourse.

Hybrid working from home improves retention without damaging performance.

Working from home has become standard for employees with a university degree. The most common scheme, which has been adopted by around 100 million employees in Europe and North America, is a hybrid schedule, in which individuals spend a mix of days at home and at work each week1,2. However, the effects of hybrid working on employees and firms have been debated, and some executives argue that it damages productivity, innovation and career development3-5. Here we ran a six-month randomized control trial investigating the effects of hybrid working from home on 1,612 employees in a Chinese technology company in 2021-2022. We found that hybrid working improved job satisfaction and reduced quit rates by one-third. The reduction in quit rates was significant for non-managers, female employees and those with long commutes. Null equivalence tests showed that hybrid working did not affect performance grades over the next two years of reviews. We found no evidence for a difference in promotions over the next two years overall, or for any major employee subgroup. Finally, null equivalence tests showed that hybrid working had no effect on the lines of code written by computer-engineer employees. We also found that the 395 managers in the experiment revised their surveyed views about the effect of hybrid working on productivity, from a perceived negative effect (-2.6% on average) before the experiment to a perceived positive one (+1.0%) after the experiment. These results indicate that a hybrid schedule with two days a week working from home does not damage performance.

Artificial intelligence and illusions of understanding in scientific research.

Scientists are enthusiastically imagining ways in which artificial intelligence (AI) tools might improve research. Why are AI tools so attractive and what are the risks of implementing them across the research pipeline? Here we develop a taxonomy of scientists' visions for AI, observing that their appeal comes from promises to improve productivity and objectivity by overcoming human shortcomings. But proposed AI solutions can also exploit our cognitive limitations, making us vulnerable to illusions of understanding in which we believe we understand more about the world than we actually do. Such illusions obscure the scientific community's ability to see the formation of scientific monocultures, in which some types of methods, questions and viewpoints come to dominate alternative approaches, making science less innovative and more vulnerable to errors. The proliferation of AI tools in science risks introducing a phase of scientific enquiry in which we produce more but understand less. By analysing the appeal of these tools, we provide a framework for advancing discussions of responsible knowledge production in the age of AI.

Online images amplify gender bias.

Each year, people spend less time reading and more time viewing images1, which are proliferating online2-4. Images from platforms such as Google and Wikipedia are downloaded by millions every day2,5,6, and millions more are interacting through social media, such as Instagram and TikTok, that primarily consist of exchanging visual content. In parallel, news agencies and digital advertisers are increasingly capturing attention online through the use of images7,8, which people process more quickly, implicitly and memorably than text9-12. Here we show that the rise of images online significantly exacerbates gender bias, both in its statistical prevalence and its psychological impact. We examine the gender associations of 3,495 social categories (such as 'nurse' or 'banker') in more than one million images from Google, Wikipedia and Internet Movie Database (IMDb), and in billions of words from these platforms. We find that gender bias is consistently more prevalent in images than text for both female- and male-typed categories. We also show that the documented underrepresentation of women online13-18 is substantially worse in images than in text, public opinion and US census data. Finally, we conducted a nationally representative, preregistered experiment that shows that googling for images rather than textual descriptions of occupations amplifies gender bias in participants' beliefs. Addressing the societal effect of this large-scale shift towards visual communication will be essential for developing a fair and inclusive future for the internet.

The cortical amygdala consolidates a socially transmitted long-term memory.

Social communication guides decision-making, which is essential for survival. Social transmission of food preference (STFP) is an ecologically relevant memory paradigm in which an animal learns a desirable food odour from another animal in a social context, creating a long-term memory1,2. How food-preference memory is acquired, consolidated and stored is unclear. Here we show that the posteromedial nucleus of the cortical amygdala (COApm) serves as a computational centre in long-term STFP memory consolidation by integrating social and sensory olfactory inputs. Blocking synaptic signalling by the COApm-based circuit selectively abolished STFP memory consolidation without impairing memory acquisition, storage or recall. COApm-mediated STFP memory consolidation depends on synaptic inputs from the accessory olfactory bulb and on synaptic outputs to the anterior olfactory nucleus. STFP memory consolidation requires protein synthesis, suggesting a gene-expression mechanism. Deep single-cell and spatially resolved transcriptomics revealed robust but distinct gene-expression signatures induced by STFP memory formation in the COApm that are consistent with synapse restructuring. Our data thus define a neural circuit for the consolidation of a socially communicated long-term memory, thereby mechanistically distinguishing protein-synthesis-dependent memory consolidation from memory acquisition, storage or retrieval.

Neural signatures of natural behaviour in socializing macaques.

Our understanding of the neurobiology of primate behaviour largely derives from artificial tasks in highly controlled laboratory settings, overlooking most natural behaviours that primate brains evolved to produce1-3. How primates navigate the multidimensional social relationships that structure daily life4 and shape survival and reproductive success5 remains largely unclear at the single-neuron level. Here we combine ethological analysis, computer vision and wireless recording technologies to identify neural signatures of natural behaviour in unrestrained, socially interacting pairs of rhesus macaques. Single-neuron and population activity in the prefrontal and temporal cortex robustly encoded 24 species-typical behaviours, as well as social context. Male-female partners demonstrated near-perfect reciprocity in grooming, a key behavioural mechanism supporting friendships and alliances6, and neural activity maintained a running account of these social investments. Confronted with an aggressive intruder, behavioural and neural population responses reflected empathy and were buffered by the presence of a partner. Our findings reveal a highly distributed neurophysiological ledger of social dynamics, a potential computational foundation supporting communal life in primate societies, including our own.

From compulsivity to compulsion: the neural basis of compulsive disorders.

Compulsive behaviour, an apparently irrational perseveration in often maladaptive acts, is a potential transdiagnostic symptom of several neuropsychiatric disorders, including obsessive-compulsive disorder and addiction, and may reflect the severe manifestation of a dimensional trait termed compulsivity. In this Review, we examine the psychological basis of compulsions and compulsivity and their underlying neural circuitry using evidence from human neuroimaging and animal models. Several main elements of this circuitry are identified, focused on fronto-striatal systems implicated in goal-directed behaviour and habits. These systems include the orbitofrontal, prefrontal, anterior cingulate and insular cortices and their connections with the basal ganglia as well as sensoriomotor and parietal cortices and cerebellum. We also consider the implications for future classification of impulsive-compulsive disorders and their treatment.

Unraveling how the third-party brain under stress responds to injustices.

How third-party individuals respond to injustices is important for resolving conflict in society. A study in PLOS Biology shows that individuals experiencing acute stress prefer to aid victims over punishing offenders, an opposite pattern to non-stress conditions.

Rationalizing risk aversion in science: Why incentives to work hard clash with incentives to take risks.

Scientific research requires taking risks, as the most cautious approaches are unlikely to lead to the most rapid progress. Yet, much funded scientific research plays it safe and funding agencies bemoan the difficulty of attracting high-risk, high-return research projects. Why don't the incentives for scientific discovery adequately impel researchers toward such projects? Here, we adapt an economic contracting model to explore how the unobservability of risk and effort discourages risky research. The model considers a hidden-action problem, in which the scientific community must reward discoveries in a way that encourages effort and risk-taking while simultaneously protecting researchers' livelihoods against the vicissitudes of scientific chance. Its challenge when doing so is that incentives to motivate effort clash with incentives to motivate risk-taking, because a failed project may be evidence of a risky undertaking but could also be the result of simple sloth. As a result, the incentives needed to encourage effort actively discourage risk-taking. Scientists respond by working on safe projects that generate evidence of effort but that don't move science forward as rapidly as riskier projects would. A social planner who prizes scientific productivity above researchers' well-being could remedy the problem by rewarding major discoveries richly enough to induce high-risk research, but in doing so would expose scientists to a degree of livelihood risk that ultimately leaves them worse off. Because the scientific community is approximately self-governing and constructs its own reward schedule, the incentives that researchers are willing to impose on themselves are inadequate to motivate the scientific risks that would best expedite scientific progress.

Acute stress during witnessing injustice shifts third-party interventions from punishing the perpetrator to helping the victim.

People tend to intervene in others' injustices by either punishing the transgressor or helping the victim. Injustice events often occur under stressful circumstances. However, how acute stress affects a third party's intervention in injustice events remains open. Here, we show a stress-induced shift in third parties' willingness to engage in help instead of punishment by acting on emotional salience and central-executive and theory-of-mind networks. Acute stress decreased the third party's willingness to punish the violator and the severity of the punishment and increased their willingness to help the victim. Computational modeling revealed a shift in preference of justice recovery from punishment the offender toward help the victim under stress. This finding is consistent with the increased dorsolateral prefrontal engagement observed with higher amygdala activity and greater connectivity with the ventromedial prefrontal cortex in the stress group. A brain connectivity theory-of-mind network predicted stress-induced justice recovery in punishment. Our findings suggest a neurocomputational mechanism of how acute stress reshapes third parties' decisions by reallocating neural resources in emotional, executive, and mentalizing networks to inhibit punishment bias and decrease punishment severity.

State-dependent connectivity in auditory-reward networks predicts peak pleasure experiences to music.

Music can evoke pleasurable and rewarding experiences. Past studies that examined task-related brain activity revealed individual differences in musical reward sensitivity traits and linked them to interactions between the auditory and reward systems. However, state-dependent fluctuations in spontaneous neural activity in relation to music-driven rewarding experiences have not been studied. Here, we used functional MRI to examine whether the coupling of auditory-reward networks during a silent period immediately before music listening can predict the degree of musical rewarding experience of human participants (N = 49). We used machine learning models and showed that the functional connectivity between auditory and reward networks, but not others, could robustly predict subjective, physiological, and neurobiological aspects of the strong musical reward of chills. Specifically, the right auditory cortex-striatum/orbitofrontal connections predicted the reported duration of chills and the activation level of nucleus accumbens and insula, whereas the auditory-amygdala connection was associated with psychophysiological arousal. Furthermore, the predictive model derived from the first sample of individuals was generalized in an independent dataset using different music samples. The generalization was successful only for state-like, pre-listening functional connectivity but not for stable, intrinsic functional connectivity. The current study reveals the critical role of sensory-reward connectivity in pre-task brain state in modulating subsequent rewarding experience.

The politics of allyship: Multiethnic coalitions and mass attitudes toward protest.

Recent work finds that nonviolent resistance by ethnic minorities is perceived as more violent and requiring more policing than identical resistance by ethnic majorities, reducing its impact and effectiveness. We ask whether allies-advantaged group participants in disadvantaged group movements-can mitigate these barriers. On the one hand, allies can counter negative stereotypes and defuse threat perceptions among advantaged group members, while raising expectations of success and lowering expected risks among disadvantaged group members. On the other hand, allies can entail significant costs, carrying risks of cooptation, replication of power hierarchies, and marginalization of core constituencies. To shed light on this question we draw on the case of the Black Lives Matter (BLM) movement, which, in 2020, attracted unprecedented White participation. Employing a national survey experiment, we find that sizeable White presence at racial justice protests increases protest approval, reduces perceptions of violence, and raises the likelihood of participation among White audiences, while not causing significant backlash among Black audiences. Black respondents mostly see White presence as useful for advancing the movement's goals, and predominant White presence reduces expectations that protests will be forcefully repressed. We complement these results with analysis of tens of thousands of images shared on social media during the 2020 BLM protests, finding a significant association between the presence of Whites in the images and user engagement and amplification. The findings suggest that allyship can be a powerful tool for promoting sociopolitical change amid deep structural inequality.

It matters what you see: Graphic media images of war and terror may amplify distress.

Media exposure to graphic images of violence has proliferated in contemporary society, particularly with the advent of social media. Extensive exposure to media coverage immediately after the 9/11 attacks and the Boston Marathon bombings (BMB) was associated with more early traumatic stress symptoms; in fact, several hours of BMB-related daily media exposure was a stronger correlate of distress than being directly exposed to the bombings themselves. Researchers have replicated these findings across different traumatic events, extending this work to document that exposure to graphic images is independently and significantly associated with stress symptoms and poorer functioning. The media exposure-distress association also appears to be cyclical over time, with increased exposure predicting greater distress and greater distress predicting more media exposure following subsequent tragedies. The war in Israel and Gaza, which began on October 7, 2023, provides a current, real-time context to further explore these issues as journalists often share graphic images of death and destruction, making media-based graphic images once again ubiquitous and potentially challenging public well-being. For individuals sharing an identity with the victims or otherwise feeling emotionally connected to the Middle East, it may be difficult to avoid viewing these images. Through a review of research on the association between exposure to graphic images and public health, we discuss differing views on the societal implications of viewing such images and advocate for media literacy campaigns to educate the public to identify mis/disinformation and understand the risks of viewing and sharing graphic images with others.

The Black Lives Matter movement mitigates bias against racial minority actors.

Watching movies is among the most popular entertainment and cultural activities. How do viewers react when a movie sequel increases racial minority actors in the main cast ("minority increase")? On the one hand, such sequels may receive better evaluations if viewers appreciate racially inclusive casting for its novel elements (the value-in-diversity perspective) and moral appeal (the fairness perspective on diversity). On the other hand, discrimination research suggests that if viewers harbor biases against racial minorities, sequels with minority increase may receive worse evaluations. To examine these competing possibilities, we analyze a unique panel dataset of movie series released from 1998 to 2021 and conduct text analysis of 312,457 reviews of these movies. Consistent with discrimination research, we find that movies with minority increase receive lower ratings and more toxic reviews. Importantly, these effects weaken after the advent of the Black Lives Matter (BLM) movement, especially when the movement's intensity is high. These results are reliable across various robustness checks (e.g., propensity score matching, random implementation test). We conceptually replicate the bias mitigation effect of BLM in a preregistered experiment: Heightening the salience of BLM increases White individuals' acceptance of racial minority increase in a movie sequel. This research demonstrates the power of social movements in fostering diversity, equality, and inclusion.

A leadership-level culture cycle intervention changes teachers' culturally inclusive beliefs and practices.

Despite an abundance of support for culturally inclusive learning environments, there is little consensus regarding how to change educational contexts to effectively and sustainably foster cultural inclusion. To address this gap, we report findings from a research-practice partnership that leveraged the Culture Cycle Framework (CCF) to expand educators' praxis to include both independent and interdependent models of self. Most U.S. schools validate independent cultural models (i.e., those that prioritize individuality, uniqueness, and personal agency) and overlook interdependent models (i.e., those that prioritize connectedness, relationality, and collective well-being), which are more common among students from marginalized racial and socioeconomic backgrounds. Using a quasi-experimental longitudinal design, we trained school leadership to integrate ideas about cultural inclusion (i.e., validating the importance of both independent and interdependent cultural models) into school-wide flagship practices. We assessed downstream indicators of culture change by surveying teachers and students across the district and found that a) leadership-level training enhanced school-wide beliefs about cultural inclusion, b) teachers' endorsement of culturally inclusive beliefs predicted their use of culturally inclusive practices, and c) teachers' use of culturally inclusive practices predicted enhanced psychosocial and academic outcomes among students. This research represents a comprehensive culture change effort using the CCF and illustrates a means of fostering inclusion-focused educational culture change and assessing downstream consequences of culture change initiatives.

Key language markers of depression on social media depend on race.

Depression has robust natural language correlates and can increasingly be measured in language using predictive models. However, despite evidence that language use varies as a function of individual demographic features (e.g., age, gender), previous work has not systematically examined whether and how depression's association with language varies by race. We examine how race moderates the relationship between language features (i.e., first-person pronouns and negative emotions) from social media posts and self-reported depression, in a matched sample of Black and White English speakers in the United States. Our findings reveal moderating effects of race: While depression severity predicts I-usage in White individuals, it does not in Black individuals. White individuals use more belongingness and self-deprecation-related negative emotions. Machine learning models trained on similar amounts of data to predict depression severity performed poorly when tested on Black individuals, even when they were trained exclusively using the language of Black individuals. In contrast, analogous models tested on White individuals performed relatively well. Our study reveals surprising race-based differences in the expression of depression in natural language and highlights the need to understand these effects better, especially before language-based models for detecting psychological phenomena are integrated into clinical practice.

Evaluating language models for mathematics through interactions.

There is much excitement about the opportunity to harness the power of large language models (LLMs) when building problem-solving assistants. However, the standard methodology of evaluating LLMs relies on static pairs of inputs and outputs; this is insufficient for making an informed decision about which LLMs are best to use in an interactive setting, and how that varies by setting. Static assessment therefore limits how we understand language model capabilities. We introduce CheckMate, an adaptable prototype platform for humans to interact with and evaluate LLMs. We conduct a study with CheckMate to evaluate three language models (InstructGPT, ChatGPT, and GPT-4) as assistants in proving undergraduate-level mathematics, with a mixed cohort of participants from undergraduate students to professors of mathematics. We release the resulting interaction and rating dataset, MathConverse. By analyzing MathConverse, we derive a taxonomy of human query behaviors and uncover that despite a generally positive correlation, there are notable instances of divergence between correctness and perceived helpfulness in LLM generations, among other findings. Further, we garner a more granular understanding of GPT-4 mathematical problem-solving through a series of case studies, contributed by experienced mathematicians. We conclude with actionable takeaways for ML practitioners and mathematicians: models that communicate uncertainty, respond well to user corrections, and can provide a concise rationale for their recommendations, may constitute better assistants. Humans should inspect LLM output carefully given their current shortcomings and potential for surprising fallibility.

The role of positive emotion in harmful health behavior: Implications for theory and public health campaigns.

Meta-analyses have concluded that positive emotions do not reduce appetitive risk behaviors (risky behaviors that fulfill appetitive or craving states, such as smoking and excessive alcohol use). We propose that this conclusion is premature. Drawing on the Appraisal Tendency Framework and related theories of emotion and decision-making, we hypothesized that gratitude (a positive emotion) can decrease cigarette smoking, a leading cause of premature death globally. A series of multimethod studies provided evidence supporting our hypothesis (collective N = 34,222). Using nationally representative US samples and an international sample drawn from 87 countries, Studies 1 and 2 revealed that gratitude was inversely associated with likelihood of smoking, even after accounting for numerous covariates. Other positive emotions (e.g., compassion) lacked such consistent associations, as expected. Study 3, and its replication, provided further support for emotion specificity: Experimental induction of gratitude, unlike compassion or sadness, reduced cigarette craving compared to a neutral state. Study 4, and its replication, showed that inducing gratitude causally increased smoking cessation behavior, as evidenced by enrollment in a web-based cessation intervention. Self-reported gratitude mediated the effects in both experimental studies. Finally, Study 5 found that current antismoking messaging campaigns by the US Centers for Disease Control and Prevention primarily evoked sadness and compassion, but seldom gratitude. Together, our studies advance understanding of positive emotion effects on appetitive risk behaviors; they also offer practical implications for the design of public health campaigns.

Bodily maps of musical sensations across cultures.

Emotions, bodily sensations and movement are integral parts of musical experiences. Yet, it remains unknown i) whether emotional connotations and structural features of music elicit discrete bodily sensations and ii) whether these sensations are culturally consistent. We addressed these questions in a cross-cultural study with Western (European and North American, n = 903) and East Asian (Chinese, n = 1035). We precented participants with silhouettes of human bodies and asked them to indicate the bodily regions whose activity they felt changing while listening to Western and Asian musical pieces with varying emotional and acoustic qualities. The resulting bodily sensation maps (BSMs) varied as a function of the emotional qualities of the songs, particularly in the limb, chest, and head regions. Music-induced emotions and corresponding BSMs were replicable across Western and East Asian subjects. The BSMs clustered similarly across cultures, and cluster structures were similar for BSMs and self-reports of emotional experience. The acoustic and structural features of music were consistently associated with the emotion ratings and music-induced bodily sensations across cultures. These results highlight the importance of subjective bodily experience in music-induced emotions and demonstrate consistent associations between musical features, music-induced emotions, and bodily sensations across distant cultures.

Americans harbor much less favorable explicit sentiments toward young adults than toward older adults.

Public and academic discourse on ageism focuses primarily on prejudices targeting older adults, implicitly assuming that this age group experiences the most age bias. We test this assumption in a large, preregistered study surveying Americans' explicit sentiments toward young, middle-aged, and older adults. Contrary to certain expectations about the scope and nature of ageism, responses from two crowdsourced online samples matched to the US adult population (N = 1,820) revealed that older adults garner the most favorable sentiments and young adults, the least favorable ones. This pattern held across a wide range of participant demographics and outcome variables, in both samples. Signaling derogation of young adults more than benign liking of older adults, participants high on SDO (i.e., a key antecedent of group prejudice) expressed even less favorable sentiments toward young adults-and more favorable ones toward older adults. In two follow-up, preregistered, forecasting surveys, lay participants (N = 500) were generally quite accurate at predicting these results; in contrast, social scientists (N = 241) underestimated how unfavorably respondents viewed young adults and how favorably they viewed older adults. In fact, the more expertise in ageism scientists had, the more biased their forecasts. In a rapidly aging world with exacerbated concerns over older adults' welfare, young adults also face increasing economic, social, political, and ecological hardship. Our findings highlight the need for policymakers and social scientists to broaden their understanding of age biases and develop theory and policies that ponder discriminations targeting all age groups.