OPR VII.20: The Evolution of Morality

Overview of §20

In §20, Gaus explores the idea—foreshadowed in §19—that not only can a selection from the socially optimal eligible set of rules be justified as a result of convergence in a ‘Kantian Coordination Game’ (see §19.2), so that “[c]oordination on a morality can occur even though no procedure of coordination has itself been publicly justified” (410), but “the process of arriving at a publicly justified morality may well be a social evolutionary one, in which people gradually come to coordinate on a common set of moral rules” (410).  

He begins (in §20.1) by arguing that a ‘more realistic’ Kantian Coordination Game, involving “agents whose actions do not track [those of] their idealized Member of the Public counterparts[s]” (410), still yields convergence on a member of the eligible set.  This result relies upon the presence in sufficient numbers among the members of the group playing the game of ‘Moralists’, whose actions do track those of their idealized Member of the Public counterparts.  (The others are ‘Simple Coordinators’, who will coordinate on any moral rule, seeing the advantages of cooperation with others but having only a vague idea of which rules would satisfy their own values and only a vague idea of the contours of the socially eligible set.)  The Moralists see granting moral authority to a rule outside the socially eligible set as no better than having no moral rule at all, and consequently adopt a strategy of acting in accordance with their favoured member of the socially eligible set even in response to Simple Coordinators’ strategies of acting in accordance with a rule which is not a member of that set.  In a basic coordination game between a Moralist and a Simple Coordinator, the Moralist cannot improve her utility by unilaterally changing her move to x (a member of the socially eligible set) from z, matching the Simple Coordinator’s adoption of that rule, which is not a member of the socially eligible set.  But she’ll do so anyway, since unilateral (non-coordinating) action on x is no worse from her point of view than coordinating on z.  As a result, the game will have only one Nash equilibrium, at x/x: so, “a large enough group of Moralists can move the entire group to an equilibrium within the eligible set” (412).

This appeal to the concept of a Nash equilibrium does not render the account practically irrelevant, moreover.  Whereas in single-move coordination games, Nash equilibrium solutions may require that each player have more information about others’ moves far more than any feasibly could, in an evolutionary account the game has room for players to learn each other’s moves slowly and adjust their moves accordingly over time.  Gaus’s conclusion is that this model may show us how our social morality has evolved “and how this process may itself be justificatory” (413, my italics).

The rest of §20 comments on features of this account, including features which distinguish it from other views which are also seen as ‘social evolutionary accounts’.  The features are as follows:

  • True morality is partly determined by positive morality.  In the Kantian Coordination Game just described, there is no member of the socially eligible set which is at the outset the correct choice.  The problem of which to choose is solved by the “actual interactions of what we might call a group of predominantly moderately goodwilled people” (414).  If the Kantian Coordination Game models the evolution of our social morality and that evolutionary process is indeed justificatory, as Gaus seems to think it is (more on this below), then “normative ethics…cannot tell us what is true morality without appeal to the actual facts about the moral equilibrium we have reached” (414).
  • Public justification is path-dependent.  It’s a feature of social evolutionary accounts of morality that the prevailing set of rules at any given time is path-dependent: the rules evolved from somewhere.  This, Gaus thinks, is a plausible way to think about publicly justified social morality.  Revisions to our social morality are informed by the current content of that morality: “[m]oral innovations and change work with what the current morality is, and so even attempts at reform are deeply influenced by the moral status quo” (416).
  • There can be good reasons to sustain systems of rules for whose adoption there was originally no good reason.  Since evolutionary success does not, on Gaus’s view, provide sufficient reason to sustain a rule that is evolutionarily successful, one might think that the public unjustifiability of the original adoption of some rule overrides any evolutionary success it might have had since then as the locus of an equilibrium.  However, “what is ultimately important are not efficient causes but sustaining reasons” (416), and a rule whose original adoption was unjustified may come to be publicly justified.  Gaus gives the example of “rules about civility and gentlemanly behaviour, which were originally introduced in part to solidify class distinctions…[but] because the basis for more generalized and justifiable norms of civility and nonoffensive behaviour” (416).  This feature of Gaus’s account serves to distinguish it from historical entitlement theories, according to which an unjustified step at any point is sufficient to show that later, resulting situations are themselves unjustified.
  • Gaus’s view is not selectionist.  In §20.2, Gaus spends quite a bit of time analysing the contrasts between his view and Hayek’s.  Hayek’s view is ‘selectionist’: it holds that moral rules are rules that are privileged by some selection mechanism(s) which favour(s) groups who coordinate on them, rather as loud male birdsong might be privileged by the mechanism of sexual selection by female birds, for example.  Moreover, Hayek thinks that the reason that we should give moral rules their special place in our deliberations is precisely that they have been privileged in this way by the selection mechanisms (Gaus calls this the evolutionist’s Sufficiency Claim).  Gaus’s own view is not selectionist: it does not suggest that justified moralities are privileged by any selection mechanism that favours groups who coordinate on them (though, he says, “it would be nice” [420] if they were).  And he rejects the Sufficiency Claim on grounds that we can coherently question evolutionarily selected moral rules, asking, for example, whether they might not be mere superstitions.

§20.2 concludes with further discussion of the Sufficiency Claim and Gaus’s rejection of it.  In reply to the objection that questioning of moral rules is only possible from within a framework of evolutionarily selected rules and reason, and so has to take for granted the justifiability of these, Gaus argues that the mere fact that the moral rules and the reason that we use to question them (and, more generally, to question the outcomes of the relevant evolutionary selection mechanism) are coeval products of that mechanism does not show that such questioning is really beyond the capacity of reason.  For a rule to be justified just because it has been selected by the mechanism requires the premiss that what is selected by the mechanism is morally worthy, and that doesn’t follow from the fact that the mechanism has, as a matter of fact, privileged our present moral rules, even if, as a matter of fact, those rules are morally worthy.  There would have to be some reason to think that our confidence in the moral worth of rules was justified by the very fact that they had been selected by the mechanism.

One possibility is that the fact that moral rules have been selected by certain evolutionary selection mechanisms shows that they solve coordination problems.  But, as Gaus points out, it looks as if what’s doing the justificatory work here is the value of solving coordination problems, not the evolutionary selection mechanisms.  An improved suggestion adds that we ourselves cannot tell by inspecting a given rule whether it can solve coordination problems; the fact that it has been selected by the relevant mechanisms gives us evidence that it does.  But this suggestion falls far short of the Sufficiency Claim.  For one thing, the fact that the rule was once selected for does not entail that it is now socially useful: “even if we make the strong assumption that the selection mechanism inherently selects rules we consider worthy, if social change is rapid we cannot conclude that our current rules are worthy unless social evolution also occurs very rapidly” (424).  For another, and more importantly, there oppressive ways of achieving social cooperation: not all ways of solving coordination problems respect our status as free and equal persons.  So, the Sufficiency Claim should be rejected.


  1. The basic idea of the first part of §20 is to show us that not only would a group of Moralists would converge upon a particular member of the socially optimal eligible set of rules, conferring upon that member the status of uniquely justified social rule (the lesson of §19), but a group of Moralists (in sufficient numbers) and Simple Coordinators would also converge upon a particular member of the socially optimal eligible set of rules, conferring upon it the status of uniquely justified social rule.  To the extent that this is just a more complex form of the Kantian Coordination Game of §19, it inherits some of the problems with that game.  In particular, I think that Colin was right last week to be worried about the ‘slippage’ from the idea that one would have sufficient reason to adopt a strategy that is a sub-optimal member of the eligible set from the point of view of one’s evaluative standards, when switching away in the context of others’ adherence to that strategy would lead to a failure to coordinate at all, to the idea that general convergence on one member of the eligible set shows that I have sufficient reason to internalise it.  After all, if the rule is, in my view, a sub-optimal one, why wouldn’t I have reason to adopt it in my interactions with others—avoiding the non-coordination baseline and the associated impossibility of moral authority—while continuing to try to persuade others of the superiority of my preferred alternatives, for example?
  2. Moreover, the introduction of the Simple Coordinators raises new questions.  The considerations that Gaus discusses at 411-2 tend to suggest that Simple Coordinators in a group of Moralists who all favour the same member of the socially optimal eligible set will end up adopting that same rule.  But what of the case where the Moralists in the game don’t all favour the same member of the socially optimal eligible set?  Do the considerations advanced in §19 give us reason to think that they will slowly converge despite the presence of the Simple Coordinators?  What will the Simple Coordinators do in the meantime, given that the Moralists that they meet will not all be playing the same rule?  And why should we accept that a game with sufficient numbers of moralists in it to produce the desired equilibrium is an adequate model of our world, which may not have or have had moralists in sufficient numbers?  I don’t mean to suggest that these questions are the foundation of an objection to Gaus’s arguments (I’m not game theorist enough to be able to say that), but what Gaus says in §20 on this matter seems to me to be too brief to be satisfactory.
  3. I don’t find much to object to in Gaus’s discussion of features of social evolution accounts of morality.  But I do wonder whether more might not be said in reply to his rejection of the Sufficiency Claim.  In particular, might not a proponent of that claim propose that rules that are selected by her favoured evolutionary selection mechanism M tend to be publicly justified rules, though we cannot know this by inspecting each rule?  Gaus’s conclusion that “there are insuperable difficulties in defending any version of the Sufficiency Claim” (424) is reached on the basis of the assumption that the evolutionary selection mechanisms favoured by proponents of the Sufficiency Claim are always mechanisms that privilege no more than coordinative function.  But I don’t see why this has to be the case.  Given the evident difficulties that attend trying to work out just whether a rule satisfies the BPPJ, one might think that we cannot know by inspecting a rule whether it’s publicly justified.  And we might also have reasons to think that Moralists will be favoured by evolutionary selection mechanisms.  Gaus’s own arguments (concerning the convergent behaviour of Moralists) suggest that in that case, something much closer to the Sufficiency Claim than he allows might be true, even if the good reason to follow an M-selected rule isn’t always overriding.
  4. A related question.  Let’s grant that Gaus is right to reject the Sufficiency Claim.  In that case, is the evolution of social morality really itself a justificatory process, as Gaus suggests it is (413)?  I’m not sure.  The models that Gaus describes show how evolutionary processes can lead to the selection of a rule from the socially optimal eligible set.  But the justification for that rule doesn’t appeal to the evolutionary processes at all.  It appeals to the fact that (according to Gaus) that rule satisfies the Basic Principle of Public Justification.  Part of the explanation of the fact that the rule satisfies the BPPJ is that evolutionary processes have generated convergence on it, which is why it meets condition (c) of the BPPJ (“moral agents generally conform to L” [263]).  But the explanation doesn’t do any normative work of its own.
This entry was posted in Books, Posts, Reading Group. Bookmark the permalink.

Leave a Reply