Monday, September 5, 2022

Pedestrians gave more than twice as much money to the guy wearing higher-class symbols than they did to the one wearing lower-class symbols, which were perceived as of elevated competence, trustworthiness, similarity to the self, and perceived humanity

The influence of signs of social class on compassionate responses to people in need. Bennett Callaghan, Quinton M. Delgadillo and Michael W. Kraus. Front. Psychol., August 25 2022. https://doi.org/10.3389/fpsyg.2022.936170

Abstract: A field experiment (N = 4,536) examined how signs of social class influence compassionate responses to those in need. Pedestrians in two major cities in the United States were exposed to a confederate wearing symbols of relatively high or low social class who was requesting money to help the homeless. Compassionate responding was assessed by measuring the donation amount of the pedestrians walking past the target. Pedestrians gave more than twice (2.55 times) as much money to the confederate wearing higher-class symbols than they did to the one wearing lower-class symbols. A follow-up study (N = 504) exposed participants to images of the target wearing the same higher- or lower-class symbols and examined the antecedents of compassionate responding. Consistent with theorizing, higher-class symbols elicited perceptions of elevated competence, trustworthiness, similarity to the self, and perceived humanity compared to lower-class symbols. These results indicate that visible signs of social class influence judgments of others’ traits and attributes, as well as in decisions to respond compassionately to the needs of those who are suffering.

General discussion

As economic inequality rises in many parts of the world, and countries such as the United States roll back social safety net programs (Piketty, 2014), the responsibility for dealing with inequality’s deleterious impacts (Wilkinson and Pickett, 2009) has increasingly fallen to economically precarious individuals themselves or to private citizens exercising compassion, defined as concern for the suffering of others and the motivation to help improve their circumstances (e.g., Goetz et al., 2010Gilbert, 2017Mascaro et al., 2020). Building on prior research and theorizing in the rich tradition of research on sympathy, empathy, and compassion (Batson et al., 1989Cialdini et al., 1997Oveis et al., 2010), the current research examined the tendency for people to respond compassionately (or not) in the presence of those who were apparently suffering or, at least, made salient a concern with suffering related to poverty and homelessness (i.e., a panhandler), in two cities in the United States. The current research suggests that people respond more compassionately, and perceive such individuals more favorably, when they signal higher–relative to lower–social status through physical appearance. This pattern of results arose even though all confederates and targets appeared to be generally low in status, and it arose in an experimental, but ecologically valid, context where participants shared their own money.

This research also contributes to a longstanding body of research suggesting that non-verbal status cues influence behavior on the part of others (e.g., Bickman, 1971Tracy et al., 2018). That symbols of high social class more than doubled the donations of pedestrians over a 4-h period indicates their power in shaping initial judgments of others’ basic human traits and in eliciting compassionate responses in everyday life. Importantly, our results align with past theory and research suggesting that high status signaling provides many direct benefits to individuals, including grooming and mating partners in non-human primates (Sapolsky, 2004). This research adds received generosity, among humans, to this list of benefits.

Interestingly, mere novelty and noticeability of the higher status confederate do not seem to explain observed differences in generosity. In the field experiment, the mere frequency of the interactions did not differ by condition; in Study 2, in fact, participants were in fact more likely to attend to the lower status target. Instead, the quality of these interactions and their outcomes (as indicated by the analysis of extreme donations) differed. Anecdotally, this qualitative distinction bears out. When people did go out of their way to speak to the confederate, the higher status one received comments such as “I usually don’t give money to people on the street, but you seem like a nice guy.” In one case, a pedestrian (also donned in a business suit) even dropped a business card into the higher status confederate’s collection cup–a tacit invitation for the confederate to seek employment, rather than a trivial one-time donation.

As discussed, the large donations of $5 or $10, given their size and exclusive presence in the relatively higher status trials, likely contribute substantially to some of the effects we observe in the field study. Much like the interactions sketched above, these donations might also represent a qualitative shift in how donors approached the situation: they may have donated $5 or $10 in the hopes of more effectively meeting the confederate’s immediate perceived needs, as such an amount would be more appropriate than more common donation amounts (e.g., $1 or less) for most self-care and survival needs, such as purchasing a meal. Thus, these donations might be particularly representative of compassionate responding insofar as they are intended to effectively and (depending on participants’ construal of the situation) immediately alleviate suffering. However, they also suggest the possibility of theoretical accounts we did not fully theorize. For instance, it is possible that status signaling is most effective at eliciting high-variance responding; in other words, signaling higher status might not strongly impact tendencies to engage in compassion in general, but, rather, impacts tendencies to engage in extreme–as defined in relation to more typical donation amounts–acts of compassion (again, however, use of the word “extreme” might be misleading, as these donations might also be described as simply independently sufficient to meeting the goals at hand).

It is also possible that this pattern of results reflects an unobservable moderation effect. Perhaps, for instance, the effects of status signaling are most pronounced among those who are more inclined to acts of extreme generosity to begin with. Alternatively, this effect might be attributable to the presence of stronger effects among participants who are higher in SES themselves. The design of the field experiment study did not allow us to assess the SES of passersby, and, thus, whether participants’ own social class characteristics contributed to decisions to respond compassionately to the confederate. As indicated by the overall low levels of subjective SES attributed to the target in the perceptual study, it is likely that the higher status confederate was perceived as closer to participants, in terms of socioeconomic standing, than the lower status confederate across the board (excluding those who are themselves poor or unhoused). Still, however, the perceptual study does suggest meaningful differences in self-other similarity according to status signaling condition, and the possibility that signalers who better “match” the status of perceivers benefit from even greater compassion than those who merely signal higher status has received mixed empirical support (see, e.g., Goodman and Gareis, 1993). Thus, it is possible that high status signals appealed specifically to passersby of particularly high SES and who, due to greater access to financial resources, may have stood to lose less through larger donations or simply regarded higher amounts of money as an appropriate default for donation (as a proportion of the money they had on hand, for instance). Though it may be difficult to measure individual differences such as predispositions toward extreme generosity in a field study context, future replications of this research might employ methods of subjectively coding participant SES (e.g., Bickman, 1971) or systematically varying the SES characteristics of the research sites (e.g., Goodman and Gareis, 1993) in order to determine the regularity with which these extreme donations occur and whether they are given disproportionately by those of higher socioeconomic standing.

Together, these qualitative experiences and extreme donation profile provide some support for the general pattern observed in Study 2, and support a central tenet of theories of compassion: that compassionate responding hinges on the reputation of targets, especially with respect to their likelihood of engaging in reciprocal cooperation with other prosocial individuals (Goetz et al., 2010). The present research adds signals of social class as a possible cue that reliably elicits such reputational perceptions.

Moreover, high status signals increased specific judgments of competence, trustworthiness, humanity, and self-other similarity. Thus, the results of the current studies suggest that poor individuals who adopt these symbols might be seen as more effective at converting gifts into intended outcomes (such as personal advancement or care), as less likely to engage in behaviors that might be seen as making them blameworthy for their plight (e.g., drug or alcohol use; see Goetz et al., 2010), and as more likely to use those gifts for intended means rather than as a strategy to accrue undeserved wealth. In short, such signals may make one appear more deserving of compassion (Goetz et al., 2010).

A closely related alternative explanation for the current set of results, which more strongly emphasizes the perceived ability (rather than the inclination) to engage in future prosocial behavior by the confederate, is that participants were more likely to see the higher status confederate’s need state as temporary, rather than chronic. Consistent with certain evolutionary accounts of reciprocal altruism (e.g., Sugiyama and Sugiyama, 2003Tracy et al., 2018), the perceived combination of high temporary need and high baseline competence may have biased individuals toward helping the higher status confederate in his time of need because he was perceived as more able to help others, or “pay it forward,” when he had the opportunity to do so. Given that the high and low status targets were strongly discriminated along the lines of competence, this alternative explanation is plausible. Future research is needed, however, to determine whether such perceptions of ability to engage in future acts influence compassionate responding independent of perceptions of deservingness.

In a similar vein, our field study operationalizes compassion as costly helping behavior–a common method of doing so within the social-psychological literature and one that avoids many of the biases inherent in self-report measures (Mascaro et al., 2020). Our second study also includes a number of social perceptions that index deservingness, an antecedent to compassion in prevailing theoretical accounts of the construct (e.g., Goetz et al., 2010). While this research demonstrates the influence of status signaling on theoretically important perceptions of a target (Study 2) and responses toward a confederate (Study 1), this research does not measure compassion, as a subjective psychological state, directly. Nor does the second study measure compassionate responding directly, as in Study 1. Thus, the two studies taken together show a pattern that is consistent with a theoretical account emphasizing compassion: one in which status signaling affects particular theoretical antecedents of compassionate responding (i.e., warmth, competence, self-other similarity, and ascribed humanity), which then influence compassion and compassionate responding. However, these results do not necessarily confirm that status signaling directly influences perceptions linked to deservingness and, subsequently, compassion and compassionate responding.

To address this theoretical gap, future research might attempt to measure compassion directly and demonstrate that signaling relatively higher (as compared to lower) status–by way of heightened perceptions of deservingness–heightens self-reported compassion for those suffering in the relevant context as well as subsequent compassionate responding (i.e., donations). In doing so, researchers should be mindful of best-practices in the measurement and definition of this complex emotion (Gilbert, 2017Mascaro et al., 2020). For instance, such research might attempt a multi-method approach to conceptualizing and measuring compassion that synthesizes quantitative reports of one’s own and others’ mental states, physiological measurements, and observations of behavior (e.g., Mascaro et al., 2020). Additionally, such research might take care to distinguish compassion from subjective and emotional states–such as distress, sadness, and love–that are sometimes used interchangeably with compassion in the literature (e.g., Goetz et al., 2010Gilbert, 2017). Second, in order to test the full theoretical model we have proposed here, future research should manipulate status signaling and measure both the antecedents we propose and compassion (or compassionate responding) within the same study. Such a study could at least determine whether the key variables related to deservingness mediate the effect of status signaling on compassion. Ideally, future research could also manipulate these mediators to establish a truly causal chain of effects (Spencer et al., 2005).

It is also possible, however, that conditional differences in confederate behavior contributed to differences in generosity on the part of passersby. The confederate was not blind to condition or hypotheses, and previous research suggests that donning high status sartorial signals can change the behavior of even naïve participants (Kraus and Mendes, 2014). Though this is a possibility, we minimized this likelihood by having the confederate behave consistent with standardized instructions. Moreover, that a follow-up study elicited theoretically relevant patterns of perception from passive observers suggests that the effect of status signaling on generosity observed in the field is at least partially driven by perceiver judgments. Finally, even if the behavior of the confederate did subtly differ between conditions, such subtle differences would need to compete with the cacophony of stimuli that individuals normally encounter when walking down a busy street in New York or Chicago, so the context in which we chose to conduct our field experiment also mitigates concerns with experimenter effects.

Indeed, it was partially because we expected multiple competing demands on the attention of passersby that we chose to manipulate comparatively obvious visual cues (combined with spoken statements to draw attention), rather than other cues that also signal status, such as vocal pitch (e.g., Gregory and Webster, 1996), accent (e.g., Labov, 2006Kraus et al., 2019), or cultural signifiers of aesthetic taste (Bourdieu, 1984). Nonetheless, these other modalities represent interesting potential avenues for future research.

Similarly, those who did attend to visual cues of status also likely perceived other superficial but potentially important characteristics, such as those that indicate membership in particular social identity groups. It is interesting to speculate about how these other characteristics of the confederate (i.e., an individual generally perceived to be White and male) may have impacted the effect of status signaling on compassionate responding. For example, membership in other social categories might modify the results observed in these experiments. Theoretical accounts suggest that symbols of social status influence perception similarly across race and gender (Major and O’Brien, 2005), but previous research also finds that social status and race or gender may interact in subtle ways to produce marked differences in status-linked outcomes, such as health and mortality rates (Case and Deaton, 2015) or experienced bias and discrimination (e.g., Goff and Kahn, 2013Rivera and Tilcsik, 2016). Future research would need to determine if high status symbols confer the same benefits to members of other intersecting social groups as they apparently do for White men.

Future research might also measure how different characteristics of the giving context moderate how status symbols influence outcomes. For instance, some research has found that in contexts where individuals are already motivated to engage in prosocial behavior and are deciding how to distribute their resources, symbols of high status are negatively related to the receipt of altruism (Tracy et al., 2018). These researchers suggest that opposite patterns of effect with respect to status and altruistic behavior might arise depending on whether potential actors are deciding to engage in altruistic behavior in the first place or are deciding how to engage in such behavior. We echo these researchers’ calls for further investigation into this distinction as a potential moderator of the effect of status signaling on compassionate responding (p. 527). We also note that our results regarding the influence of status signaling manipulations on compassionate responses are perhaps bounded to compassionate responding in contexts involving the alleviation of suffering related to poverty and homelessness and to such responses enacted through brief, interpersonal exchanges. Thus, we caution generalizing these results to compassion directed toward other ends or within impersonal contexts, such as online behavior (see, e.g., Tracy et al., 2018).

Finally, we also acknowledge some ambiguity with respect to how participants themselves interpreted donating in the field experiment. As noted, the confederate only told participants that collected funds would be donated to charity if they had asked; few people interacted directly with the confederate in this way, and this pattern did not differ by condition. However, because we were constrained by ethical considerations in terms of what we could tell participants and the field context of the experiment made us unable to probe participants about their inferences regarding the confederate at the time they decided to donate (or not), we still do not know (as discussed) whether individual participants perceived the confederate as the primary benefactor of their donations or as an intermediary.

Even for participants operating under the latter assumption, however, the relevant behavior of donating nonetheless reflects the broader construct of compassionate responding, as those who donated were either donating directly to the target or helping him in his objective to raise money for charity (a goal that is aligned with the reduction of suffering). To this point, previous research has treated explicit contributions to third-party charities as an index for helping behavior directed toward a confederate (Pandey, 1979), and even those who donated under the assumption that the funds would be donated placed significantly more trust in the higher status than the lower status confederate–despite the lack of any guarantee the money would go to charity. Again, such a result is consistent with our overall theoretical expectation that relevant compassionate responding would be directed toward those presumed to be more honest and prosocial themselves, and it would at least appear that signaling status influenced decisions to engage in costly helping behavior (likely driven by differential patterns of social perception) regardless of how participants interpreted the situation. Still, the influence of status signaling on compassionate responding might depend on whether those signaling higher status themselves or third parties are the primary beneficiaries. Future research might investigate this distinction more explicitly.

These limitations and open questions notwithstanding, this research adds to existing models that highlight compassion, sympathy, and perceptions of deservingness as primary causes of compassionate responding (e.g., Goetz et al., 2010). Importantly, our results suggest that social status–and its accompanying interpersonal judgments–enters prominently into such processes. Ironically, low status individuals who appear to need the most help may end up receiving less of it than those who appear higher in status and more abundant in resources.

These results also have direct implications for rising levels of economic inequality in society. Given research suggesting that economic inequality and its negative consequences increase when social status is more visible (Nishi et al., 2015DeCelles and Norton, 2016), the current findings suggest that status symbols expressed through sartorial displays or other non-verbal behaviors are potential mechanisms for the perpetuation of economic inequality. We found that even among those engaging in ostensibly selfless behavior, individuals were more likely to enter into economic relationships with others who appeared higher, rather than lower, in social status. Given the high degree to which neighborhoods, professional networks, and daily life are stratified by social class, behaviors guided by status signaling can accrue and concentrate wealth and opportunity among a privileged few–further perpetuating inequality (see also Kraus et al., 2019).

These results may also hold implications for addressing economic inequality on a broader societal scale. As indicated by similar research in this domain, cross-status interactions in everyday life can perpetuate inequality by impacting support for social policy aimed at addressing it (Sands, 2017). Nonetheless, such policies are arguably likely to garner the most efficient redistributive outcomes, especially when one considers the alternatives. If subtle interpersonal cues, like clothing or similar indicators of status, shape the behavior of individual actors outside the context investigated in the current research, mechanisms of redistribution that rely on idiosyncratic preferences or the behavior of well-meaning individuals more broadly–such as large donations from wealthy donors to particular individuals or organizations–may be inefficient or underserve those who need the most assistance, whether such needs are met directly or through intermediaries (e.g., charities).

Those from denigrated groups, such as those suffering from homelessness, need monetary assistance despite lacking the ability to transmit status symbols that, as our results suggest, may make certain forms of compassionate responding (i.e., spur-of-the-moment donations) more likely. Moreover, not all charitable organizations aimed at helping such individuals may be equally adept at appealing to wealthy donors or motivating such individuals to donate in the first place. Depending on how far one may extrapolate the results reported here, our research suggests that such a process might require an understanding of how to leverage high status signals (on the part of charities themselves) or how to portray those in need in ways that emphasize their humanity, warmth, competence, and similarity to potential givers. By contrast, codified inequality-reducing policies (such as progressive taxation) do not rely on the generosity of individuals to meet their aims. Unfortunately, even well-meaning generosity, if dispatched at the level of individuals, may be biased by processes of person-perception that direct resources on the basis of attributes other than who is most needy or how resources can best be distributed.

IQ is a positive predictor of participation and spending in horse wagering; in addition, high IQ is associated with choosing more complex (high variance) betting products

Does IQ predict engagement with skill-based gambling? Large-scale evidence from horserace betting. Niko Suhonen, Jani Saastamoinen, David Forrest, Tuomo Kainulainen. Journal of Behavioral Decision Making, September 4 2022. https://doi.org/10.1002/bdm.2300

Abstract: We examine how measured intelligence, referred to as IQ, predicts a consumer's decisions on whether to participate in online horse wagering, how much to spend on those bets, and which horserace betting products to consume. We combine three individual-level archival data sets from Finland, including all online horse bets during a 1-year period from the state-sanctioned monopoly operator, the Finnish Defence Forces' IQ test scores from male conscripts born between 1962 and 1990 (N = 705,809), and administrative registry data on socioeconomic status, income, and education for these men. An analysis of male bettors (N = 15,488) shows that IQ is a positive predictor of participation and spending. In addition, high IQ is associated with choosing more complex (high variance) betting products. We find that these results are driven primarily by numerical IQ.

6 CONCLUSION

6.1 Discussion

This paper demonstrates that IQ, and especially its numerical ability subcomponent, is positively associated with an individual's decisions relating to participation in and expenditure on horse betting, and with a relative preference for complex betting formats. It is plausible that intelligent persons and those with numerical ability gain satisfaction from absorbing themselves in tasks involving “crunching numbers,” such as horse wagering. Consequently, our study provides empirical support for treating decisions on gambling as aspects of consumer behavior (Conlisk, 1993), at least in the case of skill-based gambling, as opposed to the proposition that gambling stems from behavioral biases (Barberis, 2012).

Consistent with Forrest and McHale (2018), our analyses suggest that IQ, and particularly numerical IQ, predicts participation in horse wagering. To some extent, this result is also congruent with Grinblatt et al. (2011) who find that IQ is positively correlated with an individual's decision to invest in the stock market, because skill-based forms of gambling and stock markets tend to attract similar individuals in terms of motivation and personality attributes (Arthur et al., 2016). As high-IQ men tend to spend more on horse betting products than low-IQ men do, our findings may also reflect the intellectual challenge quality of horse betting, as suggested by Binde (2013).

On the other hand, our results appear to be at odds with Gong and Zhu (2019), as none of their three measures of cognitive ability were significant predictors of which gamblers choose to engage in skill-and-chance games (as opposed to pure chance games, defined by them as comprising bingo, scratchcards, lottery, and keno). However, their list of skill-and-chance products included slot machines. Slot machines typically offer games where the outcome is random and tends to be generally regarded as a chance-based game (Stevens & Young, 2010). Furthermore, their data were a self-reported survey. Consequently, these aspects warrant caution when our results are compared with those presented in Gong and Zhu (2019).

Consistent with the hypothesis that a motivation for betting is the intellectual challenge (Binde, 2013; Johnson & Bruce, 1997, 1998), our results suggest that individuals seek to match their betting choice to their own level of IQ. Since high-IQ individuals appear to respond more to price (Grinblatt et al., 2016), it is fair to assume that these individuals in our context are likely to be aware of the lower take-out on easy products and in any case the different levels of take-out are clearly signaled by the operator. In particular, this result suggests that high numerical IQ consumers enjoy the intellectual challenge provided by complex betting formats and are willing to pay a greater take-out to play them. Consequently, their stronger preference for complex bets could reflect a genuine preference, which is consistent with intelligent persons exhibiting preference for performing challenging tasks (Cacioppo & Petty, 1981).

It is also possible to define the take-out rate as one of several “structural characteristics” which distinguish different gambling products from each other (Newall et al., 2021). When viewed purely through the lens of expected value, complex betting formats might appear as less attractive than simple ones (Newall et al., 2021), particularly for those bettors with a high numerical ability. In our approach, however, take-out is not regarded as a structural characteristic of the product, because it is not inherent to the product, but rather is a price chosen by the supplier in response to the nature of demand. In our findings, bettors with higher cognitive skills choose to purchase more complex products despite their high take-out rate. This implies that degree of complexity is the “structural characteristic,” and it is one for which they are willing to pay a higher price.

The administrative registry data facilitated the inclusion of controls reflecting demographic and socioeconomic status, which allowed us to draw conclusions about horse betting. For example, spending tends to increase with income, but slowly. This mirrors previous findings for gambling spend in general (Rude et al., 2014) and for lottery games (Combs & Spry, 2019) and implies that lower income individuals tend to allocate a higher share of their income to gambling (Castrén et al., 2018). Regarding age, engagement with horse betting appears to peak in middle-age, again similar to findings about participation in gambling generally (Welte et al., 2011). But whether the relatively greater engagement with horse betting in Finland among the middle-aged represents a cohort effect or a generational effect cannot be inferred from 1 year of data.

6.2 Limitations and future research

This paper has some limitations. Foremost, our data set includes only data on horse betting. Hence, our results may not generalize to other forms of gambling, most notably chance-based gambling. Further, we are unable to analyze propensities to problem gambling. Although some studies suggest that low IQ correlates with problem gambling (e.g., Hodgins et al., 2012; Rai et al., 2014), other factors that we could not observe are also likely to be relevant. For example, Parker et al. (2008) highlighted the role of emotional intelligence as a protective factor against the risk of developing problems.

The FDF data set also has some limitations. First, individuals have different incentives to effort when completing the test, which may bias IQ test scores. That is, if a conscript wishes to avoid training for a non-commissioned officer, which is more likely if he or she performs well in the IQ test, he or she may purposely underperform in the test. In addition, our measures of IQ may not be directly comparable to other studies, as IQ or cognitive ability is often operationalized in very different ways, depending on a study and its context. Second, some males are exempt from military service and some opt for non-military service instead. Third, the sample excludes the female population. Fourth, IQ scores were measured between 6 and 34 years prior to the betting transactions recorded in the study. To the extent that IQ can change over an adult's lifetime, this introduces measurement error into the analysis. Finally, the data are from a limited time interval and from a single country.

Our results open avenues for future studies. Rather than examining only a general measure of IQ, future studies should examine how the separate subcomponents of IQ predict consumer decision making. Intelligence and numerical ability could be instrumental in decisions relating to consumption, investment, and life outcomes in general. Future studies could also yield insight into theories of risk-taking behavior by addressing correlations between IQ and a person's risk preferences.

6.3 Concluding remarks

This paper demonstrates that a person's IQ predicts his engagement with horse betting. Our results show that IQ, and especially its numerical ability subcomponent, is positively correlated with participation in and expenditure on horse betting, and a relative preference for complex betting formats. These findings are consistent with skill-based gambling, or at least horse betting, being consumption of entertainment, which intelligent individuals enjoy. Thus, it is plausible that intelligent persons and those with numerical ability gain satisfaction from absorbing themselves in tasks involving “crunching numbers,” such as horse wagering.

Exogenous administration of testosterone increased impulsivity in heterosexual men, increasing preference for the smaller-sooner option, and inducing steeper discounting for the delayed option

Exogeneous testosterone increases sexual impulsivity in heterosexual men. Yin Wu et al. Psychoneuroendocrinology, September 5 2022, 105914. https://doi.org/10.1016/j.psyneuen.2022.105914

Highlights

• Exogenous administration of testosterone increased preference for the smaller-sooner option.

• Exogenous administration of testosterone induced steeper discounting for the delayed option.

• Exogenous administration of testosterone increases impulsivity for sexual rewards in heterosexual men.

Abstract: Testosterone has been hypothesized to promote sexual motivation and behavior. However, experimental evidence in healthy humans is sparse and rarely establishes causality. The present study investigated how testosterone affects delay of gratification for sexual rewards. We administered a single dose of testosterone to healthy young males in a double-blind, placebo-controlled, between-participant design (N = 140). Participants underwent a sexual delay discounting task, in which they made a choice between a variable larger-later option (i.e., waiting longer to view a sexual picture for a longer duration) and a smaller-sooner option (i.e., waiting for a fixed shorter period of time to view the same picture for a shorter duration). We found that testosterone administration increased preference for the smaller-sooner option and induced steeper discounting for the delayed option. These findings provide direct experimental evidence that rapid testosterone elevations increase impulsivity for sexual rewards and represent an important step towards a better understanding of the neuroendocrine basis of sexual motivation in humans.

Keywords: androgenimpulsivitysexual rewardintertemporal choicemating