The revenge of Moral Twin Earth

ABSTRACTIn this paper I revisit an important response to the Moral Twin Earth (MTE) challenge: The Common Functional Role strategy (CFR). I argue that CFR is open to a revenge problem. MTE-cases allegedly show that two linguistic communities can be in genuine disagreement even when they are regulated by distinct families of properties. CFR provides a way to reconcile the intuition that the two communities are in genuine disagreement with the claim that the use of moral terms by both communities is causally regulated by different families of properties. This is done by identifying a functional role that those families of properties both fulfill. Still, even if CFR is successful, its proponents need to face a serious revenge problem. Roughly, it could be that the families of properties that regulate each community are equally perfect realizers of the relevant higher-order functional state. I suggest that the proponent of CFR faces a dilemma: either CFR has controversial implications about first-order moral theory, or CFR needs to be coupled with substantive and parochial empirical/metaphysical assumptions to avoid those implications.KEYWORDS: Moral Twin Earthcommon functional rolemoral functionalismmoral disagreementmoral realism AcknowledgementsFor written comments on previous drafts I thank Pekka Väyrynen and several anonymous reviewers from Inquiry and other journals (especially Analytic Philosophy, the Journal of Value Inquiry, and the Australasian Journal of Philosophy). I also want to thank audiences from the PG Seminar at the University of Leeds, and the Philosophy and Political Theory Seminar at Panteion University. A distant precursor of this paper was written during my time as an undergraduate at the University of Athens at Stelios Virvidakis's Metaethics seminar (whom I thank for comments and encouragement).Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Causal metasemantics are typically used by naturalists (for a recent account of moral naturalism see Stamatiadis-Bréhier Citation2022). This explains why, historically, MTE cases have naturalist versions of moral realism as their target. But such metasemantics can also be used by anti-naturalist moral realists. For example, Adams’s (Citation1999) supernaturalist moral realism has a structurally identical metasemantics to Boyd’s.2 Someone might be hesitant in combining deontology with Boyd’s moral realism (although see sec. 4). Still, note that we can construct an equally powerful MTE case involving different moral theories (e.g. Freiman [Citation2014] appeals to virtue ethics).3 For the rest of this paper, I will use the phrase “Earthlings are regulated by some properties” as a shorthand for saying that the use of their terms is regulated by these properties.4 For dialectical reasons, I will grant this controversial assumption (cf. Plunkett and Sundell Citation2013). At any rate, the same argument could be presented (by adding further epicycles) by appealing to intuitions concerning whether the two communities track the same referent (see Väyrynen Citation2018).5 The CFR response has been proposed, in different ways, by Kraemer (Citation1991, 469), Lawrence et al. (Citation1999, 158), Copp (Citation1995; Citation2000, 133; Citation2007), Brink (Citation2001, 172–173), Levy (Citation2011, 142, 142 fn. 1). It is also recognized by its critics (e.g. Rubin Citation2008, sec. 7.3.). Traces of CFR can also be found in Merli (Citation2002, 236) and Sonderholm (Citation2013, 86). It is also a natural response for moral functionalists who adopt CM (e.g. Brink Citation1984, 122–123; Railton Citation1993).6 The exact metaphysics of the realization relation are not relevant at this point. For an overview of the different options see Baysan (Citation2015).7 It could be argued that it is unclear how a realized, higher-order, property can causally regulate one’s use of moral terms. In response, all that is required is that such properties are causally efficacious. There are at least two ways this can happen. First, such realized properties can be identical to the family of properties that realize them (as per, so-called, filler functionalism). Secondly, such realized properties can be ontologically distinct from their realizers but can nonetheless inherit their causal profile. In this sense, I disagree with Rubin’s (Citation2008, 325 fn. 27) claim that proponents of CFR ignore Horgan and Timmons’s stipulation that different first-order, non-functional, sets properties causally regulate the relevant terms used by the two communities. That stipulation is compatible with the picture that CFR suggests.8 Thanks to a reviewer from Analytic Philosophy here.9 Whether deontology is consequentializable is irrelevant at this point. Instead of CP and DP one could simply appeal to two extensionally divergent first-order moral theories.10 For Boyd, figuring out the exact details of these needs is an empirical issue (Boyd Citation1988, 329). Boyd also suggests that the relevant idea of need-satisfaction applies to communities rather than particular individuals (Citation1988, 353). This allows for a picture where there are particular trade-offs in the satisfaction of the needs of individuals, even though the needs of the community as a whole are satisfied.11 So-called ‘Deniers’ (Horgan and Timmons Citation2013, 5) include Dowell (Citation2016) and Wisdom (Citation2021).12 To clarify, the scenario being presented here is not one according to which Needs is actually perfectly realized by CP and/or DP. This is because, plausibly, neither CP or DP (or whichever properties are posited by the extensionally divergent first-order theories of one’s choice) are actually instantiated. It is perfectly coherent for a community to be causally regulated by CP (qua perfect realizer of Needs) even if the full array of properties figuring in CP is not instantiated.13 The perfect-imperfect distinction does not directly map onto the standard distinction between a total and a core realizer (e.g. Shoemaker Citation2007). Total realizers are sufficient for what they produce. But notice that this leaves open the possibility of a total realizer with superfluous properties (see note 15). Similarly, a core realizer is, roughly, the most salient and central part of a total realizer. In this sense, a core realizer is a particular instance of an imperfect realizer.14 For this reason, my use of the perfect-imperfect distinction is somewhat different from Nola & Braddon-Mitchell’s version (Citation2009, 8–9). Their version identifies perfect and imperfect realizers with natural and non-natural properties (as per Lewis [Citation1983]), respectively. Not-so-natural properties are, roughly, gerrymandered and heterogeneous. But, in my sense, a realizer can be imperfect but not gerrymandered (e.g. a realizer might be homogenous while still having some missing properties which would otherwise fully realize the relevant higher-order phenomenon in its most complete sense).15 Another case might be the following: There could be a family of properties that would be sufficient for an accurate realization of some higher-order phenomenon, if it weren’t for some superfluous properties. Again, consider the pain case. There could be a scenario of pain-realization where the relevant phenomenal experience of pain is accompanied by a phenomenal experience of tasting liquorice.16 Disagreement of this sort is routinely found in the history of the sciences. For discussion in the context of a causal theory of reference see Psillos (Citation1999, ch. 12).17 For this reason, I disagree with Väyrynen’s claim (Citation2018, 5) that a scenario involving divergence alongside referential stability involves an epistemically unclear situation.18 The relevant literature is vast. For both conceptual and empirical arguments for the multiple realizability of a variety of non-moral functional kinds see Aizawa and Gillett (Citation2009; see also Citation2011). For recent doubts on the motivation of multiple realization via appeal to the neuroplasticity of the brain see Maimon and Hemmo (Citation2022) (for discussion see Bickle Citation2020).19 It should also be stressed that the relevant realizers here are perfect: there is no question about whether visual activity in macaques is ‘better’ implemented compared to visual activity in cats.20 The same point can be made more forcefully by appealing to more quixotic perfect realizers like ectoplasmic states, etc. (I return to the question of what constraints apply to multiple realization in section 5).21 This conclusion also gets additional support from the observation that there are first-order moral theories which, on their own, are plausible (recall rule-consequentialism and Rossian deontology from the previous section).22 I find Boyd’s solution to the Qua problem very convincing (Citation2012, 53). Still, I will be focusing on Revenge as, in contrast to the Qua problem, it is a problem tailored to CFR and CM, not just the causal theory of reference in general. As Copp (Citation1995, 235–236) also notes, if the Qua problem is a real threat, then it applies to a variety of cases (not just the moral case).23 In other work, Horgan and Timmons (Citation1996, 27–28, sec. 7) cash out this “weak” characterization of moral terms by appealing to a folk, common-sense, theory of moral discourse. However, this move is potentially dialectically weak given that there are sophisticated theories of folk moral functionalism which, arguably, would not be weak in Horgan and Timmons’s sense (cf. Jackson Citation1998, ch. 6).24 Rather, commonality of reference is secured in a trivial way. To compare, consider how according to a simplistic causal theory of reference the terms ‘phlogiston’ and ‘oxygen’ have the same referent if the only constraint at play is our referential intention to refer to the phenomenon underlying the uses of those terms (cf. Psillos Citation1999, ch. 12). Similarly, Kraemer’s (Citation1991, 469) suggestion that the functional role could characterized in terms of what is minimally necessary in order to define morality, also runs into the same problematic result.25 Note that I have framed Revenge in terms of the first-order divergence and second-order, functional, convergence between the two communities, rather than in terms of their putative disagreement. To see why, consider a scenario where an Earthling and a T-Earthling, after realizing that Revenge is the case, decide to update their beliefs in a way that reflects that Needs is realized by the disjunction of CP and DP (to compare: if Water is a functional type, then presumably Earthling and T-Earthlings could take Water to refer to the disjunction between H2O and XYZ). Even if we assume that their first-order disagreement is resolved at this point, as I will argue in the next section, the threat of Revenge remains unchanged (see section 5). Thanks to a reviewer from Analytic Philosophy for discussion.26 This sort of relativism should be distinguished from the kind of “innocent” relativism that naturalist moral realists typically accept. It is true that, for example, Boyd characterizes the relevant notion of needs in terms of various “nonmoral factual” factors (e.g. physical and psychological characteristics of human beings) (Boyd Citation1988, 347) and Copp relativizes societal stability in terms of the circumstances of each society (cf. Copp Citation2007, 244, note 25). But that’s not the kind of view that’s typically labeled as “moral relativism” (in this sense, “relationalism” should be distinguished from “relativism”). As it will become apparent given my upcoming remarks, what is at stake here concerns the number of perfect realizers.27 To be more precise, in such a case moral realism would be extensionally equivalent but hyperintensionally distinct from moral relativism: the grounds of moral goodness under moral realism are different than its grounds under moral relativism. In the former case, moral goodness is grounded by stance-independent natural facts, whereas in the latter case it is grounded by the contingent values and commitments of various societies. So, even though both moral realism and moral relativism accept the same moral truths (in this scenario), they still countenance different grounds for these truths.28 The pluralist account that I have sketched is similar to what Harman (Citation1978) calls meta-ethical relativism. Roughly, this version of moral relativism allows for conflicting judgments about a particular case to be all correct. In fact, I want to suggest that by appealing to something like Needs it could be said that meta-ethical relativism is strengthened. Initially, the claim that two conflicting moral judgments can be both correct, seems to entail some kind of inconsistency (Harman Citation1978, sec. 7). But there is no inconsistency if we take the relevant regulators of these judgments to be distinct perfect realizers of (say) Needs. In this sense, the two communities are both correct in their judgments in the sense that they have tracked two distinct ways of realizing the same higher-order function.29 The Disney Principle is independently plausible. Specifically, it is a direct consequence of the view that special science laws supervene upon the more fundamental laws of nature (see Stamatiadis-Bréhier Citation2023: sec. 5). Now note that typical proponents of CM propose the realization-fact ‘CP functionally realize Needs’ as a special science law (e.g. Sturgeon Citation1985). In this sense, it is expected that the multiple realization thesis of the form ‘{CP or DP or … } functionally realize Needs’ is subject to nomic constraints from more fundamental laws. A clear example is the following: a putative perfect realizer of Needs that violates the law of the conservation of energy is nomologically impossible.30 One might be tempted to apply a naturalness constraint in order to single out a particular perfect realizer. Presumably, one of the putative realizers of Needs will be more natural (in the Lewis [Citation1983] sense) than its competitors. But in whichever way one cashes out the relevant notion of naturalness, the perfect realizers at play will score the same degree of naturalness. On the classic interpretation, a property is natural when it is defined in terms of other natural properties and its definition is relatively short. But this interpretation is notoriously problematic when it is applied to phenomena that are quite distant from the fundamental level (cf. Dunaway and McPherson Citation2016, sec. 3.4.). On a context-relative account of naturalness, on the other hand, the relevant perfect realizers all turn out to be equally natural (cf. Taylor Citation2016). This is because the context at hand would surely concern the implementation of the appropriate function. But it is stipulated that perfect realizers fulfil that role in the fullest sense. For a problem involving ties in the context of naturalness see Mokriski (Citation2020, 22). See also Stamatiadis-Bréhier (Citation2023: sec. 4.2.) for a similar scenario.31 It is worth emphasizing that the severity of Revenge concerns the possibility of multiple perfect realizers of (say) Needs. Not just any multiple realization thesis will motivate Revenge. This is particularly important since some multiple realization theses are trivial and dialectically unimportant (e.g. the realization of a mental state by two similar tokens of the same neural type) (cf. Polger and Shapiro Citation2016).32 The most extreme case of this strategy is the complete rejection of multiple realization as a general phenomenon (for recent attacks see Bickle Citation2020, secs. 2, 3.2.).33 I make no claims about whether the two horns of this dilemma are equally threatening. For my purposes, it suffices to say that they are both worthy of attention.34 To compare, consider the initial MTE problem where the mere conceptual possibility of a twin Earth is prima facie problematic.35 Although, see Brock and Miller (Citation2019). At any rate, the current discussion on the notion of needs mainly focuses on their normative import, rather than their metaphysics. One notable exception here is arguably Copp (Citation1995, ch. 10).Additional informationFundingThis work was supported by the Azrieli Foundation via an International Postdoctoral Fellowship.