Short-term challenges resulting from colder temperature experienced by hominins were apparently countered via phenotypic adaptation toward larger bodies as a buffer mechanism; no such association of temperature with brain size

Different environmental variables predict body and brain size evolution in Homo. Manuel Will, Mario Krapp, Jay T. Stock & Andrea Manica. Nature Communications volume 12, Article number: 4116, Jul 8 2021. https://www.nature.com/articles/s41467-021-24290-7

Abstract: Increasing body and brain size constitutes a key macro-evolutionary pattern in the hominin lineage, yet the mechanisms behind these changes remain debated. Hypothesized drivers include environmental, demographic, social, dietary, and technological factors. Here we test the influence of environmental factors on the evolution of body and brain size in the genus Homo over the last one million years using a large fossil dataset combined with global paleoclimatic reconstructions and formalized hypotheses tested in a quantitative statistical framework. We identify temperature as a major predictor of body size variation within Homo, in accordance with Bergmann’s rule. In contrast, net primary productivity of environments and long-term variability in precipitation correlate with brain size but explain low amounts of the observed variation. These associations are likely due to an indirect environmental influence on cognitive abilities and extinction probabilities. Most environmental factors that we test do not correspond with body and brain size evolution, pointing towards complex scenarios which underlie the evolution of key biological characteristics in later Homo.

Discussion

Climatic fluctuations and ecological factors have frequently been proposed as potential drivers of brain and body size evolution within the hominin lineage8,9,10,11,12,23. This study presents the first systematic attempt to quantitatively test different environmental effects on body and brain size variation for the genus Homo during the past ~1 Ma. The climate variables we investigated are representative of the climatological mean (30-year averages) for each 1000-year period of the past ~1 Ma. Hence, climate oscillations on sub-millennial time scales, which might have had some impact on human body and brain size evolution, are not resolved, but such a finer resolution is also precluded by the inherently larger dating uncertainty of Pleistocene human fossils.

We found that MAT is uniformly associated with body size across Mid-Pleistocene Homo, Neanderthals, and Pleistocene H. sapiens. The extent of this relationship is greater than that estimated for modern humans in a recent study31. The direction of this association supports some of the predictions of the Environmental Stress Hypothesis, with temperature (i.e., thermal stress) being the key driver: larger body sizes are consistently found in colder regions, where both annual mean and mean coldest quarter temperature are lower. These findings fit the general expectations of Bergmann’s rule and are consistent with some—though not all33,38—previous studies on humans, hominins, and other animals8,10,22,31. Following this interpretation, short-term challenges resulting from colder temperature experienced by hominin populations (thermal stress) were apparently countered via phenotypic adaptation toward larger bodies as a buffer mechanism, either through natural selection, plasticity, or a combination of both. We failed to detect any effect of low rainfall or nutrient-poor environments as determinants of stress in our analyses.

Our analyses detected no such association of temperature with brain size. We did find relationships with the 10 ka-sigma of mean annual precipitation (MAP) and NPP, but the variance in brain size explained by these variables was small compared to the effect of MAT on body size. These results suggest that brain size within Homo is less influenced by environmental variables than body size during the past 1.0 Ma. Apart from other drivers being likely more relevant (see below), one factor contributing to the difficulty of detecting environmental effects lies in the strong performance of the null model (LM-T) based on taxonomic differences in brain size variations that explained much more variance (R2 = 0.47) compared to body size (R2 = 0.05). This being said, our analyses suggest that brain sizes tend to be higher in regions of low NPP and smaller in more productive regions, although this only holds for Mid-Pleistocene Homo but not for Neanderthals or Pleistocene H. sapiens. This negative correlation is not necessarily a direct effect of environments on human phenotype but can rather be interpreted as an indirect interaction of behavioral changes with environmental variables: regions with lower NPP feature more open steppe and grassland habitats with more frequent large mammals and particularly bovids (“productivity paradox”; ref. 39). As such, our findings can be related to changes in subsistence strategies toward more frequent and systematic hunting of larger-sized bovids in these environments, in association with cognitive changes toward more complex weapons and coordinated group activity. The lithic, faunal, and isotopic records show an increase of such behaviors and ecosystems inhabited by Homo throughout the Middle Pleistocene that supports this interpretation40,41,42,43. The divergent pattern in Neanderthals and Pleistocene H. sapiens might be due to an already higher established brain size close to the physiological maximum during colonization of more northern latitudes (>40°; H. sapiens: mean = 1505 cm3, n = 37; Neanderthals: mean = 1398 cm3; n = 25), while the other taxon either evolved in situ in these areas or had higher growth potential. More early African H. sapiens fossils are required to adequately test this interpretation.

Our fossil data show a relationship between long-term variation in rainfall (MAPvar10) and brain size that is of opposite sign than expected from the Environmental Variability Hypothesis11,36. Instead, this prediction is consistent with the Environmental Consistency Hypothesis: larger brain sizes occur in more stable environments across all studied Homo taxa. This result is likely an effect of brain growth being constrained by reduced resource availability and predictability over multi-millennial scales, acting as an extinction filter.

Our linear models did not find associations with 10-ka variability measures for other environmental variables in either body or brain size. We also failed to find support for the Environmental Constraints Hypothesis (Table 1). However, we need to be careful in interpreting these negative results. The fossil hominin record is scarce and patchy in space and time, confounding the ability to find patterns in our data26. We thus modeled and analyzed synthetic datasets to assess the degree to which the intrinsic nature of the fossil record biases and distorts associations of body and brain size with environmental variables. The power analysis shows that we should have been able to detect at least medium to strong associations between brain size and MAT, MAP, mean temperature of the coldest quarter, and mean precipitation of the driest quarter (Fig. 2). The synthetic data thus suggest that our negative results for these variables, and the lack of support for the Environmental Stress and the Environmental Constraints Hypothesis, are either “true negative” findings or that true effect sizes are relatively small. On the other hand, we had little power to detect associations of body and brain size with long-term climate variability (i.e., the consequences of the Environmental Variability and the Environmental Consistency Hypotheses), leaving them as potential targets for future analyses with even larger sample sizes.

There are several implications from our study for human evolution that point toward future analyses. Many standard models and recent accounts of the origins, bio-cultural evolution, and dispersal of our genus and species have invoked environmental drivers as prime movers9,44,45,46. Yet, necessary temporal correlations of paleoanthropological and archeological data with environmental information have been plagued by issues of resolution, scale, and data availability47. Using emulated global climate model data27, this study shows that different climatic variables predict human brain and body size evolution over the past 1 Ma. These findings have implications beyond human evolution. The scaling between body size and brain size is remarkably consistent across vertebrates, but increased variability in brain growth appears to underpin observed patterns of encephalization among birds and mammals48. Consistency in the observed patterns of encephalization within lineages is often attributed to developmental constraints that link the ontogenetic trajectories of brain and body size, although there is emerging evidence that deviations from the patterns found in mammals and primates may be driven by functional variation and different selective pressures49. Such adaptive mechanisms likely underpin the variation in brain development observed in Pleistocene hominins50. The demonstration that brain and body size evolution were influenced by different environmental factors supports this broader interpretation of unique selective pressures driving phenotypic diversification in the hominin lineage.

We also note that many of the environmental variables provided no detectable correlations and explained variance is often low, raising doubts about an unquestioned a priori reliance on environmental factors in explaining macro-processes in human evolution. There is a need for more quantitative tests of such hypotheses in explicitly formulated theoretical frameworks. Future work on these questions could (i) expand analyses on environmental drivers into the entire Pleistocene and Pliocene and (ii) examine other proposed drivers that are not tested here (see below). There are ample changes in the size of endocranial volumes and body mass between ~5 and 1 Ma among taxa of Ardipithecus, Australopithecus Paranthropus, and Homo that could be the result of climatic forcing and ecological adaptations, or yet other factors. This period also constitutes the focus of the original variability hypothesis11,36; however, the fossil record ~5–1 Ma has lower sample sizes per taxon and is patchier in time and space. While we gathered datasets for body and brain size back to 4.4 Ma6, we refrained from extending our analyses to this period as the current quality of the fossil data with the added uncertainty of climate models >1 Ma renders such studies more speculative. Improved paleoclimate models and new discoveries with good chronometric ages and taxonomic information will eventually allow for such studies.

In the meantime, testing other proposed drivers of human body and particularly brain size could be more fruitful. Inter-species competition and niche exclusion likely drove some of the observed significant differences in brain and body size between (sympatric) species of Homo (e.g., refs. 2,4,18), including shifts to larger social groups or communication networks driving further encephalization16. Archeologically established changes in subsistence patterns likely played a role as the nutritional basis allowing for the evolution of larger bodies and the maintenance of energetically costly brains14,15,51,52, and we have found indirect evidence to support this in our study. Yet the spatio-temporal trajectories and taxonomic associations of these behaviors in the archeological record are not well resolved. Finally, there is a long-standing debate about a feedback process between culture, cognition, and encephalization. Increased reliance on technology and material culture might have started a long-term directed evolutionary process selecting for advanced cognition and larger brains19,20,21, with greater detachment from direct environmental effects, particularly in H. sapiens. In parallel with brain size increases, stone tool technology showed major changes over the past 2 million years53 with an accelerated pace of cultural change by ~300 ka and again with the onset of the Upper Paleolithic and Later Stone Age17,54,55,56.

While many of these factors might have played a key role in body and/or brain size evolution, future models should include interacting components57,58 such as the co-evolution of changing environments, subsistence, and technology in driving brain evolution14,18,51,52,59. Such potential influences on hominin brain and body size need to be tested by formulating and testing explicit hypotheses with statistical analyses. This strategy requires innovative ways to translate the often qualitative archeological information into comparable quantitative data, potentially via machine learning methods. In this study, the support or falsification of certain environmental hypotheses to explain body and brain size changes among Homo in the past million years exemplify the usefulness of this approach.

Beauty, the feeling

Beauty, the feeling. Aenne A. Brielmann, Angelica Nuzzo, Denis G. Pelli. Acta Psychologica, Volume 219, September 2021, 103365. https://doi.org/10.1016/j.actpsy.2021.103365

Highlights

• Over 850 people report their feelings while experiencing or remembering beauty.

• Beauty experiences are intensely pleasurable, subjectively universal, and meaningful.

• Beauty expresses harmony in variety, exceeds expectation, and begs continuation.

• Beauty memories are active and social.

• Our findings are the first comprehensive, empirical description of beauty.

Abstract: Many philosophers and psychologists have made claims about what is felt in an experience of beauty. Here, we test how well these claims match the feelings that people report while looking at an image, or listening to music, or recalling a personal experience of beauty. We conducted ten experiments (total n = 851) spanning three nations (US, UK, and India). Across nations and modalities, top-rated beauty experiences are strongly characterized by six dimensions: intense pleasure, an impression of universality, the wish to continue the experience, exceeding expectation, perceived harmony in variety, and meaningfulness. Other frequently proposed beauty characteristics — like surprise, desire to understand, and mind wandering — are uncorrelated with feeling beauty. A typical remembered beautiful experience was active and social like a family holiday — hardly ever mentioning beauty — and only rarely mentioned art, unlike the academic emphasis, in aesthetics, on solitary viewing of art. Our survey aligns well with Kant and the psychological theories that emphasize pleasure, and reject theories that emphasize information seeking.

4. Discussion

4.1. Survey vs. philosophers: model explains 72% of variance

We identified 11 feeling-of-beauty dimensions in the writings of seven philosophers. Our philosophical mixed-effects model uses those dimensions to account for 72% of the variance in beauty ratings. Compared to our survey results, of the seven philosophers considered here, Kant's claims are the most correlated (r = 0.74), and have the most matches to the data: 10 out of 11 assessed dimensions (see Table 2). Kant's theory also states our participants' belief that beauty is found in nature rather than art and that both the object as well as its story contributes to beauty. Yet, contrary to Kant's theory, higher surprise is not associated with more intense beauty. However, as we will show below, there is a kind of surprise, i.e., the surprise of something exceeding one's expectation, that is indeed linked to beauty.

4.2. Survey vs. psychologists: model explains 70% of variance

We identified 10 feeling-of-beauty dimensions in the 25 papers and books (by 54 authors) on psychological theories of beauty that we cite here and in the supplement. Our psychological mixed-effects model uses 7 of those dimensions to account for 70% of the variance in beauty ratings. The ratings for images, music, and memories reveal a link to beauty for seven out of the eleven here-considered characteristics that psychologists claim are linked to beauty. In all of our surveys, intense beauty was associated with intense pleasure, as claimed by Fechner (1876). A strong link between beauty and being moved was also evident, as reported by Vessel et al. (2013). The notion that a positive prediction error contributes to beauty (Salimpoor et al., 2015) was also confirmed. Harmony in variety, the central beauty criterion in Diessner et al.'s (2018) theory was associated with beauty, too, and so was meaningfulness (see Leder & Nadal, 2014). We found mixed results regarding Berlyne's (1971) claims. While our results indeed suggest an inverted-u-shaped relation between subjective complexity and beauty, where medium complexity levels are associated with most intense beauty, we did not find such a relationship between beauty and arousal (assessed as excitement). Our music results reject Kivy's (1990) claim that musical beauty is linked to interest.

Information-seeking — i.e. learning, interest, and wanting to understand the experience— was not systematically linked to beauty in our studies, contrary to previous claims (Biederman & Vessel, 2006). Our results reject Reber et al.'s (2004) claim that ease of processing—i.e., felt understanding—of the experience is tied to its beauty, at least not in a way that is consciously accessible to people. The other cognitive dimension we included, learning, was also not associated with beauty, contrary to claims by Armstrong and Detweiler-Bedell's (2008). Taken together, these negative findings indicate that information-seeking is not important for experiencing beauty, at least not in their awareness.

One might suppose that depiction of an ugly object could only be beautiful by offering information, but we do not believe that the lack of correlation between beauty and self-reported information seeking prevents beautiful depiction of an ugly object. For instance, it seems possible that the mature Rembrandt, seen in person, might have been thought ugly. But his way of painting himself – the brush strokes, color-selection, and so on – may evoke beauty among those admiring his self-portrait.

It is also worth noting that our findings regarding people's explicit, self-reported feelings of learning progress do not exclude the possibility that beauty or pleasure serves as a signal of unconscious learning progress, e.g., in the sense of increasing long-term processing efficiency (Brielmann & Dayan, 2021).

4.3. Implications for the science of beauty

Our results, summarized in Table 5, provide an empirical characterization of the beauty experience. They will inform current theories of aesthetic appreciation (e.g., Leder & Nadal, 2014; Pelowski et al., 2017) and provide a first broad test of philosophy- and psychology-based theories of beauty. Our findings complement previous efforts that contrasted people's theoretical conceptions of beauty to other aesthetic evaluations (Menninghaus et al., 2019; see Supplementary Material for a quantitative comparison between their and our data).

Our findings are in line with the notion that beauty is a positive emotion, i.e., strongly correlated with pleasure (Armstrong & Detweiler-Bedell, 2008; Fechner, 1876) and being moved (Vessel et al., 2013). We did not find a correlation between intense beauty and information-seeking here, i.e., wanting to understand the experience, learning, or interest, despite the prevalence of these notions in several psychological theories (Armstrong & Detweiler-Bedell, 2008; Kivy, 1990; Reber et al., 2004). We did, however, find that several other features mentioned in contemporary literature are indeed correlated with beauty, such as the feeling that the experience exceeded expectation (Salimpoor et al., 2015), harmoniously combined various elements (Diessner et al., 2018), and meaningfulness (Leder & Nadal, 2014).

Our current study documents the feelings that are correlated with beauty. This list of beauty-associated characteristics offers a basis for developing a predictive model. The development of a model that can predict the beauty intensity of a given experience with as few predictors as possible would be a big step in explaining beauty.

4.4. Beauty and art

The philosopher and art critic Arthur Danto (2002) noted that until World War I and the dada movement, it was generally accepted that beauty was central to the definition of art. Danto notes that dada and the subsequent postmodern movements “disconnect[ed] beauty from art”. Art today is much more general than just beauty. Danto says that beauty is merely one of many attributes that art can have and that the only necessary one is meaning. However, Danto's strict dichotomy between beauty and meaning is undermined by finding that beauty is associated with meaningfulness in our participants' reports.

Our participants' beauty memories and beliefs assign hardly any role to art. A typical memory of an intense beauty experience was a family vacation on the beach or the mountains, rather than a museum visit. Participants agreed that beauty lies in nature, not art (see Fig. 2), and most of those who believed in a universally beautiful object thought it was part of nature, like flowers or sky. Only 6% of the intense beauty recollections (31/479) mentioned any kind of art.

Beauty is central to popular notions of aesthetics (Augustin et al., 2012), but, in academia, aesthetics and empirical aesthetics (e.g. Leder & Nadal, 2014; Pelowski et al., 2017) focus on art appreciation. Our results challenge the relevance of art, and thus art appreciation, aesthetics, and empirical aesthetics, to how most people experience beauty.

Here, we set out to characterize people's notions and experiences of beauty. In our survey sample of ordinary people—neither artists nor academics—we found that moments of intense beauty are associated with nature and social interactions rather than art. We wonder how members of the art world, e.g. art school students, might respond to a feeling-of-beauty survey like ours. Art matters in their daily lives, but the modern movement assigns little role to beauty in modern art (Danto, 2002), so, even in the art world, descriptions of intense beauty experiences might only rarely mention art. Similarly, since we got so few mentions of art in the intense-beauty recollections, we also wonder if the beauty evoked by art might be different, possibly requiring more learning and understanding (Armstrong & Detweiler-Bedell, 2008). We leave that to future work, but our results already suggest that understanding everyday beauty experiences may require a theory of beauty that deals with social and nature-related experiences.

“I Made It Work”: How Using a Self-Assembled Product Increases Task Performance

“I Made It Work”: How Using a Self-Assembled Product Increases Task Performance. Sören Köcher, Keith Wilcox. Journal of Consumer Psychology, June 14 2021. https://doi.org/10.1002/jcpy.1262

Abstract: Although it is well established that consumers have an increased valuation for self-assembled products, less is known about how using such products influences objective consumption outcomes. Across three experiments, the current research demonstrates that consumers perform better on tasks when they use a product they have self-assembled—as opposed to an identical but ready-to-use product. We show that this effect results from an increase in self-efficacy and rule out possible alternative accounts (i.e., product efficacy beliefs, performance motivation, feelings of psychological ownership, and product liking). In addition, we demonstrate that the self-assembly effect emerges only when consumers actually use the self-assembled product, is robust when product assembly requires different amounts of time and effort, and is not merely the result of a question-behavior effect. Theoretical contributions and opportunities for future research are discussed.

General Discussion

Across three experiments, we demonstrate that using a self-assembled product to execute a task improves performance on the task; an effect that results from an increase in self-efficacy. These findings contribute to prior literature on the consequences of consumers’ involvement in product creation processes (e.g., Fuchs et al., 2015; Norton et al., 2012; Walasek et al., 2017) by broadening our understanding of how product assembly influences consumption outcomes. Although our findings are consistent with research on self-customization, we find that the product assembly effect is the result of an increase in self-efficacy, rather than the desire to affirm identity as suggested by prior research on self-customization (Kaiser et al., 2017). To provide further support that the effect of product assembly operates through a different process, we conducted a follow-up experiment where we examined how using a self-assembled (vs. self-customized) product influences task performance when people can and cannot affirm identity. Consistent with the findings of Kaiser et al. (2017), the results show that consumers perform best when they use a self-customized product and can affirm (vs. not affirm) their identity. In contrast, when consumers use a self-assembled product, their performance is not affected by whether they can affirm identity (see Web Appendix F for study details). Moreover, the results show that assembling a product has a negligible effect on how self-expressive the product is (M = 1.32 on a seven-point scale). These findings suggest that self-expression and the desire to affirm identity, which are important drivers of the self-customization effect, do not appear to play a role in the self-assembly effect.

Our findings offer several directions for future research. The current work demonstrates that making a product functional by assembling it enhances consumers’ sense of self-efficacy and performance. Yet, future research is necessary to fully understand why this effect is conditioned on the actual use of the product. Although we suggest that this is because the self-assembled product is a symbol of competence, it is also possible that this finding is the result of magical thinking on the part of the consumer that only allows self-efficacy to spillover when the assembled product is used. Moreover, it remains unclear whether other actions taken to make a product usable beyond assembly could evoke a similar effect. For instance, would simply mixing the ingredients of a protein shake lead people to have a better workout? In a similar vein, prior research on ritualistic behavior (Brooks et al., 2016; Wang, Sun, & Kramer, 2021) shows that engaging in rituals prior to a task can improve task performance, which research has suggested (Hobson, Schroeder, Risen, Xygalatas, & Inzlicht, 2018), but not empirically demonstrated, could be due to an increase in self-efficacy. Through this lens, product assembly could be viewed as a form of ritualistic behavior that increases self-efficacy. Nevertheless, future research is necessary to fully examine the relationship between rituals and product assembly.

Moreover, our findings suggest that self-efficacy may not be the only mediator of the observed effect. Thus, further research could explore additional factors that may play a role in the process. For instance, it is possible that consumers feel more responsible for performance outcomes when they use a self-assembled product, which could improve their task performance.

In addition, it is worth investigating whether the strength of the effect is dependent on consumers’ a priori self-efficacy in the domain. For instance, the effect of using a self-assembled pen on consumers’ anagram task performance may be less pronounced among consumers who are generally confident in their ability to solve anagrams. In a similar vein, since some consumers may not be confident in their ability to assemble products, using a successfully assembled product could evoke stronger feelings of self-efficacy in such consumers. In addition, the time and effort required to assemble the products used in our experiments was relatively low. Thus, future research could explore conditions wherein product assembly becomes too effortful and demanding such that the favorable consequences may vanish.

Finally, throughout the studies reported in this article, we only examined first use situations, an approach that is widely used in the related literature. However, because self-assembled products are often used more than once in real life, future studies may want to examine whether the detected effects persist when products are used over an extended period. This could be particularly interesting for products that need to be assembled each time before they are used (e.g., professional pool cues). If product assembly becomes a ritual established through repetition each time the product is used, the effect on self-efficacy could be reinforced and intensified over time (Hobson et al., 2018). However, it could also be argued that consumers may habituate to the assembly process, which may limit its impact over the long term.

Nonreligiosity and Life Satisfaction: Reexamining a Supposedly Negative Relationship

Pöhls, Katharina  (2021). Nonreligiosity and Life Satisfaction: Reexamining a Supposedly Negative Relationship. PhD thesis, Universität zu Köln. Jun 2021. https://kups.ub.uni-koeln.de/50458/

Abstract: Most previous research seems to indicate that nonreligiosity is related to disadvantages, e.g., a lower level of life satisfaction. This dissertation, consisting of a literature review and two empirical studies, calls a general negative link into question. In the literature review, several conceptual and methodological issues of most previous research on this topic were identified. In contrast, no or only small differences in the level of life satisfaction between religious and nonreligious individuals can be found, if empirical studies utilize multidimensional conceptualizations of (non)religiosity, representative samples with a substantial number of nonreligious individuals, a differentiation between (non)religious subgroups, include relevant context factors, and/or test for both linear and nonlinear relationships between the variables. The first empirical study examined the relationship between (non)religious self-identification and life satisfaction in a quantitative intercultural comparison across 24 countries worldwide (N=33,879). The results indicate that only in religious societies, identifying as not religious or atheist is related to lower life satisfaction than high religiosity. When the fit between individual and country characteristics was controlled for, a curvilinear relationship between (non)religiosity and life satisfaction emerged, as only weakly religious individuals were less satisfied with life than highly religious individuals. In the second empirical study, the relationship between an individual’s kind of (non)belief, nihilism, and life satisfaction was researched in a representative German sample (N=3,212) with a quantitative comparative design between four (non)religious groups. Only uncertainty what to believe in was related to nihilism, while atheists were not more likely to indicate nihilistic tendencies than theists, suggesting a curvilinear relationship between (non)religiosity and nihilism. Atheists were slightly less satisfied with life than theists, but an individual’s (non)belief was only weakly related to his or her life satisfaction. Thus, the findings of this dissertation further challenge the idea of a universal benefit of religiosity for an individual’s life satisfaction and provide implications for future research on this topic.

What do couples’ activities and behaviors on Instagram reveal about the quality of their relationships?

Picture perfect? Examining associations between relationship quality, attention to alternatives, and couples’ activities on Instagram. Liesel L. Sharabi, Annamariah Hopkins. Journal of Social and Personal Relationships, February 3, 2021. https://doi.org/10.1177/0265407521991662

Abstract: What do couples’ activities and behaviors on Instagram reveal about the quality of their relationships? To answer this question, we surveyed couples (N = 178) about their perceptions of their relationships and analyzed 3,270 of their recent Instagram posts. Actor and partner effects were found between relational quality and engagement with the relationship on Instagram (i.e., the number of couple pictures and partner-initiated likes and comments). There were also actor effects of attention to Instagram alternatives on the perceived quality and actual pursuit of alternative partners, as well as a partner effect on alternative quality. The findings contribute to extending the investment model to the digital age and have methodological implications for understanding relationship dynamics on visual social network sites.

Keywords: Alternatives, Instagram, investment model, relationship quality, social network sites

Greater conformity to traditional gender roles predicted more frequent consumption of beef and chicken and lower openness to vegetarianism among men but offered no predictive value among women

Gender differences in meat consumption and openness to vegetarianism. Daniel L. Rosenfeld, A. Janet Tomiyama. Appetite, Volume 166, November 1 2021, 105475. https://doi.org/10.1016/j.appet.2021.105475

Abstract: Understanding gender differences in meat consumption can help strengthen efforts to improve the sustainability of eating patterns. Compared to women, men eat more meat and are less open to becoming vegetarian. Simply considering between-gender differences, however, may overlook meaningful within-gender heterogeneity in how masculine and feminine identities associate with eating behavior. Distinguishing between specific types of meat is also important, given that some meats (e.g., beef) pose greater challenges to sustainability than do other meats. Through a highly powered, preregistered study (N = 1706), we investigated the predictive value of traditional gender role conformity and gender identity centrality for meat consumption frequency and openness to becoming vegetarian. Greater conformity to traditional gender roles predicted more frequent consumption of beef and chicken and lower openness to vegetarianism among men but offered no predictive value among women. No effects were observed for pork or fish consumption. Among women, greater traditional gender role conformity and gender identity centrality were associated with openness to becoming vegetarian for health reasons. Among men, lower traditional gender role conformity was associated with openness to becoming vegetarian for environmental reasons. These findings suggest that understanding meat consumption calls for greater distinctions between specific types of meat as well as deeper consideration of within-gender heterogeneity.

Keywords: MeatVegetarianismEating behaviorGenderSustainability

Check also People tend to belittle the extent of their meat consumption to mitigate the cognitive dissonance triggered by the meat paradox:

Meat‐related cognitive dissonance: The social psychology of eating animals. Hank Rothgerber,  Daniel L. Rosenfeld. Social and Personality Psychology Compass, April 7 2021. https://www.bipartisanalliance.com/2021/04/people-tend-to-belittle-extent-of-their.html

Regularly drinking alcohol & drinking with meal is significantly associated with a lower risk of all-cause mortality, cardiovascular disease mortality, cancer mortality, or other-cause mortality

Alcohol Consumption Levels as Compared With Drinking Habits in Predicting All-Cause Mortality and Cause-Specific Mortality in Current Drinkers. Hao Ma et al. Mayo Clinic Proceedings, Volume 96, Issue 7, July 2021, Pages 1758-1769. https://doi.org/10.1016/j.mayocp.2021.02.011

Abstract

Objective: To investigate the joint associations of amounts of alcohol consumed and drinking habits with the risks of all-cause mortality and cause-specific mortality.

Patients and Methods: A total of 316,627 healthy current drinkers, with baseline measurements between March 13, 2006, and October 1, 2010, were included in this study. We newly created a drinking habit score (DHS) according to regular drinking (frequency of alcohol intake ≥3 times/wk) and whether consuming alcohol with meals (yes).

Results: During a median follow-up of 8.9 years, we documented 8652 incident cases of all-cause death, including 1702 cases of cardiovascular disease death, 4960 cases of cancer death, and 1990 cases of other-cause death. After adjustment confounders and amount of alcohol consumed, higher DHS was significantly associated with a lower risk of all-cause mortality, cardiovascular disease mortality, cancer mortality, or other-cause mortality (Ptrend<.001, Ptrend=.03, Ptrend<.001, and Ptrend<.001, respectively). We observed that the amount of alcohol consumed have different relationships with the risks of all-cause mortality and cause-specific mortality among participants with distinct drinking habits, grouped by DHS. For example, in the joint analyses, a J-shaped association between the amount of alcohol consumed and all-cause mortality was observed in participants with unfavorable DHS (Pquadratic trend=.02) while the association appeared to be U-shaped in participants with favorable DHS (Pquadratic trend=.003), with lower risks in those consuming greater than or equal to 50 g/wk and less than 300 g/wk.

Conclusion: Our results indicate that alcohol consumption levels have different relationships with the risk of mortality among current drinkers, depending on their drinking habits.

Sex differences in risk taking: Our findings suggest that sensitivity to rewards, associated with resting-state theta oscillations in the anterior cingulate cortex, is a trait that potentially contributes to those sex differences

Resting-State Theta Oscillations and Reward Sensitivity in Risk Taking. Maria Azanova, Maria Herrojo Ruiz, Alexis V. Belianin, Vasily Klucharev and Vadim V. Nikulin. Front. Neurosci., April 28 2021. https://doi.org/10.3389/fnins.2021.608699

Abstract: Females demonstrate greater risk aversion than males on a variety of tasks, but the underlying neurobiological basis is still unclear. We studied how theta (4–7 Hz) oscillations at rest related to three different measures of risk taking. Thirty-five participants (15 females) completed the Bomb Risk Elicitation Task (BRET), which allowed us to measure risk taking during an economic game. The Domain-Specific Risk-Taking Scale (DOSPERT) was used to measure self-assessed risk attitudes as well as reward and punishment sensitivities. In addition, the Barratt Impulsiveness Scale (BIS11) was included to quantify impulsiveness. To obtain measures of frontal theta asymmetry and frontal theta power, we used magnetoencephalography (MEG) acquired prior to task completion, while participants were at rest. Frontal theta asymmetry correlated with average risk taking during the game but only in the female sample. By contrast, frontal theta power correlated with risk taking as well as with measures of reward and punishment sensitivity in the joint sample. Importantly, we showed that reward sensitivity mediated a correlation between risk taking and the power of theta oscillations localized to the anterior cingulate cortex. In addition, we observed significant sex differences in source- and sensor-space theta power, risk taking during the game, and reward sensitivity. Our findings suggest that sensitivity to rewards, associated with resting-state theta oscillations in the anterior cingulate cortex, is a trait that potentially contributes to sex differences in risk taking.

Discussion

Using resting-state MEG recordings and three distinct measures of risk attitudes, we show that sex differences in risk taking are associated with reward sensitivity, which, in turn, are linked to resting-state ACC theta oscillations (Figure 8). On the behavioral level, males were more sensitive to rewards than females. Game-based measures of risk taking showed significant sex differences and also correlated with self-reported expected benefits of risky actions (DOSPERT benefits scores). On the neural level, rsFTA explained average risk taking during the repeated game exclusively in the female subsample. By contrast, in the whole sample, rsFT correlated with first-trial risk taking and also with DOSPERT benefit and risk scores, indicating an association with reward and punishment sensitivity. Finally, due to a refined spatial specificity, theta power localized to the ACC correlated with outcome sensitivities and game-based measures of risk taking even more strongly than the rsFT did. The findings suggest that resting-state ACC activity is a possible source of sex differences in reward sensitivity, and, consequently, in risk taking.

[Figure 8. Visualization of the main findings. MEG, magnetoencephalography; ACC, anterior cingulate cortex; DOSPERT, Domain-Specific Risk-Taking Scale; BIS11, Barratt Impulsiveness Scale. Significant correlations are reported after FDR adjustment.]

Behavioral Measures

The DOSPERT benefits subscale significantly correlated with both average and first-trial risk taking, converging with previous studies (Weber et al., 2002; Hanoch et al., 2006; Fukunaga et al., 2018). Further, self-assessed reward sensitivity demonstrated a greater correlation with first-trial than with average risk taking, indicating that sensitivity to outcomes could affect the former more (Erev et al., 2008; Lejarraga and Gonzalez, 2011). Absence of correlations between BIS11 scores and performance on decision-making tasks is in line with the literature (Gomide Vasconcelos et al., 2014; Lauriola et al., 2014; Reddy et al., 2014; Hüpen et al., 2019).

We observed significant sex differences in first-trial and average risk taking during the game, which was expected based on the extensive literature on sex differences in decision-making under uncertainty (e.g., Jianakoplos and Bernasek, 1998; Weber et al., 2002; Charness and Gneezy, 2012; Zhou et al., 2014). Notably, however, Crosetto and Filippin (2013) did not report significant sex differences in their versions of BRET. This discrepancy could be explained by the use of repeated trials or more salient financial incentives in our task. Furthermore, it has been observed earlier that some measures reveal that females are more risk-averse than males, while others are not (e.g., Charness et al., 2013; Filippin and Crosetto, 2016). Our finding that there was no significant sex difference in DOSPERT likelihood scores supports this observation.

As for the reward and punishment sensitivities, we observed that on average males scored higher than females on the DOSPERT benefits subscale. Previous studies also reported sex differences in outcome sensitivities based on DOSPERT (Weber et al., 2002; Hanoch et al., 2006; Lee and Jeong, 2013) and other measures (Li et al., 2007; Cross et al., 2011). In line with previous studies, we found no significant sex differences in impulsivity (Kamarajan et al., 2008; Liu et al., 2012; Lee and Jeong, 2013). Summarizing the facts presented above, behavioral evidence from the current work suggests that sensitivity to outcomes, rather than impulsivity, is a candidate trait that could explain sex differences in risk attitudes.

Frontal Theta Asymmetry (rsFTA)

Analysis of the MEG oscillatory activity showed no significant sex differences in rsFTA, converging with previous EEG work (Ocklenburg et al., 2019). Therefore, as predicted, we simultaneously observed (1) sex differences in risk taking based on game performance and (2) no sex differences in the neural trait previously associated with this decision-making characteristic.

We found a significant positive correlation of rsFTA with average risk taking exclusively in the female subsample, confirming earlier findings in female populations by Gianotti et al. (2009). Average and first-trial risk taking in the game did not correlate with rsFTA in the joint sample, which is in contrast with the result of Studer et al. (2013). However, Studer et al. (2013) did not report sex-specific results, and this study contained 70% of females, which, according to our findings, could bias the result obtained for the joint sample. Because regression analyses demonstrated significant effects of rsFTA on average risk taking in the game and because we observed correlation coefficients of rsFTA with measures of risk taking (although insignificant) comparable in magnitude to previous work with larger samples (Studer et al., 2013), a possible interpretation is that we were not able to reliably detect significant non-parametric associations between rsFTA and risk taking for the whole sample due to the limited sample size.

Higher rsFTA may be associated with the lower relative right frontal activity, and thus the prevalence of left frontal activity (Gianotti et al., 2009; Studer et al., 2013), which is partially supported by previously observed negative associations between theta power and cortical activity (Oakes et al., 2004; Scheeringa et al., 2008). Additional evidence that frontal lateralization is related to risk taking comes from stimulation studies, focused on the role of right dorsolateral prefrontal cortex in decision-making (Knoch et al., 2006; Fecteau et al., 2007a, b; Cho et al., 2010; Sela et al., 2012). Furthermore, several studies reported sex differences in the involvement of the right and left frontal cortices in decision-making (Bolla et al., 2004; Tranel et al., 2005; Neo and McNaughton, 2011). Our findings contribute to the evidence that sex may interact with frontal asymmetry in relation to risk taking, but this requires further testing.

Theta Power

We found a strong association between rsFT and theta power in the ACC. This outcome is consistent with previous dipole-fitting studies that revealed possible sources of rsFT in the ACC (Asada et al., 1999; Scheeringa et al., 2008; Clemens et al., 2010). The association between the power of neuronal oscillations and the degree of cortical activation is a subject of ongoing research, but an emerging pattern is that stronger alpha oscillations are typically associated with weaker cortical activity (e.g., Oakes et al., 2004). Previous work indicated a similar relationship for theta oscillations, but the evidence is not particularly strong. Oakes et al. (2004) showed some negative associations between EEG theta oscillations and fMRI BOLD signals. However, these associations were positive in some clusters of voxels, such as in one in the insular region. While Scheeringa et al. (2008) did find a negative association between rsFT and metabolic activity in the ACC, a more detailed interpretation of our results regarding cortical sources and their function would require follow-up studies using combined EEG-fMRI.

Risk Taking

We report on the existence of the significant positive correlation between rsFT and first-trial risk taking. Two previous studies did not find correlations of rsFT with risk taking (Massar et al., 2012; Studer et al., 2013). There are three notable similarities between their experimental designs that differentiate them from our paradigm. First, both studies introduced losses in the task either explicitly or via a safe gamble (Ert and Erev, 2013). Second, they forced participants to choose between two gambles with the same expected value (Massar et al., 2012) or with very similar expected values (Studer et al., 2013). Third, the computation of expected values of gambles in tasks used by Massar et al. (2012) and Studer et al. (2013) was straightforward. Therefore, differences in experimental designs associated with values and presentation of options might have affected the observed correlations between rsFT and risk taking.

All observed correlations for rsFT were even stronger for the ACC theta power, and it also significantly correlated with average risk taking in the game. It is an expected result. If rsFT originates at the level of the ACC (Asada et al., 1999; Scheeringa et al., 2008; Clemens et al., 2010), then the results would be more pronounced at the source level compared to the sensor level due to the contamination of sensor-level activity from other less relevant sources. Thus, our findings are aligned with the extensive neuroimaging research demonstrating the involvement of the ACC in decisions under risk (Paulus and Frank, 2006; Christopoulos et al., 2009; Hewig et al., 2009; Mohr et al., 2010; Schonberg et al., 2012; Fukunaga et al., 2018).

Reward and Punishment Sensitivity

Additionally, we found strong correlations of self-assessed punishment (DOSPERT risks) and reward (DOSPERT benefits) sensitivities with rsFT and the ACC theta power. Few previous studies examined associations between rsFT or ACC activity at rest and outcome sensitivity. Our findings contribute to the evidence that rsFT is related to outcome sensitivity (Massar et al., 2012, 2014). Regarding theta oscillations and ACC activity during tasks, both measures have previously been associated with reactions to rewards and punishments (Debener et al., 2005; Cohen et al., 2007; Kamarajan et al., 2008; Santesso et al., 2011; Crowley et al., 2014; Van Duijvenvoorde et al., 2014). Research in humans (Gehring and Willoughby, 2002; Wang et al., 2005; Iannaccone et al., 2015) and primates (Tsujimoto et al., 2010; Womelsdorf et al., 2010; Babapoor-Farrokhran et al., 2017; Taub et al., 2018) singled out theta ACC activity as a source of signals associated with feedback and behavioral adjustment. Our results further extend this literature.

Sex Differences

We found significant sex differences in rsFT. However, evidence from previous studies is mixed. Zappasodi et al. (2006) also used MEG and reported the presence of sex differences in resting-state theta power. Other studies used EEG and reported no significant sex differences (Jaušovec and Jaušovec, 2010; Gmehlin et al., 2011; Kober and Neuper, 2011; Banis et al., 2014), or higher theta power in females compared to males (Clarke et al., 2001; Kamarajan et al., 2008; Osinsky et al., 2017). We examined the demographic characteristics of participants in these studies and did not find a pattern that could account for such inconsistent results. One possibility is that sex difference in skull conductivities affects EEG recordings but not MEG (Huttunen et al., 1999).

Sex differences in the resting-state ACC theta power were even more pronounced. It is in line with the diverse evidence from previous studies that demonstrated sex differences associated with this region (Goldstein et al., 2001; Markham and Juraska, 2002; Zhou et al., 2014). These sex differences may be linked to levels of testosterone and its effects on midbrain dopaminergic pathways (Johnson et al., 2010). On the one hand, activity of the ACC is associated with dopaminergic projections from the midbrain (Holroyd and Coles, 2002), and dopaminergic genetic polymorphisms correlate with risk taking and also with amplitudes of FRN (Heitland et al., 2012). On the other hand, higher levels of testosterone are associated with risk taking (Apicella et al., 2008; Stanton et al., 2011) and also with outcome sensitivity (Van Honk et al., 2004). Therefore, baseline ACC activity may be linked to sex differences in outcome sensitivity and, consequently, risk taking. It should be noted, however, that the previous literature on sex differences in either rsFT or resting ACC theta activity is rather scarce. Accordingly, validation of the current results in future MEG and combined MRI-EEG studies will be necessary.

Noticeably, a mediation analysis allowed us to formally test the hypothesis that sex differences in risk taking may be mediated by reward sensitivity via resting-state theta oscillations in the ACC. The results revealed that reward sensitivity assessed via DOSPERT benefits scores mediated the effects of resting-state ACC theta oscillations on average and first-trial risk taking in the game. Furthermore, structural equation modeling demonstrated that the indirect pathway of the effect of sex on first-trial risk taking via the ACC theta power and DOSPERT benefits scores was significant; it fully accounted for the overall impact of sex on first-trial risk taking. Therefore, even though there may be other confounding variables, reward sensitivity is a candidate trait for explaining sex differences in risk taking where resting-state ACC activity is a potential contributing mechanism.

Finally, regression analysis demonstrated that rsFTA and sex captured significantly different portions of variance in task performance, while ACC theta power explained variance due to sex. Therefore, if we only had information about rsFTA we would not be able to explain variability in risk taking of participants associated with their sex, while this can be done based on resting ACC theta power. Interestingly, Weis et al. (2020) have recently shown that sex classification based on resting-state connectivity of ACC can be done with 74.4% accuracy. In addition, results of regression analysis showed that average performance in the game was explained best when including both rsFTA and ACC theta power before the game (as opposed to including only one of these characteristics), which further highlights a possibility for functionally distinct involvement of these neural traits in risk-taking. Future research is required to clarify this question.

Limitations

This study has several limitations. An important drawback of our experiment is not controlling for the menstrual cycle phase of female participants. This could have confounded our results because cortical activity is affected by menstrual cycle phase and blood estrogen level (Dietrich et al., 2001; Hausmann, 2005). Furthermore, we had a relatively small sample size (35), a higher number of males compared to females (20/15), and a rather young group of participants. Due to this limitation, we could not reliably detect correlations of risk-taking measures with neural traits in male and female subsamples separately (although technically it is possible). Thus, we were mostly interpreting results relating to the joint sample as a whole. Our findings necessitate further research with larger samples, separately for males and females. Nevertheless, power analysis suggests that our joint sample was sufficient to detect correlation coefficients of 0.45 or higher. Reliability of our results was further confirmed by replicating significant results based on resting-state recordings after the game. Another limitation is that we did not have individual MRI of participants which could have improved our source-modeling results even further. Finally, it must be noted that the study was correlational, and thus we could not establish direct causal links—this critique, however, applies to almost all EEG/MEG studies.

Coding ethnographic texts on reputations from 153 cultures, found that reputational domains vary cross-culturally, yet reputations for cultural conformity, prosociality, social status, and neural capital are widespread

Garfield, Zachary H., Ryan Schacht, Emily R. Post, Dominique Ingram, Andrea Uehling, and Shane Macfarlan. 2021. “The Content and Structure of Reputation Domains Across Human Societies: A View from the Evolutionary Social Sciences.” OSF Preprints. July 2. doi:10.31219/osf.io/mk4wq

Abstract: Reputations are an essential feature of human sociality, critical for the evolution of cooperation and group living. Much scholarship has focused on reputations, yet typically on a narrow range of domains (e.g., prosociality, aggressiveness), usually in isolation. Humans can develop reputations, however, from any collective information. We conducted exploratory analyses on the content, distribution, and structure of reputation domain diversity across cultures, using the Human Relations Area Files ethnographic database. After coding ethnographic texts on reputations from 153 cultures, we used hierarchical modelling, cluster analysis, and text analysis to provide an empirical view of reputation domains across societies. Findings suggest: 1) reputational domains vary cross-culturally, yet reputations for cultural conformity, prosociality, social status, and neural capital are widespread; 2) reputation domains are more variable for males than females; and 3) particular reputation domains are strongly interrelated, demonstrating a structure consistent with dimensions of human uniqueness. We label these reputational features: Cultural group unity, Dominance, Sexuality, Social and material success, and Supernatural healing. Ultimately, through this work, we highlight the need for future research on the evolution of cooperation and human sociality to consider a wider range of reputation domains, as well as their patterning by social, ecological, and gender-specific pressures.

Moral dilemmas and trust in leaders during a global health crisis

Moral dilemmas and trust in leaders during a global health crisis. Jim A. C. Everett et al. Nature Human Behaviour, Jul 1 2021. https://www.nature.com/articles/s41562-021-01156-y

Abstract: Trust in leaders is central to citizen compliance with public policies. One potential determinant of trust is how leaders resolve conflicts between utilitarian and non-utilitarian ethical principles in moral dilemmas. Past research suggests that utilitarian responses to dilemmas can both erode and enhance trust in leaders: sacrificing some people to save many others (‘instrumental harm’) reduces trust, while maximizing the welfare of everyone equally (‘impartial beneficence’) may increase trust. In a multi-site experiment spanning 22 countries on six continents, participants (N = 23,929) completed self-report (N = 17,591) and behavioural (N = 12,638) measures of trust in leaders who endorsed utilitarian or non-utilitarian principles in dilemmas concerning the COVID-19 pandemic. Across both the self-report and behavioural measures, endorsement of instrumental harm decreased trust, while endorsement of impartial beneficence increased trust. These results show how support for different ethical principles can impact trust in leaders, and inform effective public communication during times of global crisis.

Discussion

The COVID-19 pandemic has raised a number of moral dilemmas that engender conflicts between utilitarian and non-utilitarian ethical principles. Building on past work on utilitarianism and trust, we tested the hypothesis that endorsement of utilitarian solutions to pandemic dilemmas would impact trust in leaders. Specifically, in line with suggestions from previous work and case studies of public communications during the early stages of the pandemic, we predicted that endorsing instrumental harm would decrease trust in leaders, while endorsing impartial beneficence would increase trust. Experiments conducted during November–December 2020 in 22 countries across six continents (total N = 23,929; valid sample for self-report task 17,591; valid sample for behavioural task 12,638) provided robust support for our hypothesis. In the context of five realistic pandemic dilemmas, participants reported lower trust in leaders who endorsed instrumental sacrifices for the greater good and higher trust in leaders who advocated for impartially maximizing the welfare of everyone equally. In a behavioural measure of trust, only 28% of participants preferred to vote for a utilitarian leader who endorsed instrumental harm, while 60% voted for an impartially beneficent utilitarian leader. These findings were robust to controlling for a variety of demographic characteristics as well as participants’ own policy preferences regarding the dilemmas. Although we observed some variation in effect sizes across the countries we sampled, the overall pattern of results was highly robust across countries. Our results suggest that endorsing utilitarian approaches to moral dilemmas can both erode and enhance trust in leaders across the globe, depending on the type of utilitarian morality.

We designed our set of dilemmas to rule out several alternative explanations for our findings, such as a general preference for less restrictive leaders (Supplementary Note 7), leaders who treat everyone equally (Supplementary Note 8) and leaders who seek to minimize COVID-19-related deaths (Supplementary Note 9). In addition, all of our results survived planned robustness checks to account for the possibility that local policies related to lockdowns or contact tracing could bias participants’ responses. Post hoc analyses demonstrated that our findings were highly consistent across the different dilemmas for instrumental harm (Lockdown, Tracing and Ventilators) and impartial beneficence (Medicine and PPE).

While the robustness of our findings across countries speaks to their broad cultural generalizability, further work is needed to understand the observed variations in effect sizes across countries. It seems plausible that both economic (for example, gross domestic product or socio-economic inequality) and cultural (for example, social network structure) differences across countries could explain some of the observed variations. One possibility, for example, is that country-level variations in tightness–looseness72, which have been associated with countries’ success in limiting cases in the COVID-19 pandemic73, might moderate the effects of moral arguments on trust in leaders. Another direction for future research could be to explore how country-level social network structure might influence our results. Individuals in countries with a higher kinship index74 and a more family-oriented social network structure, for example, might be less likely to trust utilitarian leaders, especially when the utilitarian solution conflicts with more local moral obligations.

There are several important limitations to the generalizability of our findings. First, although our samples were broadly nationally representative for age and gender (with some exceptions; see Results), we did not assess representativeness of our samples on a number of other factors including education, income and geographic location. Second, while our results do concord with the limited existing research examining the effects of endorsing instrumental harm and impartial beneficence on perceived suitability as a leader37, and held across different examples of our pandemic-specific dilemmas, it of course remains possible that different results would be seen when judging leaders’ responses in other types of crises (for example, violent conflicts, natural disasters or economic crises) or at different stages of a crisis (for example, at the beginning versus later stages). Third, the reported experiments tested how responses to moral dilemmas influenced trust in anonymous, hypothetical political leaders. In the real world, however, people form and update impressions of known leaders with a history of political opinions and behaviours, and it is plausible that inferences of trustworthiness depend not just on a leader’s recent decisions but also on their history of behaviour, just as classic work on impression formation shows that the same information can lead to different impressions depending on prior knowledge about the target person75. Furthermore, we did not specify the gender of the leaders in our experiments (except in the voting task for China and for the Hebrew and Arabic translations, where it is not possible to indicate ‘leader’ without including a gendered pronoun; here it was translated in the masculine form). Past work conducted in the United States suggests that participants may default to an assumption that the leader is a man76, but it will be important for future work to assess whether men and women leaders are judged differentially for their moral decisions. Because women are typically stereotyped as being warmer and more communal than men77, it is plausible that women leaders would face more backlash for making ‘cold’ utilitarian decisions, especially in the domain of instrumental harm. Fourth, because the current work focused on trust in political leaders, it remains unclear how utilitarianism would impact trust in people who occupy other social roles, such as medical workers or ordinary citizens. Fifth, and finally, it could be interesting to explore further the connection between impartial beneficence and intergroup psychology, especially with regards to teasing apart ‘impartiality’ and ‘beneficence’. For example, even holding beneficence constant, a leader who advocates for impartially sharing resources with a rival country may be perceived differently from one who impartially shares with an allied country (and, while speculative, this distinction might explain why Israel was an outlier in impartial beneficence, being a country in a region with ongoing local conflicts).

Our results have clear implications for how leaders’ responses to moral dilemmas can impact how they are trusted. In times of global crisis, such as the COVID-19 pandemic, leaders will necessarily face real, urgent and serious dilemmas. Faced with such dilemmas, decisions have to be made, and our findings suggest that how leaders make these judgements can have important consequences, not just for whether they are trusted on the issue in question but also more generally. Importantly, this will be the case even when the leader has little direct control over the resolution. While a national leader (for example, a president or prime minister) has the power and responsibility to resolve some moral dilemmas with policy decisions, not all political leaders (for example, as in our study, local mayors) have that power. A leader with little ability to directly impact the resolution of a moral dilemma might consider that voicing an opinion on that dilemma could reduce their credibility on other issues that they have more power to control.

To conclude, we investigated how trust in leaders is sensitive to how they resolve conflicts between utilitarian and non-utilitarian ethical principles in moral dilemmas during a global pandemic. Our results provide robust evidence that utilitarian responses to dilemmas can both erode and enhance trust in leaders: advocating for sacrificing some people to save many others (that is, instrumental harm) reduces trust, while arguing that we ought to impartially maximize the welfare of everyone equally (that is, impartial beneficence) increases trust. Our work advances understanding of trust in political leaders and shows that, across a variety of cultures, it depends not just on whether they make moral decisions but also which specific moral principles they endorse.