H

HS

a

ARAA

KMGBRCP

1

poeGrc

2

oaai

hFa

0

Journal of Economic Behavior & Organization 126 (2016) 95–109

Contents lists available at ScienceDirect

Journal of Economic Behavior & Organization

j ourna l h om epa ge: w ww.elsev ier .com/ locate / jebo

omo Ludens: Social rationality and political behavior

erbert Gintisanta Fe Institute, 1399 Hyde Park Road, Santa Fe, NM 87501, USA

r t i c l e i n f o

rticle history:eceived 27 November 2015ccepted 19 January 2016vailable online 10 February 2016

eywords:oralityene-culture co evolutionehavioral game theoryational actor modelulturerivate and public persona

a b s t r a c t

The most straightforward defense of political democracy is grounded in rational choice.Political democracy gives people power in public life parallel to that afforded by marketsin private life: the power to turn their preferences into social outcomes. However, ratio-nal choice models of voter behavior dramatically underpredict voter turnout in all but thesmallest elections. Fiorina (1990) has called this “the paradox that ate rational choice the-ory.” This anomaly is the centerpiece of Green and Shapiro’s (1994) critique of rationalchoice methodology in political theory. This paper proposes a form of social rationality thatstrengthens the classical rational actor model. This concept explains many central statisticalregularities concerning voter turnout and the historical regularities concerning collectiveaction.

© 2016 Elsevier B.V. All rights reserved.

. Introduction

The most straightforward defense of political democracy is grounded in rational choice. Political democracy gives peopleower in public life parallel to that afforded by markets in private life: the power to turn their preferences into socialutcomes. However, rational choice models of voter behavior dramatically underpredict voter turnout in all but the smallestlections. Fiorina (1990) has called this “the paradox that ate rational choice theory.” This anomaly is the centerpiece ofreen and Shapiro’s (1994) critique of rational choice methodology in political theory. This paper proposes a form of social

ationality that strengthens the classical rational actor model. This concept explains many central statistical regularitiesoncerning voter turnout and the historical regularities concerning collective action.

. Public and private spheres

The private sphere is the locus of transactions as people live their lives in civil society. An agents private persona is the setf preferences, beliefs, and social affiliations that govern this behavior in the private sphere. The public sphere is the locus ofctivities through which the rules of the game that define society itself are created, maintained, transformed, interpreted,nd executed. An individual’s public persona is the set of preferences, beliefs, and social networks that govern his behaviorn the public sphere. Political theory includes the study of public sphere behavior.

The private and public spheres are closely interrelated in individual decision-making. A public sphere transaction mayave private sphere costs and benefits that a participant in the public sphere may take into account in deciding how to act.or instance, an individual may not vote if queues at the polling station are very long, or may decide to skip a collectivection in which the probability of physical harm is very high.

E-mail address: hgintis@comcast.net

http://dx.doi.org/10.1016/j.jebo.2016.01.004167-2681/© 2016 Elsevier B.V. All rights reserved.

96 H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109

By contrast with the private sphere, much behavior in the public sphere is fundamentally non-consequential: an agent’sactions have no discernible effect on social outcomes. Consider, for instance, voting. Estimates of the probability that a singlevoter’s decision will determine the outcome of large election are between one in ten million and one in one hundred million(Good and Mayer, 1975; Chamberlain and Rothschild, 1981; Gelman et al., 1998). In a compendium of close election resultsin Canada, Great Britain, Australia, and the United State, no election in which more that 40,000 votes were cast has everbeen decided by a single vote. In the Massachusetts gubernatorial election of 1839, Marcus Morton won by two votes out of102,066 votes cast. In the Winchester UK general election of 1997, Mike Oaten won by two votes out of 62,054 votes cast.The result was annulled and in a later by-election, Oaten won by 21,000 votes. In smaller elections, a victory by a very smallmargin is routinely followed by a recount where the margin is rarely less that twenty-five (Wikipedia, List of Close ElectionResults, November 2014).

There is thus virtually no loss in accuracy in modeling voting behavior in large elections as purely non-consequential inthe sense that a single individual’s decision to vote or abstain, or for whom to vote, has no effect on the outcome of theelection (Downs, 1957a; Riker and Ordeshook, 1968).

A canonical participant in a decision process is an individual whose choice is non-consequential: his behavior affects theoutcomes infinitesimally or not at all. According to the data presented above, voters in a large election are canonical partici-pants. Individuals who participate in large collective actions are similarly canonical participants, as are those who volunteerto fight in a war between nations. Of course, there are some public sphere activities that are potentially consequential andhence non-canonical, such as running for office, organizing a voter registration drive, or contributing considerable amountsof money to a particular party or candidate. But most activities in the public sphere are canonical. Ignoring the infinitesimalprobabilities that canonical participants affect outcomes is useful and harmless simplification, akin to ignoring the force ofgravity in analyzing the electronic circuitry of a computer or ignoring the light from distant stars in calculating the propertiesof a solar panel.

Non-consequential public sphere activities are at the center of the structure and dynamics of modern societies. If citizensdid not vote, liberal democratic societies could not operate. Even if voting were mandatory, as it is in many countries aroundthe world, liberal democracy could not operate, as classically rational canonical voters would choose to remain rationallyignorant, having no incentive to incur the costs in time and energy of becoming politically informed. Moreover, modernliberal democracy was achieved through collective actions over centuries. These collective actions have been successfulbecause of the collective impact of canonical participants who incurred significant personal costs, often death, in opposingillegitimate authority (Bowles and Gintis, 1986). These participants systematically violated the rules of classical rationality,as presented in the seminal works of von Neumann and Morgenstern (1944) and Savage (1954).

However, canonical participants generally consider their behavior as rational and goal-oriented. If one asks a voter waitingin a queue at the polling booth why he is standing there, or if one asks someone in a group protesting political corruptionwhy he is chanting and holding signs, he will think the question absurd. He is there, of course, to register his support forvarious candidates for office, or to help topple a corrupt government. If you point out that he will make no difference to theoutcome, he may plausibly respond that you are guilty of faulty reasoning because if everyone following your reasoning, noone would vote and no one would fight to topple a corrupt regime. If you persist in asking why he personally votes, notingthat the other participants do not follow your reasoning, and his abstention will not affect the decision of others, he mayplausibly respond that this reasoning is also illogical because if all citizens reasoned this way, most would conclude thefutility of voting.

The classical axioms of rational choice theory cannot explain the behavior of a canonical participant in the public spherebecause these axioms cover only situations in which choices are consequential in the sense that distinct choices lead todistinct outcomes (Savage, 1954). The behavior of canonical participants in the public sphere is not rational in this sense.We shall, however, argue that canonical participants are socially rational in an analytically clear sense.

3. Classical and social rationality

Classical rationality (Savage, 1954) fails to capture even the most elementary forms of socially rational behavior. Severaleconomists, decision theorists, and philosophers have explored a more socially relevant form of rationality that they term var-iously “we-reasoning,” “team reasoning,” and “collective intentionality” (Bacharach, 1987, 1992, 2006; Bacharach et al., 2006;Bratman, 1993; Colman et al., 2008; Gilbert, 1987, 1989; Hurley, 2002; Searle, 1995; Sugden, 2003; Tuomela, 1995). Hereare several analytically clear examples of choice behaviors that should appear in any plausible account of social rationality.

Consider the two-player game illustrated in Fig. 1. Assuming Bob is perfectly self-regarding, classical rationality suggeststhat he is indifferent between choosing Up or Down. Indeed, in animal behavior studies, even our closest relatives oftenchoose Up and Down with roughly equal frequency in such situations. But human children regularly choose the prosocialaction from a very young age (Silk et al., 2005; Vonk et al., 2008; Warneken and Tomasello, 2006). We call this elementaryform of social rationality prosocial default. A socially rational agent exhibits prosocial default if, whenever he is classicallyindifferent to choices A and B, but A Pareto-dominates B for the other players, then he does not choose B. In this case, choosing

Down is a dominant strategy for Alice, and choosing Down is the only socially rational choice for Bob.

Consider the slightly more sophisticated game depicted in Fig. 2. In this case Down is the socially rational choice for Bob,and Down, while no longer a dominant strategy, is the preferred choice for Alice provided she knows that Bob is sociallyrational.

H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109 97

Fig. 1. A simple game where Down is the prosocial default.

g{Umt

sars

a

3

tpc

eocc

Fig. 2. A more sophisticated game where Down is the prosocial default.

Bayesian rationality is also incapable of justifying the choice of a Pareto-efficient Nash equilibrium in a pure coordinationame with a unique Pareto-efficient equilibrium. For instance, consider the game depicted in Fig. 3. The Pareto-efficientDown,Down} equilibrium cannot be justified by a classical rationality argument. For instance, if Bob believes Alice will playp, his best response is Up. Bob may believe this because he believes Alice believes he will play Up. And so on. Similarly, Aliceay be perfectly classically rational in playing Up. But this is not socially rational. If social rationality is mutually known,

hen the unique Nash equilibrium is the Pareto-efficient choice {Down,Down}.More generally, we argue as follows. We say a strategy profile � = {�1, . . ., �n} in an n-player game strictly dominates a

trategy profile � = {�1, . . ., �n} if the payoff to � is strictly greater than the payoff to � for each player i = 1, . . ., n. We say that strategy profile � is socially irrational if there is a Nash equilibrium that strictly dominates �. Otherwise we say � is socialational. Finally, let us say that a set of agents is socially rational if they never choose individual strategies that not part of aocially rational strategy profile.

We shall show that the classical notion of common knowledge of rationality can imply that agents are socially irrational.It follows that it may be inadvisable to assume common knowledge of rationality in analyzing games in which players

re assumed to be socially rational.

.1. The email game

In the Email Game (Rubinstein, 1989), Bob and Alice independently choose A or B. If the state of the world is a, the payoffo both players choosing A is m > 0, and their common payoff when both choose B is zero. If the state of the world is b, theayoff to both players choosing B is m, and their common payoff when both choose A is zero. In both states, if they fail tooordinate their choices, the player choosing B loses an amount l > m, while the player choosing A receives zero.

If both players know the state of the world, which is b with probability p and a with probability 1 − p, there is a Nashquilibrium in which both players coordinate to achieve the payoff m in either state of the world. But suppose only Bob

bserves the state of the world, and if the state is b, his computer sends a message to this effect to Alice’s computer. If Alice’somputer receives this message, which occurs with probability 1 − �, where � > 0 is some small error probability, then Alice’somputer sends an email to Bob’s computer confirming that she has received his message and now she knows that the state

Fig. 3. Socially rational choice in a coordination game.

98 H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109

of the world is b. However, this message also has a probability � of going astray. Assuming Bob’s computer receives thisconfirming message, it sends a second email to Alice’s computer confirming that it has received her confirmation. The twocomputers continue to send confirming messages back and forth, each time with a probability � of going astray. Clearly,with probability one Bob’s computer will eventually fail to receive a confirming message, at which point the number ofmessages his computer sent is displayed on his computer screen, and similarly, the number of messages Alice’s computersent is displayed on her computer screen.

It will be convenient to assume � < 1/2 and � < (1 − p)l/m. Using the iterated elimination of dominated strategies, which isjustified by CKR, Rubinstein (1989) shows that there is only one Nash equilibrium of this email game in which Bob plays A in thestate of the world a. In this equilibrium both players play A independently of the number of messages sent. To prove this assertion,let tA and tB be the number of messages sent by Alice’s and Bob’s computers, respectively, and let sA(t) be Alice’s choices ofA or B in a Nash equilibrium when the number of messages her computer sent is t, and similarly sB(t) is Bob’s equilibriumchoice when his computer sent t messages. Suppose sB(0) =A. If tA = 0, then Alice did not receive a message. With probability(1 − p) the actual state is a, Bob chooses A, and Bob’s computer sent no message, while with probability p� the actual stateis b, Bob’s computer sent a message, but Alice’s computer did not receive the message. The posterior probabilities of statesof the world, p′

A and p′B, according to Alice, are then given by

p′B = Pr[B|tA = 0] = Pr[tA = 0|B]Pr[B]

Pr[tA = 0]= p�

p� + 1 − p(1)

p′A = Pr[A|tA = 0] = Pr[tA = 0|A]Pr[A]

Pr[tA = 0]= 1 − p

p� + 1 − p(2)

If Alice chooses A her expected payoff is �A = p′Am + p′

B(0). If Alice chooses B, her payoff is �B = p′A(−l) + p′

By, where y = m ifBob chooses B when the state is b, and y = 0 if Bob chooses A when the state is b. Thus Alice’s payoff in this case satisfies

�B ≤ p′A(−l) + p′

Bm = �m − (1 − p)lp� + 1 − p

<(1 − p)m

p� + 1 − p= �A.

Thus assuming Alice is rational, we have sA(0) = A.Now suppose tB = 1, so the state is surely b and Bob’s computer emailed this fact to Alice’s computer, but Bob received no

reply from Alice’s computer. Then the probability z that Alice did not receive the email sent by Bob is given by

z = Pr[tA = 0|tB = 1] = Pr[tB = 1|tA = 0]Pr[tA = 0]Pr[tB = 1]

= �� + (1 − �)�

>12

.

If Alice did not receive the message and Alice is rational, she will play A because we have already ascertained that sA(0) =A.Thus if Bob knows that Alice is rational, he knows that his payoff is zero if he plays A, and at most z(− l) + (1 − z)m, which isstrictly negative since l > m and z > 1/2, if he plays B. Thus if Bob is rational and he knows Alice is rational, then Bob will playA and sB(1) =A. Now suppose Alice sent one message but received no confirmation, so tA = 1. Then Bob failed to receive themessage with probability z and if Alice knows that Bob is rational and Bob knows that she is rational, then Alice knows thatBob plays A, so Alice calculates her payoff to A as zero (since the state is b) and her payoff to B is at most z(− l) + (1 − z)m < 0, sosA(1) =A. It is clear that this argument can be repeated for all t > 1, in each iteration assuming another higher level of mutualknowledge of rationality. So assuming CKR, we have shown that sA(t) = sB(t) =A for all t ≥ 0.

However, if Bob and Alice are rational and they both know that each sent out at least one message (tB ≥ 2, tA ≥ 1), the stateis surely b, each player knows that this is the case, and each player knows that other knows that this is the case. Therefore thestrategy profile � in which sB(0) = sB(1) = sA(0) =A and sB(tB) = sA(tA) =B for tB ≥ 2, tA ≥ 1 form a Nash equilibrium that strictlydominates the Nash equilibrium derived from the CKR assumption. To see this, suppose first that Bob plays his part in thestrategy profile �. Then if tA = 0, Alice receives no signal, so she knows that with probability (1 − p)/(1 − p + p�) the state is aand Bob’s computer sent no signal, and with probability p�/(1 − p + p�) the state is b and Bob’s signal went astray. Her payoffto A is thus zero and her payoff to B is (− (1 − p)l + p�m)/(1 − p + p�) < 0, so sA(0) =A. If tA = 1, then Alice knows the state issurely b and she knows that Bob knows this as well. Her computer automatically sends a message to this effect to Bob, whicharrives with probability 1 − �. If this message arrives, then tB ≥ 2, so Bob plays B. Otherwise Bob plays A. Thus Alice’s payoffto A is zero and her payoff to B is (1 − �)m − �l > 0, so sA(1) =B. Similarly, sA(t) =B for all t ≥ 2.

Now suppose Alice plays her part in the strategy profile �. If tB = 0, then tA = 0 so Alice plays sA(0) =A, to which sB(0) =Ais a best response. If tB = 1, then Bob’s computer sent a message to Alice. Conditional on the fact that Bob received no replyfrom Alice’s computer Alice received the message with probability pB = �(1 − �)/(� + �(1 − �)) and plays sA(1) =B or Bob’s replywent astray, so Alice plays sA(0) =A with probability pA = �/(� + �(1 − �)). Because pA > 1/2 and l > m, Bob’s best response issB(1) =A. If tB = 2, then Bob knows Alice plays sA(1) =B, to which his best response is sB(2) =,B. Similarly reasoning for tB > 2and tA > 0 complete the proof.

The recursive argument for the Rubinstein equilibrium breaks down for tA = 1 because we no longer assume that Alice

knows that Bob knows that she is rational, hence she is not obliged to conclude that Bob will play A even if he received herconfirmation.

The expected payoff to the Rubinstein equilibrium is (1 − p)m, while the payoff to strategy profile � is (1 − p(1 − �)�)m forBob and (1 − p(1 − �)�)m − p(1 − �)2�l for Alice. Because these payoffs are larger than those of the Rubinstein equilibrium,

rs

3

aetx

dtk

tutx>a

atxhbw

4

aAii

wa

a

T

Taa

H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109 99

ational players will reject reasoning based on CKR in favor of the simple equilibrium: play B if and only if you know thetate is b. CKR in this situation implies socially irrational behavior on the part of the agents.

.2. Global games

Global games are games in which players receive correlated but noisy signals of payoff structure. We consider Carlessonnd van Damme’s (1993) two-player global game in which Alice and Bob have two strategies, and ˇ. Either player cannsure a payoff of x by playing ˛, and each receives a payoff of 4 if they coordinate on ˇ. If they fail to coordinate, however,he ˇ-player receives zero. If both players see x, there is a Pareto-efficient equilibrium in which both players choose when

> 4 and when x ≤ 4.The game is a global game, however, because Alice observes x with a small error �A, so that her observation xA is uniformly

istributed on the interval [x − �, x + �]. Bob similarly observes xB, which is uniformly distributed on [x − �, x + �]. We assumehe game, including the maximum error � and the fact that the actual errors xA − x and xB − x are uncorrelated, are commonnowledge.

There is a natural Nash equilibrium of this game in which each player i chooses if xi < 4 and chooses if xi ≥ 4. To seehat this is a Nash equilibrium, suppose Bob chooses this strategy. Note that Alice’s posterior for x when she observes xA isniformly distributed on [xA − �, xA + �], which implies that xA − 2� ≤ xB ≤ xA + 2�. Clearly, then if Alice observes xA ≥ 4 +2�,hen she knows that xB ≥ 4, so Bob plays ˛, to which her best response is ˛. If she observes xA < 4 −2�, then she knows thatB < 4 so Bob plays ˇ, to which her best response is ˇ. Suppose she observes xA such that 4 ≤ xA < 4 +2�. Then x has mean4, so the probability the Bob chooses is p˛ > 1/2. Hence Alice’s best response is ˛. Finally, if 4 − 2� < xA < 4, then a parallelrgument shows that Bob chooses with probability p˛ < 1/2, so Alice’s best response is to choose ˇ.

Note that the payoffs to the natural Nash equilibrium approach the full information payoffs as � → 0. However, if wessume CKR and thus reason using the iterated elimination of dominated strategies, Carlsson and van Damme (1993) showhat there is a unique Nash equilibrium in which player i chooses after observing xi ∈ (0, 2) and choosing after observingi ∈ (2, 4). This equilibrium is clearly Pareto-dominated by the natural Nash equilibrium. Moreover, in many cases Alice mayave a subjective prior that places a high probability on Bob’s playing his part in the natural Nash equilibrium, and Bob mayelieve that Alice will act similarly. In such a situation, rational Bob and rational Alice will play the natural Nash equilibriumhereas CKR implies they play a strictly dominated strategy profile. Hence CKR implies Alice and Bob are socially irrational.

. The social rationality of voter turnout

Following Palfrey and Rosenthal (1985), suppose an election is determined by majority rule, with a tie vote decided by fair coin toss. There are m voters who choose simultaneously to vote for alternative 1, vote for alternative 2, or abstain.lternative 1 is preferred by m1 > 0 voters and alternative 2 is preferred by m2 = m − m1 > 0 voters. We assume m1 ≥ m2. If bi

s the payoff to agent i if his alternative wins, with payoff zero if his side loses, and if di is the cost of not abstaining for agent, the i will vote rather than abstain if

bi

(pi

2 + 12

pi1

)≥ di, (3)

here pi1 is the probability that i’s vote leads to a tie, and pi

2 is the probability that i’s vote breaks a tie. We write ci = di/bi,nd refer to ci as the cost of voting. It is clear that agent i will vote precisely when ci excesses a threshold ci given by (3).

Suppose qj is the probability of voting for an agent preferring alternative j, j = 1, 2. The probability pi(k1, k2) that kj votesre cast for alternative j, j = 1, 2, not including the vote of agent i (should he vote) is given by

pi(k1, k2) =(

m1 − 1

k1

) (m2

k2

)qk1

1 qk22 (1 − q1)m1−k1 (1 − q2)m2−k2 . (4)

he probability that alternative 1 wins when agent i, whom we assume prefers alternative 1, votes is then

m1−1∑i=1

k1−1∑k2=0

pi(k1, k2) +m2∑

k1=0

pi(k1, k1) + 12

m2−1∑k1=0

pi(k1, k1 + 1). (5)

he first (double) summation in (5) is the probability i is not pivotal, the second summation is the probability that i breaks tie, and the final summation is the probability i creates a tie and the coin flip favors alternative 1. The probability thatlternative 1 wins when this agent i abstains is similarly

m1−1∑i=1

k1−1∑k2=0

pi(k1, k2) + 12

m2−1∑k1=0

pi(k1, k1 + 1). (6)

100 H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109

Fig. 4. The size effect: voter turnout rates decline as electorate size increases.

Subtracting (6) from (5) we see that the largest cost of voting ci for which agent i will vote rather than abstain is given by

2ci =m1−1∑k1=0

pi(k1, k1) +m2−1∑k1=0

pi(k1, k1 + 1). (7)

Because pi(k1, k2) is the same for all alternative 1 supporters, there is a maximum cost of voting c∗1 such that

c∗i = 1

2

⎛⎝ m2∑

k1=0

pi(k1, k1) +m2−1∑k1=0

pi(k1, k1 + 1)

⎞⎠ . (8)

Repeating this argument for the case where agent i prefers alternative 2, there is a maximum cost of voting c∗2s such that

c∗2 = 1

2

⎛⎝m2−1∑

k1=0

pi(k1, k1) +m2−1∑k1=0

pi(k1, k1 + 1)

⎞⎠ . (9)

Suppose the cost of voting has cumulative distribution Fj(c) for agents who prefer alternative j, j = 1, 2. For ease of expo-sition, we assume the cost of voting is uniformly distributed on the unit interval [0, 1]. Then we can replace ci with qi in Eqs.(8) and (9), giving two equations in two unknowns for the Nash equilibrium values of voter turnout (q1, q2).

This Nash equilibrium predicts several major empirical regularities of voter turnout, both in small-scale real-worldelections and in the laboratory (Levine and Palfrey, 2007). Rather than deriving the comparative statics of this voter turnoutmodel analytically, which may or may not be possible, I have generated numerical solutions using the Mathematica softwareprogram (Wolfram Research Inc., 2014).

The first regularity is the electoral size effect: voter turnout declines with increasing size of the electorate (Lijphart, 1997).For instance, national elections have larger turnout rates that state elections in the United States. Fig. 4 illustrates thisphenomenon in our model with electorate sizes of 18–36, where the difference between the majority and minority voters

is two. Similar results hold, however, for higher ratios of majority and minority voters. Note that these are clearly smallelection results because voter turnout declines rapidly towards zero for even moderately large electorate sizes.

The second and third electoral regularities are the voting cost effect and the importance of election effect: when the costof voting increases, fewer people vote, and when the stakes are higher, more people vote. The reason given by our model is

H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109 101

Fig. 5. The voting cost and Importance of election effects: In an electorate of size 20, a higher cost to benefit ratio leads to lower turnout for all sizes of thems

tlro

rcn

vsao1BhpiTgdr

haeut

ajority faction. The vertical axis shows the percentage increase in turnout when the cost of voting is lowered by one third. The horizontal axis shows theize of the majority coalition.

hat an increase in the cost of voting, ceteris paribus, entails a higher cutoff cj , and a higher benefit to a voter from winningowers the cost cutoff cj . We can represent both of these effects, assuming they are experienced equally by all agents, byeplacing cj by ˛cj in Eqs. (5) and (9), where < 1 corresponds to a decreased cost-benefit ratio. The results for an electoratef size 20 where the majority consists of between 11 and 20 voters is given Fig. 5.

The fourth electoral regularity is the competition effect: turnout is higher when the election is expected to be close. Theeason in our model is that the probability that there will be a pivotal voter is higher when the election is expected to belose (Shachar and Nalebuff, 1999). It should be noted that this appears to be true, as illustrated in Fig. 6, but it is certainlyot obvious.

These comparative static voter turnout phenomena, verified analytically for a very small electoral size but reflectingoter behavior in large elections, lend support for the simple observation that people behave in large elections rationally andtrategically as though they were actually involved in very small elections. Humans appear to follow a logic that may be describeds distributed-effectivity: in public life, choose a rule that like-minded people might plausibly choose, and if followed by manyf us, will likely lead to the most desirable outcome. Distributed effectivity has been termed rule-consequentialism (Harsanyi,977; Hooker, 2011; Roemer, 2010), and team-based reasoning (Gilbert, 1989; Tuomela, 1995; Searle, 1995; Bacharach, 2006;acharach et al., 2006). Each of these terms captures part of the somewhat subtle, yet substantively important, notion thatuman minds often act together even in their separateness. Distributed effectivity is the principle that distinct minds canroduce real-world effects by acting in parallel using similar and conformable decision processes. Rule-consequentialism

s the principle that like-minded agents become consequential by virtue of the common decision algorithms and values.eam reasoning carries the connotation that the agents is both privileged to be part of a social process, so that the collectiveoal becomes internalized as a personal goal, and yet is duty-bound and committed to performing his part in the socialecision process even when this conflicts with personal goals. In terms of our model of voter turnout, distributedly effectiveeasoning takes the form of each member of a coalition preferring a distinct alternative recognizes that all voters

The validity of distributed effectivity implies that people are perfectly reasonable in assenting to such assertion as “I amelping my candidate win by voting” and “I am helping promote democracy by demonstrating against the dictator.” Thesessertions are rigorously correct, despite the non-consequential nature of the acts of individual minds. Because distributed

ffectivity is so ingrained in our public persona, people untrained in traditional rational decision theory simply cannotnderstand the argument that it is irrational to vote or to participate in collective actions, even when they can be persuadedhat their actions are non-consequential.

102 H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109

Fig. 6. The competition effect: the closer in size the majority and minority coalitions, the higher the probability of voting. The election size is twenty.

Of course, in a very large election, both probabilities would be so small as to be negligible. But with distributed effectivity,agents reason that their choice is diagnostic of what other voters will do, so the relevant probabilities become consequential.The fifth is that strategic voting is widely observed, such as ignoring candidates that have no hope of winning, and voting foran unwanted candidate in order to avoid electing an even less preferred candidate (Niemi et al., 1992; Cox, 1994). Finally,in many circumstances there is a strong social network effect: individuals who are more solidly embedded in strong socialnetworks tend to vote at a higher rate (Edlin et al., 2007; Enos, 2012). This is because the power of distributed effectivity isa function of the complexity of the network of distributed cognition to which the individual belongs.

5. Comparison with other models of voter turnout

How might we account for canonically non-consequential political behavior? One possible answer is that people believetheir public sphere behaviors are consequential even when they are not, so they act as though their actions effectivelydetermine the outcome, at least with some substantially positive probability. From my reading of the literature on canonicalpolitical behavior, this is the most common, though rarely explicitly stated, assumption. For instance, Duncan Black’s famousmedian voter theorem (Black, 1948) implicitly assumes that a self-interested citizen will vote and this vote will register hispersonal preferences. Similarly, Anthony Downs, a pioneer in the application of the rational actor model to political behavior(Downs, 1957a) describes his model as follows:

Every agent in the model—whether an individual, a party or a private coalition, behaves rationally at all times; thatis, it proceeds toward its goals with a minimal use of scarce resources and undertakes only those actions for whichmarginal return exceeds marginal cost. Downs (1957b, p. 137)

And yet, almost immediately after stating this assumption, he writes:

[We assume that] voters actually vote according to (a) changes in their utility incomes from government activity and

(b) the alternatives offered by the opposition Downs (1957b, p. 138).

These two assumptions are compatible only if agents believe that their votes are consequential.But in fact canonical participants generally do not believe that their behavior is consequential. For instance Enos and

Fowler (2010) report a study in which the median respondent to the question as to the chance their vote will change the

o1

Sor6tn

oj7vatP

awmatep

rboTini

coopio

a1taa

6

rmwaca

rn

H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109 103

utcome of a presidential election gave the answer 1 in 1000, which although small, is in fact too large by a factor of at least0,000. The authors write:

However. . .over 40% of regular voters know that the chances of a pivotal vote are less than one in amillion.. . .[Moreover], the less likely you are to think your vote will actually matter, the more likely you are to vote.

imilarly, in a Wisconsin survey reported in Dennis (1970), 89% of respondents disagree with the statement “So manyther people vote in the national elections that it doesn’t matter much to me whether I vote or note.” Moreover, 86% ofespondents agree with the statement “I generally get a feeling of satisfaction from going to the polls to cast my ballot,” and5% of respondents disagree with the statement “A person should only vote in an election if he cares about how it is goingo come out.” Thus, although voters behave strategically (Fedderson and Sandroni, 2006), they know that their behavior ison-consequential (Edlin et al., 2007; Hamlin and Jennings, 2011).

Another possible answer is that people consider voting a duty or social obligation, so that not voting is an unethical actf free-riding on the altruism of others. When asked if the good citizen must always vote, American level of agreement isust slightly lower than obeying the laws and not evading taxes (Dalton, 2008). In an Annenberg study of the 2000 election,1% of Americans agreed that they felt guilty when they failed to vote. Even among those who reported that they had nototed, nearly half said they felt guilty. Blais (2000) reports that more than 90% of respondents in two Canadian provincesgree that “it is the duty of every citizen to vote.” Clarke et al. (2004) similar findings for British voters. However, the dutyo vote cannot explain important regularities in voter behavior, including the close election and underdog effects (Levine andalfrey, 2007).

A third possibility is that many voters are altruistic and vote out of concern for the well-being of others who will beffected by the outcome of the electoral process. Although if voting is non-consequential, a single voter cannot affect theell being of others, in this case, where there may be many millions of others, the one in ten million or one in a hundredillion chance of changing the outcome of the election, when multiplied by the number of people thereby affected, becomes

significant quantity. Certainly, however, no voter thinks in this bizarre way, and many canonical participants have interestshat are far narrower than the citizenry as a whole, and often act to promote the interests of one group of citizens at thexpense of another. Indeed, it is common to hear a small group of voters deemed “selfish” because they promote their ownarochial interest above the good of society as a whole.

To model the rationality of the canonical participant in the political sphere, we must revise the standard axioms ofational choice (Savage, 1954). In a related paper, I have explored the implications of replacing Savage’s assumption thateliefs are purely personal “subjective probabilities” with the notion that the individual is generally embedded in a networkf social actors over which information and experience concerning the relationship between actions and outcomes is spread.he rational actor thus draws on a network of beliefs and experiences distributed among the social actors to which he isnformationally and socially connected. By the sociological principle of homophily, social actors are likely to structure theiretwork of personal associates according to principles of social similarity, and to alter personal tastes in the direction of

ncreasing compatibility with networked associates (McPherson et al., 2001; Durrett and Levin, 2005; Fischer et al., 2013).To this principle of distributed cognition we may add a principle of distributed effectivity in which canonical participants

onsider their actions as effective contributions to social outcomes when they do their part as a member of a networkf loosely linked individuals with consonant objectives distributed across the network. In this framework, the behaviorf a canonical participant is not a costly choice with no benefits, but rather a voluntary, costly but personally enriching,articipation in a collective effort. With this notion in place, a theory of canonical participant behavior must elucidate how

ndividual preferences over political outcomes are formed, and how agents trade off between public and private spherebjectives.

The principle of distributed effectivity appears closely related to the notions of team reasoning and team intentionalitys developed in an extensive philosophical and economics literature (Bratman, 1993; Bacharach, 1987, 1992, 2006; Gilbert,987, 1989; Tuomela, 1995; Searle, 1995; Hurley, 2002; Sugden, 2003; Bacharach et al., 2006; Colman et al., 2008). However,he behavior explored in these contributions is generally socially structured cooperation and collaboration, in which actionsre highly consequential. We will approach the problem of distributed effectivity by considering canonical participation as

form of moral behavior.

. Self-regarding, other-regarding, and universalist rational action

Rational actors exhibit three types of motives in their daily lives: self-regarding, other-regarding, and universalist. Self-egarding motives include seeking personal wealth, consumption, leisure, social reputation, status, esteem, and otherarkers of personal advantage. Other-regarding motives include valuing reciprocity and fairness, and contributing to theell-being of others. Universalist motives are those that are followed for their own sake rather than for their effects. Chief

mong universalist goals are character virtues, including honesty, loyalty, courage, trustworthiness, and considerateness. Ofourse, in the private sphere such universalist goals have consequences for those with whom one interacts, and for society

s a whole. But one undertakes universalist actions for their own sake, beyond any consideration of their effects.

Agents will generally trade off among these various motives. For instance, being honest may be personally costly oreputationally rewarding, and may either hurt or benefit others whose well-being one values. Universalist motives thus doot reduce to self- or other regarding motives, but they do trade off against these other motives.

104 H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109

Self- and other-regarding behavior is well documented in the literature, but universalist behavior is far less so. I willpresent an example, as revealed by laboratory experiments using experimental game theory.

6.1. The universalist principle of honesty

Universalist moral actions are performed, at least in part, because it is virtuous to do so, apart from any effects theseactions have on oneself, others, or society in general. For instance, one can be honest in dealing with another agent withoutcaring at all about the effect on the other agent, or even caring about the impact of honest behavior on society at large.Similarly, one can be courageous in battle because it is the right thing to do, independent from the effect of one’s actionson winning or losing the battle. Of course, the value of honesty in a transaction may be lessened or turned negative if it ispersonally costly or it harms others about whom one cares.

Many studies have show that people exhibit considerable degrees of honesty even when this is costly and there is nochance that they could be discovered cheating (Bucciol and Piovesan, 2011; Houser et al., 2012; Fischbacher and Follmi-Heusi, 2014; Mazar et al., 2008; Shalvi et al., 2012; Cohn et al., 2014). A particularly clear example of the value of honestyis reported by Gneezy (2005), who studied 450 undergraduate participants paired off to play three games of the followingform, all payoffs to which are of the form (a, b) where player 1 (Alice) receives a and Player 2 (Bob) receives b. In all games,Alice was shown two pairs of payoffs, A:(x, y) and B:(z, w) where x, y, z, and w are amounts of money with x < z and y > w, soin all cases, B is better for Bob and A is better for Alice. Alice could then say to Bob, who could not see the amounts of money,either “Option A will earn you more money than option B,” or “Option B will earn you more money than option A.” The firstgame was A:(5,6) vs. B:(6,5) so Alice could gain 1 by lying and being believed, while imposing a cost of 1 on Bob. The secondgame was A:(5,15) vs. B:(6,5) so Alice could gain 10 by lying and being believed, while still imposing a cost of 1 on Bob. Thethird game was A:(5,15) versus B:(15,5), so Alice could gain 10 by lying and being believed, while imposing a cost of 10 onBob.

Before starting play, the experimenter asked each Alice whether she expected her advice to be followed, inducing honestresponses by promising to reward her if her guesses were correct. He found that 82% of Alices expected their advice to befollowed (the actual result was that 78% of Bobs followed their Alice’s advice). It follows that if Alices were self-regarding,they would always lie and recommend B to their Bob.

The experimenters found that, in game two, where lying was very costly to Bob and the gain to lying for Alice was small,only 17% of subjects lied. In game one, where the cost of lying to Bob was only one but the gain to Alice was the same as ingame two, 36% lied. In other words, subjects were loathe to lie, but considerably more so when it was costly to their partner.In game three, where the gain from lying was large for Alice, and equal to the loss to Bob, fully 52% lied. This shows thatmany subjects are willing to sacrifice material gain to avoid lying in a one-shot, anonymous interaction, their willingness tolie increasing with an increased cost of truth-telling to themselves, and decreasing with an increase in their partner’s costof begin deceived. Similar results were found by Boles et al. (2000) and Charness and Dufwenberg (2006). Gunnthorsdottiret al. (2002) and Burks et al. (2003) have shown that a social–psychological measure of “Machiavellianism” predicts whichsubjects are likely to be trustworthy and trusting.

7. The public sphere

Humans evolved in hunter-gather societies consisting of a dozen families or so (Kelly, 1995), in which political behaviorwas an part of daily life, involving the sorts of self-regarding, other-regarding, and universalistic motivations described above(Gintis et al., 2015). In particular, political activity was strongly consequentialist: a single individual could expect to make adifference to the outcome of a deliberation, a conflict, or a collaboration, so that our political morality developed intimatelyentwined with material interests and everyday consequentialist moral sentiments (Boehm, 1999).

In the transition from small-scale hunter-gatherer societies to modern mass societies with millions of members, thepublic sphere passed from being closely embedded in daily life to being a largely detached institutional arena, governed bycomplex institutions controlled by a small set of individuals, and over which most members have at best formal influencethrough the ballot box, and at worst no formal influence whatever. Political activity in modern societies has thus becomepredominately non-consequentialist.

Canonical participants in the public sphere appear to follow a non-consequentialist logic that may be summarized asdistributed effectivity: in public life, choose a rule that like-minded people might plausibly choose, and if followed by all suchlike-minded people, will lead to the most desirable outcome (Harsanyi, 1977; Hooker, 2011; Roemer, 2010). Distributedeffectivity explains why people are perfectly reasonable in assenting to such assertion as “I am helping my candidate winby voting” and “I am helping promote democracy by demonstrating against the dictator.” Because distributed effectivity isso ingrained in our public persona, canonical participants untrained in traditional rational decision theory simply cannotunderstand the argument that it is irrational to vote or to participate in collective actions, even when they can easily bepersuaded that their actions are non-consequential.

Distributed effectivity can also explain many stylized facts of voter behavior. First, when the cost of voting increases,fewer people vote. The rule here is something like “My unusual personal situation means voting would be very costly to metoday. I would not expect anyone in my position to vote, so I am comfortable with not voting.” Second, it explains why voterturnout is higher when the issues to be decided have greater social impact. Third, it explains why turnout is higher when

H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109 105

tvtc

ba

8

prrtp

roaHR

eoKb2

rpsa

9

bfWoatc

eceit

Fig. 7. A typology of human motivations.

he election is expected to be close. Finally, it explains why, in a two-party election, turnout is likely to be higher amongoters for the side that is not expected to win. Indeed, it is reasonable to speculate that distributed effectivity leads voterso act in very large elections in much the same way they would in very small elections, although in very small electionsonsequentialist issues (e.g., self-interested) may trump the non-consequentialist rule.

We conclude that the individual immersed in consequentialist everyday life expresses his private persona, while hisehavior in the public sphere reveals his public persona. Individuals acting in the public sphere, are, then a different sort ofnimal, one which Aristotle called zoon politikon in his Nicomachean Ethics.

. Private and public persona

The concept of a non-consequentialist public persona suggests a two by three categorization of human motivations, asresented in Fig. 7. In this figure, the three columns represent three modes of social interaction. The self-regarding modeepresents the individual whose social behavior is purely instrumental to meeting his personal needs, while the other-egarding represents the individual who is embedded in a network of significant social interactions with valued others, andhe universal represents the individual who values moral behavior for its own sake. The two rows represent the agent’srivate persona of social relations in civil society, and the agent’s public persona of political relationships in the public sphere.

Homo Economicus is the venerable rational selfish maximizer of traditional economic theory, Homo Socialis is the other-egarding agent who cares about fairness, reciprocity, and the well-being of others, and Homo Vertus is the Aristotelian bearerf non-instrumental character virtues. The new types of public persona are Homo Parochialis votes and engages in collectivection reflecting the narrow interests of the demographic, ethnic and/or social status groups with which he identifies. Finally,omo Universalis acts politically to achieve what he considers the best state for the larger society, perhaps reflecting Johnawls’ (1971) veil of ignorance, John Harsanyi’s (1977) criterion of universality, or John Roemer’s (2010) Kantian equilibrium.

I have left the public persona/self-interested box empty because there are no self-interested canonical participants. Muchmpirical evidence supports the notion that few voters base their choices on the personal economic experience, but rathern the general experience of their relative demographics or that of the nation as a whole (Sears et al., 1980; Kinder andiewiet, 1979; Markus, 1988; Fowler and Kam, 2007), Moreover, many studies show that voters are strongly influencedy the social networks to which they belong (Knack, 1992; Jankowski, 2007; Schram and Sonnemans, 1996; Abrams et al.,011).

Non-canonical agents acting in the political sphere, for instance politicians, are properly located in the private personaow. The individual whose private persona is other-regarding is generally considered altruistic, whereas the individual whoseublic persona is other-regarding (Homo Parochialis) is often considered selfish, acting in a partisan manner on behalf of thepecific interests of the social networks to which he belongs. In terms of our typology, however, Homo Parochialis is in factltruistic, sacrificing on behalf of the interests of members of these social networks.

. The evolutionary emergence of private morality

By cooperation we mean engaging with others in a mutually beneficial activity. Cooperative behavior may confer netenefits on the individual cooperator, and thus can be motivated entirely by self-interest. In this case, cooperation is aorm of mutualism. Cooperation may also be a net cost to the individual but the benefits may accrue to a close relative.

e call this kin altruism. Cooperation can additionally take the form of one individual’s costly contribution to the welfaref another individual being reliably reciprocated at a future date. This is often called reciprocal altruism (Trivers, 1971),lthough it is really just tit-for-tat mutualism. However, important forms of cooperation impose net costs upon individuals,he beneficiaries many not be close kin, and the benefit to others may not be expected to be repaid in the future. Thisooperative behavior is true altruism.

The evolution of mutualistic cooperation and kin altruism is easily explained. Cooperation among close family membersvolves by natural selection because the benefits of cooperative actions are conferred on the close genetic relatives of the

ooperator, thereby helping to proliferate genes associated with the cooperative behavior. Kin altruism and mutualismxplain many forms of human cooperation, particularly those occurring in families or in frequently repeated two-personnteractions. But these scenarios fail to explain two facts about human cooperation: that it takes place in groups far larger thanhe immediate family, and that both in real life and in laboratory experiments, it occurs in interactions that are unlikely to

106 H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109

be repeated, and where it is impossible to obtain reputational gains from cooperating. These forms of behavior are regulatedby moral sentiments.

The most parsimonious proximal explanation of altruistic cooperation, one that is supported by extensive experimentaland everyday-life evidence, is that people gain pleasure from cooperating and feel morally obligated to cooperate with like-minded people. People also enjoy punishing those who exploit the cooperation of others. Free-riders frequently feel guilty,and if they are sanctioned by others, they may feel ashamed. We term these feelings social preferences. Social preferencesinclude a concern, positive or negative, for the well being of others, as well as a desire to uphold ethical norms (Bowles andGintis, 2011).

9.1. The roots of social preferences

Why are the social preferences that sustain altruistic cooperation in daily life so common? Early human environments arepart of the answer. Our Late Pleistocene ancestors inhabited the large-mammal-rich African savannah and other environ-ments in which cooperation in acquiring and sharing food yielded substantial benefits at relatively low cost (Boyd and Silk,2002). Human longevity, including an extended period of dependency of the young also made the cooperation of non-kin inchild rearing and provisioning beneficial (Hrdy, 1999). As a result, members of groups that sustained cooperative strategiesfor provisioning, child-rearing, sanctioning non-cooperators, defending against hostile neighbors, and truthfully sharinginformation had significant advantages over members of non-cooperative groups.

There are several reasons why these altruistic social preferences supporting cooperation outcompeted amoral self-interest. First, human groups devised ways to protect their altruistic members from exploitation by the selfish. Prominentamong these is the collective punishment of miscreants (Boyd et al., 2010), including the public-spirited shunning, ostracism,and even execution of free-riders and others who violate cooperative norms.

Second, humans adopted elaborate systems of socialization that led individuals to internalize the norms that inducecooperation, so that contributing to common projects and punishing defectors became objectives in their own right ratherthan constraints on behavior. Together, the internalization of norms and the protection of the altruists from exploitationserved to offset, at least partially, the competitive handicaps born by those who were motivated to bear personal costs tobenefit others (Gintis, 2003).

Third, between-group competition for resources and survival was and remains a decisive force in human evolutionarydynamics. Groups with many cooperative members tended to survive these challenges and to encroach upon the territory ofthe less cooperative groups, thereby both gaining reproductive advantages and proliferating cooperative behaviors throughcultural transmission. The extraordinarily high stakes of intergroup competition and the contribution of altruistic cooperatorsto success in these contests meant that sacrifice on behalf of others, extending beyond the immediate family and even tovirtual strangers, could proliferate (Choi and Bowles, 2007; Turchin and Korotayev, 2006; Bowles, 2009).

Between-group competition accounts for the fact that humans are extraordinarily group-minded, favoring cooperationwith insiders and often expressing hostility toward outsiders. Boundary-maintenance supported within-group coopera-tion and exchange by limiting group size and within-group linguistic, normative and other forms of heterogeneity. Insiderfavoritism also sustained the between-group conflicts and differences in behavior that made group competition a powerfulevolutionary force (Choi and Bowles, 2007)

In short, humans have social preferences because in the course of our evolution as a species, cooperation was highlybeneficial to the members of groups provided they were able to construct social institutions that compensated for thedisadvantages of those with prosocial preferences regarding fellow group members, while heightening the group-leveladvantages associated with the high levels of cooperation that these prosocial preferences generated. These institutionsproliferated because the groups that adopted them secured high levels of within-group cooperation, which in turn favoredthe groups’ survival as a biological and cultural entity in the face of environmental, military and other challenges.

10. Evolutionary emergence of the public sphere

Like Aristotle’s zoon politikon, we are political nature, and are unique in this respect. How might a species incorporatingan important public sphere have arisen?

The social life of most species, including mating practices, symbolic communication, and power relations, is expressedin genetically-grounded stereotypical form (Alcock, 1993; Krebs and Davies, 1997). Homo sapiens is unique in adapting itssocial life in highly flexible and deeply ways to environmental and social challenges and opportunities (Richerson and Boyd,2004). This flexibility is based on two aspects of our mental powers. The first is our ability to devise new rules of social life,and to base our social interactions on these new rules. This capacity, absent in other species, makes us Homo Ludens: manthe game player. This capacity is possessed even by very young children who invent, understand, and play games for fun. Inadult life, this same capacity is exercised when people come together to erect, maintain, and transform the social rules thatgovern their daily transactions. Broadly speaking, then, we see the public sphere as the arena in which society-wide rules

of the game are created, evaluated, and transformed, and politics as the cooperative, conflictual, and competitive behaviorsthrough which rules are established and individuals are assigned to particular public positions.

The emergence of bipedalism, cooperative breeding, and lethal weapons (stones and wooden spears) many hundreds ofthousands of years ago in the hominin line, together with favorable climate change, made the collaborative hunting and

sact

aiwpaohle

staaact

1

hmtpac

•

•

•

R

AAABB

BBBBBBBB

H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109 107

cavenging of large game fitness enhancing (Whiten and Erdal, 2012; Tomasello, 2014; Gintis et al., 2015). Lethal weaponsre the most unique of these innovations, for other predators, such as lions, tigers and other big cats, wolves, foxes and otheranines, use only their natural weapons—sharp claws and teeth, powerful jaws and great speed—in hunting, while none ofhese endowments was available to early hominins.

Lethal hunting weapons, moreover, transformed human sociopolitical life because they could be applied to humans justs easily as to other animals. The combination of the need for collaboration in hunting and the availability of lethal weaponsn early hominin society undermined the social dominance hierarchy characteristic of primate and earlier hominin groups,

hich was based on pure physical prowess: the alpha male was the strongest member of the group and had dictatorialowers. The successful sociopolitical structure that ultimately replaced their ancestral social dominance hierarchy wasn egalitarian political system in which lethal weapons made possible group control of leaders, group success dependedn the ability of leaders to persuade and motivate, and of followers to contribute to a consensual decision process. Theeightened social value of non-authoritarian leadership entailed enhanced biological fitness for such leadership traits as

inguistic facility, ability to form and influence coalitions, and indeed for hypercognition in general (Boehm, 1999; Gintist al., 2015).

This egalitarian political system persisted until some 10,000 years ago when cultural changes in the Holocene involvingettle trade and agriculture entailed the accumulation of material wealth, through which it became possible once againo sustain a social dominance hierarchy with strong authoritarian leaders who could buy a modicum of protection andllegiance from well-rewarded professional soldiers and clansmen (Richerson and Boyd, 2001). Yet, despite the power ofuthoritarian states, the zoon politikon that social evolution had nourished over tens of thousands of years was not erased by

few thousand years of Holocene history. Indeed, the extremely high level of tribal and clan warfare prevalent until recententuries doubtless favored groups whose members conserved the hunter-gatherer mentality of political commitment andhe desire for personal political efficacy (Pinker, 2011).

1. Conclusion

This paper has provided evidence for a model of human behavior based on the rational actor model, in which individualsave both private and public persona, and their preferences range over self-regarding, other-regarding, and universalistodes in both the private and the public sphere. Morality in this model is defined in behavioral terms: moral choices are

hose made in social and universalist modes. The public sphere in this model is an arena where preferences and actions arerimarily non-consequentialist. The other-regarding preferences of Homo Socialis and the character virtues of Homo Vertusre underpinnings of civil society, while Homo Parochialis and Homo Universalis make possible the varieties of political lifeharacteristic of our species.

This taxonomy of human motives has several important implications for a theory of political behavior.

Despite the ubiquity of the assumption that rational individuals have personal interests which they register throughelectoral processes and collective actions, the notion is incoherent, ineluctably entailing faulty reasoning. Private personaindividuals, whether Homo Economicus, Homo Socialis, or Homo Vertus, will simply not participate in such processes, andthose who do are canonical participants whose political preferences are constituted by the social networks in which theyare embedded as Homo Parochialis, and the higher-level moral principles to which they adhere as Homo Universalis.Private sphere costs and benefits may play a large role in whether an individual participates in electoral processes orcollective actions, but they have little or no effect on his electoral preferences or which collective actions he supports.Thus we should not be at all surprised when abstract moral principles appear to trump economic interests in individualeconomic decisions.The fact that the canonical participants are a mix of Homo Parochialis and Homo Universalis explains why political move-ments are sensitive to issues of justice and fairness and insensitive to issues of social efficiency when the latter conflictwith the former. For instance, voters typically care about corruption, workers’ rights, graft, and unemployment but notrates of economic growth or measures of wealth dispersion.

eferences

brams, S., Iversen, T., Soskice, D., 2011. Informal social networks and rational voting. Br. J. Polit. Sci. 41, 229–257.lcock, J., 1993. Animal Behavior: An Evolutionary Approach. Sinauer, Sunderland, MA.ristotle, 2002. [350BC]. Nicomachean Ethics. Focus Publishing, Newburyport, MA.acharach, M., 1987. A theory of rational decision in games. Erkentness 27, 17–56.acharach, M., 1992. The acquisition of common knowledge. In: Bicchieri, C., Chiara, M.L.D. (Eds.), Knowledge, Belief, and Strategic Interaction. Cambridge

University Press, Cambridge, pp. 285–316.acharach, M., 2006. In: Gold, N., Sugden, R. (Eds.), Beyond Individual Choice: Teams and Games in Game Theory. Princeton University Press, Princeton.acharach, M., Gold, N., Sugden, R., 2006. Beyond Individual Choice: Teams and Frames in Game Theory. Princeton University Press, Princeton.lack, D., 1948. On the rationale of group decision-making. J. Polit. Econ. 56, 23–34.

lais, A., 2000. To Vote or Not to Vote. University of Pittsburgh Press, Pittsburgh.oehm, C., 1999. Hierarchy in the Forest: The Evolution of Egalitarian Behavior. Harvard University Press, Cambridge, MA.oles, T.L., Croson, R.T.A., Murnighan, J.K., 2000. Deception and retribution in repeated ultimatum bargaining. Org. Behav. Hum. Decis. Process. 83, 235–259.owles, S., 2009. Did warfare among ancestral hunter-gatherer groups affect the evolution of human social behaviors. Science 324, 1293–1298.owles, S., Gintis, H., 1986. Democracy and Capitalism: Property, Community, and the Contradictions of Modern Social Thought. Basic Books, New York.

108 H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109

Bowles, S., Gintis, H., 2011. A Cooperative Species: Human Reciprocity and its Evolution. Princeton University Press, Princeton.Boyd, R., Gintis, H., Bowles, S., 2010. Coordinated punishment of defectors sustains cooperation and can proliferate when rare. Science 328, 617–620.Boyd, R., Silk, J.B., 2002. How Humans Evolved, Third Edition. W. W. Norton, New York.Bratman, M.E., 1993. Shared intention. Ethics 104, 97–113.Bucciol, A., Piovesan, M., 2011. Luck or cheating? A field experiment on honesty with children. J. Econ. Psychol. 32, 73–78.Burks, S.V., Carpenter, J.P., Verhoogen, E., 2003. Playing both roles in the trust game. J. Econ. Behav. Org. 51, 195–216.Carlsson, H., van Damme, E., 1993. Global games and equilibrium selection. Econometrica 61, 989–1018.Chamberlain, G., Rothschild, M., 1981. A note on the probability of casting a decisive vote. J. Econ. Theory 25, 152–162.Charness, G., Dufwenberg, M., 2006. Promises and partnership. Econometrica 74, 1579–1601.Choi, J-k, Bowles, S., 2007. The coevolution of parochial altruism and war. Science 318, 636–640.Clarke, H.D., Sanders, D., Stewart, M.C., Whitely, P., 2004. Political Choice in Britain. Oxford University Press, Oxford.Cohn, A., Fehr, E., Maréchal, M.A., 2014. Business culture and dishonesty in the banking industry. Nature 516, 86–89.Colman, A.M., Pulford, B.D., Rose, J., 2008. Collective rationality in interactive decisions: evidence for team reasoning. Acta Psychol. 128, 297–387.Cox, G.W., 1994. Strategic voting equilibria under the single nontransferable vote. Am. Polit. Sci. Rev. 88, 608–621.Dalton, R.J., 2008. The Good Citizen: How a Younger Generation is Reshaping American Politics. Sage, New York.Dennis, J., 1970. Support for the institution of elections by the mass public. Am. Polit. Sci. Rev. 64, 819–835.Downs, A., 1957a. An Economic Theory of Democracy. Harper & Row, Boston.Downs, A., 1957b. An economic theory of political action in a democracy. J. Polit. Econ. 65, 135–150.Durrett, R., Levin, S.A., 2005. Can stable social groups be maintained by homophilous imitation alone? J. Econ. Behav. Org. 57, 267–286.Edlin, A., Gelman, A., Kaplan, N., 2007. Voting as a rational choice: why and how people vote to improve the well-being of others. Ration. Soc. 19, 293–314.Enos, R., Fowler, A., 2010. Why Did You Vote. Daily Dish.Evren, Ö, 2012. Altruism and voting: a large-turnout result that does not rely on civic duty or cooperative behavior. J. Econ. Theory 147, 2124–2157.Fedderson, T., Sandroni, A., 2006. A theory of participation in elections. Am. Econ. Rev. 96, 1271–1282.Fiorina, M.P., 1990. Information and rationality in elections. In: Ferejohn, J., Kuklinski, J. (Eds.), Information and Democratic Processes. University of Illinois

Press, Urbana, IL, pp. 329–342.Fischbacher, U., Follmi-Heusi, F., 2014. Lies in disguise – an experimental study on cheating. J. Eur. Econ. Assoc. 11, 525–547.Fischer, I., Frid, A., Goerg, S.J., Levin, S.A., Rubenstein, D.I., Selten, R., 2013. Fusing enacted and expected mimicry generates a winning strategy that promotes

the evolution of cooperation. Proc. Natl. Acad. Sci. U.S.A. 110, 10229–10233.Fowler, J.H., Kam, C.D., 2007. Beyond the self: social identity, altruism, and political participation. J. Polit. 69, 813–827.Gelman, A., King, G., Boscardin, J., 1998. Estimating the probability of events that have never occurred: when is your vote decisive? J. Am. Stat. Assoc. 93,

1–9.Gilbert, M., 1987. Modeling collective belief. Synthese 73, 185–204.Gilbert, M., 1989. On Social Facts. Routledge, New York.Gintis, H., 2003. The hitchhiker’s guide to altruism: genes, culture, and the internalization of norms. J. Theor. Biol. 220, 407–418.Gintis, H., van Schaik, C.P., Boehm, C., 2015. Zoon Politikon: the evolutionary origins of human political systems. Curr. Anthropol. 56, 327–353.Gneezy, U., 2005. Deception: the role of consequences. Am. Econ. Rev. 95, 384–394.Good, I.J., Mayer, L.S., 1975. Estimating the efficacy of a vote. Behav. Sci. 20, 25–33.Green, D.P., Shapiro, I., 1994. Pathologies of Rational Choice Theory: A Critique of Applications in Political Science. Yale University Press, New Haven, CT.Gunnthorsdottir, A., McCabe, K., Smith, V.L., 2002. Using the machiavellianism instrument to predict trustworthiness in a bargaining game. J. Econ. Psychol.

23, 49–66.Hamlin, A., Jennings, C., 2011. Expressive political behavior: foundations, scope and implications. Br. J. Polit. Sci. 41, 645–670.Harsanyi, J.C., 1977. Morality and the theory of rational behavior. Soc. Res. 44.Hooker, B., 2011. Rule Consequentialism. The Stanford Encyclopedia of Philosophy.Houser, D., Vetter, S., Winter, J., 2012. Fairness and cheating. Eur. Econ. Rev. 56, 1645–1655.Hrdy, S.B., 1999. Mother Nature: A History of Mothers, Infants, and Natural Selection. Pantheon Books, New York.Hurley, S.L., 2002. Consciousness in Action. Harvard University Press, Cambridge.Jankowski, R., 2007. Altruism and the decision to vote explaining and testing high voter turnout. Ration. Soc. 19, 5–34.Kelly, R.L., 1995. The Foraging Spectrum: Diversity in Hunter–Gatherer Lifeways. The Smithsonian Institution, Washington, DC.Kinder, D.R., Kiewiet, D.R., 1979. Economic discontent and political behavior: the role of personal grievances and collective economic judgments in

congressional voting. Am. J. Polit. Sci. 25, 495–527.Knack, S., 1992. Civic norms, social sanctions, and voter turnout. Ration. Soc. 4, 133–156.Krebs, J.R., Davies, N.B., 1997. Behavioral Ecology: An Evolutionary Approach, fourth ed. Blackwell Science, Oxford.Levine, D.K., Palfrey, T.R., 2007. The paradox of voter participation? A laboratory study. Am. Polit. Sci. Rev. 101, 143–158.Lijphart, A., 1997. Unequal participation: democracy’s unresolved dilemma. Am. Polit. Sci. Rev. 91, 1–14.Markus, G., 1988. The impact of personal and national economic conditions on the presidential vote: a pooled cross-sectional analysis. Am. J. Polit. Sci. 32,

137–154.Mazar, N., Amir, O., Ariely, D., 2008. The dishonesty of honest people: a theory of self-concept maintenance. J. Mark. Res. 45, 633–644.McPherson, M., Smith-Lovin, L., Cook, J., 2001. Birds of a feather: homophily in social networks. Annu. Rev. Sociol. 27, 415–444.Niemi, R.G., Whitten, G., Franklin, M.N., 1992. Constituency characteristics, individual characteristics and tactical voting in the 1987 British general election.

Br. J. Polit. Sci. 22, 229–240.Palfrey, T.R., Rosenthal, H., 1985. Voter turnout and strategic uncertainty. Am. Polit. Sci. Rev. 79, 62–78.Pinker, S., 2011. Our Better Angels: Why Violence Has Declined. Penguin Books, New York.Rawls, J., 1971. A Theory of Justice. Harvard University Press, Cambridge, MA.Richerson, P.J., Boyd, R., 2001. Institutional evolution in the Holocene: the rise of complex societies. In: Runciman, G. (Ed.), In: Proceedings of the British

Academy: The Origin of Human Social Institutions, 110, pp. 197–220.Richerson, P.J., Boyd, R., 2004. Not by Genes Alone. University of Chicago Press, Chicago.Riker, W.H., Ordeshook, P.C., 1968. A theory of the calculus of voting. Am. Polit. Sci. Rev. 62, 25–42.Roemer, J., 2010. Kantian equilibrium. Scand. J. Econ. 112, 1–24.Rubinstein, A., 1989. The electronic mail game: strategic behavior under almost common knowledge. Am. Econ. Rev. 79, 385–391.Savage, L.J., 1954. The Foundations of Statistics. John Wiley & Sons, New York.Schram, A., Sonnemans, J., 1996. Why people vote: experimental evidence. J. Econ. Psychol. 17, 417–442.Searle, J., 1995. The Construction of Social Reality. Free Press, New York.Sears, D.O., Lau, R.R., Tyler, T.R., Allen, H.M., 1980. Self-interest vs. symbolic politics in policy attitudes and presidential voting. Am. Polit. Sci. Rev. 74,

670–684.

Shachar, R., Nalebuff, B., 1999. Follow the leader: theory and evidence on political participation. Am. Econ. Rev. 89, 525–547.Shalvi, S., Eldar, O., Bereby-Meyer, Y., 2012. Honesty requires time (and lack of justifications). Psychol. Sci 23, 1264–1270.Silk, J.B., Brosnan, S.F., Vonk, J., Henrich, J., Povinelli, D.J., Richardson, A.S., Lambeth, S.P., Mascaro, J., Shapiro, S.J., 2005. Chimpanzees are indifferent to the

welfare of unrelated group members. Nature 473, 1357–1359.Sugden, R., 2003. The logic of team reasoning. Philos. Explor. 6, 165–181.

TTTTvV

WWW

H. Gintis / Journal of Economic Behavior & Organization 126 (2016) 95–109 109

omasello, M., 2014. A Natural History of Human Thinking. Harvard University Press, Cambridge.rivers, R.L., 1971. The evolution of reciprocal altruism. Q. Rev. Biol. 46, 35–57.uomela, R., 1995. The Importance of Us: A Philosophical Study of Basic Social Emotions. Stanford University Press, Stanford.urchin, P., Korotayev, A., 2006. Population dynamics and internal warfare: a reconsideration. Soc. Evol. Hist. 5, 112–147.on Neumann, J., Morgenstern, O., 1944. Theory of Games and Economic Behavior. Princeton University Press, Princeton.

onk, J., Brosnan, S.F., Silk, J.B., Henrich, J., Richardson, A.S., Lambeth, S.P., Schapiro, S.J., Povinelli, D.J., 2008. Chimpanzees do not take advantage of very

low cost opportunities to deliver food to unrelated group members. Anim. Behav. 75, 1757–1770.arneken, F., Tomasello, M., 2006. Altruistic helping in human infants and young chimpanzees. Science 311, 1301–1303.hiten, A., Erdal, D., 2012. The human socio-cognitive niche and its evolutionary origins. Philos. Trans. R. Soc. B: Biol. Sci. 367, 2119–2129.olfram Research, Inc., 2014. Mathematica 10.0. Wolfram Research, Champaign, IL.