Economic behaviours among non-human primates

1. Statement

In October 2018, we organized an exploratory workshop at the Villa Finaly (Sorbonne in Florence) entitled ‘Economic behaviour of non-human primates’. This workshop served as a preparatory event in view of this theme issue and the shaping of a community of primatologists, neuroscientists and behavioural economists who could start a dialogue on whether economic behaviours occur in non-human primates, to what extent and in which sense. We were driven by the conjecture that some decision-making abilities and observed exchanges among non-human primates are the evolutionary precursors of more elaborate economic abilities we find in humans. We hope the current theme issue provides substantial elements to evaluate how far this conjecture can be accepted.

Human economies are complex systems made of intricate relations between productive activities and consumption patterns and are unfolding most of the time in a context of radical uncertainty. In the face of such uncertainty, it would be reasonable to say that most of our decisions are noisy, if not predominantly indecisive. Moreover, from a cognitive point of view, the economic environment that humans have created through their repeated interactions has eventually produced a set of laws, mechanisms and institutions that most lay persons do not or even cannot understand [1,2]. Confronted with this cognitive gap between the human mind and the economy that the human mind itself and human activities have generated, economists have proposed two main theoretical solutions to make sense of how we still manage to make our way, through our individual and collective economic lives, with more or less adaptive success. But these solutions are largely mute on how the human mind, from infancy, to childhood, adolescence and adulthood, finally reaches a certain level of adaptation to economic decisions and environments. This developmental approach is rarely taken into account in economics, if one excepts the important work on this aspect of things by Heckman (e.g. [3]). The evolutionary and trans-species comparison provides an effort in the same direction.

The first solution—the one supported by a so-called orthodox approach in economics—is to assume that individuals are endowed with cognitive abilities that are sufficient for them to maximize their individual welfare through their decisions in their typical environments. It is not necessary, at this stage, to define these cognitive abilities as fully rational or boundedly so. The bottom-line, between orthodox and behavioural economics, is that we have, in our psychological mindset and neurobiological fabric, the naturally selected resources to adaptively respond to our economic environments. The fact that the human mind is endowed with a self-contained rational decision module (such as formally described by the axioms of decision-theory under uncertainty—see [4], and for a recent discussion of the psychological relevance of axiomatic decision-theory, [5]) or with a less coherent panoply of biases and heuristics [6,7] does not make any substantial difference as long as the relationship between resources and behaviour leads to utility maximization in the long run. The assumption that primates have developed these abilities that make economic activities possible in principle is discussed in many of the articles collected in this special issue, as we detail below.

The second type of solution economists have proposed does not require a particular mental structure or stringent conditions on cognitive abilities, provided the minimal responsive behaviour to one's environment that defines a living organism. It is interesting to note, in that respect, that the expected empirical realization of the parameters of a paradigmatic behavioural model of decision-making, namely prospect theory [8], does not make any sort of specific predictions about the type of neurobiological underpinnings leading to the realization of the model. It is in this sense purely behavioural and is actually, by design, compatible with a wide array of biological implementations. Suffice it to mention the seminal study by Dener et al. [9], which shows that chick-peas display the risk-sensitivity patterns—risk-aversion in gains and risk-seekingness in losses—emblematic of prospect theory. This observation shifts, by consequence, the focus from individual cognitive conditions to the analysis of objective features in the decision environments that make economic behaviour possible and also identifiable as such.

For us, this second line of approach to the gap between cognitive abilities and environments raises the hardest questions. We might need a prior view of what can be reasonably counted as non-human primate economic behaviours, so that we do not become too undiscriminating in defining as ‘economic’ any animal behavioural pattern that fits the parameters of a model of decision-making. A recent attempt in this direction has been pursued by Addessi et al. [10]. To qualify as being non-trivially ‘economic’, behaviour must be mediated by some form of supra-individual settings or, in other terms, by specific institutions, the latter notion taken in a broad and minimal sense to mean frameworks through which both individual cognition and social interactions are mediated. It is the ability to deal with and within these institutions that jointly defines adaptive economic behaviour and the ability, for an animal species, to have generated those institutions. In that sense, the exchange of goods (through barter or by using some medium of exchange), organization and division of productive tasks, collective efforts to generate social welfare, ‘business’ activities that go beyond the satisfaction of immediate needs, etc. are definite markers of an economic system that encompasses a variety of behaviours that can be properly defined as ‘economic’.

The common ground of the papers in this theme issue is that they address the question as to whether the roots of human economic behaviour can be traced back to non-human primates. Studying non-human primates, rather than any other animal (or vegetal) species, is a way to reconcile the two approaches we have just emphasized. On the one hand, we can point to evolutionary continuities and discontinuities between cognitive abilities in human and non-human primates in a way that would be less conclusive if we studied more phylogenetically distant animal species. On the other hand, the use of decision-making (individual or collective, i.e. for instance, game-theoretical) experimental paradigms to investigate the non-human primates' abilities to instantiate economic behaviour—in a model-dependent sense, then—provides a set of analytical and empirical tools to examine the extent of these cognitive and behavioural continuities or discontinuities and the factors that underlie them. A similar conclusion can also be reached in a few studies or theoretical positions that are model-independent and yet can document or assume the existence of emerging economic behaviour among human primates.

Frans de Waal's statement in his preface ‘How animals do business’ [11] that opens this series of edited papers exactly assumes not only the conceptual plausibility but the actual prevalence of economic behaviour among non-human primates. This contribution asks the hard question as to how to define economic behaviour with respect to animals and evaluate their frequency among ape and monkey societies. The better answer is maybe by pointing at some ‘obvious’ instances such as the famous colony of temple macaques in Bali involved in stealing and ‘blackmailing’ with visitors' paraphernalia in exchange for food [12]. A fully fledged trading system has developed and apt behaviours to sustain it are culturally transmitted from one generation of macaques to the next. For de Waal, there are other arguments that can support the relevance of thinking of animal behaviour in economic terms. But these stem from the fact that economic sciences have evolved in the past decades and have undertaken a behavioural turn that led to the relaxation of fully fledged rationality assumptions that could be empirically vindicated only if we observed thus-envisioned economic agents constantly maximizing their utility, which we apparently do not do even among most profit-oriented humans. This remark would be a bit vacuous if it were not accompanied by a characterization of why humans and other animals do not behave according to an ideal model of rationality. Interestingly, what de Waal underlines is the role and importance of emotions to shape actual economic behaviours. A phylogenetic continuity in the selection of emotional mechanisms within the Primate order has been documented across a series of experimental games or field situations in which, for instance, fairness and reciprocity are involved, and justifies the labelling of a broader field of non-human primate behavioural economics. Despite methodological constraints that differ between humans and non-humans, the behavioural parameters that enrich core decision-theories can be largely considered common across the Primate order. The behavioural turn in economics, which has jointly led to the adoption of theoretical models sensitive to biases and emotions and the exploration of choice behaviour within typical experimental paradigms, has definitely made economics and biology closer disciplines (see [13]).

2. Emerging economic behaviour in non-human primates: reality or fantasy?

One of the most obvious examples of an economic organization among monkeys is given by the already quoted above example of the Balinese Uluwatu temple macaques. New data and interpretation are provided by Jean-Baptiste Leca [14] and his colleagues in their paper ‘Acquisition of object-robbing and object/food-bartering behaviours: a culturally maintained token economy in free-ranging long-tailed macaques'. This field study will be very satisfying for anyone who is seeking a model-independent demonstration that properly labelled economic behaviours can emerge among non-human primates outside laboratory-stylized experiments. Many limitations and wrong conclusions can arise owing to the way an experiment has been designed and run or because the parameters of a model have been mis-specified. Leca et al. [14] suggest a different and most convincing ecological validation of an emergent and sustaining primate model of a trading system. If we look closely at its social and material organization, we see that this trading system is not random. Some objects (such as cameras and glasses) are preferably stolen and bartered for food because they present some physical affordances but also because it is learnt and then anticipated that they bear some intrinsic value for their initial owner. They stand for media of exchange as they have clearly no intrinsic value for the macaques. Leca and colleagues insist on the cultural transmission of apt behaviour (when to steal, how to initiate the exchange) and cognitive adequacy (recognition of the good media of exchange, possible anticipation of the acceptable amount and type of food to be accepted during the exchange) between trained adults and youngsters. The data analysis provided in this paper shows the conjunction of two processes: (i) experiential learning, proved by the fact that young and old subjects' performance, in exchange success rate, significantly differs, and (ii) associations between the perceived intrinsic value (for the robbed human) of the item and the quality–quantity of food obtained in exchange. Subjects clearly engage in utility-maximizing behaviour based on the perception of the relative value of the token and the reward. It is thus possible to affirm that an economic system and basic cognitive abilities to maximize one's utility (even through very basic cognitive aptitudes) are mutually reinforced and that, in this sense, this study points to the fundamental interface between a mind and an emerging institution in which economic science is primarily interested.

How far is it plausible than non-human primates reach into the emergence of a typical economic institution like a monetary mechanism on top of a trading system, given that even in the most favourable ecological case, as described by Leca and colleagues, we still cannot speak of such an emergence? This is where it becomes relevant to couple field-studies with laboratory-based attempts at fixing behavioural and cognitive limitations that explain the actual non-emergence of a proto-money among non-human primates but also assess whether it is in principle impossible that they could do so, and why. This theme issue contains two articles that precisely address this question. Michael Beran and Audrey Parrish [15] show that the token-exchange paradigm, which is widely used to investigate the ability to abstract an exchange performance from the obtaining of potentially suboptimal immediate primary rewards, is instrumental in identifying the cognitive structure that underlies monetary behaviour and that non-human primates fail to fully exemplify. Several non-human primate species are remarkably gifted in exchanging tokens against primary rewards (food). The ability to associate a token with no intrinsic value to a primary reward is the basic requirement on the path towards monetary abstraction, or what Beran and Parrish call the emergence of a currency. They report that some monkey species, for instance capuchins, are able to implement timely choices in order both to obtain valuable tokens and to exchange them to gain optimal reward. They thereby display a clear understanding of the exchange environment and its inherent contingencies and opportunities, and engage in sophisticated probabilistic inferences (as shown elsewhere in [16]). They are also capable of delaying gratification by accumulating tokens, finely exemplifying Keynes' notion of a precautionary demand for money [17]. Nothing is missing, even though cognitive control and impulsivity could be evoked to eliminate the possibility of a medium for exchange among non-human primates [18]. But Beran and Parrish conclude with a distinct hypothesis. No lack of particular cognitive abilities in principle prevents, in their view, the possibility of non-human primates selecting and consistently adopting a token-based currency; however, what could explain why they do not do so is a motivational, rather than a cognitive, factor. The differential quantity judgement performances are poorer with secondary rewards (tokens) than with primary rewards (food), at least in some of the tests that are administered. By contrast, it is interesting to acknowledge that money in our human species is a highly motivational stimulus.

Another study in this issue, by Eleonora Quintiero and colleagues [19], addresses the problem regarding which features in token-exchange structures are required for the systematic adoption of one token, among a variety thereof, as a systematic means of exchange. The focus here is on the association by tufted capuchin monkeys (Sapajus spp.) between the token and the primary reward according to different qualitative–quantitative trade-offs. They test whether capuchins choose optimally, or whether they display violations of rationality accountable in terms of the less-is-more effect [20] or selective-value effect [21,22]. Both effects boil down to the preference for smaller option sets as long as the larger set does not contain more of a highly valued item. It means that the evolution of the cognitive abilities to discriminate qualitatively probably preceded the cognitive abilities to discriminate quantitatively. The authors investigate, through a series of original experiments, how the processing of quality–quantity trade-offs in the context of token exchanges is a potentially relevant factor underlying the emergence of money. They refine previous studies of the less-is-more or selective-value effects by using more combinations than in extant similar studies. Their variable of interest is the impact of the food quantity and quality on subjects' choices. More specifically, they test the impact of non-variability (in terms of qualitative associations) versus a higher variability of a food–token association on the selection of the token in exchange rounds. One main result they bring about is that capuchins seek variety in their food choice even though a less ‘variable’ token yields a more constant intake of a highly preferred reward. Such a result, which does not align with the above-mentioned psychological effects, is fully compatible with an optimal foraging approach, common to a lot of animal species. Moreover, it provides a further crucial element in the analysis of which cognitive or structural factors are required to assess the plausibility of money-mediated exchanges among non-human primates. The ability to generalize from internally varied bundles of goods and quantitative–qualitative trade-offs has probably been crucial in the apparition of money for humans. This paper therefore offers a perspective for future studies in non-human primate economics that could validate not only the relevance of behavioural models to account for seemingly economic behaviours in non-human animals, but also, more specifically and less expectedly, the cognitive and evolutionary microfoundations of search-theoretical models of money emergence (e.g. [23,24]), a question that some of the authors of this article had initiated in the token-exchange literature [25].

Chung et al. [26] present a review paper on the conditional valuation for the combination of goods in primates that directly relates, in fundamental microeconomic terms, to the concerns raised by the previous paper in the series. Namely, particular goods acquire their consumption value by reference to other available goods that are not consumed. This phenomenon is called conditional valuation of goods and it is an effect that is well known by microeconomists when goods are categorized not in reference to their independent value but in relation one to the other. Economic value is a relative concept in both ways, first by reference to the quantity of the same good (leading typically to a marginal diminution of its utility) and, second, by reference to some complementary or substitute goods. This second aspect, far more neglected in neuroeconomics and animal behavioural studies, is felicitously highlighted by Chung and colleagues. Complementary goods mutually enhance their respective values: an ice cream is better with a sun-bath and a sun-bath is better with an ice cream. By contrast, goods with a similar or quasi-similar purpose do not pool their value. The authors assume that complementary goods tend to reduce their respective value when included together in the same bundle. This is an effect that has not been widely noticed in human microeconomics but that they report here in non-human primates. This last point seems crucial. On the whole, human and non-human primates display the same increasing-value pattern for complementary goods offered together and a tendency not to increase or, thus, to decrease the value of a bundle when substitute goods are presented together. Think again of our token issues above and how a medium of exchange is related to the acquisition not of a single good in a piecemeal way but to the possibility of acquiring bundles of goods presenting different internal characteristics. A token that is systematically associated with an irrelevant bundle of goods from the point of view of their inherent complementarity is less likely to be adopted than one that correlates with the type of sought-for variety between goods. As the plausibility of abstractly token-mediated trading in monkeys depends on a correct apprehension of the intrinsic characteristics of the commodities thereby obtained, an attention to those relevant characteristics, as offered by the authors, is a most welcome way to ground in relevant microeconomic concepts token-studies in non-human primates.

3. Looking for an optimal fit between cognitive abilities and environmental factors

Economic behaviour is basically the problem of obtaining resources in order to sustain one's existence. Particular economic organizations and systems, primitive or highly evolved, developed because of some environmental constraints (generically embraced through the term of scarcity) and of cognitive constraints (generally subsumed under the label of bounded rationality). It is mostly assumed that decision processes are consistent across various environments and constraints. However, this assumption has to be qualified in terms of the precise processing of environmental and contextual factors. Only in the interplay between specific cognitive abilities and specific environmental factors can we assess the optimality of non-human primates' economic behaviour.

A fundamental aspect of this interplay is that it is dynamic and that the life of economic agents has a minimal duration involving repeated interactions with the environment. Analysis at this very level is undertaken by Cécile Garcia, Sébastien Bouret, François Druelle, and Sandrine Prat [27] in their review paper ‘Balancing costs and benefits in primates: ecological and paleoanthropological views. Their central point is that optimal cost–benefits computations presuppose the building of a long-term mental representation of the decision environment. These representations must bear on the opportunities to obtain food, services and partners, which are costly to get but highly beneficial resources. They are costly to the extent that they necessitate significant investment in time and energy with no immediate feedback about the return on investment, hence the need of representations or models of the world that relate the selection of actions and their long-term outcomes. Garcia and colleagues review available data across the primate order, interestingly including extinct primates (hominins). One key feature they uncover is that primates have in general been endowed with high cognitive ability, making them able to manage a multiplicity of resources of different values. There is a leading thread, surfacing here again, tending to demonstrate that the reliable perception of value-contrasts associated with potential sets of resources and rewards is the basic determinant of economic behaviour and organization. Garcia and colleagues indeed emphasize how that individual cognitive capacity also induces primates to organize their behaviour in space and time, individually and collectively. This is what an economist, of the second brand we have distinguished above (such as [28]), would consider a naturally ingrained tendency to generate institutional frameworks. Institutions—here spatial and temporal foraging patterns—are definitely cognitive extensions or mediations in the view of the individual and the group's survival and reproduction.

A similar analytical framework and underlying question motivates the contribution by Elsa Addessi and colleagues [29] on the sensitivity of capuchin monkeys to opportunity costs and its influence on their intertemporal choice behaviour. An opportunity cost amounts to the fact of foregoing a reward by accepting another, or by renouncing a certain action by the mere fact of engaging in another one. A choice necessarily implies an opportunity cost, in the very basic sense that choosing A is not choosing B. This involves a trade-off, a cost in terms of alternative opportunities that I could have enjoyed. For this reason, opportunity costs are perhaps the most elementary concept of economic analysis (e.g. [30]). It expresses the basic relationship between scarcity and choice itself. Scarcity implies that certain options will be given priority, that an order or a plan of consumption must be set up. In the absence of scarcity, no sacrifice of opportunities takes place and no need to develop apt intertemporal choice behaviour is required. So it is extremely interesting and potentially fecund, as a means of investigating the interplay of the objective constraints making up a resource environment with the behavioural characteristics of choice patterns, to experimentally modulate the magnitude of opportunity costs incurred by individual subjects and observe the influence of that manipulation on their tolerance to delay of reward obtention. Higher opportunity costs, in the sense of having to forego consumption for a longer period in order to obtain a highly desirable good in the end, drive individuals towards less patient intertemporal choices. The results of Addessi and her colleagues confirm the conjecture that the resourcefulness of an environment shapes behaviour and possibly long-term cognitive features and thereby provide valuable insights into the evolutionary precursors of human economic behaviour.

Francesca De Petrillo and Alexandra Rosati [31] scrutinize the import of another pervasive environmental feature that is most likely to have influenced the evolution of proto-economic behaviour among non-human primates: the radical uncertainty under which most of our decisions are made. The development of decision-theory in the second part of the twentieth century is driven by an effort to reconcile the possibility of optimal choices and our relative ignorance of the probability of their expected outcomes. This development has been a way to find formal solutions to the problem seminally raised by Frank Knight in his 1921 [32] book wherein he precisely spells out the eminently problematic connection between uncertainty and the possibility of profit, in sheer economic parlance. Since then, economists tend to distinguish between degrees in which the probabilities of outcomes are known, in particular proposing the standard distinction between risk (those probabilities are in fact known, which concerns only some varieties of lottery games) and uncertainty proper or ambiguity decision contexts in which only vague subjective estimations are possible. The less classical notion of ‘radical uncertainty’, recently thoroughly surveyed by Kay and King [33], referring to contexts in which no knowledge about the utilitarian consequences of our choices (how much, and even what, will be obtained) and the probability of that indefinite outcome is available, is also met in nature. In response to these various types and degrees of uncertainty, non-human primates have adapted corresponding behavioural strategies. De Petrillo and Rosati [31] convincingly develop an argument according to which these differential implementations of types of uncertainty in natural environments biologically transpose, along an evolutionary scale, to individual biological, including neurophysiological and hormonal, traits as well as in typical social traits (dominance, gender differences, cooperative styles, etc.) in non-human primates—traits that are most certainly lying at the foundation of human economies.

From a very close perspective, Amélie Romain et al. [34] investigate the particular cognitive aptitudes and mechanisms that are involved when primates make decisions under ambiguity, in the Knightian sense mentioned above. Romain et al.'s very suggestive experimental study shows how some special mode of cognitive flexibility is involved when some species of non-human primates (especially orangutans, gorillas and chimpanzees, and to a lesser extent capuchins, macaques and bonobos) face ambiguous choices. In their setting, some rewards were partially covered by a cache, implementing the notion of ambiguity. In those cases, the subjects behaved as if they could guess the size of the covered rewards. Those subjects form expectations based on the available information, or, in other words, the degree of predictability of each lottery. This involves cognitive flexibility on the part of those subjects because modulating one's behaviour according to the predictability of a consequence assumes the incorporation of hidden information in one's decision. Alternatively, individuals failing to do so behave as if what they see is what they will get. In Romain and colleagues' interpretation, those latter individuals lack cognitive flexibility and use a unique blind decision rule, instead of a flexible, complex one. This result is also suggestive as it potentially connects up to some prevailing behavioural models in decision-theory, in particular regret theory [35], which models the formation of rational expectations on the basis of processing counterfactual information about foregone options. Individuals failing to anticipate regret when facing uncertain choices are more likely to fall into irrational and suboptimal choice patterns [36,37].

Social information is also ambivalent and social adaptation relies on the capacity to process the social signals from one's group. Using natural viewing behaviour, Adams et al. [38] postulate that, to make decisions in a social context, individuals rely on information processed in the orbito frontal cortex and the lateral prefrontal cortex. To test this working hypothesis, they developed an elegant video-watching task in which rhesus macaques face a series of video clips. The videos display monkeys engaged in a variety of natural behaviours that span a wide array of ethological possibilities. The rhesus macaques decode the social information involved in these different situations. The authors' findings suggest that both areas encoded the content and value of visual information driving gaze decisions during natural viewing behaviour. These results support an underlying cognitive endorsement of social decision-making compatible with orthodox economics.

4. Experimental settings and the relevance of behavioural economics models

Stefanie Keupp et al. [39] studied risk-attitudes in a population of chimpanzees (Pan troglodytes). The experimental design and procedure they opted for are perfectly standard. The subjects were tested individually in familiar locations. The choices are binary, the rewards hidden in cups on a table. There is a safe option consistently offering a fixed amount of reward and a risky one presenting a half chance of gaining twice the quantity of the sure reward or nothing. Given this simple design equalizing the expected value of both options, the relative frequency of choice of an option is an index of the risk-aversion of the subject. There is an innovation here. In most studies, the authors report, the risky option yields a minimal reward, whereas in their study, there is the risk (this is where the risk lies) that the chimpanzee having chosen that option gets strictly nothing. Chimpanzees have been considered risk-seeking in previous studies that did not include this design feature, which happens to decrease the risk-seekingness of Keupp et al.'s subjects. Zero-values have a special, even a mythical, status in behavioural economics. The classical 1953 Allais paradox [40] showed for the first time how a small risk of getting nothing in a lottery yielding high gains otherwise shifts the subjects' preferences towards a stochastically dominated lottery if there is the certainty of getting something. Not only do Keupp et al.'s results show that a zero-value effect is present in some non-human primates, but they incidentally illustrate the fact that it provokes an accentuation of risk-aversion as well as a diminution of risk-seekingness, a welcome generalization. These authors draw a general lesson that invites researchers to be cautious about generalizing their experimental results. The small methodological change they introduce in their experimental setting routinely used for the measure of risk-attitudes reverses a typical conclusion. It becomes illusory to qualify chimpanzees, as a species, as being ‘risk-seeking’ or ‘risk-averse’.

The risk of over-generalization and essentialization of experimentally elicited behavioural features is clearly accentuated by the coupling of a routine experiment and a routine behavioural model instrumental for data-fitting. The degree to which the free behavioural variables of the model are satisfied by the collected data is supposed to tell us something about the phenotypic dispositions of an individual or a group of individuals. There are several ways to exert continued caution and avoid such hasty conclusions. One is to submit the behavioural variables of a model to a robustness check before submitting them to the verdict of experimentation over a (usually) small population sample. Is it in fact plausible that, if by ‘any chance’ the degree of fit between the model and the data is significant, such a correlation can be soundly interpretable? This is clearly the concern that Nioche et al. [41] had in mind when, following a recent multiplication of studies showing to what extent different species of monkeys behaved as predicted by Prospect Theory (e.g. [42–44]), they wondered whether the variables of that model could be expected to have some evolutionary meaning, in which case it would strengthen the impact of the above-mentioned studies. They specifically focus on investigating whether probability distortion and the inverted S-shape pattern of the prospect-theoretical utility function could represent an evolutionary advantage when individual decisions conform to these features. To answer this question Nioche et al. [41] used a genetic algorithm, simulating the evolution of a population of agents across several generations. The algorithm actually selected individuals that exhibited risk-seekingness in the domain of losses and a typical distortion of the probability function. In parallel, Nioche et al. [41] collected data on nine macaques performing bets over gains and losses. The overall results are fully consistent with the results of the simulation and partially consistent with the predictions of Prospect Theory (risk-seekingness in losses but no risk-aversion in gains), which is where lies the crux of their message. The conflation of the simulation and the experimental data helps to determine which behavioural traits, rather than an overall model, are likely to have been selected as adaptive across a species' evolution.

The same methodological prudence is adopted by Sarah Brosnan [45] in an article that explores what behaviour in economic games tells us about the evolution of non-human species' economic decision-making behaviour. Instead of ‘just’ applying a game-theoretical framework to probe the factors that influence decision-making, or to engage into inter-species comparisons through these stylized strategic situations or even to hope to gain direct knowledge on propensions to competition or cooperation in those same species, it is possible, Brosnan argues, to use game-theoretical experimental methods to engage in a reconstructive study of what behavioural ingredients are selectively required to make up fully fledged economic behaviour as we know it in humans. Games are malleable tools for the experimentalist who is especially interested in the almost infinite variability of their internal structure (a set consisting of available actions, a number of players, different payoffs assigned to the actions and the players, and further rules such as the number of periods, of moves allowed, the modalities of encounters between players, and the degree of mutual information they can share). This indeed allows a stepwise analysis of the performance of different individuals of possibly different species when some of the components of a game are independently manipulated. Brosnan reviews the results that have been obtained through a large number of such structural manipulations. She is interested in the typical results obtained in those experiments and interprets them as assessments of the propensity to cooperate or the ability to coordinate actions, but not in a typical way. In particular, just to mention a central fact, she emphasizes that an individual action can be optimally aligned with another and yielded a Nash equilibrium of a game without the players having been aware of playing their part of a coordinate strategy. It could just happen that, by simply looking for their own payoff, those self-centred individuals coordinated their action in an optimal way. It requires quite another understanding of the game to reach coordination by taking into account the motives and incentives of other players. Chimpanzees understand that their action is part of a collective strategy, capuchins monkeys do not. Through game-theoretical manipulations one can thus directly compare, across species and types of economic problems (competition, cooperation, coordination), not only the phylogenetic evolution of economic performances—as far as they are captured by game-theoretical structures—but, most importantly, the mental representation that each species is likely to have of those structures. This touches the core of economic theory: institutions can be defined as game-forms and their efficiency is directly related to the proper representation individuals have thereof [46].

Institutions are sets of rules, which assimilates them to games, and they are, by the same token, frames within which individuals make their decisions. It is at this specific juncture that Jérôme Sallet and colleagues' [47] contribution to this theme issue articulates. They test the responsiveness of macaques' decisions to contextual manipulations of sequences of rewards. They proceed to two main types of such manipulations: the order in which a preferred and a non-preferred reward is presented and the order in which a reward for oneself (self-reward) only or the same reward for oneself but in a package involving a reward for a conspecific (prosocial reward) is also presented. In both cases, the individual has to choose only one option. The first overall result is that animals are not indifferent to these sequential framings. The second result, most interestingly, shows that the presentation of the self-reward before the prosocial reward tends to favour the choice of the self-reward over the prosocial one, although the same amount of reward is obtained in both cases. The authors, after having rigorously controlled for possible confounding factors, such as delay-discounting, concluded that prosociality can be dependent on some basic framing-effects. The results of this study therefore add up to one of the main messages that can be retrieved from this theme issue, namely that behavioural conclusions are finely dependent on the implementation of experimental paradigms. This was expected, but the papers presented in this section, each one in its own way, offer constructive solutions to avoid undue scepticism about the interpretability of the results thereby obtained with a view to delineating plausible evolutionary precursors of human economies.

5. Methodological challenges in non-human primate economic behaviour research

The views presented in this theme issue are also relevant for the epistemological foundations of the field of neuroeconomics, where direct access to neural computations is almost uniquely provided by non-human primate models. One aspect to be discussed, for instance, is the so-called descriptive-experience gap (see [48]) that highlights the problem of the comparability of results obtained by means of ‘descriptive’ (explicit) stimuli tasks with humans and more experience-based (and implicit) tasks in non-human primates. It is most often the case that experiments run with non-human primates are of an experience-based nature rather than of a descriptive one. This polar distinction, however, requires more nuance and qualification, which motivated Basile Garcia et al. [49] to review the methodological and epistemological challenges that are raised by the translation of experimental paradigms and comparison of results from human to non-human primates.

As the authors rightly note: ‘most non-human primate neurophysiological research relies on behavioural designs that share features of both description and experience-based choices. As a consequence, it is unclear whether the neural mechanisms discovered in non-human primate electrophysiology may be linked to description-based or experience-based decision-making biases’. We could not agree more and this state of affairs casts a shadow over the soundness of laboratory-based comparative results obtained by means of typical decision-tasks. Basile Garcia and colleagues elaborate, in their paper, a programmatic solution to regain confidence in comparative neuroeconomics, by finding ways to palliate the respective lack of knowledge we have about the neural correlates of experience-based learning and decision-making in humans and the even scarcer identification of brain mechanisms and mere behavioural aptitudes possibly underlying the processing of symbolic information (in description-based tasks) in monkeys. This required translational approach will contribute to the building of a general trans-species neural model of decision-making under uncertainty.

The natural decision environments are evidently not scattered with symbolic representations waiting for interpretation but are experienced through repeated trials and errors. But what Basile Garcia and colleagues make clear about experimental designs in neuroeconomics—i.e. that they generally lie somewhere over a continuum between experience-based and description-based design—is in fact also true of natural cultural environments. The cultural components in an environment can be typically defined by the symbolic artefacts and activities it elicits among subjects. An encompassing experimental methodology across the experience–description divide will therefore help us to understand, when adopted in neuroeconomics, what the neural correlates are, across individuals and species, that underlie the emergence of symbolic capacities in economic decision-making contexts. The idea is certainly to follow the recommendation of either getting experimentally closer to ecological reality or, at least, being aware of the incontrovertible gap between experiments and the field. Nature is not carved out as stylized experimental settings. Temporal features (the duration of the experiments does not reflect actual and variable reward obtention delays in the wild), spatial features, the nature and variety of the very rewards and the format of the rewards (in terms of their constancy, homogeneity, frequency, magnitude, etc.) are mostly abstracted out in the laboratory.

Yoo, Hayden and Pearson [50] deal, in their contribution, with one crucial problem of this nature—one that again lies at the core of how we should conceive, model and test the interplay between decisional capacities and the environment outside, in order to be as close as possible to valid ethological observations. Real-world decisions, the authors point out, have no temporal or spatial boundaries. Animal brain organization has developed according to this ecology, not that of artificial experimental settings. The dynamics of brain states, in decision-making, respond to a continuous experiential flow and include a complex neural network (responsible for sensorial, motor and cognitive responses)—not alleged isolated correlates of a ‘pure’ capacity of decision-making. Yoo and colleagues envision the implications of the interplay between these two dynamics, in the outside world and within neural networks, on a radical revision of decision-modelling strategies. They argue that behavioural decision-theoretical models should substitute alternative constructs such as reward rates, online algorithms and control systems to the concept of value and utility functions. This approach would reflect the continuous nature of stimuli and the architectural complexity of naturally embedded decision-making. The methodological suggestions offered by this article entail a major consequence for the future of non-human primate behavioural economics. For the authors, who are neuroscientists, that field should get closer to, if not absorbed by, behavioural ecology, and they explicitly assume that this can be done if standard ‘paramorphic’ economic concepts are altogether renounced in favour of ‘homeomorphic’ ones¹. Yoo, Hayden and Pearson call for more homeomorphism in the study of non-human and human primate decision-making. The type of analysis proposed by the authors offers a possibility of carrying over economic analysis to animal studies with a view to better understanding the evolutionary foundations of institutions such as markets, division of labour, or trade and monetary exchange. Those institutions share essential characteristics that Yoo and colleagues assign to natural environments: continuous decisions, the complexity of decision processes, variability and volatility of rewards, which makes a behavioural ecological understanding of these very environments quite relevant. But the applicability of standard decision-models can also be valid and serendipitous, provided some methodological cautiousness that has been emphasized and complied with by the articles included in this theme issue.

6. Conclusion and the future of a subfield of behavioural economics

In the light of the contents of the articles that make up this theme issue and according to the two main approaches in economics that we distinguished at the beginning of this introduction, we can conclude that an investigation of the cognitive gap between individual cognitive abilities and a complex decision environment—considered as the main topic of investigation of economic science, especially in its behavioural vein—aims at either:

We focused on economic behaviour in non-human primates both because of their obvious phylogenetic closeness to humans and because they have so far been the most studied species in this area of research. However, we expect that our collective work will be inspiring for scientists studying non-primate taxa with a view to assess behavioural convergences and thus obtain a more comprehensive picture of the evolutionary basis of economic behaviour. In that sense, the present theme issue could serve as an anchor for a research programme laying out the foundations of an evolutionarily oriented behavioural economics. The discussion arising from this interdisciplinary exchange will foster the debate on the evolutionary origins of economic behaviour within a wider scientific community of behavioural economists, psychologists, and evolutionary and cultural anthropologists. It will also depend on an extra effort on the part of these several communities. The exchange between decision-making neuroscience and ethologists is not so frequent but we have found, in the genesis of this theme issue, that it can easily be kindled. After all, both communities are interested in the interface between neurobiologically wired cognitive and behavioural abilities. It could be more challenging to deepen the dialogue between economists and biologists [52]. A retrospect look at the flourishing of neuroeconomics in the past 20 years leads to the following ambivalent observation: economic conceptual constructs borrowed by neuroscientists and other biologists have either been taken in an a-theoretical way (utility, value, decisions and their likes are in fact heavily loaded theoretical terms that are often mistakenly taken as primitives) or have been unduly simplified, owing to their wide divulgation out of the initial decision-theoretical context in which they were born (especially Prospect Theory). It is possible that we have reached a stage at which the benefits taken from this conceptual or model translation from economics to biology are marginally diminishing. A deeper understanding of what economic theory of human behaviour and institutions is about could yield a deeper epistemological integration of biology, psychology and economics. A deep satisfaction comes from the fact that most of the articles presented in this theme issue, having been invited upon the expression of this concern, have made more than a minimal effort to defuse it.

Data accessibility

This article has no additional data.

Competing interests

We declare we have no competing interests.

Funding

This study was supported by Agence Nationale de la Recherche (Macaques40 AAPG-2006)