Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 20 February 2019

Experimental Economics—A Philosophical Perspective

Abstract and Keywords

The originally Hobbesian ideal of twentieth-century neoclassical economics as a discipline that studies human interaction “more geometrico” as a scenery of interactive rational decision making is rejected. “Explaining” overt behavior as (if it were) the equilibrium outcome of opportunity seeking rational choices is impossible if the requirement of approximate truth of the explanans is upheld. Stylized accounts of some central experiments (prisoner’s dilemma, ultimatum, dictator, impunity games, double oral auctions) show why this is so and illustrate basic contributions of experimental economics in an exemplary manner. A somewhat detailed account of an experiment concerning “equity” shows the explanatory potential and “workings” of experimental economics and how its findings can contribute to traditional philosophical and psychological discussions. Why the Humean “attempt to introduce the experimental method of reasoning into moral subjects” must remain incomplete until experimental economics and experimental psychology become fully complementary research strategies is indicated as well.

Keywords: experimental economics, behavioral economics, rational choice, Hobbes, Hume, equity, “as-if” explanations

David Hume’s Treatise of Human Nature advertises itself as “an attempt to introduce the experimental method of reasoning into moral subjects” (Hume, 1739, 1978).1 Our subsequent efforts to put experimental economics into philosophical perspective are located within this Humean research program. Neither the relevance of Hume’s arguments nor that of the findings of modern experimental economics can be fully appreciated without going back to Hobbes, though. The criticisms of the Hobbesian rational choice paradigm (shared by neoclassical economics) that have been around since the times of the British moralists2 become increasingly compelling through experimental economic corroboration. But the contributions of experimental economics are not only critical. Complementary to experimental psychological research economic experiments have also identified many phenomena and regularities that advance our understanding of social interaction and open up new avenues of research—present and future.

In the next section we focus on the critical implications of experimental economic findings and outline why a nonpsychological (Hobbesian) “economic” approach must be substituted by a psychologically founded (Humean) approach to “human behavior.” After this we turn to some problems of “equity theory” to illustrate concretely, if in an exemplary manner, how experimental economics can contribute to the “discussion of moral subjects.” Because economics is not only about individual behavior but also about institutions, we indicate briefly how experimental economics contributes to economics as a theory of “market” institutions. A glimpse of topics of future research concludes this article.

Hobbesian General Economic Theory and Experimental Economics

It is useful to distinguish between rational choice theory (RCT) and rational choice modeling (RCM). RCT is the modern “worldly philosopher’s”3 favorite substantive explanatory and normative theory. RCM is a language derived from neoclassical economic modeling and noncooperative game theory that enables researchers to express certain crucial empirical assumptions about human decision making explicitly and precisely. We believe that the merits of RCM are beyond doubt, whereas—due to its Hobbesian heritage—the status of RCT is precarious at best and requires extended attention.4

Three Hobbesian Principles Underlying Modern Rational Choice Theory

As laid out in Leviathan (1651, 1978), Hobbesian social theory is guided by three basic principles (see also Kliemt, 2016):

First, theories of human (inter-)action should be derived “more geometrico” from first “self-evident” a priori principles of individual behavior (“a priorism”).

Second, overt human behavior is to be explained as serving individual actors’ interests5 (“self-regarding individualism”).

Third, behavior can be explained exclusively by relating it to (future) causal consequences of each (overt) act taken separately within consistent (opportunity seeking) individual choice-making6 (“causal incrementalism”).

With only slight exaggeration one can state that the great camps of modern social theory defined themselves either in opposition or adherence to the foundational a priorism, self-regarding individualism and causal incrementalism of the Hobbesian paradigm. “Predisciplinary” approaches that would eventually become modern political, (social) psychological, sociological theory, and empirical legal theory opposed the three Hobbesian ideals of RCT; neoclassical microeconomics, decision theory, and game theory endorsed the ideals of Hobbesian RCT and spelled them out more fully and precisely in mathematical terms.7

The Centrality of Hobbesian Rational Choice Theory

Since the mid-twentieth century, experimental economics and experimental psychology have accumulated empirical counterevidence against the constitutive principles of the Hobbesian research program. In particular, experimental game theory as a subfield of experimental economics corroborated many objections that had been put forward already by Hobbes’s fellow British moralists. In the process, economists had to face anomalies that forced them to patch up their research program so much that its progressiveness became doubtful.8

What at its pinnacle in the 1970s was celebrated as “the new world of economics” (McKenzie and Tullock, 1978) by its adherents and criticized as economic imperialism by its foes has lost much of its unity and its original grand empirical claims by now. Experimental economics turned away from testing general economic theories of the RCT type. Its modus operandi became increasingly similar to the familiar piecemeal experimental fact-finding process of experimental psychology.9 Clinging to the Hobbesian conception (or illusion) of universal RCT became increasingly difficult. Yet neoclassical economics had been under “empirical siege” for a long time anyway.

First Principle Under Attack

For Hobbes, as for most ancient to early modern philosophers, geometry is the ideal type of “scientific” argument.10 Because Hobbes did not yet understand the difference between definitions and axioms, he referred to axioms as self-evidently true definitions.11 Despite this error, within a benevolent interpretation Hobbes’s aspiration of developing a coherent and universal theory of human behavior “more geometrico” makes some sense.12 If there were indeed a priori knowledge that is neither subject to tests nor in need of corroboration but could be captured in axioms, then the ideal of theorizing about the world “more geometrico” would be sound. Therefore, though incompatible with the “experiental spirit of science,” we cannot dismiss such theorizing “more geometrico” out of hand. We need to deal with it in particular in case of modern neoclassical economics in which the original Hobbesian epistemic spirit, and not only the harmless ideal of mathematization, lingers on.13

Economists often think that RCT can be useful regardless of the fact that it rests on unrealistic principles. They support this view by invoking what they take to be the message of Friedman’s essay on positive—that is, descriptive-explanatory—economics (see Friedman, 1953). In this essay, Friedman, along with more doubtful theses, rightly emphasizes the “testing view” of empirical research. In this regard Friedman’s views are based on the Popperian idea that any theory with empirical content must imply potentially falsifiable predictions: To test and—to possibly—refute or corroborate a theory, it is necessary that testable predictions of observable phenomena be made.14

Yet Friedman’s essay can be and has been (ab-)used to defend the economists’ favorite strategy of immunizing their theories against empirical criticism by reinterpreting them as giving rise to “as-if explanations.” As-if explanations are “technologically” useless. This is true in particular for the outcomes of interactive-choice making that can be described “as if” all individuals had been reacting optimally in an “equilibrium” (in which none could improve his or her own results given the behavior of all others). If it comes to intervening into the world rather than merely representing worldly phenomena knowledge of factually corroborated causal relations and mechanisms is necessary. An effect on the world can only be brought about by a real mechanism. If one can merely describe something as if it were brought about by, say, rational choice making but cannot claim that rational choice making was in fact the operative causal mechanism, then that information is useless for purposes of causally intervening into the world. Moreover, it forms an explanandum and not an explanans.

The view that thought experiments of the “as-if” kind—entertaining and elegant as they may be—are sufficient for explanations has lost its popularity among most modern economists. Many no longer insist that “as-if” arguments have explanatory force. Without findings of empirical law-like regularities, even the exercise of teasing out interesting results from weaker and weaker premises becomes useless for purposes of practical intervention.15

Friedman is right in requiring that in an empirical theory certain (potential) observations must be specified as incompatible with the theory. This has immediate implications for traditional neoclassical economics when it operationalizes its hypotheses with the measuring rod of money: rendering the potentially falsifying situations observable typically relies on specifying a relation between observable substantive, in particular monetary “payoffs” and overt (human) choices.16 Regrettably, with this operationalization of its “predictions” neoclassical economics did not fare well in experimental tests. In particular, the second (self-regarding individualism) and third (causal incrementalism) of the Hobbesian principles underlying RCT do not pass critical examination.

Second and Third Principles Under Attack

As a matter of fact, humans often opt for alternatives that—as far as substantive pecuniary payoffs are concerned—are dominated by other alternatives. It is well corroborated by experiments that even in fully anonymous one-off choice problems in which “practically all conceivable” causal future consequences of their choice acts are eliminated (or “controlled for” by the laboratory setting), human subjects frequently choose alternatives that according to the natural ordering of substantive monetary payoffs should not be chosen.17

As long as RCT is formulated in substantive terms, economic experiments speak loudly against at least two foundational assumptions of neoclassical economics: going for the self-regarding better outcomes (“self-regarding individualism”),18 choosing in view of causal effects of each act taken separately (“causal incrementalism”). Subjective rankings may diverge from the material or monetary outcomes. But this re-evaluation would have to be brought about by the very cognitive and emotional processes that economists want to avoid in their explanations.

Choosing According to the Natural Order of Pecuniary Self-Regarding Rewards

Even in cases in which it might seem safe to “predict” that practically everybody will choose according to the natural ranking of pecuniary rewards, this may not be so (and may be regarded as paradoxical in cases like the so-called prisoner’s dilemma to which we shall turn in the next subsection). To illustrate, consider the choices of a personal choice maker “Row” (all monetary payoffs measured in experimental currency units [ECU]) in Table 1.

Table 1 A dominant choice alternative




ECU 100



ECU 101


The choice of D by Row leads to higher pecuniary results for Row than the choice of C no matter what. If we rely on extrinsic motivation exerting the causal influence by a choice of C, though possible, is not compliant with the “natural” ordering of monetary payoffs. Still, human subjects who are exposed to choices as in Table 1 in a laboratory might have some “second thoughts.” They might think that the experimenter has some hidden agenda because “on face value” the problem is “too” simple.19

What can be said is therefore only this: As long as it is assumed by the choice-making entity that the Choice of D or C will not influence whether x or y occurs (i.e., that x, y are causally independent of C, D) and, to the extent that the entity is interested in monetary rewards only, the “prediction” that in situations like Table 1 Row would choose D should still stand—if at the price of rendering that prediction almost tautological.

As can be learned already from this extremely trivial choice situation, it is not accidental that already Hobbes is ambiguous about the role of subjective perceptions of choice situations.20 It is practically impossible to live up to the ideal of providing purely externalist (in particular behaviorist) accounts of human behavior that do not make assumptions about mental processes internal to the choice maker.

Conceding that at least in laboratory experiments with rather small choice sets and for reasonably high stakes we should expect that a pecuniary dominant alternative like the one in Table 1 will predominantly be chosen, we should be aware that even here context may matter.21 In particular, the mere presence of other individuals—like experimenters in the preceding example—seems to influence choices. As experiments show, it can in fact significantly change behavior, even though in pecuniary terms the dominance of a “D-” over a “C-”alternative is not altered.

Choosing According to the Natural Order of Pecuniary Rewards to Self and Other

Imagine two actors who each have two choices of the preceding kind. One is referred to as “Row” (indexing choice alternatives by “row”). The other one is referred to as “Column” (indexing choice alternatives by “col”). As long as the choices would be presented to the two actors separately as in Table 1, the basic RCT prediction would be Drow and Dcol even if Row and Column knew of each other’s existence. As long as Column does not expect that his choice causally affects the choice of Row and vice versa, the overt consequences expected by the two actors should—according to the principle of causal incrementalism—be unaltered.

Now, let the presentation by two separate tables be changed. Let the interaction be represented by a 2 x 2 player game form with monetary payoffs.22

Table 2 A 2 x 2 Prisoner’s Dilemma






ECU 100, ECU 100

ECU 0, ECU 101


ECU 101, ECU 0

ECU 1, ECU 1

Note. Row’s payoffs are first in each cell; column’s payoffs are second.

Table 2 is referred to as a “game form” because, strictly speaking, it does not represent a “game” in the game theoretic sense. In a game as understood in orthodox game theory, the “payoffs” are representing the preferences of the players “all things considered” by so-called utilities. Such utilities represent rankings without restriction to a dimension of value. Due to processes internal to the choice maker, these rankings can well deviate from the “natural” substantive or “pecuniary” rankings. Other than monetary payoffs or, say, classical (hedonistic) measures of pleasure and pain, utilities in the modern sense are not reasons for ranking alternatives. They are merely representing rankings of alternatives that come about for whatever reasons.23

In orthodox game theory the rankings emerge after mental processes concerning the evaluation of alternatives have—starting with a “common prior”—run their course. They are assumed to be “commonly” known among the interacting “players.”24

Because the internal mental processes are beyond appropriate control of the experimenter, it can be guaranteed in the laboratory only that the game form is commonly known among the players. To that effect, a table like Table 2 can, for instance, be introduced by a “public event” to players in a laboratory;25 for example, via projecting it on a screen that can be seen by all participants of the experiment while all are aware of each other’s presence and awareness (with awareness of that awareness up to any level). Alternatively, the instructions for an experiment concerning interactive choice making could be read out to the participants who are aware of the presence of the other participants while listening and thus know that all know what has been communicated “in public” to all participants.

After such a public event it is still not commonly known, however, how the individuals perceive and rank their choice alternatives. They could, for instance, rank the results of the interaction according to some function of the monetary payoffs accruing to “self and other(s)”—that is, they might be using the sum of monetary payoffs aggregated over all participants (utilitarianism) or some measure of the relative size of monetary payoffs (fairness). Cooper and Kagel (2012) provide an informative survey of such and related other concerns, often referred to as “social preferences” (models).

The public event disseminates the knowledge of the presence of (at least) one other individual and the knowledge that external effects are causally exerted on self and other(s). But if substantive extrinsic motives alone could explain overt behavior, then this knowledge should not play a role for choice making. According to the second and third Hobbesian foundational assumptions of RCT, choices should be the same in terms of individuals’ aspiration for higher monetary or substantive payoffs.

Yet in cases in which a game form like that of Table 2 is common knowledge, it may seem intuitively implausible that individuals will choose with respect to their own monetary payoffs only. Numerous experimental studies show that as a matter of fact in cases like the commonly known game form of interactions as depicted in Table 2, many participants seem to be motivated by “efficiency considerations”—(ECU 100, ECU 100) being so much greater than (ECU 1, ECU 1)—in that they seem to go for the higher monetary sum in particular if the difference between C and D choices to themselves is small (e.g., only a single ECU).26 They may, of course, also be guided by what philosophers call “we intentions” and similar motives.27

It is one of the merits of experimental economics and, for that matter, experimental psychology that they allow for systematically checking on the influence of making, say, the difference in monetary sums between the (C, C) and the (D, D) result of Table 2 less distinct and at the same time increase the “temptation” to unilaterally deviate from (Crow, Ccol) as, for instance, in Table 3.

Table 3 Another 2 x 2 Prisoner’s Dilemma






ECU 120, ECU 120

ECU 0, ECU 200


ECU 200, ECU 0

ECU 80, ECU 80

Note. Row’s payoffs are first in each cell; column’s payoffs are second.

In a game form like Table 3 we would expect to observe more D choices than in case of an interaction corresponding to Table 2. The scale of the payoff differences seems to matter even if the dominance relations are unaltered between Table 2 and Table 3. Yet whether we have changed only one or several parameters at a time is open to doubt. It is often surprisingly difficult to define precisely what an isolated experimental change is.

For instance, when going from Table 1, we have Row confronted with a choice of x, y by another individual and at the same time Row was exerting an external effect on another individual by choosing either D or C. Table 2 and Table 3 differ in payoff size. But again different presentations of payoff equivalent game forms may lead to differences in experimentally observed behavior, too.28

Presentation Effects

Because in Table 3 (and Table 2) the differences between the monetary payoffs of choosing C respectively D are identical, ECU 80 (ECU 1), the payoffs can be split into sums (see Pruitt, 1967). For instance, we could split payoffs to yield the following Table .29

Table 3ʹ 2 x 2 Prisoner’s Dilemma of Table 3 With Row’s Monetary Payoffs “Decomposed”






ECU120 = 50 + 70, ECU120

ECU0 = 50 + (–50), ECU200


ECU200 = 130 + 70, ECU0

ECU80 = 130 + (–50), ECU80

In the first row we have: 120 = 50 + 70, and, 0 = 50 + (–50),

In the second row we have: 200 = 130 + 70, and, 80 = 130 + (–50)

One can interpret this as

“Row brings about 50 by choosing Crow and Column “adds” to it either 70 by choosing Ccol or (–50) by choosing Dcol

“Row brings about 130 by choosing Drow and Column “adds” to it either 70 by choosing Ccol or (–50) by choosing Dcol

Identical tables can be presented to participants of experiments to inform them about payoff effects of C respectively D choices on self and other (see Table 4).

Table 4 Decomposition of Table 3

On Self

On Other

On Self

On Other













Note. R = 7, R+ = 17, and x = 6.

Obviously (indefinitely many) other such decompositions of what choices of “self” and what choices of “other” contribute to always identical final monetary payoffs may be chosen by the experimenter: Subjects’ payoffs are not the splits separately but only their sums. Therefore, to the extent that they are interested in final payoffs (sums), they should not be influenced by variations in the decomposition. However, as experimental results show, subjects’ behavior is influenced by alternative decompositions (see Pruitt, 1967, for some experimental results).

That different presentations of the substantively same situation can lead to different decisions is presumably more familiar from Kahneman’s and Tversky’s example of the “Asian disease.” In this example, there are two A, B pairs of options A = {A-1, A-2} and B = {B-1, B-2}, respectively, from which exactly one is to be chosen:

In the so-called Gain-Frame (“rescue”) the pair of options A is

A-1: 200 (of 600) are rescued for sure and 400 are not rescued for sure

to be compared with

A-2: 600 will be rescued with 1/3 and none will be rescued with 2/3.

In the so-called Loss-Frame (“die”) the pair of options B is

B-1: 200 (of 600) will not die and 400 will die for sure

to be compared with

B-2: none will die with 1/3 and 600 will die with 2/3.

In the Gain-Frame A, option A-1 tends to be preferred over A-2. In the Loss-Frame B, option B-2 tends to be preferred over B-1. Because substantively A-1 and B-1 as well as A-2 and B-2 are equivalent, this effect of framing should not occur among rational actors focusing on substantive results and probabilities.30

Many experimental economists tend to discount such framing effects as irrelevant. They believe that adding of details of situations of interactive decision making that do not influence the monetary end results should be treated as irrelevant. A presentation effect like the one demonstrated by Pruitt is harder to ignore precisely because it arises from alternative presentations of the game form’s monetary payoffs without resorting to uncontrollable influences arising from admitting elements of “thick descriptions.”

Controlling Influences

In experimental situations the experimenter can try to cut off the so-called shadow of the future. Causal effects of choices that trigger responses of others reaching beyond the experiment itself can be eliminated. To accomplish this, subjects can be and are regularly brought together in ways that make it common knowledge among them that their choices will most probably remain anonymous to each participant (under double-blind conditions even to the experimenter). By such anonymity one hopes to cut off the (rational) expectation of future specific reactions to individual choices made in the experiment.

In terms of the previous examples, imagine n >>1 actors in the “row role” and the same number of actors in the “column role” brought into two separate rooms connected by a door in which an experimenter visible to both groups stands. Let it also be made clear that in the other room there are real people (one could, for instance, let individuals in each room sing briefly a certain song together). The experimenter reads the instructions to both groups. Then participants make their choices, knowing from the instructions that they are randomly paired with somebody about whose identity they will never learn anything beyond the fact that the person is in the other room, and so on.

In view of such measures, choices will only affect what other(s) and self get in the given interaction. Deviations from what extrinsic incentives controlled by the experimenter suggest must be the result of some kind of motivation internal to the player.

Beyond efficiency, so-called equity concerns seem to matter most as sources of such motivation. There is a long tradition of studying them in social psychology (see originally Adams, 1965; Homans, 1961; Mikula, 1973) and philosophy (see Frankena, 1966, going back to Aristotle).31 We focus here on the experimental economic side of the street.32

The Example of Equity Theory

Economists tend to approach problems in a top-down manner, starting from general theories and overt behavior, whereas psychologists tend to proceed in a bottom-up manner, focusing on perceptions of specific situations and cognitive processes. Experimental economists insist on providing substantive incentives for participants and on real choice making, whereas psychologists rely more strongly on hypothetical, imagined choice situations and self-reports concerning perceptions. Economists emphasize inferring preferences from choice acts (allegedly “revealing” preferences), whereas psychologists let participants state their preferences.

The preceding list could easily be extended and discussed further in general terms.33 Instead of this, the following subsections try to present experimental economics “at work” by an account of how equity issues can be approached by it.

The Classical Core of Equity Theory

To allocate individual “shares” of a collective reward R among potentially entitled individuals, i = 1,2, …, n(>1), “equitably” according to each individual’s, s, i, contribution Ii(>0) is a recurrent social problem. It is typically solved according to the “equity norm” of allocating rewards proportional to contributions


   RiIi= RjIj, for all i,jN.


N = {1,2,…,n} the set of individuals contributing to some project;

Ii(>0) the individual contribution of i ∈ N to the project;

Ri(≥0) the individual reward for i ∈ N with iNRi=R.

The norm (*) applies to a wide range of cases,34 including standard laboratory experiments in which a “pie,” typically a positive monetary amount R, is to be divided. Even in cases in which an asymmetry of the allocation of decision ‘rights’ prevails—for example, cases in which one individual alone decides over the allocation in a dictatorial way—the substantive allocation of resources seems to closely follow the suggestions of equity theory. This illustrates how deeply the equity norm is rooted in behavior and judgment. We observe:

(O-1) If the relevant parameters of (*) are clearly defined and commonly known, the basic equity norm tends to prevail (in spite of inequitable role assignment in many allocation games).

The early reward allocation experiments in social psychology (see, e.g., Mikula, 1973) assign the allocator role to one participant with “dictatorial powers.” For instance, in a two-actor variant of such a game form one actor is entitled to assign R1, R2 to individuals 1, 2, respectively, while the second has no say and no direct causal influence on the final allocation, R = R1 + R2. By experimental design (anonymous matching for one-off interactions) there should be no strategically relevant shadow of the future reaching beyond the lab interaction. Anticipation of causal influences of his present behavior on actions of others that will have future material consequences for the allocator cannot explain deviations from what his own monetary payoffs suggest.

The observed allocations are typically in clear contradiction to the second and third Hobbesian foundational assumption of neoclassical economics. The deviations of overt allocator behavior from assigning the highest possible material payoff to himself must be explained as a result of some kind of intrinsic motivation triggered by the experienced past rather than expectations of the future.

How behavior in violation of the third Hobbesian foundational assumption is “pushed by the past rather than pulled by the future” is illustrated by the design of the original psychological experiments: Typically the dictatorial allocation takes place after individuals contributed to some (real effort) task. Dictators almost without exception then assign rewards according to relative sizes of contributions, that is, corresponding to (*): There are practically no deviations from what the basic equity norm (*) demands.35 The norm (*) applies not only to contributions like “work” but beyond.

More generally speaking, even though equity demands proportional treatment of individuals, there is a degree of freedom concerning the choice of the dimension to which proportionality applies. Some actors may focus on subjective effort, others on objective contributions, and still others on some prior claims. The overall outcome can vary under the central proportionality requirement (*).

It is interesting to observe that a combination of being rule-guided and opportunistic seems to be very frequent. Dimensions that form the basis of equity considerations may be chosen opportunistically with an eye on the implications for the choosing individual. Once a dimension has been chosen, the actor will feel bound by the equity norm. That is, human actors may anticipate their own commitments to certain nonopportunistic acts and make opportunistic choices in view of this.

As the preceding indicates, the focus of experimental economics on substantive self-interest proves fruitful in directing attention to the interplay of motives that guide adoption of long-term commitments to rules (self-serving) and behavior under rules (rule or norm guided). The game theory background of experimental economists alerts them to other, in particular, information aspects of interaction that may more easily escape the attention of other theorists (though early on a standard textbook on game theory was written by psychologists, see Luce and Raiffa, 1957). To such aspects we turn next with a more detailed discussion of a typical concrete example of experimental economic research. We do so with an eye on illustrating the experimental economics modus operandi more generally.

Stochastic and Strategic Uncertainty in Equity Problems

Standard dictator game experiments restrict themselves to dyadic interactions in which the fundamental parameters are commonly known; for example, in the dyad, actors i, j commonly know (Ii, Ij, R). But such an assumption of common knowledge may not, and in real life often will not, be fulfilled. This suggests experiments including informational uncertainty to tease out the behavioral influence of informational restrictions.

For the sake of specificity, let the dictatorial game form be modified by the assumption that the actor at the receiving end of the allocation can reject a proposal made by the first mover and thereby causally influence the material payoffs that she herself and the allocator respectively receive. Then a so-called ultimatum game form emerges (see Güth et al., 1982). The allocating individual, 1, proposes (rather than imposes) an allocation R = R1 + R2 for individuals 1, 2 (he is referred to as a proposer). The other individual (the responder) may respond with acceptance of the proposal. Then the two will receive R1, R2, respectively. When the responder rejects, the material payoffs will be R1 = 0, R2 = 0.

By relying on variants of the ultimatum game form, the role of private information can be studied in the laboratory. Assume that the joint reward, R, of cooperation can be either high or low. If only the proposer knows that the joint reward is high, a rather pronounced proclivity to offer merely the equitable (half-)share of the low reward expresses itself. The mere fact that the other individual cannot know whether or not the equity norm (*) is fulfilled36 induces the allocator to behave in ways that may merely appear compliant with the norm but are not if the larger reward is in fact accessible.

This response to asymmetry in knowledge suggests that individuals are happy to make use of “moral wiggle room” (see Dana et al., 2007) if it is on offer. Individuals seem to be concerned more strongly with how they are regarded by others than with their self-respect. The old suspicion that having a conscience amounts to no more than the fear that somebody might watch seems to be experimentally corroborated.37

Economists tend to infer from observations like the preceding that it is in the end only extrinsic motivation and reputation that matter. Though this conclusion goes too far, we may note:

(O-2) Private information and strategic uncertainty that rule out the monitoring of (*) render the prediction of equitable overt behavior on the basis of an intrinsically motivating equity norm (*) more precarious.

Again we are confronted with the interplay of selfish, extrinsic motives and intrinsic ones. It is one of the merits of experimental (economic and psychological) methods that they allow us to scrutinize such Humean “moral subjects” in a controlled experimental rather than merely speculative way. We shall illustrate this again in terms of a specific follow-up class of experiments (like the preceding mostly reviewed in Güth and Kocher, 2014).

Generosity Game Forms

Game forms concerning generous behavior are not only useful to characterize conflicts between selfish and other motives. They can “reveal” the operation of principles that guide decision makers in adjudicating between competing moral concerns like equity (“fairness”) and efficiency (“common good or collective welfare”).

A Two-Person Ultimatum Variant of a Generosity Interaction

For instance, in a two-person ultimatum-variant, U, of a generosity game form, a proposer, X, is informed that he can earn a fixed monetary amount x>0 should the responder, Y, accept a proposed monetary amount, R−x. The proposer is asked to choose R[ R, R+ ] with 0<x<R<2x< R+. The proposer can determine R−x such that R−x = x or R−x< x or R−x> x If the responder rejects R−x, actors X and Y both get nothing; if the responder accepts, the payoff vector will be (x, R−x).

Indicating responder acceptance by δ(R) = 1 and responder rejection by δ(R) = 0, the substantive payoffs can be expressed as δ(R)x, for the proposer, and δ(R)(R−x), for the responder. The proposer X will get x or nothing. His choice of R from [R, R+] with 0<x<R<2x<R+ will not affect x but merely δ(R) and thereby δ(R)x and, of course, (R−x). Choosing R he can go below or beyond 2x but also, in line with (*), propose equal monetary payoffs.

Due to the ultimatum structure of the game form, all reward choices R[ R, R+ ]  and general acceptance are equilibria in terms of the material payoffs. If the proposer X values efficiency and assumes that the responder Y is not insisting on equality—R = 2x so that R – x = x—but possibly accepts a share R – x ≥ x, he should choose R = R+. This choice would indicate that the proposer does not regard equity as a paramount consideration and does not expect the responder to treat equity as supreme.

A proposer X who is most strongly interested in receiving relatively more than the responder should opt for the smallest possible, R, that he believes to be acceptable for the responder. If for such a choice of R¯[ R, R+ ] we have δ(R) = 0 if R<R¯ and δ(R) = 1 if RR¯, the choices would be in equilibrium.

An Experimental Study of a Two-Person Generosity Interaction

In an experiment with 33 randomly formed pairs, individuals were randomly assigned to the proposer or the responder role. Proposers could choose only integers R with R = 7, R+ = 17, and x = 6. Responders endowed with veto power for all possible R[ R=7, R+=17 ]  (integers 7, 8 …,17) had to choose a full response strategy; that is, a response δ(.){ 0,1 } to any possible R choice. Figure 1 (adapted from Güth, Levati, and Ploner, 2012) represents the frequencies of δ(R) = 1 choices on the ordinate, whereas the possible R values are listed on the abscissa.

Experimental Economics—A Philosophical PerspectiveClick to view larger

Figure 1 Responders’ acceptance rates in the U(ltimatum) treatment.

Responders tend to reject proposals only when they receive less than the proposer, that is, R < 12. Once they get at least as much as the proposer, rejection is very rare. Response behavior shows that responders do not resent generous choices of R > 12 favoring themselves. But they also accept the nongenerous equal allocation resulting from R = 12 and partly (at least 50 percent) a “spiteful” R from the range R − x < x.

Further questions concerning possible motives can be discussed by comparing cases in which the responder can, U, with cases in which the responder cannot, D, veto the distribution. The aggregate distribution of proposer choices R is displayed in Figure 2 for two treatments (U, D) with and without veto power of responders.

Experimental Economics—A Philosophical PerspectiveClick to view larger

Figure 2 Absolute frequencies of proposer X’s choices, separately for both treatments (U and D).

The two modes are R = R+ = 17 and R = 12 with the former dominating the latter: In the two-person generosity setting proposers tend to choose the generous offer R = R+ more often than the equitable one, R = 2x irrespective of treatment U or D. This seems to indicate that the considerations of proposers are not governed by strategic anticipations of the causal consequences on response behavior. The relevant opportunity costs of achieving more of the one moral aim are measured in terms of foregoing the other moral aim rather than in terms of incurring less or more material opportunity costs to self (as brought about by rejection). Efficiency can obviously be more important than fears of rejection or, say, the desire to receive more than the co-player.

Digging Deeper Into Equity Concerns

The potential to explore dimensions of human behavior systematically in the light of former explorations is an invaluable advantage of controlled experiments. This crucial methodological point can be concretely illustrated by further experiments that were suggested by preceding experimental results concerning aspects of equity theory.

For instance, the motivation of actors can be explored by admitting a strategically irrelevant third “dummy” player. In this vein, consider a three-person generosity game form with a proposer X, a responder Y, and a dummy player Z. As before, the pecuniary agreement payoff of the proposer X is exogenously fixed. Concerning the payoffs of Y and Z, two alternative game forms are considered:

  1. (y) the responder’s agreement payoff is exogenously fixed at y(>0)

    while the dummy collects (R−x−y);

  2. (z) the dummy’s payoff is exogenously fixed at z(>0) in case of agreement

    while the responder collects (R−x−z).

Also, let all three agreement payoffs be positive. Whether or not agreement payments are achieved depends on δ(.), indicating Y’s acceptance (δ(R) = 1) or rejection (δ(R) = 0).

Again proposer X chooses R. Responder Y can accept or reject the emerging distribution. But now proposals as well as responses affect also Z. In the (y)-games—in which the payoff for the responder is fixed along with that of X—as well as in the (z)-games—in which the payoff for the dummy Z is fixed along with that of X—the residual goes to the remaining participant. The monetary payoffs of y-game forms are δ(R)x for X and δ(y)y for Y, whereas Z receives δ(R)(R−x−y). The monetary payoffs of z-game forms are δ(R)x for X and δ(R)(R−x−z) for Y, whereas Z gets δ(z)z.

Specifying the parameters in this way, it can be studied how the presence of an affected (nonstrategic) third party influences the decisions of X and Y: if, for example, proposer X can choose a pie size from R ∈ [12,22]; that is, R = 12 and R+ = 12. Depending on whether the (y) or (z) treatment applies, the pie size will directly affect the monetary payoff of the dummy responder as residual claimant. Acceptance rates of individuals in the responder or Y role are displayed graphically in Figure 3.

Experimental Economics—A Philosophical PerspectiveClick to view larger

Figure 3 Responder acceptance rates for the 11 possible pie sizes p, separated by treatments R = 12, R+ = 22.

Acceptance rates of 100 percent do exist in particular for R ≥ 18. The lowest rejection rates are observed in the (Z b) treatment for R = 18. For R < 18, in particularly in the (Z b) treatment the acceptance rates most dramatically decline.

Figure 4 displays proposer behavior. Note first, that whenever an equal agreement payoff is possible for all players—whether it be in the (y) or in the (z) case—equal sharing is the prevailing choice and x = y or x = z lead to the same modal R = 18 choices. If due to x ≠ y or x ≠ z general equality is impossible, efficiency seeking, R = R+ = 22, becomes modal.

Experimental Economics—A Philosophical PerspectiveClick to view larger

Figure 4 R choices of proposers for upper (y) and lower (z) treatments (exogenously fixed monetary payoffs in brackets).

The preceding specific examples illustrate the process of exploring aspects of equity theory. They show how experiments may be linked together in chains of “one-step modifications.” However, care is required when it comes to assess what should and what should not count as a single step.

To illustrate the rather subtle issues that may arise here, consider the differences between the U (ultimatum) and the D (dictator) variants of the game forms with only two players X and Y as discussed earlier. In a dictator variant, D, of such a game form the payoff vector is (x, R−x) regardless. In the ultimatum variant it depends on δ(R) according to (δ(R)x,δ(R)(R−x)). This seems to suggest that eliminating the dependence of payoffs on response behavior is a one-step modification of the ultimatum game form. Yet this view, though obviously common among experimentalists, may be contested. One step away could mean that the responder could have influence only on her own payoff (i.e., if being able to reject a proposer’s allocation for herself but not for him). Such an impunity game form is more plausibly one step away from the ultimatum game form than a dictator game form.

Experimental results and evidence often depend on subtleties, as the preceding example does. However, there are more robust results of experimental economics once certain institutional settings are factored in.

Markets in the Laboratory

So far, our account of experimental economics highlighted tensions between experimental economic results and the Hobbesian premises of RCT. It also indicated ways of improving theories of “moral science” on the basis of experimental results. In view of these results, it seems appropriate that experimental economists like Vernon Smith (Smith, 1976, 2003) are emphasizing the Adam Smithian tradition of “moral science” as laid out in the theory of moral sentiments (see A. Smith 2002). Though they normally do not explicitly acknowledge this, they thereby opt for behavioral foundations that are incompatible with the Hobbesian strand of standard neoclassical economics. On the other hand, market experiments of economists seem in fact to corroborate the central hypothesis that market institutions quite robustly can guide individuals toward efficient outcomes (see Smith, 1962; Smith and Williams, 1990).38

For the sake of specificity, imagine that there are 30 participants, of whom 15 are randomly assigned the role of seller and 15 to the role of buyer. Each of the sellers has a yellow card with a monetary amount on it. The monetary amount indicates how much the seller must at least receive to avoid losses. When selling at a higher price than his induced monetary “value” (see Smith, 1976; 2003), the difference between the sale price and the value is his. Buyers have red cards with a monetary amount indicating how much they can pay at most before incurring a loss. Again, if they buy cheaper than this, the difference is theirs.

Assume for illustrative purposes that the amounts noted on the cards of the sellers are in ascending order 1, 2, …, 15, respectively, while the buyers have on their cards in descending order 17, 16, …, 4, 3. Initially also assume that there is an auctioneer offering to match sellers and buyers. If the buyers and sellers would inform the auctioneer truthfully about their “bid” and “ask prices” and the auctioneer would in line with welfare economic theory maximize collective welfare, he would determine the market-clearing price of 9 and let all trades take place at it. Figure 5 illustrates this standard economic account. All traders to the left of the two marginal traders39 turn a profit and should trade, whereas those to the right should not trade. Moreover, an appropriately motivated auctioneer can see to it that all trades are taking place at a uniform price.

Experimental Economics—A Philosophical PerspectiveClick to view larger

Figure 5 Market equilibrium

Vernon Smith did not assume the presence of an auctioneer determining the market-clearing price. He organized a sequential bidding process. Letting as in a “double oral auction,” buyers and sellers call their asks and bids publicly, and the process robustly tends toward generating efficient market-clearing outcomes:40 All and only those individuals trade who “should” do so to realize all gains from trade. The marginal trade will take place at the equilibrium price; but other than in case of a well-meaning auctioneer who imposes such nondiscrimination, there need not be a uniform price for all trades.

The experimental results of Vernon Smith have been corroborated in many experiments. Moreover, the original methodology is underlying many of today’s experiments in preparation of institutional design (ranging from eBay to governmental frequency auctions). This practical use of experiments is noteworthy. Yet for the present purposes it is more important to emphasize that efficient results are predicted to emerge not only rather independently of RCT (see also Gode and Sunder, 1993); they are also predicted in a methodologically nonindividualist manner based on a hypothesis akin to: “For all institutions of a double oral auction type, if … then efficient solutions will emerge.”

How this law-like relationship can be explained to “work” in individualist behavioral terms is an interesting explanandum for the methodological individualist. We note only that it is implausible that the methodologically individualist solution will be in terms of (standard) RCT based on rational expectations of other individuals’ behavior.

The Open-Ended Quest of Experimental (Economic) Research

For the methodological individualist, laws that generalize across institutions (for all double oral auctions, if … then …) rather than individuals and their decision making (for all individual decisions, if … then …) must be reduced to hypotheses concerning individual motivation and individual cognitive processes. This, we believe, will inevitably lead to a Humean research program that focuses on psychological laws. In this way experimental economics shows that fences erected between economics and psychology are artificial (as artificial as those erected between sociology and psychology).

Though leading toward “economic psychology” as basic empirical theory of economics, experimental economics does not show that the economic approach to human behavior is not useful. Quite to the contrary, when digging a tunnel, it is good strategy to start from two sides. Given the scarcity of all resources, including those of doing research, we must try to make ends meet. An eclectic way of combining many approaches and seeking common ground rather than keeping things separate is in all likelihood the right thing to do. At the same time, though we can make philosophical use of tools from many toolboxes, real progress will be made only if disciplinary specialization is allowed for as well.

To what extent the philosopher regards the toolbox of experimental economics as well as the results created by its use as philosophically relevant depends obviously on the fundamental conception of philosophy she endorses.41 In any event the ongoing Humean research program should, among other topics, address the following:


Test Focus on

Theoretical Aspects

Individual (one-person) choice task


Consistent EU/prospect theory/case-based decision theory (CBDT)

With and without risk

Lotteries/portfolio selection/(non)additive probabilities

Variation of complexity

Reinforcement/directional learning/evolutionary dynamics, restart effects; cyclicity of dynamic adaptation

Individual (one-person) choice task in social contexts

Reward allocation/dictator

Central concern for social interactions

Other regarding concerns (the dark and bright ones)

Lying, cheating, bribing, colluding but also altruism, efficiency seeking

Variation of “other(s)”

(Non-) asymmetry, kin-, group related

Strategic interaction (multiperson games)

Zero vs. variable sum

Only pure or mixed, only one or multiple equilibria

One-off vs. repeated

“Folk theorems” in the lab, reputation equilibria

Various equilibrium refinements/satisficing

For deterministic, stochastic, and incomplete information games


Convergence to equilibria and, if so, which type

Perfect/imperfect (information)

Board games in the lab, independent choice acts, private information

Complete/incomplete (information)

All sorts of strategic market games (game-theoretic IO)

Static vs. sequential

“Hot,” “cold,” and “lukewarm”

Nonstrategic social interaction (cooperative games predicting outcomes only)

Competitive markets (double auction, clearing house, intermediaries)

Double oral auction/clearing house, stock, and commodity exchange Internet markets/trading platforms run by firms like Amazon and e-Bay or other intermediaries (market makers)

Characteristic function (face-to-face and free communication or more restricted and computerized)

Tests of cooperative game solutions (variants of core and bargaining sets, internally and externally stable sets, … as set solution and value (Shapley) concepts, as point solutions)

Unstructured bargaining

Back to cooperative traditions, no reduction to strategic interaction

  • “Good”: no common knowledge of subtle details

  • “Bad”: problems with methodological individualism


Useful Handbooks

Kagel, J. H., and Roth, A. E. (1995). The Handbook of Experimental Economics (Princeton, NJ: Princeton University Press).Find this resource:

Kagel, J. H., and Roth, A. E. (2016). The Handbook of Experimental Economics (vol. 2) (Princeton, NJ: Princeton University Press).Find this resource:

Kolm, S.-H., and Ythier, J. M. (2006). Handbook of the Economics of Giving: Altruism and Reciprocity. Volume 1: Foundations, Vol. 2: Applications (Amsterdam: Elsevier Science & Technology).Find this resource:

Plott, C. R., and Smith, V. L. (2008). Handbook of Experimental Economics Results (vol. 1) (Amsterdam: North-Holland).Find this resource:

References Cited in Text

Adams, J. S. (1965). “Inequity in Social Exchange.” Advances in Experimental Social Psychology 2: 267–299.Find this resource:

Albert, H. (1967). Marktsoziologie und Entscheidungslogik (Neuwied/Berlin: Luchterhand).Find this resource:

Binmore, K. (1987). “Modeling Rational Players I.” Economics and Philosophy 1987 (3): 179–214.Find this resource:

Binmore, K. (1988). “Modeling Rational Players II.” Economics and Philosophy 1988 (4): 9–55.Find this resource:

Bolton, G. E., and Ockenfels, A. (2000). “ERC: A Theory of Equity, Reciprocity, and Competition.” American Economic Review 90: 166–193.Find this resource:

Campbell, R., and Sowden, L. (1985). Paradoxes of Rationality and Cooperation: Prisoner’s Dilemma and Newcomb’s Problem (Vancouver: UBC Press).Find this resource:

Chamberlin, E. H. (1948). “An Experimental Imperfect Market. “The Journal of Political Economy 56(2): 95–108.Find this resource:

Cooper, D., and Kagel, J. (2012). “Other-Regarding Preferences.” In J. H. Kagel and A. E. Roth (Eds.), The Handbook of Experimental Economics, vol. 2, pp. 217–289. (Princeton, NJ: Princeton University Press).Find this resource:

Dana, J., Weber, R. A., and Kuang, J. X. (2007). “Exploiting Moral Wiggle Room: Experiments Demonstrating an Illusory Preference for Fairness.” Economic Theory 33(1): 67–80.Find this resource:

Debreu, G. (1959). Theory of Value. Cowles Foundation monograph 17 (New Haven, CT: Yale University Press).Find this resource:

Easley, D., and Ledyard, J. (1993). “Theories of Price Formation and Exchange in Double Oral Auctions.” In D. Friedman and J. Rust (Eds.), The double auction market: institutions, theories, and evidence: proceedings of the workshop on double auctions markets held June, 1991, in Santa Fe, New Mexico, pp. 63–97. (Cambridge, Mass.: Perseus Publ.).Find this resource:

Fagin, R., Halpern, J. Y., Moses, Y., and Vardi, M. Y. (1995). Reasoning about Knowledge (Cambridge, MA/London: MIT Press).Find this resource:

Fehr, E., and Schmidt, K. M. (1999). “A Theory of Fairness, Competition, and Cooperation.” Quarterly Journal of Economics 114: 817–868.Find this resource:

Frankena, W. K. (1966). Some Beliefs About Justice (Lawrence: The University of Kansas).Find this resource:

Frey, B. S. (1999) Economics as a Science of Human Behaviour: Towards a New Social Science Paradigm (Berlin et. al.: Springer Science & Business Media).Find this resource:

Frey, B. S., and Bohnet, I. (1996). “Experiments, Theory—And Reality?” In P. Weise (ed.), Ökonomie und Gesellschaft, pp. 340–359 (Frankfurt: Campus).Find this resource:

Friedman, M. (1953). “The Methodology of Positive Economics.” In M. Friedman (ed.), Essays in Positive Economics, pp. 3–46 (Chicago: Chicago University Press).Find this resource:

Gilboa, I., and Schmeidler, D. (2010). A Theory of Case-Based Decisions (Cambridge: Cambridge University Press).Find this resource:

Gode, D. K., and Sunder, S. (1993). “Allocative Efficiency of Markets with Zero-Intelligence Traders: Market as a Partial Substitute for Individual Rationality.” Journal of Political Economy 101: 119–137.Find this resource:

Güth, W., and Kirchkamp, O. (2012). “Will You Accept Without Knowing What? The Yes-No Game in the Newspaper and in the Lab.” Experimental Economics 15(4): 656–666.Find this resource:

Güth, W., and Kliemt, H. (2007). “The Rationality of Rational Fools: The Role of Commitments, Persons and Agents in Rational Choice Modeling.” In F. Peter and H. B. Schmid (eds.), Rationality and Commitment, pp. 124–149 (Oxford: Oxford University Press).Find this resource:

Güth, W., and Kocher, M. G. (2014). “More Than Thirty Years of Ultimatum Bargaining Experiments: Motives, Variations, and a Survey of the Recent Literature.” Journal of Economic Behavior & Organization 108: 396–409.Find this resource:

Güth, W., Levati, M. V., and Ploner, M. (2012). “An Experimental Study of the Generosity Game.” Theory and Decision 72(1): 51–63.Find this resource:

Güth, W., Schmittberger, R., and Schwarze, B. (1982). “An Experimental Analysis of Ultimatum Bargaining.” Journal of Economic Behavior & Organization 3(4): 367–388.Find this resource:

Hausman, D. M. (2011). Preference, Value, Choice, and Welfare (Cambridge: Cambridge University Press).Find this resource:

Heilbroner, R. L. (2011). The Worldly Philosophers: The Lives, Times and Ideas of the Great Economic Thinkers (New York: Simon and Schuster).Find this resource:

Hertwig, R., and Ortmann, A. (2001). “Experimental Practices in Economics: A Methodological Challenge for Psychologists?” Behavioral and Brain Sciences 24(3): 383–403.Find this resource:

Hobbes, T. (1651/1968). Leviathan. edited by M. Oakeshott, (Harmondsworrth: Penguin).Find this resource:

Homans, G. C. (1961). Social Behavior: Its Elementary Forms (London: Rutledge & Kegan Paul).Find this resource:

Hume, D. (1739/1978). A Treatise of Human Nature (Oxford, UK: Clarendon).Find this resource:

Kliemt, H. (2009). Philosophy and Economics I: Methods and Models (München: Oldenbourg).Find this resource:

Kliemt, H. (2016). “Economics and Philosophy.” In Faccarello, G., and Kurz, H. (eds.), Handbook on the History of Economic Analysis: Great Economists Since Petty and Boisguilbert, pp. 161–174 (Northampton, Mass.: Edward Elgar).Find this resource:

Koopmans, T. C. (Ed.). (1951). Activity Analysis of Production and Allocation (No. 13) (New York: Wiley).Find this resource:

Lakatos, I. (1978). The Methodology of Scientific Research Programmes (Cambridge: Cambridge University Press).Find this resource:

Lewis, D. (1969). Convention Cambridge (Cambridge, Mass.: Harvard University Press).Find this resource:

Luce, R. D., and Raiffa, H. (1957). Games and Decisions: Introduction and Critical Survey: A Study of the Behavioural Models Project, Bureau of Applied Social Research (New York: Columbia University/Wiley).Find this resource:

Mackie, J. L. (1980). Hume’s Moral Theory (New York: Routledge).Find this resource:

Mayo, D. G. (1996). Error and the Growth of Experimental Knowledge (Chicago: University of Chicago Press).Find this resource:

McCabe, K., Houser, D., Ryan, L., Smith, V., and Trouard, T. (2001). “A Functional Imaging Study of Cooperation in Two-Person Reciprocal Exchange.” Proceedings of the National Academy of Sciences 98(20): 11832–11835, doi: 10.1073/pnas.211415698.Find this resource:

McClennen, E. F. (1990). Rationality and Dynamic Choice: Foundational Explorations (Cambridge: Cambridge University Press).Find this resource:

McKenzie, R., and Tullock, G. (1978). The New World of Economics: Exploration Into the Human Experience (Irwin: McGraw-Hill).Find this resource:

Mikula, G. (1973). “Gewinnaufteilungsverhalten in Dyaden bei variiertem Leistungsverhältnis.” Zeitschrift für Sozialpsychologie 3: 126–133.Find this resource:

Nozick, R. (1969). “Newcomb’s Problem and Two Principles of Choice.” In N. Rescher (Ed.), Essays in Honor of Carl G. Hempel, pp. 114–146. Synthese Library 24. Springer Netherlands. Retrieved from this resource:

Pearl, J. (2000). Causality: Models, Reasoning, and Inference (Cambridge: Cambridge University Press).Find this resource:

Pruitt, D. G. (1967). “Reward Structure and Cooperation: The Decomposed Prisoner’s Dilemma Game.” Journal of Personality and Social Psychology 7(1p1): 21.Find this resource:

Raphael, D. D. (Ed.) (1969). Moralists: 1650–1800 (Oxford: Clarendon Press).Find this resource:

Raphael, D. D. (2003). Hobbes, Morals and Politics (London: Routledge).Find this resource:

Rapoport, A., and Chammah, A. M. (1965). Prisoner’s Dilemma: A Study in Conflict and Cooperation (Vol. 165) (Ann Arbor: University of Michigan Press).Find this resource:

Robison, W. (1994). “Hume and the Experimental Method of Reasoning.” Southwest Philosophy Review 10(1): 29–37.Find this resource:

Rousseau, D. M. (Ed.). (2012). The Oxford Handbook of Evidence-Based Management (Oxford: Oxford University Press).Find this resource:

Samuelson, P. A. (1938). “A Note on the Pure Theory of Consumer’s Behaviour.” Economica 5(17): 61–71.Find this resource:

Sauermann, H. (ed.) (1967–1972). Beiträge zur experimentellen Wirtschaftsforschung [Contributions to experimental economics] (vol. 13) (Tübingen: Mohr-Siebeck).Find this resource:

Schmölders, G. (1953). “Ökonomische Verhaltensforschung.” ORDO: Jahrbuch für die Ordnung von Wirtschaft und Gesellschaft 5: 203–244.Find this resource:

Schneider, L. (1967). The Scottish Moralists on Human Nature and Society (Chicago and London: Chicago University Press).Find this resource:

Selten, R. (1978). “The Equity Principle in Economic Behavior.” In H. W. Gottinger and W. Leinfellner (eds.), Decision Theory, Social Ethics, Issues in Social Choice, pp. 289–301 (Dordrecht/Holland: D. Reidel).Find this resource:

Sen, A. (1973). “Behaviour and the Concept of Preference.” Economica 40(159): 241–259.Find this resource:

Smith, A. (2002). The Theory of Moral Sentiments (Cambridge: Cambridge University Press).Find this resource:

Smith, V. L. (1962) “An Experimental Study of Competitive Market Behavior.” Journal of Political Economy 70:2 (April): 111–137.Find this resource:

Smith, V. L. (1976). “Experimental Economics: Induced Value Theory.” The American Economic Review 66(2): 274–279.Find this resource:

Smith, V. L. (2003). “Constructivist and Ecological Rationality in Economics.” The American Economic Review 93(3): 465–508.Find this resource:

Smith, V. L., & Williams, A. W. (1990). “The boundaries of competitive price theory: Convergence, expectations, and transaction costs.” Advances in behavioral economics 2: 31–53.Find this resource:

Stodder, J. (1998). “Experimental Moralities: Ethics in Classroom Experiments.” The Journal of Economic Education 29(2): 127–138.Find this resource:

Strawson, P. F. (1962). “Freedom and Resentment.” Proceedings of the British Academy 48: 187–211.Find this resource:

Sugden, R. (1993). “Thinking as a Team: Towards an Explanation of Nonselfish Behavior.” Social Philosophy and Policy 10(1): 69–89.Find this resource:

Tetlock, P., and Gardner, D. (2015). Superforecasting: The Art and Science of Prediction (New York: Cornerstone Digital of Random House).Find this resource:

Thaler, R. (1991). The Winner’s Curse: Paradoxes and Anomalies of Economic Life (Princeton: Princeton University Press).Find this resource:

Thaler, R. H. (2015). Misbehaving: The Story of Behavioral Economics (New York: Norton & Company).Find this resource:

Tuomela, R., and Miller, K. (1988). “We-Intentions.” Philosophical Studies 53(3): 367–389. doi: 10.1007/BF00353512.Find this resource:


(1.) At the time of Hume, experimental meant “experiental,” comprising empirical and experimental in the more narrow present use of terms. Here the focus is on “experimental” in the modern sense. Interesting philosophical discussions of related issues are found in Robison (1994) and Mayo (1996).

(2.) In essence, the lecture that Vernon Smith gave on the occasion of receiving the Nobel Prize was making a similar point by framing Adam Smith as a behavioral economist; see Smith (2003). But Adam Smith was only one of many British moralists who were responding to Hobbes’s rational choice approach to human behavior; for collections of texts, see Raphael (1969) and Schneider (1967), and as introduction, see the first two chapters in Mackie (1980).

(3.) Alluding to Heilbroner (2011).

(4.) For details, see Güth and Kliemt (2007).

(5.) This leaves open whether a subjective/internalist/teleological or an objective/externalist/evolutionary perspective is adopted. As we shall point out in some detail, it is an insufficient separation between looking at the world through the window of cognitive processes of the actor herself from a kind of participant’s point of view and looking at the actor from an objective point of view as an “entity” showing behavior that leads to ambiguities in foundational controversies of social theory in general and in economics in particular.

(6.) The British moralists’ discussion of the three Hobbesian principles anticipated many of the foundational discussions among economists after neoclassical economics established itself as a proper discipline; see the references in note 2. On the concept of consistency employed here, see Samuelson (1938) and Sen (1973); on causal modeling, see Pearl (2000), and for further philosophical comments supporting incrementalism, see Kliemt (2009) as opposed to McClennen (1990).

(7.) Adherents of the Hobbesian paradigm subscribe implicitly to all three premises, whereas its opponents go at least against one of them. Because proponents as well as opponents are not always coherent in what they say, classification becomes somewhat messy. But because the three premises clearly describe central dimensions of controversy, we use them as guidance and as organizing principles of our discussion.

(8.) In Lakatosian categories (see Lakatos, 1978) the research program characterized by the three ideals may not be progressive, but the experimental research that sprang up since the 1950s in psychology and economics in opposition to the implied behaviorism of the basically Samuelsonian, neoclassical approach is clearly progressive. We are quite willing to concede that the progressive elements would not exist had the empirical and observational focus of behaviorism and the analytical discipline of economics not prepared the ground for it; for an impressive list of “anomalies,” see Thaler (1991).

(9.) The search for a kind of “new synthesis” of psychological and economic models of individual behavior became increasingly influential through the ascent of behavioral economics; for a popular but reliable and impressive account, see Thaler (2015). It is undeservedly forgotten that this has been at the top of the agenda of the German school of experimental economics since the 1960s—see Sauermann (1967–1972) but also Schmölders (1953) and the seminal methodological work of Hans Albert (1967); with Bruno Frey and his students adopting a kind of intermediate role, see Frey (1999) and Frey and Bohnet (1996).

(10.) An explicit statement to the effect can be found in Leviathan, § 4 (1651, 1968).

(11.) Quite surprisingly, only in the twentieth century could the distinction between definitions (noncreative and eliminable) and axioms be made in a precise way. Generally until well after the rise of non-Euclidean geometries in the nineteenth century the status of the axioms of geometry—and a fortiori axioms in other fields—was unclear.

(12.) Hobbes intended to base his theory on simple “laws of motion” summed up in a “more geometrico” version of a “push-pull” mechanics. He thought that all there is matter in motion. Matter tends to preserve its state of motion. It is either, in the case of inanimate matter, merely externally moved by forces (pushed or pulled) or, in the case of animate matter, internally moved to keep up its state of movement; see Raphael (2003). Despite its deficiencies, Hobbes’s view was clearly a laudable step toward modern systematic theorizing about (human) nature. It may be noted in passing that Newtonian mechanics as developed shortly after Hobbes accomplished the same with respect to all “matter in motion” but gave up the pull-push model.

(13.) Even if the authors of the most influential Cowles Commission series in mathematical economics themselves seem to have been endorsing Bourbaki-type views of mathematics, the reception of their exercises in axiomatic method, for example, Koopmans (1951) or Debreu (1959), was often inspired by classical views of the axioms as representing “self-evident” facts of the world. Without an implicit claim to self-evidence—or at least fundamental intuitive plausibility—the economists’ insistence that explanations had to be framed in their maximization under constraints framework would be unintelligible.

(14.) How evidence-based research gets into fields like evidence-based management is illustrated in Rousseau (2012). With respect to prediction in politics, see Tetlock and Gardner (2015).

(15.) Gilboa and Schmeidler (2010) endorse a concept of “theory” as a purely conceptual—classificatory—rather than nomological scheme. Due to this acknowledgment of the exclusively conceptual nature of the exercise, this philosophically very sophisticated book on “decision-representation” (“decision-logic”) is impervious to our criticism of making a priori claims about a posteriori facts.

(16.) For the sake of simplicity we restrict ourselves merely to monetary incentives, but what we say would apply as well to other substantive payoffs.

(17.) In such cases it becomes impossible to claim that observable, extrinsically motivating monetary incentives are sufficient to explain and predict overt choice making: a lesser amount of money accrues though a larger amount could have been causally brought about by the choice maker through an alternative choice (refuting the pursuit of individual interest); this takes place without thereby incurring additional future consequences that could explain choosing the smaller amount to secure better future outcomes (refuting causal incrementalism).

(18.) Of course, this is a special variant of individualism. For the sake of brevity, we lump individualism and the pursuit of self-regarding motivations together here.

(19.) Some theories concerning the nature and significance of mental representations are already lurking in the background if such “second thoughts” are abstracted away as irrelevant for explaining overt choices. They may even impute a kind of Newcomb problem; see Nozick (1969) and Campbell and Sowden (1985).

(20.) In the central introductory sentence of § 10 of Leviathan, Hobbes (1651, 1968) speaks of “any future apparent good” (italics added) expressing the role of subjectivity along with future directedness of rationally seeking for increasing one’s potential of attaining rewards.

(21.) To the best of our knowledge, an experiment demonstrating the truth of this has not been run. Given the complexities of the human mind, it might actually lead to “strange” results.

(22.) As mentioned before in an intuitive way already, a game form is a representation of the “physical” move structure of an interactive situation without assuming that the preferences—and the utility functions representing them—are common knowledge among the players.

(23.) This explains the conventional remark that according to classical utility conception, alternatives had been preferred because they were creating higher utility, whereas according to modern utility conceptions, they are assigned higher utility because they are preferred (for whatever reasons).

(24.) That this raises problems if mental processes of actors who are reasoning about the game are taken into account should be obvious. The reasoning about knowledge approach—as spelled out in detail, for instance, in Fagin et al. (1995) from a formal and in Lewis (1969) from a philosophical point of view—cannot be based on preferences “all” things considered if the reasoning is treated as included in the considerations. Then we might have to say that the preferences are fixed only after the game is solved through reasoning. This would mean that analysis on the basis of preferences would become circular. This problem may be avoided if we take preferences as representing choices—as in standard revealed preference conceptions. However, this way out would not fit well with a reasoning about knowledge or “eductive” approach. It would be congenial to an adaptive externalist perspective only. See on related issues Binmore (1987, 1988) and Hausman (2011).

(25.) A game form has basically the same move structure as the game but other than in a game the payoffs are substantive and not necessarily representing the choices or preferences over those substantive results.

(26.) See an early discussions of prisoner’s dilemma game results in this vein in Rapoport and Chammah (1965).

(28.) Playing against a machine or another individual may have effects on brain activity even if the game forms are payoff equivalent; see McCabe et al. (2001) corroborating the view that there is a distinct “participant’s” besides an “objective attitude” toward the world; see on this, of course, Strawson (1962).

(29.) Restricting ourselves to the monetary payoffs of Row because the payoffs of Column could be split symmetrically.

(30.) Game theory is based on subjective rankings of alternative “all things” rather than merely substantive orderings considered. It could incorporate the other effects but at the price of eliminating the crucial aim of explaining overt choices in terms of likewise observable extrinsic motivators.

(31.) Experimental studies of equity problems serve subsequently as illustrations of what experimental economic studies can conceivably contribute to debates dating back at least to Aristotle. Rather than talking about the results of such moral phenomenological research, we will present—if in a very stylized manner—how experimental economics studies such issues. We intend to present some of the “know how” rather than the “know that” of experimental economics (experimental game theory). Those who are interested in accounts of relevant empirical results might want to consult handbooks such as Kagel and Roth (1995), Plott and Smith (2008), and Kolm and Ythier (2006).

(32.) For an early account of equity seeking and its experimental confirmation by an economic experimentalist, see Selten (1978). Fehr and Schmidt (1999) and Bolton and Ockenfels (2000) are popular, experimentally informed accounts of how to integrate equity theory into standard economic utility theory.

(33.) For an even-handed account of incentivized and nonincentivized studies and other relations between experimental economics and experimental psychology, see Hertwig and Ortmann (2001).

(34.) One further (implicit) assumption seems to be that R depends nearly linearly on the sum of the individual contributions.

(35.) If work is more of a fun type and the dictator or some other individual contributed more than the others or, in the two individuals’ case, if a dictatorial allocator contributed more than half, he might nevertheless allocate equally.

(36.) In “yes-no games” responders have veto power but do not know which allocation the proposer has selected. Anticipating the nonmonitoring by responders, proposer participants offer less to responders than in ultimatum but still more than in dictator experiments; see Güth and Kirchkamp (2012).

(37.) The relation to a superior omniscient being was often seen as a comparative advantage of religious individuals who by signaling their religiosity apparently signaled reliability.

(38.) Other than in Chamberlin’s (1948) Harvard classroom experiment in which participants had to imagine monetary incentives to simulate market behavior, participants in Smith’s experiment were earning real money. Though Smith was originally referring to his studies as simulating markets, he later on insisted that he created markets in the laboratory. The following account shares this view. But it is not faithful in detail to the original experiments. It emphasizes certain aspects in a stylized and simplified form; for a fuller account true to the original approach, see Easley and Ledyard (1993).

(39.) The two who could buy and sell at ECU 9 with zero profit and loss.

(40.) That the double auction is oral is not necessary. A computerized setup in which all are informed about the bid and ask prices at which trades take place will do.

(41.) That economic classroom experiments can be used to address “moral subjects” in most instructive ways is constructively shown in Stodder (1998). This should not be neglected as a technology of “teaching ethics” with real rather than merely thought experiments.