Irrationality in mate choice revealed by túngara frogs

See allHide authors and affiliations

Science  28 Aug 2015:
Vol. 349, Issue 6251, pp. 964-966
DOI: 10.1126/science.aab2012

A third option leads to poor mates

A “decoy” effect decreases rational decision-making in humans. Irrational decisions in this case are a result of a choice between two options being affected by the introduction of a suboptimal third choice. Lea and Ryan show that tungara frogs are also subject to a decoy effect, choosing a male with a less appealing call when presented with a third, inferior calling male. These results suggest that the choice of mates by animals may be context dependent. It appears that rational choice may not always drive sexual selection.

Science, this issue p. 964


Mate choice models derive from traditional microeconomic decision theory and assume that individuals maximize their Darwinian fitness by making economically rational decisions. Rational choices exhibit regularity, whereby the relative strength of preferences between options remains stable when additional options are presented. We tested female frogs with three simulated males who differed in relative call attractiveness and call rate. In binary choice tests, females’ preferences favored stimulus caller B over caller A; however, with the addition of an inferior “decoy” C, females reversed their preferences and chose A over B. These results show that the relative valuation of mates is not independent of inferior alternatives in the choice set and therefore cannot be explained with the rational choice models currently used in sexual selection theory.

Evolutionary theory is tightly linked to traditional decision theory, which predicts consumer behavior by assuming individuals’ decisions will lead to outcomes that maximize the chooser’s subjective utility, in terms of satisfaction or benefit (1, 2). A rational individual is one who makes choices that obey the simple mathematical axioms of transitivity (if A > B and B > C, then A > C) and regularity (if A > B in the absence of C, then A > B in the presence of C). The simple nature of rational models is intuitively attractive; however, there has been much recent debate as to how valuable they are in predicting actual behavior (35). Human behavior commonly deviates from what is predicted by rational choice models, with individuals making seemingly suboptimal decisions regardless of outcome importance. One well-known violation of regularity is the “decoy effect” (68). For example, while shopping for a used vehicle, the buyer may value both low price and fuel efficiency. Of the two vehicles considered, one has a higher price tag but also better efficiency (A), whereas the second has a lower price but also lower efficiency (B). The buyer decides that he or she values lower prices over higher efficiency and so chooses B. At this point, the salesperson mentions that there is a third vehicle (C), which also has good fuel efficiency but a much higher price than both A and B. This causes the buyer to reconsider, despite no interest in the higher-priced vehicle. To the salesperson’s delight, the buyer ultimately chooses A, spending more money for better fuel efficiency. This irrational behavior has been produced by the decoy effect.

Mate choice is one of the most important decisions an animal makes. In many species, these critical decisions occur in dynamic social environments (such as leks) containing multiple potential mates with complex traits. Comparable to human consumers maximizing utility, we expect animals to maximize their Darwinian fitness by making rational mate choices. The preference function concept, central to sexual selection theory (9), assumes that mate choice rules obey formal rationality (2, 7). The results of the scant empirical studies that have tested this assumption were either inconclusive (2) or failed to reject the axioms of transitivity (10) and regularity (11). Thus, we designed a study to address the axiom of regularity using a decoy paradigm (68). We hypothesized that females exhibit decoy effects similar to those commonly observed in humans, whereby the probability of choosing stimulus A over B is dependent on the presence of the inferior third option C [P(A|B) ≠ P(A|B,C)].

Mate choice behavior in our subject, the túngara frog (Physalaemus pustulosus), has been thoroughly studied for three decades (12). Males form lek-like aggregations and produce advertisement calls to attract females. Females exhibit a highly stereotyped and robust phonotactic response to speakers broadcasting stimuli that mimic male advertisement calls. Females generally prefer “static” call characteristics such as low dominant frequency and longer call durations, which are largely replicable within individuals, in addition to faster call rates, which change dynamically with the social conditions. For the current study (13), potential mates were represented by three acoustic stimuli (A, B, C) varying in two traits under selection by females: (i) static attractiveness [dimension 1 (DIM-1)] and (ii) call rate (DIM-2). The subjective values of each of these independent traits were determined by the actual measured preferences of 78 females from the same population in preliminary choice trials, which were completed before beginning our decoy experiments (Fig. 1, A and B; fig. S1; and tables S1 and S2). To represent variation in static attractiveness, we chose three natural male call variants from a previous study that demonstrated the range of multidimensional acoustic variation present in our study population (13, 14). Static attractiveness and call rate were inversely combined to produce an inferior decoy stimulus (C) among the final three stimuli (Fig. 1, C and D). We then measured females’ relative mate preferences for each set of paired calls [(A, B), (A, C), and (B, C)] and for the trinary choice set (A, B, C). Based on compromise effects in other decoy experiments, we predicted that the relative preference for the intermediate stimulus A (target), with respect to the highest-value stimulus (competitor), will increase when C (decoy) is present.

Fig. 1 Independent valuation of call traits.

(A) Static attractiveness (DIM-1) is a composite trait of acoustic characters inherent to a given natural call variant (B1, A1, C1). (B) Calls were presented at three different call rates (DIM-2): C2 = 1/4, A2 = 1/2, B2 = 1/1 (in calls per second). Females’ relative preferences for the three variants of each trait were measured in preliminary phonotaxis trials: (At, Bt), (Bt, Ct), (At, Ct). The trait value is the proportion of females’ choices out of the total possible. Each stimulus variant (e.g., At) was presented in two of the three pairwise tests; thus the sum of proportions for the three stimuli was 1.50, whereas the maximum possible for a given stimulus is 1.0. Error bars indicate ±SE of the binomial distribution. (C and D) Differences in DIM-2 are discriminated more by females than differences in DIM-1, leading to an asymmetric relation among final stimuli (A, B, C) in total value, as calculated by the sum of binomial z ratios for each trait.

In experiment one, we tested 40 females repeatedly on each of the four choice sets, using a grouped-speaker configuration that allowed females to choose the decoy option during the trinary test (Fig. 2A). Experiment two was identical to experiment one, with the important exception that during the trinary set, the decoy stimulus was broadcast from a speaker directly above the females’ starting position to make it perceptible but inaccessible (Fig. 2B). The influence of unavailable decoys, referred to as the phantom decoy effect (15), has been shown to shift preferences toward the asymmetrically dominated target; thus, the prediction is the same as in experiment one. This experimental design also makes it possible to avoid complications potentially arising from the relative spatial positions of the speakers, such as a preference for centrally positioned males. Eighty females completed both the binary (A, B) and trinary (A, B; in the presence of C) choice sets. Fifty of these 80 females were also tested with the two additional binary tests [(A, C) and (B, C)] to verify the inferior status of stimulus C in this configuration (experiment two).

Fig. 2 Mate preference reversal for two stimuli (A, B) with the addition of an inferior decoy option (C).

(A) Experiment one: The decoy was available as a third option in the trinary test. (B) Experiment two: The decoy was broadcast from a ceiling speaker and so was perceptible but unavailable. (C and D) Relative preferences were dependent on the presentation of the decoy in both experiments. Experiment one: binary (n = 40 females) versus trinary (n = 33, excludes choices to stimulus C), Χ2 = 3.765, P < 0.05; experiment two: binary versus trinary (n = 77, excludes no choice data), Χ2 = 14.700, P < 0.001. Error bars indicate ±SE of the binomial distribution. ns, not significant. *, P < 0.05; ***, P < 0.001.

Results were consistent in both experiments. As predicted by independent trait valuation of the three stimuli (Fig. 1, C and D), stimulus C (decoy) was inferior to both A and B in all tests [experiment one: C < B, P = 0.007; C < A, P < 0.001; P(C, trinary) = 0.175, P = 0.042; all n = 40 females; experiment two: C < A, P < 0.001; C < B, P = 0.034, all n = 50]. In all binary tests, but not the trinary tests, the preferred stimulus in the pair was that with the fastest call rate [DIM-2] (Fig. 2, C and D). In the trinary tests, females were significantly more likely to choose the intermediate target A in the presence of the decoy regardless of the availability of the decoy as an option (Fig. 2, C and D). Female túngara frogs reversed their preferences in the presence of an irrelevant alternative in two separate experiments and thus violate a key assumption of mate choice models derived from decision theory.

In decision theory, predictions are based on the expected outcomes of individual behavior, and no assumptions are made about the underlying cognitive processes. Given the similarity of our results to systematic biases exhibited by human consumers, mate choice models might benefit from behavioral economics insights that employ psychological concepts such as perceptual biases or limited cognitive resources to explain irrational choices. Economic rationality does not account for reference dependence, which is inherent to perceptual systems during rapid comparative evaluations (16, 17). In socially complex situations such as frog choruses, rational decisions could be time-consuming, potentially resulting in lost mating opportunities or the risk of further exposure to predators. Decision rules might evolve to include loss aversion (4, 18, 19), mitigating the risk of costly errors, which are more likely when there are extreme alternatives and in uncertain environments. Such heuristics could lead to stabilizing selection on male traits and maintenance of genetic variation. Moreover, as human consumers are susceptible to manipulation by salespeople, context-dependent choice rules may make female frogs vulnerable to behavioral exploitation by competing males; for instance, if males are selective of their nearest neighbors (20, 21).

Although it is clear that female choice patterns do not coincide with the consistent valuation predicted by traditional models in sexual selection, it is far from clear whether perfect formal rationality is mutually compatible with optimal evolutionary fitness (2224). Closer inspection is required to determine whether inconsistencies revealed by decoy effects are, in fact, suboptimal in the context of fitness maximization. Variation of female mate choice in different social contexts might reflect adaptations for using additional sources of information (25), resulting in the expression of more complex but predictable choice patterns.

Growing empirical evidence suggests we lack a coherent understanding of the decision rules governing mate choice. Rational choice models have proven fruitful for static or otherwise simple selection scenarios, and yet may prove inadequate for generating accurate predictions of how sexually selected male traits evolve by female mate choice in the socially dynamic sexual marketplace. Further elucidating complex but predictable mate choice patterns would generate valuable insight into the evolution of decision-making and the coevolutionary processes of sexual selection.

Supplementary Materials

Materials and Methods

Fig. S1

Tables S1 and S2

References (2628)

Additional Data Table S1

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank the Autoridad Nacional del Ambiente for permission to do field work in Panama; the Smithsonian Tropical Research Institute for logistical support; M. Still, S. Beckett, and A. Smejdir for assistance; and M. Jennions and an anonymous reviewer for helpful suggestions on the manuscript. M.J.R. thanks M. Bateson for early conversations on decoys. This study was supported by NSF Integrative Organismal Systems grant 1120031 to M.J.R., R. Taylor, and R. Page and an NSF Predoctoral Fellowship to A.M.L. Procedures were approved by the Institutional Animal Care and Use Committees at The University of Texas at Austin and the Smithsonian Tropical Research Institute (AUP 2011-0825-2014-02). Experimental data are provided in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article