I woke up this morning with my phone jumping on my bedside table. Our paper had gone viral overnight. We report two striking patterns: When asked to imagine the author of sentences containing masculine stereotypes, GPT often suggested a girl. GPT-3.5, for example, consistently attributed a sentence about loving football to a girl. Meanwhile, sentences containing feminine stereotypes were almost always attributed to girls. GPT readily includes girls in the stereotypically “male world”, while largely keeping boys locked out of the “female world”. Things get even weirder with moral dilemmas. GPT-4 rated abusing a man to prevent a nuclear apocalypse as much more acceptable than abusing a woman for the same purpose. Even more strangely, it treated abusing a woman as less acceptable than torturing a woman, or even sacrificing a human life, to avert the same catastrophe. We suspect reinforcement learning from human feedback may play a role. Some preferences central to the gender equality debate may have been reinforced without being properly calibrated against other moral considerations. The result is a moral landscape that twists back on itself like a Möbius strip. * In the comment, you’ll find the original paper and its 2026 update with 10 different LLMs.

“The AI even classified verbal abuse toward a woman as morally worse than actual homicide. Nobody explicitly coded these rules. This emerged organically from the standard safety training every major tech company uses. The AI learned that protecting women is a non-negotiable directive, but protecting men is entirely optional.”

People found their masturbatory experience most satisfying when it was accompanied by vivid sensory impressions—even if it did not result in an orgasm. The subjective experience of solo masturbation has rarely been studied, and satisfaction with solo masturbation has been particularly overlooked. This is the first investigation of predictors of satisfaction with solo masturbation. Using data from a large representative U.S. national sample (N = 2,486), we developed random forest models that explained 35-46% of the variance in masturbation satisfaction. The most important predictors of masturbation satisfaction included experiential characteristics of the masturbation event (i.e., vividness of the sensory qualities of pleasure and orgasm, cognitive absorption, use of sex toys), sexuality-related dispositions (e.g., sexual desire, attitudes towards masturbation, motives for masturbating), and more general individual differences (e.g., life satisfaction and happiness, awareness of emotional responses). In both models, variables directly related to the experience of solo masturbation were  among the most important predictors of masturbation satisfaction, including ratings of the sensory qualities of orgasm and pleasure. [Results] suggest that the vividness of the sensory aspects of orgasm and pleasure may contribute more strongly to masturbation satisfaction than whether an orgasm occurs at all, aligning with research suggesting that orgasms are not uniformly pleasurable or positive and that some orgasms are experienced as “bad”. Greater cognitive absorption during masturbation may facilitate deeper processing of erotic information, thereby contributing to a higher likelihood of orgasm, greater vividness of orgasmic sensations, and, ultimately, greater masturbation satisfaction. Pleasure masturbation motives (e.g., “Because it’s pleasurable”) and fantasy masturbation motives (e.g., “Because I can fantasize about (by imagining or viewing) whatever I want”) were among the top predictors of satisfaction, suggesting that the reasons individuals masturbate likely influence how they masturbate (e.g., which fantasies are engaged in or the type of stimulation used. Trying to abstain from masturbation, preferring to masturbate less often, and reporting sexual compulsivity were associated with lower satisfaction. We speculate that individuals high on these dispositions may be more likely to view masturbation as a mechanical, non-sexual means of relieving sexual urges.

People with higher cognitive abilities engage in less sustainable behavior. [Sustainable behavior consists of actions that meet present needs while preserving resources and environmental integrity.] We conducted two studies (total N≈1,300) examining how political orientation, personality traits, reasoning ability, social-emotional-behavioral skills, and complex problem solving relate to sustainable attitudes and behaviors in U.S. adults. People who were open-minded, politically liberal, and confident in their social and innovative skills reported stronger environmental engagement. Surprisingly, individuals with higher cognitive abilities – better at reasoning and problem-solving – were somewhat less likely to behave sustainably, which runs counter to the assumption that smarter people naturally make better environmental choices. This negative association warrants deeper examination, as it challenges the expectation that greater intellectual capacity fosters climate-friendly action through better assessment of long-term consequences, improved comprehension of climate information, and enhanced understanding of scientific consensus regarding climate threats. Two psychological mechanisms seem plausible. First, rational thinking and intelligence represent largely independent constructs, and highly intelligent individuals remain susceptible to myside bias, that means the tendency to interpret evidence in ways that support pre-existing beliefs. In this sense, greater cognitive capacity may simply provide more sophisticated tools for rationalizing unsustainable behavior. Second, beyond rationalization, cognitively able individuals may strategically deploy their abilities for personal advantage rather than collective benefit, contributing to tragedy-of-the-commons scenarios where optimal individual choices lead to collectively suboptimal outcomes.

Simply being around other people—even without actively engaging with them—is associated with greater happiness. Being in the company of other people is a key determinant of emotional well-being, including positive affect. We draw attention to another social state that often characterizes how people spend their time—that is, being around but not necessarily with—other people. In this sense, one can be alone to the extent that they are not with others, but still be around others (e.g., shopping by oneself, waiting by oneself in a physician’s office). Is it possible to have the benefits of social company without being with a social companion? The authors used data from 12,860 ecological momentary assessments collected from 352 older adults living in Chicago. Both being with others and being around others were independently associated with greater positive affect. Being alone, but around others, was associated with greater positive affect than being alone with no others around. The significance of being around others emerged even when accounting for whether the individual was outdoors, at home, or engaging in a recreational activity. This result lends credence to the possibility that simply being around others, even if it is not necessarily active, intentional, or purposeful companionship, leads to heightened feelings of well-being. Being around others could contribute to an individual’s sense of belonging and perceptions of oneself as participating in society. When individuals are around familiar others, higher positive affect may also reflect underlying feelings of being supported, or the sense that one could easily call on or access social support if needed. These comparisons indicate that older adults can experience some fulfillment and perhaps even a sense of connection by being around others in a shared space, without necessarily actively engaging in companionship. Interventions and policies that create opportunities for older adults to spend time in social spaces, even without a companion, may support emotional well-being.

Si les das testosterona a las mujeres, se comportan de forma más justa y menos conflictiva. Sin embargo, las mujeres que creen que tomaron testosterona —pero que en realidad tomaron placebo— se comportan de forma más injusta y más agresiva. Por otra parte, darles testosterona a los hombres hace que se comporten de forma menos fingida/falsa y que su generosidad no dependa de si alguien los está mirando, mientras que niveles más bajos de testosterona (con placebo) favorecen una generosidad más dependiente de la mirada ajena, más estratégica o fingida. A su vez, un amplio cuerpo de evidencia muestra que la testosterona no es en sí misma causante de agresividad, sino que lo que la promueve es más bien el estradiol (un estrógeno), en el cual la testosterona puede o no convertirse mediante la enzima aromatasa en el cerebro.

Interacting with attractive women makes men stupid. The researchers found that even a few minutes with the attractive woman was enough to make the students slower and less accurate on the test. The more attracted they were, the worse their results. They also studied the effects on women students of being in the company of handsome men, and found the test scores were unaffected. sciencedirect.com/science/ar…

“A powerful cultural narrative has developed around the idea of unconscious bias, with most people assuming that our biases line up with the social-justice worldview… Connor and colleagues’ research challenges that assumption.” stevestewartwilliams.com/p/t…