Many objectives are achieved more effectively if everyone cooperates (e.g. group hunting). However, collaboration requires complicated decisions on how resources (e.g. prey) should be divided among collaborators. The current research focuses on decisions about accepting or rejecting a division of resources. Fairness considerations reduce the likelihood of accepting unfair offers (Brosnan & de Waal 2003), whereas visceral factors, such as hunger or thirst, may increase the acceptance likelihood of unfair distributions of resources (Loewenstein 1996). Evidence points out that testosterone may be related to increased fairness considerations (high testosterone males are more likely to reject unfair distributions; Burnham 2003), while an increase in testosterone levels after exposure to potential mates (Roney et al. 2003) may be associated with a higher likelihood of accepting small offers (Wilson & Daly 2004). We propose that sex cues moderate the relationship between testosterone and decisions about accepting or rejecting a distribution of resources.

The most widely used instrument in research on bargaining is the ultimatum game, developed by Güth et al. (1982). In the ultimatum game, two players are given the opportunity to split a sum of money. One player, the proposer, offers a fraction of the money to the other player, the responder. If the responder accepts the offer, the money is distributed in agreement with the proposer's offer. If the responder rejects the offer, neither player gets anything (Page et al. 2000). Proposers should realize that responders are willing to accept any positive take-it-or-leave-it offer since any amount of money is better than no money. Thus, standard economic theory predicts that proposers should offer very small amounts and responders should be willing to accept them (Rubinstein 1982). Nevertheless, systematic deviations from game theoretic equilibria are frequently observed. Proposers make generous offers and responders often reject small, albeit positive, offers (e.g. Thaler 1988; Nowak et al. 2000; Henrich et al. 2004). We hypothesize that exposure to sex cues will impact ultimatum game behaviour.

Sex cues lead to changes in what is perceived ‘just’ (Loewenstein et al. 1997; Ariely & Loewenstein 2006). Fairness norms prescribe a fifty-fifty split in ultimatum games (Nowak et al. 2000; Henrich et al. 2004). We hypothesize that the preference for this normative fair split is reduced after exposure to sex-related stimuli. Also, exposure to highly attractive women increases male subjects' preference for smaller, earlier rewards (Wilson & Daly 2004). We propose that the presence (versus the absence) of sex cues increases the likelihood of accepting small and unfair offers made by proposers because sex cues reduce fairness considerations and lead to a greater preference for small, immediate rewards.

We conjecture that this general effect will be more pronounced among men with lower digit ratios. Evidence suggests that ratio between the length of the index finger (2D) and the ring finger (4D) reflects the degree to which humans are exposed to prenatal androgen. 2D:4D reliably differs by sex: males tend to have lower values of 2D:4D than females (Manning 2002). Exposure to high prenatal androgens, low prenatal estrogens, or both, may lead to lower digit ratios (Manning et al. 1998; Manning 2002; Lutchmaya et al. 2004; Cattrall et al. 2005; van Anders et al. 2006). 2D:4D partly clarifies individual differences in aspects of mate quality in men. Lower digit ratios are associated with greater sperm numbers (Manning et al. 1998), good health (Manning 2002), higher physical attractiveness (Roney & Maestripieri 2004; but see Neave et al. 2003), greater number of sexual partners (Hönekopp et al. 2006b) and greater number of children fathered (Manning et al. 2000). Superior athletic and musical ability are associated with lower digit ratios (Sluming & Manning 2000; Manning & Taylor 2001; Hönekopp et al. 2006a) and these qualities may be regarded as honest signals of male fitness. The finding that males with low 2D:4D display higher levels of courtship behaviour when in the presence of potential mates (Roney & Maestripieri 2004) is particularly relevant for the current study. It is known that non-tactile visual exposure to females is associated with transient and short-onset increases in male testosterone levels (e.g. Purvis & Haynes 1974; Harding 1981; Hellhammer et al. 1985; Roney et al. 2003). Since lower digit ratios are associated with greater sensitivity to testosterone (Manning et al. 2003) and higher circulating testosterone levels (Manning et al. 1998), visual exposure to females may particularly affect male subjects with low digit ratios. Because 2D:4D is related to all kinds of different proxies of mate quality (e.g. Manning et al. 2000; Wood et al. 2003; Roney & Maestripieri 2004), ‘mating’ contexts might provide powerful cues particularly for males with lower digit ratios. We predict that responder behaviour in ultimatum games after exposure to sex cues will be associated with 2D:4D.

Subsequently, participants engaged in a seemingly unrelated task and were informed about the rules of the ultimatum game: the proposer had to suggest a split of €10 and the responder had to accept or reject the proposer's offer. Each participant made two relevant and consequential choices. They recorded both, a maximum offer as proposer and a minimum offer they were willing to accept as responder. To avoid anchoring effects, subjects recorded their choices on a visual analogue scale with endpoints €0 and €10. If the maximum offer equalled or exceeded the responder's minimum demand, the responders would accept the offer, and both the proposer and responder would be paid according to their choices. If the proposer's offer was less than the responder's demand, the responder rejected the offer and no deal was made, leaving both the proposer and the responder empty handed. Participants were informed that 20% of participants would be randomly chosen, that they would be designated either the role of a proposer or a responder and that they would be partnered with another individual participating in the same experiment. After the game, participants entered their email address that was used to inform the participants several weeks later about their eventual gains.

After the ultimatum game, the right-hand was scanned by a male experimenter to evaluate digit ratios (we used the right-hand because relations with right-hand digit ratio are typically stronger; Manning 2002). Participants positioned their hand palms on the glass plate of a scanner. Lengths of the second and fourth digits were gauged from the ventral proximal crease of the digit to the finger tip by means of an Adobe Photoshop tool (‘measure tool’). We measured from the most proximal crease when there was a band of creases at the base of the digit. The lengths of index (2D) and ring (4D) fingers were measured twice by the same rater with a time-lapse of several weeks. Digit ratio measures had a high level of repeatability (intraclass correlation coefficient (ICC)=0.99). In the analysis, we used the average between the two measurements.

Figure 1 Scatterplots of right-hand 2D:4D ratio and minimum acceptance level in an ultimatum game. (a) Study 1, landscape condition. (b) Study 1, sexy women condition. (c) Study 2, T-shirt condition. (d) Study 2, bra condition.

Figure 2 Scatterplots of right-hand 2D:4D ratio and minimum acceptance level in an ultimatum game in Study 3. (a) Elderly women condition. (b) Young women condition. (c) Sexy women condition.

The ultimatum game may be considered as a lab model of cooperative hunting or reflect dilemmas of food sharing (Page et al. 2000). Lower minimum acceptance levels in ultimatum games are thus not trivial. Three experimental studies demonstrate that a subtle cue related to sex is able to impact decision-making. Consistent with the predictions we derived from Wilson & Daly (2004) and Ariely & Loewenstein (2006), we found that heterosexual males are more likely to accept unfair offers in an ultimatum game after exposure to sex cues. In three experiments, 2D:4D was reliably and consistently related to demand in ultimatum game behaviour. To our knowledge, this is the first demonstration of a Person×Context interaction (e.g. Mischel & Shoda 1995) for 2D:4D. The significance of this interaction between digit ratio and context is best exemplified by the meta-analysis of our three experiments. In control conditions, males with lower digit ratios state higher minimum acceptance levels, which is consistent with the finding that males with lower digit ratios have a preference for the normative fair split (Millet & Dewitte 2006) and with the finding that males with higher testosterone levels are more likely to reject an unfair split in the ultimatum game (Burnham 2003). However, in sex-cue conditions, men with lower digit ratios state lower minimum acceptance levels. This is in line with the finding that exposure to highly attractive women causes men to choose a smaller, sooner reward over a larger, later reward (Wilson & Daly 2004) and with the finding that sex cues lead to changes in the perception of what is normatively ‘just’ (Loewenstein et al. 1997; Ariely & Loewenstein 2006). We strongly advocate future research on 2D:4D×Context interactions. The present study stresses the importance of contextual variables that might impact associations between 2D:4D and aspects of phenotype/personality. Context effects may elucidate puzzling inconsistencies. Our results may partly clarify why some reported relationships with 2D:4D are hard to replicate (e.g. the association between testosterone concentrations and digit ratios; Manning et al. 1998, but see Neave et al. 2003). This study suggests that even the physical appearance of the researcher who interacts with respondents, measures digit lengths, etc. might affect the relation between 2D:4D and other variables.

B.V.d.B. is supported by a Junior Researcher (Aspirant) grant from the Fund for Scientific Research—Flanders, Belgium (FWO). S.D. is supported by grant G.0391.03 from the Fund for Scientific Research—Flanders, Belgium (FWO) and a research grant from the Katholieke Universiteit Leuven (OT 03/07). Financial support by Censydiam-Synovate is gratefully acknowledged.