Nature Human Behaviour最新文献

Dishonest behaviours such as tax evasion impose significant societal costs. Ex ante honesty oaths—commitments to honesty before action—have been proposed as interventions to counteract dishonest behaviour, but the heterogeneity in findings across operationalizations calls their effectiveness into question. We tested 21 honesty oaths (including a baseline oath)—proposed, evaluated and selected by 44 expert researchers—and a no-oath condition in a megastudy involving 21,506 UK and US participants from Prolific.com who played an incentivized tax evasion game online. Of the 21 interventions, 10 significantly improved tax compliance by 4.5 to 8.5 percentage points, with the most successful nearly halving tax evasion. Limited evidence for moderators was found. Experts and laypeople failed to predict the most effective interventions, though experts’ predictions were more accurate. In conclusion, honesty oaths were effective in curbing dishonesty, but their effectiveness varied depending on content. These findings can help design impactful interventions to curb dishonesty.

Public concerns about the societal effects of generative artificial intelligence (AI) are shaped by narratives that have the potential to influence research priorities and policy agendas. Understanding the origins and dynamics of these narratives is crucial to effectively address the actual impacts of AI and ensure a constructive discourse about its risks and potential.

This shift in media coverage points to the need for a closer examination of the underlying discourse. We currently see four main types of narratives around generative AI:

1. (1)

   The ‘existential risk’ narrative contends that existential risks from artificial superintelligence or artificial general intelligence could stem from the next generations of generative AI-type systems. As generative AI systems become more sophisticated, their capabilities could surpass human control and lead to potentially existentially catastrophic consequences. Strong versions of this narrative raise the concern that artificial superintelligence or artificial general intelligence technologies could lead to human extinction³.

2. (2)

   The ‘effective accelerationist’ narrative champions the rapid development of AI. Proponents argue that its potential benefits for solving complex global problems far outweigh the risks, and the existential risks from advanced AI are zero or near zero and so can be dismissed⁴. This narrative is driven by a strong belief in the power of AI progress to bring about substantial positive change.

3. (3)

   The ‘real, immediate societal risks’ narrative focuses only on the tangible, immediate societal risks of generative AI. It emphasizes issues such as the creation of deepfake pornography, unjust capability distribution or the growing environmental effects of generative AI, and argues that these present-day concerns are much more pressing and relevant than speculative existential risks. Proponents of this view argue that focusing on distant existential threats distracts us from addressing the real and present dangers of AI⁵.

4. (4)

   The ‘balanced risks’ narrative advocates for an approach to AI risk governance that acknowledges both the existential and immediate societal risks posed by AI. It encourages finding meaningful connections between these two classes of risks, and suggests that addressing them in tandem can lead to more comprehensive and effective risk mitigation strategies and policies⁶.

Executive function (EF) delays are well documented in paediatric neurodevelopmental conditions (NDCs). There is no consensus about whether EF delay represents a transdiagnostic feature of NDCs. This systematic review and meta-analysis synthesized 180 studies reporting two or more NDC comparisons on EF, examined differences between NDCs, and the moderating effects of gender, age, publication year, DSM editions and assessment types. Studies using established EF measures across seven domains (attention, fluency, set-shifting, set-switching, response inhibition, planning and working memory) in participants under 18 were included. Summary effects were compared: (1) for all reported NDCs relative to control, (2) for each individual NDC relative to control and (3) between NDC groups. Results confirmed that EF delay was a transdiagnostic feature of neurodevelopmental delay, with a moderate effect size of delay across all NDCs (g = 0.56, 95% confidence interval (CI) 0.49–0.63) compared with control. This effect increased with comorbidities (g = 0.72, 95% CI 0.59–0.86), DSM-5 criteria and informant measures. Comparisons between NDCs revealed few differences: children with tic disorders (TD) showed smaller EF delays, children with attention-deficit/hyperactivity disorder (ADHD) showed larger delays in attention, response inhibition, planning and working memory compared with TD and specific learning disorders, while children with autism spectrum disorders showed greater delays on set-switching compared with ADHD. Findings support transdiagnostic models of neurodevelopment to further a developmentally sensitive science that can reveal how EF delays contribute to brain circuitry, symptom profiles and functioning, and ultimately support early interventions and outcomes for all children with NDCs.

Maintaining COVID-19 vaccine demand was key to ending the global health emergency. To help do this, many governments used chatbots that provided personalized information guiding people on where, when and how to get vaccinated. We designed and tested a WhatsApp chatbot to understand whether two-way interactive messaging incorporating behaviourally informed functionalities could perform better than one-way message reminders. We ran a large-scale preregistered randomized controlled trial with 249,705 participants in Argentina, measuring vaccinations using Ministry of Health records. The behaviourally informed chatbot more than tripled COVID-19 vaccine uptake compared with the control group (a 1.6 percentage point increase (95% confidence interval, (1.36 pp, 1.77 pp)) and nearly doubled uptake compared with the one-way message reminder (a 1 percentage point increase (95% confidence interval, (0.83 pp, 1.17 pp)). Communications tools designed with behaviourally informed functionalities that simplify the vaccine user journey can increase vaccination more than traditional message reminders and may have applications to other health behaviours.

A sniff in humans typically lasts one to three seconds and is commonly considered to produce a long-exposure shot of the chemical environment that sets the temporal limit of olfactory perception. To break this limit, we devised a sniff-triggered apparatus that controls odorant deliveries within a sniff with a precision of 18 milliseconds. Using this apparatus, we show through rigorous psychophysical testing of 229 participants (649 sessions) that two odorants presented in one order and its reverse become perceptually discriminable when the stimulus onset asynchrony is merely 60 milliseconds (Cohen’s d = 0.48; 95% confidence interval, (55, 59); 120-millisecond difference). Discrimination performance improves with the length of stimulus onset asynchrony and is independent of explicit knowledge of the temporal order of odorants or the relative amount of odorant molecules accumulated in a sniff. Our findings demonstrate that human olfactory perception is sensitive to chemical dynamics within a single sniff and provide behavioural evidence for a temporal code of odour identity.