Tuesday, December 29, 2020

We found a reminiscence bump in adolescence (peaking around age 14) for both ratings of the autobiographical salience of songs featured in the charts during that period and the familiarity of these songs

A Cross-Sectional Study of Reminiscence Bumps for Music-Related Memories in Adulthood. Kelly Jakubowski et al. Music & Science, October 23, 2020. https://doi.org/10.1177/2059204320965058

Abstract: Music is often intimately linked to identity, as evidenced by the high value many people place on musical activities and the way in which music can become seemingly effortlessly coupled to important memories from throughout one’s lifespan. Previous research has revealed a consistent reminiscence bump in autobiographical memory—the disproportionate recall of memories from between ages 10 to 30 years in comparison with other lifetime periods—which also appears to extend to music-related memories. The present study represents one of the largest explorations of the musical reminiscence bump across adulthood to date. Participants (N = 470; ages 18 to 82 years) were shown the titles and artists of 111 popular songs that had featured in the charts between 1950 and 2015 and rated the degree to which they had autobiographical memories associated with each song, as well as the degree to which they were familiar with and liked the song. We found a reminiscence bump in adolescence (peaking around age 14) for both ratings of the autobiographical salience of songs featured in the charts during that period and the familiarity of these songs. Liking ratings showed more divergent results depending on a participant’s current age, including evidence for a cascading reminiscence bump, in which liking ratings from young adults increased for music from their parents’ adolescent years. We also revealed new evidence that music-related autobiographical memories appear to invoke similar retrieval processes to the common methodology of eliciting autobiographical memories via word cues. We contextualize these results in relation to general theoretical accounts of the reminiscence bump, and age-related differences in the bump are discussed in relation to various sociocultural and technological changes in music listening habits.

Keywords: Aging, autobiographical memory, musical memory, music-evoked autobiographical memory, reminiscence bump


In this article, we investigated the reminiscence bump for popular music from across a 65-year period (songs featured in the charts from 1950 to 2015) in participants across the full range of adulthood (ages 18 to 82 years). This represents one of the largest investigations of the musical reminiscence bump to date and, to our knowledge, the first extensive exploration of this phenomenon in the French population.

We found consistent evidence for a reminiscence bump for two aspects of the songs used here: ratings of the degree to which the songs evoke autobiographical memories, and familiarity ratings for the songs themselves. Reminiscence bumps for these two dependent variables were evidenced in all four age groups. The results for autobiographical salience ratings of the songs were broadly aligned with general theoretical conceptions of a reminiscence bump that occurs between ages 10 to 30 years (Rubin et al., 1998). The familiarity ratings exhibited largely the same response pattern as the autobiographical salience ratings, but the familiarity bump in the two middle age groups began somewhat earlier than expected (around 5 years before participants were born, although this result was statistically significant following correction for multiple comparisons only for the 42–55 age group). This suggests that songs that featured in the charts just before participants were born continued to be popular for some time beyond their initial release, and also demonstrates that the correspondence between the year a song is featured in the charts and the year at which a participant is first/most often exposed to it are not always equivalent. This result parallels the findings of Rathbone et al. (2017), who measured both age at release and age at which a pop song was rated as most important, and found the average age at release to be approximately 5 to 6 years earlier than the average age at which songs were rated as most personally important to participants.

We also found some slight variations in the reminiscence bump between age groups. The peak (highest point) in the bump for the oldest age group for both autobiographical salience and familiarity ratings was 5 years later than the other age groups. It is possible that this represents a cohort effect; for instance, the oldest age group may have engaged with music in different ways during adolescence than the younger groups as a result of both sociocultural and technological factors, causing them to discover their most autobiographically salient music later in their teenage years. It could also be that the bump may shift its peak later in time as people age (and potentially reengage with the music of their youth in different ways across the lifespan), such that the three younger groups may eventually show a similar pattern when they reach the age of the oldest group. Another notable difference between the four age groups was that the youngest group showed a less pronounced reminiscence bump (lower peak) than the other three age groups. This aligns with previous literature showing a stronger reminiscence bump effect in older than younger adults (Janssen et al., 2005). This may be related to reengagement with favorite music over the course of a lifetime, which can strengthen the link between the music and associated memories via regular retrieval and rehearsal (see also the work of Janssen et al., 2007 on memory re-sampling effects, which are stronger for music than other cultural products, such as films and books).

The youngest group showed evidence of a cascading reminiscence bump for music released up to two decades before they were born. This bump we found here is similar, although slightly earlier in time, in comparison with the bump occurring 8 to 12 years before participants were born reported by Krumhansl and Zupnick (2013). This difference may be partially attributed to the fact that the average maternal age is approximately 2 years older in France than the US,2 but could also reflect some cultural variations in music listening habits or sharing of music between generations. In our study, the cascading reminiscence bump was particularly evidenced in higher than average liking ratings for music from this period, which were even higher than for current music, that is, music of the participants’ own reminiscence bump period. It would be interesting to follow up this finding in longitudinal research, for instance, to test whether this cascading reminiscence bump is maintained throughout a participant’s lifetime, or whether liking ratings for current pop music might increase as these young participants age and begin to look back on this period in a more nostalgic light. Future cross-sectional research could also compare how younger versus older adults respond to music released before they were born, as the present design primarily allowed for this particular factor to be investigated in the youngest two groups.

In general, liking ratings showed the least consistent evidence of a reminiscence bump. This was particularly due to differences between the two oldest and two youngest groups—the two oldest groups showed liking responses that were more consistent and more correlated with their autobiographical salience and familiarity ratings than the two youngest groups. These dissimilarities in liking ratings could be related to generational differences in the way people engage with music. For instance, younger adults tend to listen more to music via streaming and online services in comparison with older adults, which might allow for access to a wider range of music in terms of its release date, in comparison with older groups who more often make use of CD collections (Krumhansl, 2017). If young adults are listening to a broader array of music from across many decades,3 this may increase their openness to music from different eras and result in less varied liking ratings across the songs used in this study. The effects of such shifts in listening methods and technologies on the formation and lifetime stability of the musical reminiscence bump have yet to be fully explored. The results presented in Figure 4 also show evidence of some intergenerational shared preferences, with the three youngest groups all giving their highest liking ratings for music from the late 1970s to early 1980s, indicating that particular stylistic conventions or features of the songs themselves may also play a notable role in shaping preferences for the pop songs presented in this study.

Finally, we found that the reminiscence bump for autobiographical salience ratings of the songs aligned relatively well with previous findings on the reminiscence bump for autobiographical memories evoked via word cues. The results of our analysis presented in Figure 5 indicate that songs featured in the charts during adolescence are more likely to elicit strong autobiographical associations in older adults, peaking around age 14. Adolescence is a key period in terms of identity development (e.g., Erikson, 1956Meeus, 2011), and previous research also suggests that musical tastes are developed around this period (Holbrook & Schindler, 1989Lamont & Loveday (2020)North & Hargreaves, 2002). However, Holbrook and Schindler’s (1989) findings indicate that musical preferences peak for music released around age 24, suggesting that the reminiscence bump in music-related autobiographical memories may not be entirely explained by the crystallization of musical tastes. Our analysis also contributes the novel insight that the shape of the music-related reminiscence bump is well-characterized by a gamma distribution, which has implications for future research in terms of informing sampling decisions and making more precise assumptions about the predicted associations between age and music-related memories.

The comparatively earlier reminiscence bump evidenced for both music- and word-related memories indicates that the associative retrieval processes underlying these two tasks are accessing a somewhat different set of memories than the top-down retrieval method of asking participants to recall particularly important memories. Such findings are important for informing the development of interventions that aim to elicit memories via specific types of retrieval cues. For instance, this finding suggests that using musical or word cues to elicit memories in people with Alzheimer’s disease might be particularly effective for bringing back memories from adolescence. The “important memories” method may be less effective in this population in general, due to the impairments in strategic retrieval that are common to this disease. Therefore, it is important to investigate whether other associative cue types, beyond music and word cues, may be found to be effective in eliciting memories from other lifetime periods (e.g., early adulthood), or whether, on the other hand, all associative retrieval tasks show a bump in the same temporal location. Studies that compare autobiographical memories evoked by different cue types across the same sample of participants using the same data collection protocol should be conducted as a matter of priority, in order to ensure the differences in the reminiscence bump location seen here cannot be attributed to methodological differences between studies.

In general, further research is needed to isolate the mechanisms underlying these different types of retrieval tasks, and theoretical accounts of the reminiscence bump require revision to incorporate explanations for these differences in temporal location of the bump. In particular, the majority of existing theories provide a more sufficient explanation for memories accessed via the “important memories” method than via word cues or music (Koppel & Berntsen, 2015). It should also be noted that our analysis of the word-cued reminiscence bump gave some indication that there may actually be two bumps in the word-cued memories distribution (see Figure 5). An initial comparison of the studies producing earlier versus later word-cued bumps does not reveal any systematic difference in methodology; rather, it may be that the cue word method invokes several possible retrieval strategies, including a combination of both top-down and associative processes that may vary from one participant or one cue to another.

Future research on this topic should further compare the methods of soliciting ratings of songs via visual presentation of the title to auditory presentation of the song. In particular, our method can be potentially useful in clinical settings in which family or carers may be asked to select personally-relevant music to be used with patients from a written song list; as such, it is of both theoretical and practical importance to fully explore any potential limitations of this method. Our study is based on the assumption that a song title can elicit phenomenologically similar autobiographical memories to hearing the song, following the results of Cady et al. (2008). An additional important finding from Cady et al. (2008) is that seeing the song title and hearing the music both elicited a high degree of mentally “hearing” the song during the subsequent autobiographical recall task, which did not significantly differ between these two retrieval conditions. It is likely that the approach used in the present study also elicited a high degree of musical imagery for the songs whose titles were presented, although this factor was not explicitly measured. Subsequent research should investigate the degree to which musical imagery mediates the relationship between being presented information about a song (e.g., title, printed lyrics) and subsequent retrieval of autobiographical memories associated to that song. This is particularly important for understanding how different types of retrieval cues may affect the retrieved memory content and differentially impact the shape and temporal location of the reminiscence bump.

Our study has followed a similar approach to most previous research in this domain by focusing solely on popular music, which we acknowledge represents only one of many genres of music that may be autobiographically relevant to participants. In addition, technological advances such as music streaming services now offer researchers the opportunity to monitor and measure the listening histories of participants, which may be utilized to map the relationship between date of first exposure, number of total exposures, and the autobiographical salience of a song in a more precise manner than ever before (see Stephens-Davidowitz, 2018, for an initial exploration of the popularity of songs on Spotify by age and gender of the listener). Online streaming frequency or number of radio plays of particular songs, for instance, may also be used as proxy measures for likelihood of familiarity with a song, although such figures may be more representative of certain demographic groups than others (e.g., Spotify usage in particular is still skewed toward younger adults)4.

In conclusion, the results of this study indicate the presence of a reminiscence bump in adolescence for both the familiarity of songs featured in the charts during that period and ratings of the autobiographical salience of these songs. Evidence for this bump was found consistently across participants currently aged 18 to 82 years. Liking ratings showed more divergent results depending on the participants’ current age, with evidence for a cascading reminiscence bump, in which liking ratings from young adults increased for music of their parents’ generation. We also revealed new evidence that music-related autobiographical memories appear to invoke similar retrieval processes to the common methodology of eliciting autobiographical memories via word cues. These findings contribute to both theoretical and practical discussions around the extent to which music is intrinsically linked to personal memories from throughout the lifespan, and provide further impetus for exploring the efficacy of music to cue vivid autobiographical memories in everyday and clinical settings.

Norwegian mating: Men were more selective in physical appearance, whereas women were more selective in all the other mate preferences (e.g., understanding, dominant, kind, intellectual, etc.)

Norwegian Men and Women Value Similar Mate Traits in Short-Term Relationships. Mehmet Mehmetoglu, Ilmari Määttänen. Evolutionary Psychology, December 29, 2020. https://doi.org/10.1177/1474704920979623

Rolf Degen's take: https://twitter.com/DegenRolf/status/1343866293712261121

Abstract: Previous research has provided evidence that females are generally the more selective sex in humans. Moreover, both sexes have been found to be more selective in long-term mating compared to short-term mating. In this study, we have examined the effects of sex, mating strategy (preferred relationship length) and their interaction on mate preferences (i.e., mate selection criteria) in an egalitarian Nordic society, namely Norway. The study sample consisted of 1,000 individuals, 417 of whom were male and 583 female respondents. According to our findings, men were more selective in physical appearance, whereas women were more selective in all the other mate preferences (e.g., understanding, dominant, kind, intellectual etc.). The respondents that were seeking short-term relationships had higher preference for physical appearance, humorousness and sociability. On the other hand, the respondents that were seeking long-term relationships were more selective in most of the other mate preferences (i.e., understanding, kind, cultivated, domestic, reliable, and similar). Interestingly, no interaction effect was found between sex and mating strategy in that differences between long-term and short-term seekers in mate preferences did not change depending on sex. This suggests that men and women value the same traits in short-term relationships.

Keywords: mate preference, sex differences, sexual selection, long-term mating, short-term mating

To recap, we found evidence for sex differences in mate selection criteria: men were more selective with respect to physical attractiveness and women were more selective with respect to all the other mate preference criteria. This was an expected result in light of previous research with similar findings (Buss & Schmitt, 1993Castro & Lopes, 2011Regan et al., 2000Shackelford et al., 2005). The respondents that were searching for a short-term partner had a higher preference for physical attractiveness, humorousness and sociability. The respondents that were searching for long-term relationships were more selective in most of the other mate preference criteria. Perhaps surprisingly, no interaction effect between mating strategy and sex was found. This was contrary to what was predicted, based on Sexual Strategies Theory.

The respondents that were searching for long-term relationships were more selective in most of the other mate preference criteria (see also Castro &Lopes, 2011Stewart et al., 2000). An issue with previous studies on the topic of short/long term relationship and sex differences is that typically the groups have been analyzed separately while often implying that there is an interaction between the sex and relationship length.

The results also suggested sex differences in preferences depending on the relationship length, but a relationship duration-sex interaction was not explicitly presented (Stewart et al., 2000). Thus, it is not completely clear, whether relationship length and sex interact with each other when they are analyzed together in a single analysis. This is a major question when resolving the hypotheses around this issue.

So, does each sex have also their particular preference when it comes to short-term mating (compared to long-term mating), or do both sexes have the same predictable pattern of preferences? Direct evidence for such an interaction-effect is relatively scarce in general. One exception was a study, in which sex and relationship length had an interaction in which women displayed a higher preference for partner’s sexual passion and desire for short-term partner than long-term partner, whereas there was no such difference among men (Regan et al., 2000). Another study found a sex-relationship length interaction in which both sexes had a similar high preference for attractiveness in short-term relationships but not in long-term relationships, in which women did not pay as much attention to attractiveness (Li & Kenrick, 2006). At least one study found no relationship length-sex interaction and interpreted this as evidence against Sexual Strategies Theory and in favor of Attachment Fertility Theory (Pedersen et al., 2014). Similarly, our results did not support such interaction effect, and thus underlying sex difference in any of the preferences.

One issue that may make interpreting the results more difficult may be the reporting style and underlying choosiness of each sex. For instance, commonly found self-reported preference for physical attractiveness may be influenced by different perception of attraction among different sexes: it is possible that women are more critical in their evaluations.

This study was conducted in an egalitarian, Nordic society, which may be relevant in the study of sex differences in preferences, as they are influenced by social change and societal norms (Bech-Sørensen & Pollet, 2016). Gender equality and strong social safety nets provided by the government may unmask preferences, which might in other environments be hidden under the most urgent materialistic needs. Chinese women, especially those with high socioeconomic status or who lived in cities, preferred “good father” over “good genes” or “good provider” in a self-report study (Lu et al., 2015). Some studies have provided evidence of change in preferences over time (Souza et al., 2016). Studies utilizing personality traits have provided evidence that people prefer traits that are associated to their own traits even in more traditionalistic societies such as Islamic countries (Atari et al., 2020).

Our results lacked the hypothesized interaction-effect, and thus did not support Sexual Strategies Theory, but it is not clear whether or not the results can be interpreted as supporting Attachment Fertility Theory (Pedersen et al., 2014) or some other existing theory. It is also worth remembering that not all traits are adaptations. Some features or traits may be a result of selection for that trait in the other sex (e.g., male nipples) or may otherwise be byproducts of an adaptation (Gould & Lewontin, 1978). It is possible that a similar issue may arise with preferences that are interpreted to be sex-specific or not sex-specific. As an example, it is possible that short-term mate preferences are actually adaptations in men but not in women.

Several studies have studied long- and short-term mating preferences via several different research methods, often in conflicting choice-situation (see Conroy-Beam & Buss, 2019Cottrell et al., 2007Mogilski et al., 2019Perilloux & Cloud, 2019). As their experimental designs and methods differ from the current study, their use as a comparison against the results for this study is not completely straight-forward.

There were some limitations in the sample. The data was self-reported. However, self-reported preference measures are the most commonly used method in other studies of human mate preferences as well. It is also possible that people who are seeking a short-term relationship differ in their attractiveness from the ones who are seeking a long-term relationship. This, in turn, might have an influence on the preferences of the individuals. One final limitation of the study is that for interactions statistical power depends on the number of observations in the smallest cell, which in our case, corresponds to women respondents seeking short-term relationship (n = 63). There were 520 women respondents seeking long-term relationship, 92 men respondents seeking short-term relationship, and 325 men respondents seeking long-term relationship. The number of women seeking short-term relationship was low, as such, power to detect interactions if they exist was low, thus, the non-significant interactions should be interpreted cautiously. Future studies should pay attention the interaction-result that we presented in this study. In ideal case, a large number of women seeking for short-term relationships should be recruited for the study. Perhaps some innovative experimental design could also study this issue in the future.


Work perceived as immoral is associated with higher wages; the least moral types are more likely to be employed in such work

Sorting and wage premiums in immoral work. Florian H. Schneider, Fanny Brun and Roberto A. Weber. University of Zurich, Department of Economics Working Paper No. 353, June 29 2020. https://www.econ.uzh.ch/static/wp/econwp353.pdf

h/t Vitalik Buteryn Endnotes on 2020: Crypto and Beyond (vitalik.ca)

Abstract: We use surveys, laboratory experiments and administrative labor-market data to study how heterogeneity in the perceived immorality of work and in workers’ aversion to acting immorally interact to impact labor market outcomes. Specifically, we investigate whether those individuals least concerned with acting morally select into jobs generally perceived as immoral and whether the aversion among many individuals to performing such acts contributes to immorality wage premiums, a form of compensating differential. We show that immoral work is associated with higher wages, both using correlational evidence from administrative labor-market data and causal evidence from a laboratory experiment. We also measure individuals’ aversion to performing immoral acts and show that those who find immoral behavior least aversive are more likely to be employed in immoral work in the lab and have a relative preference for work perceived as immoral outside the laboratory. We note that sorting by “immoral” types into jobs that can cause harm may be detrimental for society. Our study highlights the value of employing complementary research methods.

Keywords: Wage premium, immoral behavior, sorting, experiments

JEL Codes: C92, J31, D03


7. Discussion and Conclusion

We investigate whether heterogeneity in individual preferences for avoiding immoral work and the perceived immorality of work influence the jobs that individuals select and individuals’ earnings. Our study employs a laboratory experiment, surveys and administrative data to identify heterogeneity in concerns for morality and to create (or, measure) variation in the immorality of jobs. We use these different kinds of data to test two main hypotheses— first, that jobs generally perceived as immoral will yield a wage premium and, second, that individuals less concerned with moral behavior will be more likely to sort into such jobs.

In a laboratory setting, we use a simple behavioral task to classify individuals as “moral” or “immoral” types. We then show that this characteristic predicts the outcomes that individuals obtain as we experimentally vary only the immorality of work. We find support for both our hypotheses. Labor markets for immoral work yield significantly higher wages. Moreover, immoral workers are significantly more likely to be hired for immoral work than are moral workers; but this relationship disappears in a labor market for neutral work. We also find that a market for immoral work benefits immoral types, particularly when they compete with fewer other immoral types.

We separately use survey responses to classify the immorality of real-world firms and industries and show that industries classified as immoral pay higher wages. We also use surveys to obtain a separate measure of workers’ moral types. This individual characteristic is correlated with the moral type measured in the laboratory and predicts subjects’ behaviors in the laboratory labor market. Moreover, both the survey-based and lab-based measures of morality also predict stated preferences for working in firms and industries outside the laboratory that vary in their perceived morality. Using either measure, workers less concerned with morality are more willing to work for firms and industries generally perceived as less moral.

Given the significance of many social ills produced by immoral work practices, such as deceptive marketing of socially harmful products, our study sheds important new light on the interaction between individual’s types, their willingness to do immoral work and the resulting labor-market outcomes. Our work also has several important policy implications.

First, in those jobs and industries with the greatest potential to do societal harm, social welfare will often be higher when workers voluntarily internalize the negative impacts of their actions and forgo potentially profitable opportunities. For instance, a weapons manufacturer may restrict sales to conflict areas if top management has a moral aversion to the social harm caused by such sales. However, our evidence suggests that it is the least moral types who will sort into these industries and that, therefore, labor market sorting will make it less likely that such internalization will occur. This also creates an important contrast between our findings and related work on sorting by mission-oriented types into firms or jobs with a pro-social orientation. In such cases, sorting may often be beneficial for society, as those who care about a cause become the ones who impact it; in our case, however, those who care about doing “good” may avoid the opportunity to determine how much “bad” takes place.

Second, another implication of our empirical findings is that the perception that a firm, industry or type of work is immoral may be self-reinforcing. If, as our results indicate, the perception that work involves immoral acts leads people less concerned with acting morally to differentially opt into such work, then the end result of such sorting may be a workforce more likely to commit immoral acts. Even if some of the firms and industries that we study do not actually involve any inherently immoral activities in their line of work, the fact that they are disproportionately more attractive to people more willing to do immoral things may result in a self-confirming greater prevalence for immoral behavior. Thus, firms and industries that regularly confront the perception that they involve immoral work—such as the banking sector—may need to be particularly attuned to such selection in their hiring.

Finally, our theoretical analysis predicts—in line with our experimental data—that the least moral types are overcompensated by the immorality premium. This is in stark contrast to Mankiw’s (2010) “just deserts theory”—that is, everybody should receive his or her contribution to society. Our work suggests a perverse case in which those willing to do the most socially harmful acts may instead benefit from doing so. Moreover, this benefit is the direct result of the actions by others who are concerned with behaving morally. Indeed, we provide evidence that a shift in preferences toward a greater aversion to performing immoral work may reward those individuals who are least concerned with morality. Of course, our work leaves open many important questions regarding the precise characteristics that lead some kinds of work to be differentially perceived as immoral and the specific nature of the preference underlying workers’ market behavior. Nevertheless, as the above examples make clear, the differential sorting by people more or less concerned with immoral behavior into different lines of work has important implications for the extent to which market activity yields beneficial social outcom


Feminist men report having sex with women more recently than those who do not call themselves feminists; our study cannot answer whether some men adopt a feminist identity to increase their access to sexual partners

Feminist Men and Sexual Behavior: Analyses of Men’s Sex with Women. Max Stick, Tina Fetner. Men and Masculinities, December 24, 2020. https://doi.org/10.1177/1097184X20980789

Abstract: Men’s identification with and support for feminism has attracted the interest of masculinity scholars. This study explores an under-researched dimension of this phenomenon, investigating the relationship between feminist identification and sexual behavior. In heterosexual encounters, do feminist men report having sex more recently than those who do not call themselves feminists? During sexual encounters, do feminist men behave differently than non-feminists? In particular, do feminist men organize their sexual behavior in a way that prioritizes their partners’ sexual pleasure to a greater extent than non-feminists? Using representative survey data of Canadian adults, we examine the self-reported sexual behavior of heterosexual Canadian men. We find that self-identifying feminist men report having sex more recently and are more likely to report engaging in breast stimulation and performing oral sex on their partners than non-feminists. We discuss the implications of these findings on the sociological literature on gender and sexuality.

Keywords: masculinity, feminism, sexuality, sexual behavior, Canada


The Aesthetic Self. The Importance of Aesthetic Taste in Music and Art for Our Perceived Identity

The Aesthetic Self. The Importance of Aesthetic Taste in Music and Art for Our Perceived Identity. Joerg Fingerhut,  Javier Gomez-Lavin, Claudia Winklmayr and  Jesse J. Prinz. Front. Psychol. Dec 2020, doi: 10.3389/fpsyg.2020.577703

Abstract: To what extent does aesthetic taste and our interest in the arts constitute who we are? We present a series of empirical findings of an Aesthetic Self Effect supporting the claim that our aesthetic engagements are regarded as a central component of our identity: we have aesthetic selves. Counterfactual changes in aesthetic preferences, for example, moving from liking classical music to liking pop, are perceived as altering us as a person. This Aesthetic Self Effect is as strong as the impact of moral changes, such as altering political partisanship or religious orientation, and significantly stronger than for other categories of taste, such as food preferences (Pretest/Study 1, n = 251/359). Using a multidimensional scaling technique to model perceived aesthetic similarities among musical genres, we determined that aesthetic distances between genres correlate highly with the perceived difference in identity (Pretest/Study 2, n = 45/364). Further studies generalize the Aesthetic Self Effect beyond the musical domain: general changes in visual art preferences, for example from more traditional to abstract art, also elicited a strong Self Effect (Study 3, n = 237). Exploring the breadth of this effect we found an additional Anaesthetic Self Effect. That is, hypothetical changes from aesthetic indifference to caring about music, art or beauty are judged to have a significant impact on identity. This effect is stronger for aesthetic fields compared to adopting leisure activities, such as hiking or playing video games (Study 4, n = 305). Indeed, across our studies the Anaesthetic Self Effect turns out to be even stronger than the Aesthetic Self Effect. Taken together we found evidence for a link between aesthetics and identity: when our taste in music and the arts or our aesthetic interests change, we take these changes to severely transform us.


Group size is more frequently two; four mechanisms that may help explain this finding are reciprocity, coordination, social exclusion, and reproduction

Peperkoorn LS, Becker DV, Balliet D, Columbus S, Molho C, Van Lange PAM (2020) The prevalence of dyads in social life. PLoS ONE 15(12): e0244188. Dec 2020. https://doi.org/10.1371/journal.pone.0244188

Abstract: A salient objective feature of the social environment in which people find themselves is group size. Knowledge of group size is highly relevant to behavioural scientists given that humans spend considerable time in social settings and the number of others influences much of human behaviour. What size of group do people actually look for and encounter in everyday life? Here we report four survey studies and one experience-sampling study (total N = 4,398) which provide evidence for the predominance of the dyad in daily life. Relative to larger group sizes, dyads are most common across a wide range of activities (e.g., conversations, projects, holidays, movies, sports, bars) obtained from three time moments (past activities, present, and future activities), sampling both mixed-sex and same-sex groups, with three different methodological approaches (retrospective reports, real-time data capture, and preference measures) in the United States and the Netherlands. We offer four mechanisms that may help explain this finding: reciprocity, coordination, social exclusion, and reproduction. The present findings advance our understanding of how individuals organize themselves in everyday life.


Discussion

The results of the present studies provide strong evidence for the prevalence of the dyad in daily life. Our data show that dyads are most common across a range of activities (e.g., conversations, projects, holidays, movies, sports, bars) obtained from three time moments (past activities, present, and future activities), sampling both mixed-sex and same-sex groups, with three different methodological approaches (retrospective reports, real-time data capture, and preference measures) in the United States and the Netherlands. With some exceptions, we also found similar patterns for men and women.

These results are in line with classic research conducted in the 1950s and 1970s [1417] which also found dyads to be the most common group size at local sites with local participants (also see recent work by Dunbar and colleagues for conversation sizes specifically, [1821]). Yet how do dyads relate to other, larger, group sizes? What theory can be used for understanding the prevalence of the dyad in daily life? To integrate the dyad with other group sizes, we briefly review Caporael’s Core Configuration Model [5,6] and although we did not test causal mechanisms in this study, we subsequently provide four possible reasons for the predominance of the dyad in everyday life.

The core configurations model

Caporael [5,6] proposes that face-to-face groups are hierarchically structured in four core configurations. These represent kinds of interdependent interactions between people. The interdependency is determined by the type of task to be performed, the situated environment, and the physical attributes of the participants (i.e., “body morphology”). Group size is an important feature of the model but the core configuration sizes have little meaning without invoking the associated activities to be performed. The notion of being “core” configurations reflects the idea that these group configurations can be repeatedly observed in hunter-gatherer groups (and hence presumably across human evolutionary history), across an individual’s lifespan, and in daily life. These configurations occur for a specific purpose and the successful accomplishment thereof accounts for their continued recurrence. Moreover, it is proposed that cognitive processes have evolved and developed in the context of the core configurations and thereby also cause their recurrence over time.

The first core configuration is the dyad, an interaction between two entities (e.g., two humans, one human and an animal, or human-AI interaction). Tasks addressed by the dyad are, for example, internal fertilization in the context of sex, providing infant nutrition while the mother is breastfeeding the infant, or a child’s interaction with a peer or adult. One proposed function of the dyad is microcoordination (e.g., during facial imitation, gait adjustment between two people, or interactional synchrony during courtship initiation).

Second, the work group, family group, or team has a modal group size of 5 individuals and a range of 3–7 people. It refers to interactions in small face-to-face groups that have a common task orientation. Examples of small group tasks are foraging, hunting, and gathering. A modern world example would be the completion of an assignment by ad-hoc groups of students working together for brief periods in class (e.g., devising and assembling a basic technological device). The workgroup affords the function of distributed cognition across group members. This refers both to the sharing of cognitive resources (e.g., perception, knowledge, cues, focus of attention, inference, classification) in the pursuit of a shared representation of the task or problem and to a division of cognitive effort where there are specialties in the group over time allowing role-based trust.

Third, the deme, band, microband, or conceptual deme has a mode of 30 individuals with a group size range of 25–50 people. Note that Marilynn Brewer (a close collaborator of Caporael) proposed a group size range of 50–200 people for the deme/community in her 2015 keynote at the ICSD conference in Hong Kong, China [24]. The deme is similar in size to the extended family and modern-day classrooms. Common tasks of the deme are the movement from one location to another in hunter-gatherers, providing college students skill-based education (physical or cognitive), and the integration and redistribution of resources retrieved from smaller workgroups (e.g., meat from the hunt or research results from the lab). This configuration has the function of constructing a shared reality or worldview, a common bond identity and common knowledge. It also allows cooperative alliances to emerge which may lead to the breakaway of group members to form their own group in case of conflict or when the environment’s carrying capacity is reached.

Finally, the macrodeme or (seasonal) macroband has a modal group size of 300 individuals with a range of 100–500 individuals. Brewer [24] has instead provided a range of “300–1000 and beyond” people. The task of the macrodeme is the (seasonal, annual) gathering of people (bands of hunter-gatherers, scientists, businessmen) in the pursuit of exchanging resources, people, or information about more distant places or groups. The between-group mobility of people may involve mates in the case of hunter-gatherers or staff in the case of business or science. The function of the macrodeme has been proposed to be the stabilization and standardization of language. Stabilization indicating words referring to the same thing and standardization indicating that the members often use and understand the same words.

Whereas the dyad can rely on synchronization and reciprocity to obtain goals [25,26] and whereas workgroups can rely on mutual performance monitoring [27] and self-regulation [28] to achieve their objectives, members of the deme have to resort to other mechanisms to sustain cooperation and coordination. Several mechanisms have been proposed in the literature. These include informal sanctions [29]; indirect reciprocity [30]; and descriptive norms [31]. However, whereas in demes people are all individually known, this is not the case in macrodemes. As such the likelihood of knowing someone’s reputation may be insufficiently high for indirect reciprocity to serve as an effective mechanism to foster and maintain cooperation and coordination in very large groups [32]. In addition, it has been argued that there are mental constraints preventing people to keep track of the reputation of a large number of people at the same time [13,33]. Hence, other mechanisms such as proscriptive norms, institutions, and formal sanctions [34,35] may be necessary for people to work effectively together toward the accomplishment of their goals in macrodeme configurations.

Four reasons for the prevalence of the dyad

More than larger sizes, dyadic interactions enable benefits from direct reciprocity [26]. In repeated dyadic interactions, the threat of non-reciprocation supports cooperative strategies such as tit-for-tat [26,36], but this mechanism breaks down in larger groups where freeriding is possible [37]. In a dyad, such strategies are efficient, as one’s choices are noticeably affecting one’s own and other’s outcomes. The partner may readily perceive that the other prefers stable cooperation over mutual selfishness, where exploitation would meet swift retaliation. However, even the small step from dyads to triads causes complexities in the workings (and effectiveness) of reciprocity, as one’s retaliatory action can no longer be delivered to the desired target only, the third person is equally affected [38]. When group size increases further, one is increasingly less able to induce cooperation in others through strategic signalling. One individual’s actions are both less likely to be perceived, and less likely to affect the others’ outcomes [37]. Thus, being part of a dyad (compared to larger sizes) allows for more control over the social situation toward the accomplishment of mutually beneficial outcomes.

Second, the detection of emotions and mental states via nonverbal cues is most likely to occur in dyads. Yet as group size increases, it becomes increasingly challenging to attend n-1 communication channels in a group comprising n members [39]. Moreover, as group size increases, the number of interpersonal linkages along which coordination may be required increases sharply [39]. Indeed, coordination with multiple individuals is computationally complex and therefore individuals should prefer interaction partners with whom coordination is easier [40]. Coordination is more efficient with familiar others as learning about others’ preferences, intentions, and traits allows for improved behavioural anticipation [41]. As mentioned above, dyads are expected to be in a better position to attribute mental states (e.g., intentions, emotions, beliefs, desires, knowledge), to be clearer communicators, and hence to be behaviourally more predictable than larger groups. Thus, if people prefer those with predictable behaviours as interaction partners, such as familiar others [41], people may also have a preference for dyadic activities given their relatively predictable form and hence lower cognitive effort, compared with larger group sizes.

Third, group living provides various benefits. These include cooperation in the pursuit of difficult tasks, protection against danger, directing and receiving altruistic acts toward and from kin, the availability of allies, and high concentrations of mates [42]. However, group life also comes with liabilities, mostly in the form of conflict and competition. Examples include competition over material resources, high-status positions, or romantic partners. As such, individual competitors may want to exclude others who pose a threat to their interests (e.g., through derogating one’s competence and appearance or through spurious accusations [42]). Being excluded in an ancestral environment yielded dire prospects for survival and reproduction. Therefore, it is not surprising that (the threat of) social exclusion gives rise to anxiety [43] and social pain [44] and that people may be afraid to deviate from the group. Moreover, it has been argued that humans are highly sensitive to actual and threatened rejection and may possess an ostracism-detection system biased toward overdetection [45,46]. Whereas in larger groups social exclusion is a possibility, sometimes a threat, in a dyad one cannot exclude the other person without bearing the cost of becoming alone oneself. Seeing others together when one is alone may trigger affective distress, being a reminder of (the threat of) social exclusion [43] and as such is uncomfortable. Thus, even if one is generally embedded in a larger group and as such reaping its benefits, the dyad specifically may provide a relatively comfortable unit for social interaction in which self-monitoring and self-censorship can be somewhat relaxed, allowing for more authentic behaviour (i.e., in line with one’s own preferences and idiosyncrasies).

The final argument that may help explain the prevalence of the dyad in daily life concerns reproduction and infant-rearing activities, which take place mostly in pairs [5,6]. As mentioned above, this includes interactional synchrony during the initiation of courtship, which is a process confined to the mating couple [47]. For instance, it is difficult to conceive how three people would be able to nearly continuously look into each other’s eyes. Moreover, most people choose to engage in sexual activity with one other person during a sexual encounter. Next, the provision of infant nutrition through breastfeeding typically involves the mother and the infant in a breastfeeding dyad. Finally, alloparents (e.g., the grandmother [48], great-aunt or an older sibling [49]) can assist the parents in infant-rearing by watching over or feeding the child, allowing the parents to allocate their time in the pursuit of other activities. We argue that even with extensive alloparenting, the dyad may still be the most functional unit for interaction between caregiver and young child (e.g., allowing for facial imitation [5,6] and more efficient feeding).

Yet how do these theoretical reasons connect to the activities that were probed in the present studies? Our activities may have had low risk of exploitation, low costs in terms of commitment, and involved the exchange of resources likely of little consequence. However, the four proposed mechanisms we suggest are all attuned to the two-person interaction in daily life. This applies to reciprocity where the interactants engage in exchange and turn-taking. For instance, paying for the drinks at the movies, one may expect the other to pay next time. Reciprocity can happen “on the spot” (e.g., mutual self-disclosure), but more frequently unfolds in a step-by-step manner. As social transactions recur and gradually expand in significance, reciprocity results in fortified interdependent social bonds [10] (in the words of Allen et al. [50]: strong pair-wise ties). Besides trust, these processes are often accompanied by a series of emotions that may serve as internal pressures to maintain interpersonal cooperation: feelings of indebtedness, personal obligation, appreciation, and gratitude after receiving a favour [51]; feelings of guilt when failing to reciprocate [52]; anger when receiving substandard exchange [53], and forgiveness to pardon having been short-changed [54]. If the above line of reason is correct, it suggests the interesting hypothesis that various feelings and emotions such as indebtedness and gratitude are experienced most commonly and most strongly toward one other person instead of toward groups.

Second, in terms of coordination, in a dyadic interaction one enjoys relatively noise-free information. Compared to larger group sizes, dyadic interactions offer less room for hiding (e.g., question evasion, self-concealment [55]) and ambiguity, and this facilitates coordination. For instance, when a group of friends are planning a holiday trip, it may require significantly more effort to agree on a departure date and destination, not to mention all the small decisions to be made once there, compared with a dyad planning a similar trip together. Third, in a range of situations (e.g., holidays, projects, bars), being in a dyad may also entail being tied to another person, incentivizing both persons to be accommodating and cooperative to prevent social exclusion and having to fend for oneself in a sometimes daunting world. In a bar or club, for instance, it may feel uncomfortable to have no one to fall back on if needed. Finally, dyadic interaction also allows for relatively unconstrained flirting behaviour compared to situations where third-parties are present [56], reflecting the role of reproduction in the prevalence of the dyad in daily life. During a dyadic conversation, for example, a subtle courtship attempt (e.g., a wink or prolonged eye contact) can more freely be sent without potential interception costs from third parties. These mechanisms help to explain the primacy of the dyad because each of them supports cooperation in smaller groups–by which all individuals involved benefit. This is not to imply that cooperation does not occur in larger groups. We suggest that it is more challenging, however, and it involves qualitatively different mechanisms, such as indirect reciprocity [30] or third-party punishment [57]. In this empirical paper we echo the importance and prevalence of the (cooperative) dyad in society that has recently been demonstrated in mathematical and simulation analyses by Allen et al. [50].

Sex differences

For our mixed-sex diary data (Study 5) we did not find significant sex differences. There were some significant differences for mixed-sex group activities sampled in Studies 1–3, but these were generally small. However, for same-sex groups in Study 4, the sex differences were somewhat more pronounced, where women are more dyadic than men. This applies for instance to sports (36.6% vs. 21.1%) or going to the movies (55.7% vs. 42.2%) and to a lesser extent going on holiday (49.9% vs. 43.1%) or out for dinner (37.8% vs. 32%). This latter finding is consistent with research on same sex friendship [58]. Using a sample of 112,000 Facebook profile pictures the authors found women to favour dyadic relationships, while men preferred larger male groups. Moreover, it has been demonstrated that men and women process social information differently in line with these differences in social structures, where women focus more on individuals while men focus more on groups [59,60]. Although the exact mechanism is unknown, various complementary explanations have been put forward to account for these sex differences (e.g., patrilocality [61]; by-product of pair-bonding [62]; maternal caregiving and empathic potential [63,64]).

Limitations

Studies 1–4 have three limitations in common that should be taken into account. The situations that were sampled may not have been exhaustive or fully representative of daily life. Although we asked for the last time someone engaged in the activity, for some people it may have been weeks ago. Participants may also selectively remember an instance they particularly enjoyed. These concerns were addressed by Study 5, which used a real time approach in which the reference period is the present, or the last hours, and in which situations are sampled randomly. A limitation of Study 5, however, is that we did not code the situations that people reported. This is a fruitful avenue for future research given that situation taxonomies are currently lacking consensus and may be advanced by diary studies.

The current research focused on social activities and hence we have not probed individuals conducting activities alone [14,15]. Therefore, we do not know whether dyads would also be most frequent compared to individuals for various different tasks. In our studies, most participants engaged in activities with a few other people. However, there are certainly activities that require larger groups (e.g., barn-raisings). Future research could investigate whether even there, dyads may form the most effective subcomponents (“you hold the spike while I swing the hammer”). Indeed, given the aforementioned arguments (reciprocitycoordinationsocial exclusionreproduction), these activities may also predominantly yield the dyad as the most common subunit in a larger collective. Related to this, because we focused on direct interaction, and used a definition of a group as “two or more persons who are interacting with one another in such a manner that each person influences and is influenced by each other person” [23], our results do not address the fact that groups can have psychological significance beyond direct interaction [65] and measuring groups differently may yield different results.

Future research

One issue remaining for future research is to provide empirical evidence that the dyads that are so prevalent in various domains are also stable over time. Moreover, future research would do well to investigate the prevalence of the dyad in collectivistic daily life, to determine whether dyad inflation is generated by cultural factors as a necessary requirement for the effect to occur or whether an evolutionarily ingrained predisposition is sufficient in and of itself. If the former holds, then the effects should be observed only in individualistic countries, whereas if the latter holds, as argued by Caporael’s core configuration model [5,6], then the effects should be observable across individualistic and collectivistic cultures, and more generally, around the world.