The Priority of Social Morality

David Estlund (2011, 2014) explicitly accepts that the true theory of justice may well have no practical value.

Shaun Nichols and I have argued for this, with special reference to Bodo ethics, in Gaus and Nichols (2017).

For a different sort of claim that purely instrumental reasoning is in some way more basic than the notion of a rule-based social morality, see Moehler (2014).

Moving a bit beyond classic coordination problems, think of a “coordination” interaction such as a Stag Hunt. Even if we have achieved “hunt stag” equilibrium—which might be maintained simply by self-interest—a rule that makes it a moral requirement to hunt stag may stabilize cooperation in the face of trembling hands and other uncertainties.

Let us say that for a rule R to be a genuine social-moral rule for Betty, Betty must (i) recognize R as rule that applies to C circumstances; (ii) typically have motivating reason to conform to R rather than act simply on her own goals in C circumstances; (iii) her personal normative convictions endorse R; (iv) she believes that a sufficiently large subset of her group G conforms to R; (v) she believes that a sufficiently large subset of G expects her to conform to R. See further Gaus (2011: 163–181). In this chapter, I shall not distinguish the rules of social morality from social norms. They are not, however, equivalent; the rules of social morality are parts of practices of accountability and sustain the moral emotions of guilt, resentment and indignation; not all social norms do so.

See also Hardin (2013: 411ff; 2003: 98).

The rules are external in the sense that while agents understand them to be social guidelines that serve a purpose, their motivation to comply is simply that sanctions will be applied by others. I consider the contrast to “internalized” moral rules in more detail in section “The Internalization of Moral Rules” below.

See, however, the wider characterization of a convention in Bowles and Gintis (2011: 111).

Much depends on what is meant by a “closed” society. Marriage networks, for example, can make the group much more porous than first inspection would indicate.

This is one reason why “direct reciprocity” (e.g., “tit-for-tat” responses) is often a poor basis for social cooperation, engendering cycles of conflicts. See further, Boehm (2012: 60ff).

Boehm reports that in his database about half the hunter-gather societies are coded as having practiced capital punishment; there is strong reason to think that the number may be much higher, as central governments treat band and tribal executions as murder (2012: 84).

While females seldom participate in the executions, they do typically participate in the deliberation leading to execution.

As Bowles and Gintis point out, in large-scale societies too, anti-social punishment (counter-punishment) is real: experiments show great differences in societies to the extent to which punishment is accepted or evokes counter-sanctioning (Ibid.). As we shall see below that in experiments in “Power-to-Take” games, Takers who were sanctioned by their partners for taking the partner’s endowments but who did not see these takings as unfair, did not decrease their takings in a second round; in contrast, those who were sanctioned and did think their initial taking unfair (but hoped to get away with it) responded to sanctioning by decreasing their takings.

Boehm (2012: 44–45) muses that Cephu may have been something of an amoral psychopath, and so unable to internalize moral rules.

For a striking case, see Boehm (1999: 51–59).

In Boehm’s database, of the societies that engaged in capital punishment, a repeat murder was the second most reported capital offense.

It is generally thought that young children see harm to others as violating a basic moral requirement. See Turiel et al. (1987: 174). Guilt is especially associated with violation of rules against harm and the rights of others (Prinz 2007: 77).

Bowles and Gintis devote much care to analyzing how internalization of social morality can be modeled (Chap. 10). As they stress, the internalization of norms is an aspect of cultural transmission that affects preferences or values. On the general phenomenon of cultural transmission, see Richerson and Boyd (2005).

Daniel Friedman points to 150, with much larger numbers when groups fused (Friedman 2008: 16). See also David C. Rose, who mentions 200 as the typical size of the groups in which humans evolved (Rose 2011: chp. 3). Closer examination shows that group size may be understood differently: average band size may differ from typical group size (Bowles and Gintis 2011: 95).

For doubts, see Richerson and Boyd (2013).

On the other hand, it could well have been such instability that increased the benefits of cooperation (Bowles and Gintis 2011: 93ff).

To what extent genes have evolved during this period is a highly controversial question. 10,000 years is far less than the 1000 generations, which is the rule-of-thumb for the evolution of major traits. But this is a highly controversial matter that is being debated (Cochran and Harpending 2009). For a rather more widely accepted view of the relation of genetic and cultural evolution, see Henrich (2016).

Though I have argued that it is surprisingly so (Gaus 2015).

Thus the common depiction of Hobbes as somehow the father of rational choice theory (even though Hobbes himself had a much more sophisticated view of human motivation). See, for example, Hartmut Kliemt, Philosophy and Economics I: Methods and Models (Munich: Oldenbourg, 2009), pp. 46ff.

On the importance of reasons, see Bicchieri and Mercier (2014).

Plausible models of internalization often yield polymorphic results, with a population divided between internalizers and more opportunistic types. See, for example, Andrighetto et al. (2010).

See, for example, Gaus (2016: chap. IV).

Bicchieri and her co-workers have shown how subjects exploit normative ambiguity in order to provide wiggle room to advance their interests. See Bicchieri and Chavez (2013a) and Bicchieri and Mercier (2013b).

As in the folk theorem, Binmore (2005: chap. 5).

Indirect reciprocity, or reputation, might seem to underwrite cooperation in larger groups by encouraging “boycotts” of violators, but indirect reciprocity turns out to be very sensitive to the quality of information about people. See Henrich and Henrich (2007: chap. 4), Bowles and Gintis (2011: 68–70) and Vanderschraaf (2007: 167–195).

In Bowles and Gintis’s agent-based modeling allowing even for small rates of errors in reciprocation, groups over 10 seldom, and over 15 essentially never, evolved cooperation (2011: 64–68). Even in small group forager bands, direct reciprocity does not explain most cooperation (Boehm 2012: 179–180).

The experimental work on strong reciprocity and altruistic punishment is now extensive. The pioneering work was done by Ernst Fehr and his colleagues. See, for example, Fehr and Fischbacher (2005) and Fehr and Gächter (2000a, b).

The now famous Ultimatum Game is a single-play game between two anonymous subjects, Proposer and Responder, who have X amount of some endowment (say, money) to distribute between them. In the classic version of the game, Proposer makes the first move, and gives an offer of the form, “I will take n amount of X, leaving you with X−n,” where n is not greater than X. If Responder accepts, each gets what Proposer offers; if Responder rejects, each receives nothing. For a recent overview see Eric van Damme et al. (2014).

Here some small-scale societies are outliers. See Heinrich and Smith (2004).

See Hoffman et al. (1996).

See, for example, Slonim and Roth (1998). In one study with an endowment worth three month’s wages still displayed Responder rejection of lower offers (Bicchieri 2006: 114n).

One possible explanation—one that Russell sees as partaking of the magical—is that people may be moved by a sense of justice (Rawls 1999: chap. VIII). I do not think it is magical, and some evidence indicates that impartial concern for justice may be a motivational factor (Carlsmith et al. 2002: 284–299). Third-party punishment might be seen as based on an impartial sense of justice, and there is certainly considerable evidence for such punishment. See also Fehr Fischbacher (2004). However, I do not think the evidence indicates this to be a critical factor, once we have factored out the reactive moral emotions, such as anger. In an interesting experiment Simon Knight sought to determine whether Responders were upholding such a sense of justice—whether “the concern is with unfair offers in general”—or were responding not to the Proposer’s general status as a sharer or miser, but specifically what the Proposer did to her—whether the Proposer gave her a high or low offer. Knight finds that Responders’ behavior supports the latter hypothesis—that Responder Betty’s action is more strongly influenced by what has been done to her, so she will be apt to accept a high offer from a generally unfair Proposer or reject one from a generally fair one (Knight 2012).

See, for example, Hopfensitz and Reuben (2009).

Thus my focus at present is second-party, not third-party, punishment.

Another cost to which punishers appear insensitive is the number of violators; even if defection is “the norm”—there are many defectors—punishment does not generally decrease (Bone, Silva, and Raihani).

See, for example, Knight (2012: 7–8). As we shall see in the next section, expectations count.

To drastically oversimplify, The Reactive Emotion View can be modeled as claiming the decisions are based on a two-part value function. Letting X−n be an offer in an Ultimatum Game, where X is the total endowment and n is the amount Proposer reserves for himself, then Responder’s total value of the X−n offer will be V _MG−V _RE, where V _MG is the value of the absolute monetary gain, and V _RE is the value based on the reactive emotions, a value arising from the negative emotions, which focus on the ratio of X to n, as mediated by expectations of what is to be expected. A Responder will accept if total value is positive, reject if it is negative.

In the so-called “Dictator Game” Proposer simply decides on the two shares, and that’s the end of the game (not much of a game).

The variability of destruction is meant to uncover the relation of degree of emotional response to degree of punishment; I discuss presently a version of Power-to-Take that gives only limited punishment options which, not too surprisingly, considerably blunts the importance of emotions.

This is typical of takings in Power-to-Take Games; see Reuben and van Winden (2010: 912.)

“In both conditions, the sequence of actions was as follows. Before subjects played the one-shot PTT-game, they were randomly divided into two groups. One group was referred to as participants A (the take authorities) and the other as participants B (the responders). Subsequently, random pairs of a responder and a take authority were formed by letting take authorities draw a coded envelope from a box. The envelope contained a form on which the endowment of both participant A and participant B was stated. The take authorities then had to fill in a take rate and put the form back in the envelope again. After the envelopes were collected, we asked the take authorities to report their emotions as well as their expectation of what the responder would do. The envelopes were brought to the matched responders who filled in the part of their endowments to be destroyed. The envelopes containing the forms were then returned to the take authorities for their information. Meanwhile, responders were asked to indicate which take rate they had expected and how intensely they had experienced several emotions after having learned about the take rate. After completing the questionnaires and collecting all envelopes, subjects were privately paid outside the laboratory by the cashier who was not present during the experiment. Experimenters were not able to see what decisions subjects made in the game and how much they earned” (Bosman et al. 2005: 415).

That contempt is a significant emotion in almost all experiments suggests that pride is an important explanatory character trait.

On the relation of guilt to interpersonal harm, see Berndsen et al. (2004).

It is important that Thulin and Bicchieri’s target emotion appears distinctly moral; in one study emotions were measured, for example, on a 7-point scale from “Strongly Disagree” to “Strongly Agree” with statements such as “I feel angry when I learn about people suffering from unfairness” and “I think it’s shameful when injustice is allowed to occur.” These emotions are moral emotions, presupposing a normative content, thus in my terms they appear to function as moral rules.

The tale of Cephu seems to manifest both the “steaming up” and “cooling down” dynamics.

In some contexts, Russell intimated that the problem with all rule systems is that, because they depend on identification of a set of act-types, they cannot be usefully scaled up to regulate dynamic societies with constantly changing act-types. Nichols and I analyze this idea in Gaus and Nichols (2017).

I have expanded upon this point in Gaus (2018).

In addition to Mackie (2017), see Bicchieri (2016) and Bicchieri and Mercier (2014).

For a short description of this experiment, see Mockus (2012). For an in-depth treatment, see Mockus (2017).