Analytical Ethical Naturalism – A Defence

For present purposes, the natural will be characterised in terms of what is amenable to scientific explanation, and thus the expansion of such explanation is coextensive with the expansion of the domain of 'natural facts'.
This ontological naturalism may be contrasted with methodological naturalism. Following Schmitt, I would the latter as the view that the way to properly investigate a given subject-matter (e.g., social institutions) is to adopt the methods of natural science, although that subject matter might not be ontologically reducible to the posits of natural science (Schmitt, 1995).
Predicates are those features of sentences which express the concepts of the properties, or relations, which are attributed to the objects in a sentence. Thus, "was loud" is a predicate ascribing loudness to "the chanting", which is the object of the sentence.
When concepts are assembled into a proposition, the proposition's extension is said to be its truth-conditions, not its reference per se (unless we say that it refers to "the true").
Following the conventions used by Cuneo and Shafer-Landau, single quotation marks will be used to signify concepts while italics will be used to signify properties (Cuneo & Shafer-Landau, 2014).
I assume here for the sake of simplicity and illustration that there is universal agreement regarding what is loud. In reality, loudness will probably have to be relativised to individual sensitivity levels, so that we end up with a class of loudness-properties composed of properties like loud-for-Peter, loud-for-Paul, and so on.
Coextensiveness over all possible worlds is required since even if all actual things which satisfy one concept satisfy the other, just one possible case in which this is not so shows that the two must be distinct.
By 'grounded' I mean justified by, and by 'dependent' I mean metaphysically contingent on. In referring to 'moral truths', I refer to propositions which are both true and may be characterised as moral. I leave aside the question of how to characterise the moral.
Examples: "A rectangle has four sides" is true in virtue of the meanings of the terms alone, and is therefore analytic. "Obama is five feet tall" is true or false in virtue of the meanings of those terms plus facts about how tall Obama is. In other words, the concept 'Obama' does not semantically entail (i.e., imply due to its internal meaning) anything about how tall Obama is, so we would have to know something in addition to the meaning of 'Obama' in order to know whether the statement in question is true or false. It is therefore synthetic.
Alternatively, AEN might claim not that the two concepts are identical, but that the one semantically entails the other.
In this context, a cognitivist is one who takes moral claims to express truth-apt moral propositions. I assume a theory of properties as truth-makers for simplicity.
By 'concept' I mean a constituent of a proposition expressed by predicates in a sentences. They may be thought of as the meanings of those predicates. I will also take such meanings to determine reference. If two predicates express the same concept, then they have the same meaning (semantic value), and therefore the same reference, or "topic", to use the Schroeter's more neutral term (Schroeter & Schroeter, 2014).
For example, it seems open to doubt that the concept 'self' is equivalent to 'the brain' because just by reflecting on what 'self' involves seems to yield content over and above that delivered by 'the brain'.
Although important terminological qualification are required, particularly for 'moral realism'.
This network is a network of "platitudes" or "commonplaces" about how to characterise a given moral concept, which will have a 'function' therein (Jackson & Pettit, 1995b, pp. 22-23). That is, it will have a defining role within the various interconnected commonplaces that make up the network. These commonplaces will be both descriptive and normative, and will often involve other moral concepts. For instance, 'rightness' relates to other descriptive and normative beliefs such as 'pertains to actions', 'generally promotes happiness', 'is the mark of virtuous actions'.
The descriptive is therefore broader than the natural. Descriptive predicates might include those falling within the domains of theology, mysticism, phenomenology, metaphysics and occultism. So long as they have no normative content, they are descriptive. In this article, descriptive properties and predicates may be understood as natural, and vice versa, given my antecedent commitment to naturalism.
A descriptive property is one which can be ascribed descriptively (Streumer, 2013).
This description of how synthetic identities are discovered is due largely to Street (Street, 2013a).
"A rigid designator designates the same object in all possible worlds in which that object exists and never designates anything else." (LaPorte, 2011)
'M' stands for a moral predicate, and 'D' a descriptive predicate.
In this way, I advocate Streumer's position that properties can and should be sharply distinguished from the predicates which ascribe them (Streumer, 2013). It is also to advocate Jackson's understanding of properties as features of realty rather than of "discourse" (Jackson, 1998, p. 126).
This is a paraphrased adaptation of Copp's characterisation of wrongness: "to be wrong is (roughly) to be precluded by the moral code the currency of which in society would do most to ameliorate the problem of sociality" (Copp, 2012, p. 51).
Standard descriptivism is a theory of reference which understands certain words/phrases as picking out their referents by means of indirect descriptions, rather than direct denotation.
E.g., conceptual identity holds between 'hard' and 'difficult'; conceptual entailment holds between 'loves' and 'cares about'.


19

Analytical Ethical Naturalism – A Defence

Rory Torrens

Submitted for the degree of Bachelor of Arts with Honours in Philosophy in the School of Social Sciences, The Australian National University.

10/2014





















Unless otherwise indicated, this thesis represents my own work. Sources used have been cited.
14,940 words



























Table of contents
I: The dialectic of the bate and why we should care p.4
I.I: What is the issue? p.4
I.II: An analogy with auditory properties p.7
I.III: Moral realism p.9
I.IX: Ethical naturalism p.12
II: Analytical moral functionalism p.19
III: Identifying ethical and natural properties p.25
III.I: A genuine property identity? p.26
III.II: The triviality objection p.31
IX: Assimilating SEN to AEN p.33
X: Defending the analyticity of the naturalist's identity claim p.35
X.I: Summary of the connectedness model p.39
X.II: The argument from ideal disagreement p.42
XI: Reconciling AMF with TCM p.49
Bibliography p.53













I: The dialectic of the debate and why we should care
This thesis concerns the metaphysical requirements for there being moral truth. In particular, it examines whether, and if so, how, a purely naturalistic ontology is sufficient for some moral claims being true. The debate I am interested in can be seen as taking place three levels down into metaethics. On the first level is the question of whether to treat moral discourse as involving at least some claims that are true. Here we find a debate between so-called moral realists and anti-realists. I assume a species of realism. On the next level down, there is a debate among the realists themselves concerning the nature of the moral facts that they mutually allege to exist. This debate is usually drawn between ethical naturalists, who take moral facts to be identical with natural facts, and ethical non-naturalists, who deny this. I assume naturalism. Finally, ethical naturalists are themselves divided on whether or not these identity statements involving moral and natural facts may be taken as analytically or synthetically true. That is, they divide on whether moral terms have the same meaning as natural (or descriptive) terms, and therefore whether our moral language can be reduced to – theoretically replaced by – purely natural (or descriptive) language. I argue that realist analytic ethical naturalism (AEN) is a tenable position and defend it from several objections.
I.I: What is the issue?
The expansion of scientific explanation of the world we live in has caused doubt regarding what were, for many, self-evident truths. The existence of the soul, of God(s) and spirits, and of a world existing independently of us just as we ordinarily imagine it to be are some of the often-cited examples of such self-evident truths, among which I include the existence of moral facts. The discrepancy between the kind of picture of the world delivered by science and the more familiar, everyday picture is captured by Wilfred Sellars' distinction between the scientific and manifest images (deVries, 2014). The manifest image – the world as it ordinarily appears to us – seems to contain moral facts, although we might not have a clear sense of what they are. This manifest moral reality has sometimes been formulated in terms of natural law and natural rights, although the constitution of the manifest image is historically changeable (deVries, 2014). In this conception, the world is embedded with an objective order consisting of, or implying, normative facts (those concerning what ought to be done in general), including moral facts. Aquinas, for instance, argues that "the whole community of the universe is governed by Divine Reason" (Murphy, 2011). In this case, and more broadly in the monotheistic tradition, the objective order has been underwritten by theology. The Abrahamic tradition's legal-providential model of natural law ethics implies that the moral facts are bound up with, and inherit their authority and objectivity from, divine power. That is, the authority of moral facts (the quality of 'brute oughtness') as well as their objectivity (their being true regardless of what we think) are supposedly explained by their being derived from divine commands. The scientific image – the world as conceptualised in terms of the theoretical posits of science – seems lacking in such facts. They are not part of, or needed for, naturalistic explanations of phenomena. If we take the scientific image to describe reality as it truly is, then insofar as it fails to describe, involve or imply moral facts, the latter appear to be under existential threat. Furthermore, even if somehow moral facts could be extracted from the scientific image, the fact that it lacks the posits of traditional theology undermines the way in which they are traditionally invested with authority and objectivity.
In this paper, naturalism is understood as a type of metaphysical monism which accords ontological weight exclusively to that which is contained within the scientific image (our best current scientific description of the world). As a category it therefore evolves with science. It states that only natural facts and properties may be said to fundamentally obtain or exist, with other kinds of facts and properties somehow built up out of/derived from them. If the naturalist wants to be a realist in the domains of normativity and ethics, the challenge is to give an account of normative and moral facts that does not violate a naturalistic ontology. I take this to be part of my burden here, since I assume naturalism.
Whilst the ethical naturalist is committed to both scientific and ethical discourse, the position entails privileging the former whenever insurmountable incompatibilities are encountered. In such cases, the question will be: how much of ethical discourse can be salvaged by making it consistent with naturalism? For instance, perhaps ethical facts can be reduced to natural facts, but only on the condition that they shed some of the properties taken to be ascribed to them by ethical discourse (and argue that these are non-essential to it).
Before addressing the puzzle of matching natural to ethical facts, the notion of a fact must be spelled out. A fact may be understood as a true proposition. Propositions are what our sentences express – they embody their meaning or content. Propositions consist of concepts. For example, the proposition "the chanting was loud" consists, at minimum, in the concepts of 'chanting' and the predicative concept 'the property of being loud' (or 'loudness'). Similarly, the proposition "Alex's murder was wrong" consists, at minimum, in the concepts of 'Alex's murder' and 'the property of being wrong' (or 'wrongness'). The extension of a concept is the set of items that it refers to. So the concept of 'the property of being loud' includes within its extension all of those things which are loud, if there are any. Similarly, 'the property of being wrong' picks out all of those things which are wrong. The question, then, is whether 'wrongness' has an extension – that is, whether there are any wrongness properties; whether it is the case that the property of being wrong attaches to anything in the world.
I.II: An analogy with auditory properties
To understand what is going on here, consider the case of 'loudness'. Presumably, we want to include loudness properties in our ontology. We want to be able to say that some things are loud. If we suppose for the purposes of this illustration that science does not describe a world featuring the property of loudness, then the extension of 'loudness' would appear to be empty, in which case all attributions of the property being loud are mistaken. Taking "the chanting was loud", which is true enough according to the manifest image, and for realists about loudness who want to respect such claims, we find that this claim is false. Since all such claims will be false, there will be no facts about loudness if naturalism is true.
The typical way the naturalist can escape this kind of conclusion without abandoning their ontology is to find another concept which does refer (i.e., a successful natural concept) and show that it picks out precisely all those things we ordinarily take to be loud. Here, she might find that the concept 'the property of emitting decibels within [a specified range]' meets these criteria. Given this, it can be said that:
L) Loudness is the emission of decibels within [a specified range].
That is, insofar as 'loudness' picks out exactly those things which the perfectly natural concept picks out, we can say that 'loudness' does have an extension. If these two concepts necessarily have the same (purported) extension, then they ascribe the same property (since they are predicative concepts). If 'being loud' and 'being such as to emit decibels within [a specified range]' are necessarily coextensive, then loudness and emitting x decibels are the same property. Thus, the naturalist proposes a naturalistic reduction which secures a place for this particular 'manifest concept' in a naturalistic ontology. She thereby gets to be a realist about loudness.
This example illustrates a theoretical template which ethical naturalists can use to vindicate moral realism – i.e., realism about moral properties. However if naturalists cannot integrate moral properties into their ontology, then the moral realist will be forced to accept some non-natural account of those properties.
I.III: Moral realism
But what precisely am I advocating in defending 'moral realism'? In this paper, realist views in metaethics are taken to subscribe to the thesis that ethical sentences express truth-apt propositions, and that some of these propositions are made true by features of the world. Such views comes in many shades, and I wish to defend something I will call 'weak moral realism'. In this section, I clarify the general view by describing its opponent, and then explore what weak moral realism involves.
Moral antirealism denies this thesis that some moral claims are true. Two important anti-realist views are non-cognitivism and error theory. Non-cognitivists are those who doubt whether it makes sense to treat moral claims as truth-apt. There are many varieties of non-cognitivism. In Ayer's classic version ('emotivism'), "ethical concepts" are "unanalysable" "pseudo-concepts" (Ayer, 1936, p. 116), whereas Blackburn's more recent 'quasi-realism' attempts to pay ethical discourse more respect by defending the existence of standards of "correctness", rather than truth, that govern its use (Blackburn, 1993, p. 55). Nevertheless, all such varieties assert that, in the final analysis, moral sentences do not express propositions; when we make moral judgements, we give voice to certain affective states (such as desires), rather than cognitive states (such as beliefs). To claim that 'Alex's murder was wrong', for instance, is not to represent the world being a particular way (the act being wrong rather than permissible, say). It is instead to express our distaste or disapproval regarding the action. Error theorists are those who doubt that, although truth-apt, moral claims are ever true, since our universe does not contain the requisite truth-makers for such claims. Joyce and Mackie are important proponents of this view (Joyce, 2001; Mackie, 1977). Though their accounts differ – Joyce's 'fictionalism' treats statements as neither true nor false, albeit truth-apt – all error theorists take the following view: moral judgements express propositions with uniquely moral representational content. They voice beliefs which characterise features of the world with moral properties such as wrongness. However, according to error theory, all such representations are mistaken. Both non-cognitivism and error theory have been associated with naturalism. Their proponents have found the attempt by ethical naturalists to insert moral properties into the natural order unconvincing, and have concluded that antirealism is therefore preferable.
The distinction I wish to draw between weak and robust realism has to do with what features of the world make moral claims true. In particular, it has to do with the degree to which moral truths are grounded in, and dependent on, facts about beings sensitive to them, such as humans. According to robust realism, moral truths are not grounded by, and exist independently of, the morally-sensitive beings who might exist and detect them. They are embedded in the fabric of the universe such that, in Bernard Williams' phrase, they are "there, anyway", and provide us with what Sharon Street calls "external reasons" (Street, 2013b, p. 2; Williams, 1986, p. 138). Conversely, weak realism posits that moral truths are grounded and dependent on morally-sensitive beings. Precisely in what way they are dependent and grounded will distinguish between varieties of weak realism. For example, one form of weak realism might argue that moral truths are a function of the actual desires of morally-sensitive individuals; another might argue that they are a function of the conclusions of morally-sensitive communities. These two forms of weak realism are, I suggest, importantly different in how much they live up to some of the core expectations of realism, as set up by the manifest moral image. It is with respect to this dimension of assessment – how much a given metaethics upholds criteria and intuitions important to moral realists – that I distinguish robust from weak realism. And it is in this way that the distinction admits of degrees. We can say, for example, that the first form of weak realism just sketched is weaker than the latter because it fails to meet more of the realist's core expectations. These expectations include the ability to argue over moral facts as well as the imperviousness of moral truths to the momentary feelings and interests of subjects (thereby respecting the seriousness of moral deliberation).
I defend ethical naturalism on a weak realist reading; moral truths and properties are not independent of us. I thereby embrace Street's advice regarding the need for naturalism to be up-front about the metaphysics of the ethical realism that it can plausibly give us (Street, 2013a). On this point, the present thesis distinguishes itself from that advanced by, for instance, Cuneo and Shafer-Landau, who appeal to something like a Platonic realm of subject-independent conceptual truths (Cuneo & Shafer-Landau, 2014). Meanwhile, it aligns itself to the weak brand of realism vindicated by the Schroeters (Schroeter & Schroeter, 2014). I note, however, that the present thesis pays some respect to the robust realist intuition, since the moral truths are determined by a special subclass of attitudes, and to that extent do not display the kind of radical inconsistency, arbitrariness and relativity that repel realists.
I.IX: Ethical naturalism and its varieties
In this section, I describe naturalism and its counterpart, non-naturalism, in the context of metaethics, advocating the former. I then distinguish two kinds of ethical naturalism (henceforth EN) – analytic and synthetic.
As a moral realist, I take the concept of 'being wrong' to be a predicate ascribing a certain property which may or may not obtain (thereby confirming or falsifying the example proposition). Once again, the challenge is to combine this with naturalism – that is, to defend EN. If moral properties such as being wrong cannot be 'naturalised', this would, following the cognitivist example, entail that the naturalist is committed to an ethical error theory: all moral claims purport to ascribe moral properties, but no such properties exist, therefore all moral claims are false. The alternative would be to reject naturalism, and conclude that moral properties are non-natural. Many have gone this way, and found ethical non-naturalism convincing.
Non-naturalists claim, at minimum, that there are non-natural moral truths – proposition that are not made true by any natural features of the world. For instance, Moore's seminal position views moral judgements as underivable from judgements concerning 'nature' (Moore, 1903). But there is a further question regarding whether there are non-natural moral properties which constitute the truth-makers of moral propositions. Cuneo and Shafer-Landau take themselves to be non-naturalists, but remain neutral on this question; on their view, moral truths are conceptually true, where these 'truth-making concepts' are themselves irreducible to any natural concepts (Cuneo & Shafer-Landau, 2014). Enoch, on the other hand, insists on the existence of non-natural moral properties – properties which are "not reducible to other, naturalist… properties" and "do not constitutively depend on us" (Enoch, 2011, p. 819). In what follows, I defend the tenability of a view which makes the invocation of both irreducible moral concepts and properties unnecessary to uphold moral realism. This means that we get a brand of what Jackson calls "serious metaphysics" – that which is concerned with accounting for what exists 'at bottom', and not simply with listing everything that exists but which may be in an important sense derived from stuff at this bottom level – which coheres with our best scientific knowledge about the world, as well as moral realism (Jackson, 1998, p. 4).
There are competing visions of how to go about advancing ethical naturalism. According to EN, the moral truths are true in virtue of certain natural truths. In other words, naturalist moral realism argues that
P: a moral propositions M is true iff some natural proposition N is true.
Using the language of properties, we might say that moral propositions contain predicates ascribing moral properties, while natural propositions contain predicates ascribing natural properties. In both cases, if those properties obtain, then the propositions ascribing them are – to that extent – true. The naturalist argues that P is true because
P': moral properties are identical with natural properties.
Given P', in ascribing a moral property, a moral proposition thereby ascribes a natural property, and is therefore true just in case the corresponding natural proposition is true.
Ethical naturalism includes those who think the relevant identity statements are analytic (AEN) and those who think they are synthetic (SEN). Analytic statements are true or false in virtue of meaning. Synthetic statements are true or false in virtue of meaning as well as other facts.
Continuing with the example of loudness, the naturalisation strategy just outlined contains an ambiguity whose resolution brings out the divide within EN. This ambiguity was left unresolved in the identity statement L) above; it was not signified whether "loudness" and "emitting-x-decibels" were meant as concepts or properties. This provokes a question: Should we understand the identity statement between the relevant manifest and scientific concepts to constitute an analytic truth, or not? There are at least two resolutions:
LS) Loudness is the emission of decibels within [a specified range].
Here, it is asserted that the two properties are identical. This is the position of the synthetic naturalist. Alternatively:
LA) 'Loudness' is 'the emission of decibels within [a specified range]'.
Here, it is asserted that the two concepts are identical. This is the position of the analytic naturalist. The concept of 'loudness' semantically entails the latter concept.
Given that identical concepts pick out identical properties, the analytic naturalist is committed to property identities just like the synthetic naturalist. Yet SEN is not committed to conceptual identities, since, it is argued, identical properties may be ascribed with distinct concepts. (For example, we might suppose that the property being a sensation can be ascribed with the concept/predicate 'sensation' and/or 'some pattern of neural stimulation in the brain'.)
Turning to the ethical case, consider again the example: "Alex's murder was wrong". What would make this proposition true would be, for the cognitivist, that Alex's murder has the property of being wrong. The puzzle for the naturalist is, once more, to find a natural property whose extension matches wrongness, where wrongness is taken to be absent in a naturalistic ontology (qua moral property). Many natural properties of this kind have been proposed. For example, the utilitarian naturalist might suggest that utility-diminishment satisfies the criteria of being both naturalisable and coextensive with wrongness. That is, it is proposed that being utility-diminishing can be given a purely natural characterisation (for instance, in terms of subjective states of well-being, which might be further reduced to certain classes of brain activation), and that all those things which are utility-diminishing have the property ascribed when we properly call something wrong. (Note also that naturalism can provisionally 'speak the language' of moral discourse and assume the existence of moral properties qua moral properties in order to ascertain whether those properties have ontologically respectable correlates.)
The question is now whether the statement "wrongness is utility-diminishment" is true in virtue of the meaning of 'wrongness', or whether we have here two distinct concepts whose extensions coincide.
Synthetic ethical naturalism (SEN) takes up the latter option, according to which it is not possible to know, by reflecting on the meaning of 'wrongness', that it picks out all and only those things which are utility-diminishing. In order to know that the two concepts do corefer in this way, we must know certain facts about the world: we must, for example, find a suitably large set of cases in which we ascribe wrongness and work out what natural features they have in common. If we find that some natural feature (like being utility-diminishing) is common, then, it is argued, we have a property identity: the moral and natural predicates in question ascribe what is in fact the very same property.
Why should they differ in meaning if they ascribe the same thing? At a general level, SEN can explain this in terms of the manifest/scientific image distinction. Synthetic but not analytic reducibility is to be expected in any case where we attempt to graft the concepts native to our 'manifest lifeworld' onto the objective world as described by natural science. 'Loudness' just as much as 'wrongness', on this account, concern how objective features of the world appear us, or what they mean to us. The concepts we use to understand and describe such features-as-experienced need not have a tight semantic relationship to the concepts we use to understand such features independently of such experience (or as independently as we can manage).
This claim is often supported by analogy with the discovery that water is H2O (Street, 2013a). Here, it seems that we have a property identity without a corresponding predicate identity, showing that it is at least conceptually possible that the same may be true in the moral domain. In the case of water, the thought is that "water is H2O" expresses a necessary synthetic truth because both predicates ascribe that same property, even though they have distinct semantics. The identity statement is not true in virtue of meaning; it is not analytic that "water is H2O". We had to discover it through empirical chemistry. SEN proposes that similar surprising property identities may be found in the moral domain.
Analytic ethical naturalism, on the other hand, wants more than just property identity. AEN argues that there is an analytic connection between the predicates involved in true identity statements between moral and natural properties. That is, they are true in virtue of meaning. This may be because the concepts involved are identical or because one semantically entails or contains another (as the concept of 'bachelor' entails or contains the concept 'man'). I argue that while the AEN thesis might appear to merely add a dubious requirement on SEN, any synthetic identification in fact implicitly rests on the kinds of analytic connections AEN proposes. In the absence of any such analysis, the SEN project seems doomed to stipulation.
I will suggest, however, that the analytic connections underlying EN need not furnish a direct, apriori route from individual moral terms to their discrete denotations for ordinary users of those terms. Instead, analysis may be pictured as a bridge which leads onto an island populated with the more 'direct' answers we were expecting all along – with the journey an aposteriori one.
Apart from the contention that SEN ultimately rests (or ought to rest) on the kind of conceptual claims characteristic of AEN, the latter also gives us a simple explanation for the identity of natural and moral properties. For example, an AEN might argue that to have 'good' in mind is to have 'utility-maximising' in mind, even if one does not realise it, since they express the same concept. If we suppose that extension is determined by semantic values, then semantic equivalence will guarantee extensional equivalence; the ontological identity of moral and natural properties follows from the semantic identity of moral and natural predicates. That is, what moral concepts pick out (moral properties) may also be picked out by the natural concepts to which they are identical, or reducible. Showing this to be the case for a given moral claim may be called an analytic reduction.
While this may be an attractive feature of AEN, it engenders a puzzle which has driven a large number of ethical naturalists to retreat to SEN. This puzzle is most well-known in the form of Moore's open question argument (OQA), which he thought showed that for any natural (non-moral) definition of a moral term, one could always "sensibly" ask if something exhibiting those natural features also exhibited the moral feature in question (Shafer-Landau, 2003, p. 56). A contemporary descendent of the OQA is presented by Nuccetelli & Seay (which they call the "OQA*") (Nuccetelli & Seay, 2012). The OQA is a device for questioning the soundness of a proposed analytic reduction. The OQA* strengthens Moore's own, which permits the response that a statement may be analytic but not "obvious" (Nuccetelli & Seay, 2012, p. 138). For the OQA*, the test for whether or not one predicate can "be replaced (without significant loss)" by another consists in whether it is "open to doubt on a priori grounds" that the two are "content equivalent" (Nuccetelli & Seay, 2012, p. 133). If it is, then we have reason to doubt the proposed reduction. At minimum, it imposes on AEN the burden of explaining why the reduction is open to apriori doubt. Given that in many cases naturalist reductions do appear to be open to such doubt, such a burden does seem to weigh upon AEN. Yet there are ways in which it may be discharged or negated.
One way of negating the burden would be to propose a revisionary account of moral concepts on which semantic content outstripping the proposed reduction is disowned. A utilitarian reduction of 'goodness' to 'happiness' might dispute certain common judgements of retributive justice, but the revisionist can say that this is justified because focusing on utilitarian value provides a systematic, unified account of the overwhelming majority of our moral lives. This is a live option, but in what follows I will concentrate on a different strategy: discharging the burden by questioning the nature of the doubt involved. I propose that it is only naïve linguistic intuition which renders 'ideally indubitable' content equivalence dubious.
In sum, this thesis is aimed at supporting the tenability of maintaining both moral realism and naturalism by defending a particular version AEN. It thereby seeks to secure what it takes to be 'truths of the manifest image' from plausible threats to it posed by the scientific image. That is, it seeks to help vindicate the existence of moral facts and properties in a purely natural world.
II: Analytical moral functionalism
The ethical naturalism I wish to defend is based heavily on analytical moral functionalism (AMF). AMF is a leading contemporary version of AEN propounded chiefly by Jackson, (Jackson, 1998). It proposes to give a functionalist (or conceptual role) analysis of moral terms which facilitates an identification of moral properties with natural (or "descriptive") properties (Jackson, 1998, pp. 113-139). For the purposes of this paper, its core claims may be set out as follows:
Moral concepts refer to what plays the corresponding role in mature folk morality.
The concept of rightness is the concept of what plays the rightness role in mature folk morality.
The property of rightness is the descriptive property DR.
The first claim pertains to the domain in which the functionalist analysis is being applied, namely the moral domain. The second gives an example of the functionalist analysis within the moral domain. The third gives us the template for the property identity statement that falls out of such an analysis. 'DR' (qua predicate) stands for a disjunction of descriptive predicates ascribing the property picked out by mature folk morality – DR qua property. 'DR' states that an action – if actions are the things which can be called right – 'satisfies either Dr1, or Dr2, or Dr3, etc.', where Dr1 is, in turn, a predicate which states that a particular action that mature folk morality deems right, say, A1, has such-and-such descriptive properties (and describes the world surrounding world in general) (Streumer, 2013). The list of descriptions of actions (and how the rest of the world is with respect to each) includes all possible actions (and entire worlds). So any possible 'right' action must satisfy 'DR'. Thus, 'right iff DR' (Jackson & Pettit, 1995a, p. 85). In other words, 'DR' states that an action is such as to be among the actions deemed by mature folk morality to be right.
Propositions (2) and (3) are identity statements, and are necessarily true if true at all. Moreover, both propositions will be analytic and apriori, if one is. The idea here is that, for any given moral concept, it is part of its meaning that it is defined not by our present substantive understanding, but by that of some more fully reflective and informed agent or community of agents (Jackson, 2012, p. 88). This mature understanding, argues AMF, is best formulated in terms of networked attitudes, just as folk theory is. The network provides the 'cross-hairs' for determining the reference of moral concepts: 'right' refers to whatever satisfies the functional role specified by the network.
There are several points to emphasise about these claims. The first is the descriptive nature of DR. In this context, a descriptive term is to be understood negatively as not an ethical term – although, as Jackson notes, the boundary is not always clear (Jackson, 1998, p. 120). Given a network of ethical commonplaces, a Ramsey sentence may be constructed for a moral term which substitutes all ethical terms with variables. Thus a term such as 'right' may be said to ascribe the property such that 'promoting equal opportunity is right, virtuous people perform right actions…' where 'virtuous' is replaced with a variables such as 'v', which is in turned given its own Ramsey sentence to give it (descriptivised) content. 'DR' gives a complete description of all actions which satisfy the Ramsey sentence for rightness. For example, if an action promotes equal opportunity it is right. Because 'DR' characterises actions in purely descriptive terms, the property DR is a descriptive property. If something is 'right', then, given supervenience, it satisfies 'DR'. That is, we ask what it is which fills the role of 'right' as given by its Ramsey sentence. The answer is DR. Any possible world where some action satisfies 'DR', is a world in which that action is right. Since coextensive predicates ascribe the same property, and one of these is descriptive, rightness is a descriptive property (remembering that what makes a property descriptive is the fact that it can be descriptively ascribed).
The second point to emphasise is that moral dispute involves disagreement over how to characterise mature folk morality. It concerns how the network of beliefs that is mature folk morality is built, and therefore what the functional roles that determine the reference of moral concepts are. Every contending view of the network will then yield, for any given moral concept, a claim of the form 'X is DX', and yet 'DX', once fully characterised by the mature folk morality being proposed will, in each case, have a distinct reference. So although the bold among us might deny that we are mistaken in our moral judgements, this may be construed as confidence that our own descriptions match those of the mature tradition.
Lastly, 'DR' can be distinguished from its corresponding predicate whose variables are cashed out in informative descriptive sentences. This predicate will not be identical to DR, and, I take it, is what is knowable only aposteriori, according to Jackson (Jackson, 1998, p. 151). Call it '[cashed-in DR]'. The corresponding identity statement may be put thus:
Right is [the cashed-in version of DR].
Note that the identity statement involving 'right' and '[cashed-in DR]' is not prima facia analytic in the way that (3) is. Yet in what follows I shall argue that it is by invoking the resources of a plausible semantic theory – the connectedness model (TCM).
In the first instance, the difference between (3) and (4) has to do with informativeness. Simply stating that 'x satisfies DR' is uninformative insofar as it remains absolutely unclear just what the latter involves; it would be more-or-less equivalent to stating (2). Identity statements of this form will apply to all moral concepts, and thereby fail to state anything interesting about how they might differ. Nevertheless, we can ascribe the relevant descriptive property because we may suppose that mature folk morality is already fixed. This claim might be grounded either by further metaphysical assumptions about the existence of a future time at which such a morality is comprehended, or by the presumption that it is a matter of detailed empirical psychology what conclusions maximum critical reflection will produce. In the latter case, it is a further question whether introspectively conducted reflective equilibrium is sufficient to uncover the content we are after. In sum, I argue that we are licenced in supposing that (4) is true on the basis of (2).
It is worth noting an ambiguity in AMF regarding the analyticity and apriority of (4). Jackson and Pettit claim that propositions like (4) are not apriori (Jackson & Pettit, 1995b, p. 28). On the one hand, this might mean that even if we had both concepts at our disposal, we could not know, prior to investigating features of the world, whether they denoted the same property. On the other hand, it might just mean that, prior to acquiring the concept uniquely involved in (4), there is no way for us to know whether such an identity statement is true. In other words, the acquisition of the concept is aposteriori – we have to find the natural facts about what mature folk morality entails (through psychology or history, as suggested above). I think the latter interpretation is the one to make.
According to Jackson and Pettit, "the network postulated is complicated and in the process of articulation, under the drive for reflective equilibrium", and may be "something we shall never reach but rather shall converge on over time" [my bolding] (Jackson & Pettit, 1995b, p. 28). The latter feature of the network mentioned here – its being under negotiation – suggests my interpretation. It speaks to the fact that the 'final results aren't in'; we simply do not know, for empirical reasons, what it is to fill the rightness role, and so cannot assess the claim in question because we have yet to acquire the relevant concept. The other feature – its complexity – merely suggests difficulty in ascertaining whether the particular action qualifies as satisfying a normative predicate, according to the huge set of inter-related commonplaces that is folk morality. Difficulty does not entail the absence of apriority. This is why I shall later invoke ideal agents to clarify the question of apriority and analyticity.
But there is a further ambiguity lodged in the claim that 'the final results aren't in'. It maybe that we presently do have, within our cognitive-dispositional semantic template, all the content necessary to arrive at the 'final results' – not through derivation (that would make the process apriori) but through a certain kind of empirical inevitability. That is, given enough empirical information and time to engage in reflective equilibration, it is an empirical fact that we would converge on a particular set of normative conclusions. On the other hand, the determination of the relevant platitudes may be construed as a social affair, worked out over historical time and not necessarily ascertainable by individuals with enough time and information. I leave this matter open. Either way, we have a situation in which the relevant conclusions are located at the limit of (historical or psychological) processes, where ordinary agents have no access.
III: Identifying ethical and natural properties
In this section I will address two puzzles for AMF (although they apply to any version of ethical naturalism): an argument disputing the existence of a genuine property identity, and 'the triviality objection'. The latter refers to an argument put forward by Parfit and given that label by Copp, who presents it as one among five non-naturalist objections (Copp, 2012). I see it as crucial to understanding the semantics not just of moral sentences (which the "argument from because", as named by Copp, concerns), but of the metaphysics of the position itself (Copp, 2012, p. 37).
To begin the discussion, I introduce a comparison of normative terms and the term 'water', which will help to both illustrate and inspire confidence in the approach being argued for here, despite being the trigger for the first objection.
Consider the following claims:
(W1) 'Water' is 'what plays the water-role in (cleaned-up) folk physics'
(W2) Water is DW
We may take these statements as analytic, and not very informative. The property of 'being DW' (i.e., what fills the water-role) may be, as far as we know, satisfied in a variety of ways prior to knowing what does in fact satisfy the functional role. If we say of a particular object that it is water, we could just as well say that it is DW. These would amount to the same thing, at least as far as pure semantics is concerned; there may be, as Dowell and Sobel point out, important pragmatic differences (Dowell & Sobel, 2013 (Draft publication)). In doing so, we commit ourselves to a claim about what in fact satisfies the water-role. For example, we might say that "this pond contains water". These kinds of claims provide the basis for the more 'interesting' variety of identity claims. That is, after assembling and bringing coherence to a set of claims such as "this pond, this tree, this cup… contain water", we are put in a position to form identity claims such as:
(W3*) Water is H2O.
(W3*) is not semantically equivalent to (W2) – it 'cashes in' DW with a substantive and informative rather than merely formal claim.
Similarly, we may find 'immature' and incomplete materials for formulating mature identity-statements in the moral domain. For example, "repealing South African apartheid laws was the right thing to do". According to AMF, the 'mature' counterpart of such claims (one analogous to "water is H2O") will be specified by mature folk morality a la proposition (4).
III.I: A genuine property identity?
Assuming for the moment the plausible analyticity of (2), I have suggested that we can conclude (4), at least as a synthetic proposition. But it may be doubted that this strategy yields a genuine property identity at all. In particular, although I have drawn confidence in the derivation of normative identity claims from this analogy with water, it may be alleged that no such confidence is warranted, since "water is H2O" does not express a genuine property identity proposition. Note that, while I have framed the problem as following from a particular form of analytic naturalism, it also applies to any variety of ethical naturalism which proposes moral-natural property identities.
For convenience, here are the two sets of claims for comparison:
Morals:
Moral concepts refer to what plays the corresponding role in mature folk morality.
The concept of rightness is the concept of what plays the rightness role in mature folk morality.
The property of rightness is the property DR.
Right is [the cashed-in version of DR].
Water:
(W1) 'Water' is 'what plays the water-role in (cleaned-up) folk physics'
(W2) Water is DW
(W3) Water is [the cashed-in version of DW]
(W3*) Water is H2O
The criticism is as follows. 'Water' may be understood to refer to the property of being what fills the water role, whilst 'H2O' refers to the property of being of such-and-such a chemical composition. These are apparently different properties. This is demonstrated by the fact that, on places like Twin Earth, what fulfils the water role is XYZ, not H2O. A statement such as (W3*) is, then, to be read either as false or as saying something like "what fulfils the water-role in this world is H2O". The moral drawn from this is that the property of being water is inclusive of the property of being H2O; they might be co-instantiated in this world, but not in others, and are thus not identical. In a similar way, perhaps 'right' and 'DR' do not ascribe the same property. Therefore, naturalism is not vindicated.
I think that this objection brings out an important disanalogy between natural kind concepts and normative concepts, and thereby helps to refine rather than falsify the naturalist thesis. Whether or not (W3*) is true qua identity-statement depends on whether we take 'water' to be a rigid designator or not. Presumably we do want an identity-claim in the water case, and think that (W3*) is true. In order for this to be so, we must index 'water' to the actual world. That is, we must specify that part of the 'water role' is that 'water' picks out actual-world water ("the watery stuff around here"). Thus there is an identity, since the property of satisfying the water role will be identical to the property of being H2O (if this is what actually fills the water role). This is to suggest that 'water', like perhaps all natural kind terms, behaves as a rigid designator.
When it is claimed that (W3*) is false qua identity claim, even though (W2) is true, 'water' is being read as being un-indexed to the actual world (i.e., as non-rigid). It is then concluded that the properties in question are non-identical. This just illustrates that, on a non-standard understanding of 'water', (W3*) is simply the wrong cashing-in of (W3). Some other way of cashing-in DW will be turn out to be true, rather than H2O – presumably something much closer to DR, as AMF understands it (i.e., disjunctively). In that case, the non-identity again evaporates.
What we can draw from this is that we are unlikely to end up with something looking like (W3*) in the case of normative concepts because the latter are non-rigid designators. DR is the property of fulfilling the rightness role where this does not include a commonplace indexing rightness to the actual-world. It describes as right actions which may fulfil the rightness-role in counterfactual worlds, even where those same actions would fail to fulfil the rightness-role in the actual world. I take this as one lesson to draw from "moral twin earth" scenarios (Horgan & Timmons, 2010). Taking the twin planet as a possible world, there is no intuition that, if some natural property occupies the role of 'goodness' which is different from that which actually obtains, then it is not good.
The contrast with water is then clear: DW is the property of fulfilling the water role where this does include indexing. Whether a substance is counted as water will depend on what is taken as the actual world. This is what makes typical water-claims aposteriori synthetic necessities; (W3*) will be synthetic. Even if we have each concept, it will not be possible to say that the one property is identical to the other without knowing which world is actual world. We need extra empirical information. But if rightness-claims do not depend on what is taken as the actual world, varying which among all the possible worlds is actual does not vary verdicts regarding what is right; if an action satisfies the rightness role, then it is right.
This may appear to generate counterintuitive results. Take the claim 'if some action reduces unnecessary suffering, then it is not bad'. (This is not a biconditional (those will involve something like 'DR'), but it is nonetheless illustrative.) If 'not bad' is a non-rigid designator, there is no world in which an action which reduces unnecessary suffering is bad. But this may make it the case that 'jamming knives into the stomachs of acquaintances is good', if there is some possible world in which things are so arranged as to make this behaviour satisfy the goodness-role. I claim that this not so counterintuitive if we consider what is important in our concept of goodness and the fact that the role thereby defined is satisfied in this (peculiar) world, even if it is not in our actual world. The goodness role may involve platitudes such as 'is conducive to welfare'. The peculiar world in question may be populated with humans psychologically disposed to welcome violent gorings, and physiologically adapted to rapid recovery from them, such that they are conducive to welfare, and are (at least to that extent) good. Knowing these facts, we become much more willing to apply the concept of goodness, regardless of whether we are considering the case counterfactually or as actual.
If moral concepts are non-rigid, then claims such as 'M iff D' are apriori; it is impossible to imagine any circumstance in which M but not D, or D but not M. Regardless of which possible world turns out to be actual, we know it to be true and do not need to investigate the world to rule out various falsifying possibilities. The best explanation for its apriority is its analyticity: it is part of the meaning of M that D satisfies it.
Note that this thesis is likely to vindicate many of the judgements we find in open question arguments (OQA). When we are asking about whether this or that is the correct analysis of 'right', for instance, we likely to correctly deny that it is. This is because the correct analysis of 'right' is an impossibly unwieldy disjunction, unlikely to turn up as an object of scrutiny in any actual philosophical examination. The disjunction's length is multiplied beyond comprehension by the fact that it ranges over all possibilities, taking each possible action and asking of it "does x satisfy the rightness role?"
The main point in connection with property-identity is that the disanalogy with water does not rule out the naturalist conclusion. If 'right' picks out the same set of objects as the descriptive predicate 'DR', then they ascribe the same property, just as 'water' and 'H2O' ascribe the same property if 'water' is read as a rigid designator. A background assumption here is that a property is the set of its instances. If the extension of M and the extension of N precisely overlap across possible worlds, making the biconditional 'M iff N' necessarily true, then the properties they ascribe are identical.
It looks to me as though what the objector (for instance, Parfit) is doing is tying the notion of 'property' too closely with semantics, and letting our meanings determine what properties are (rather than how properties are understood).
III.II: The triviality objection
Even if the naturalist is being coherent in proposing a kind of property identity, is the identity claim they arrive at trivial? Again, although I will frame the problem in terms of AEN, any EN must address it. David Copp has called this the "triviality objection" (Copp, 2012). The objection is that claims such as (4) are not informative or interesting, insofar as they remain naturalistic. This worry might be put in terms of a semantics according to which the meaning of a predicate is given by the property it ascribes. I will ultimately reject this semantics, but I think the objection fails even if it is granted.
If M and D ascribe the same property, then they have the same meaning. So AMF (and the ethical naturalist more generally) is faced with a dilemma: either normative and natural predicates do not ascribe the property (undermining naturalism), or they ascribe the same property (rendering the identity statement trivial). The first horn of the dilemma strikes at the core EN thesis that the properties ascribed by moral predicates can be understood in natural terms. The second is supposed to render the EN's claim "trivial"; EN claims become comparable to uninteresting statements of synonymy. It adds nothing, for example, to point out that a 'hard' puzzle has the feature of being 'difficult', if both words refer to the same thing/property.
The "fact-stating argument" is designed to bring out the worry: normative facts cannot be stated without normative predicates (Copp, 2012). If M=D is equivalent to D=D, then it does not state a normative fact. Another way to put it is that naturalisation (or, indeed, any 'descriptivisation') drains the identity claim of interest. It might seem hardly worth mentioning that some natural property is identical to some moral property, if what is meant is that some natural property is identical to some natural property.
In reply, I would bite the bullet, and remark upon its softness. Granting the semantics proposed by the objector, the normative identity claim is trivial in the sense that the two terms have the same semantic value. This is because there is really only one property in question – namely, that ascribed by DR. That is, triviality seems to entail the truth of the EN identity claim. There is no possible world in which some action is 'right' and yet is not 'DR'. Even so, the naturalist's identity claims are not entirely without interest, since they illuminate the naturalistic constitution of the properties we ascribe with our normative predicates. The latter are short-cuts which ordinary users deploy without apprehending their full content. To make sense of this apparent anomaly, I will invoke the Schroeters' connectedness model for normative concepts (Schroeter & Schroeter, 2014).
IX: Assimilating SEN to AEN
Having defended AMF, I now suggest that (plausibly reconstructed) synthetic versions of EN rest on analytic proposition similar to those advanced by AEN. In light of this, the call to defend the kind of AEN I advocate becomes imperative. I illustrate the point with David Copp's SEN.
According to Copp, "normative facts are facts that concern solutions to, or ways to ameliorate, certain generic problems faced by human beings in the circumstances they face in their ordinary lives" (Copp, 2012, p. 32). In the case of normative moral facts, the relevant problem is the "problem of sociality". Supposing that there is a best way to solve these problems, what we are morally required to do is what "the ideal code requires us to do" (Copp, 2012, p. 38). This claim is analogous to (1) above. I suggest that we could spin this out into further claims about specific moral concepts (using the AMF model). Thus we can get claims analogous to (2), (3) and (4) above:
(1*) We are morally required to do what the ideal codes requires of us.
(2*) The concept of rightness is (roughly) the concept of what is promoted by the ideal moral code.
(3*) The property of rightness is the property DR*.
(4*) Right is [cashed-in version of DR*].
But does this vitiate the spirit of Copp's avowedly synthetic thesis? Are these claims analytic, as (1)-(4) are intended to be? I want to say that at least (1)-(3) are, on pain of making the underlying methodology here either obscure or stipulative. To elaborate: if we imagine what the SEN investigation would look like, it seems that the most plausible methodology to adopt would start with some form of conceptual analysis, rather than starting with an arbitrary stipulation, or no clear orientation at all.
In seeking to locate the properties that we ascribe when we make moral judgements within the natural world, SEN needs a theory of reference. That is, if moral properties refer to natural properties, the moral predicates we use must (sometimes, when correct) refer to things that we can refer to using natural predicates, even if the latter two are not interchangeable/semantically equivalent. To arrive at any synthetic identity statement (M=N), therefore, we need to know the conditions under which M refers to N. Otherwise, it could always be the case that all moral terms refer to non-natural phenomena.
Assuming that a roughly 'descriptivist' theory is correct – even if it integrates insights from causal theories ('causal descriptivism') – to begin to get a hold on what it is we refer to when using moral terms will involve inquiring into their meaning (i.e., conceptual analysis, broadly construed) (Psillos, 2012). In this case, the foundational piece of conceptual analysis yields (1*), which gives us the conditions under which 'right' refers to [cashed-in version of DR*] – a natural property.
Consider the water analogy. If we are to find out, aposteriori, the chemical property or properties which answer to our ascriptions of 'water', we must first of all know what the concept of 'water' consists in. That is not to say that we need to specify its application conditions in advance. Indeed, such explicit conceptual analysis was almost surely absent in the relevant scientific discovery. However, as Jackson argues, we are licenced in thinking that there must be some conceptual structure underlying and explaining our talk regarding water, given that it displays a degree of systematicity without which it would become nonsensical (Jackson, 1998). A thoroughgoing philosophical defence of any apparently synthetic property identification will thus require some kind of conceptual analysis which yields an analytic connection between moral and natural predicates.
X: Defending the analyticity of the naturalist's identity claim
As I have argued, SEN and AEN both need conceptual-analytic starting points of the form given by AMF. It seems, then, that the ethical naturalist must make good on there being such analytic truths such as (1)-(3). For AMF, these propositions are what get us to (4) and other substantive identity statements between moral and natural properties. I argue that (1)-(3) are plausibly analytic, and that, by invoking the Schroeter's semantic theory, so is (4).
With respect to propositions (1)-(3), I will take (2) for simplicity. If (2) is true, then, on a descriptivist construal of what constitutes mature folk morality, our moral concepts are just like our natural concepts – they pick out purely natural properties. This seems to show that it is possible to ascribe moral properties with descriptive predicates. If so, then moral properties are not irreducible. Naturalism is true.
Prima facia, we might doubt (2), and in two ways: first, insofar as it is a general claim about the necessity of idealisation in our moralising. According to accounts of moral concepts involving idealisation, when we ordinarily use a concept like 'right', we are giving an approximation of what we ourselves implicitly consider to be the real concept of 'right', which may not be exactly how we currently picture it. We do not usually take ourselves to know, absolutely and exhaustively, what is right. We might have maximal confidence in certain platitudinous cases, but on contentious matters where there is abundant disagreement and deep complexity, it is sensible to withhold such confidence. As 'intuitive moral realists', we imagine, however, that there is a right answer that we are searching after (Brink, 1989, pp. 23-24).
The present objection is that, after reflecting on what 'right' means, we see no appeal to anything other than our present substantive understanding of what constitutes rightness; mature, ideal or community standards never enters our minds. My reply to this objection is that, firstly, idealisation does seem part of our usual meaning, and, secondly, its absence precludes genuine moral disagreement. So, first, in terms of our 'semantic experience' with moral terms, there are cases of serious internal moral deliberation in which we seek to change our present substantive understanding of what constitutes, say, rightness. In doing so, we do not take ourselves to be forming a distinct concept, but rather refining our appreciation of the very same concept. For instance, I might think rightness is essentially related to impartiality and fairness, but come to question this understanding after an argument has been presented to me showing that there are no impartial grounds on which to maintain the judgement that my having children is the right thing to do. If I am committed to this judgement – that is, if I insist that it is nonetheless right – then my only option is to revise my 'impartialist' understanding of rightness. Secondly, we ordinarily take ourselves to be engaged in genuine debate on moral questions, and if the meaning of moral terms is restricted entirely to the individual's understanding, then this is impossible. Moral idealisation therefore seems analytic of the moral domain.
But even if this is so, the particular kind of idealisation proposed by AMF might be doubted. This is the second way in which the analyticity of (2) can be attacked, and whether it succeeds depends on the theoretical strength of functionalism. That is, it depends on whether or not functionalist semantics is the best way to construe the 'ideal moral understanding' that has been admitted to play an (analytic) role in our metasemantic moral thinking. While this dispute should be taken seriously, I do not have the space to enter into it
Another, extrinsic reason that we might doubt (2) is that (3) is doubtable. Since (3) follows analytically from (2), then if it is not analytic, then neither is (2). On this objection, we might think that simply attributing to some action a disjunction of descriptions of possible actions (one of which satisfies it, if it is right) does not count as attributing rightness to it. But as Jackson argues, this is just because of the "complexity of the moral functionalist story" (Jackson, 1998, p. 151). It is because the 'doubters' fail to appreciate how long and inclusive the descriptive sentence would have to be - in order to correctly assign moral properties - that they think there must be something left out. Analogously, we might argue that it is analytically true that "the concept of green is the concept of what plays the green-role in ideal colour theory", and that, following from this, it is analytic that "the property of being green is the property of being DG". DG would be the property of 'being viridian, or verdigris, or emerald, or chartreuse, or…'. We might take a 'short-cut' as ascribe the property in question using the word "green", but we could just as well have used the long disjunctive sentence which DG stands for. But we do not expect this to be obvious. After all, no such sentence token exists – and even if it did, it might be so long as to prevent its articulation by mortal (or perhaps even just very patient) creatures. Thus, such prima facia doubt may be explained away as due to cognitive limitations on conducting the relevant apriori thought experiment.
Lastly, are claims like (4) analytic, and why should we suppose that they are? I argue that an AEN which would regard (4) as synthetic would not be adequate. If such substantive identity claims were synthetic, then they would become "hostage to actual world contingencies", as Chappell points out (Chappell, 2006). We take moral truths of the form 'if x maximises happiness, then it is pro tanto good' to be true regardless of how the actual world is arranged. In other words, they are necessary truths. Such truths must be apriori, since we do not need to rule out any contingent possibilities to discover that they hold (Chappell, 2006).
In Jackson's view, it seems that (4) is not apriori. According to Jackson, "what is apriori… is not that rightness is such-and-such a descriptive property, but rather that A is right iff A has whatever property it is that plays the rightness role in mature folk morality, and it is an aposteriori matter what that property is" (Jackson, 1998, p. 151). In other words, (3) is apriori but (4) is not. Presumably, this is because (3) is analytic but (4) is not. I wish to resist this thesis. However my resistance turns on the semantics I adopt for normative concepts. Consequently, the real underlying disparity between AMF and the version of AEN I will adopt is, I would argue, slight.
First, I want to distinguish two ways in which a proposition may be analytic: it might be that there is a conceptual identity involved (i.e., they might be read as biconditional definitions), or it might be that there is only semantic entailment involved. In the latter case, a statement such as (2) is true in virtue of the meaning of 'right', but there might be something more in the concept of 'right' than this, such that it is not identical with 'the concept of what plays the rightness role in mature folk morality'. Perhaps this extra has to do with the "special emphasis and special feeling" that supposedly attaches to moral concepts – a quality which cannot be captured by any non-moral concept (Anscombe, 1958, p. 5). On the other hand, conceptual identity entails that when we think of this latter concept, we thereby think of the concept of 'right' (and vice versa).
Conceptual identity might seem too strong. But I think insights from "the connectedness model" (TCM) lend it considerable plausibility (Schroeter & Schroeter, 2014). Intuitions to the contrary may be seen as the by-product of an inadequate Fregean model of concepts and criterion of concept identity. TCM provides, I think, an enrichment of the AEN/AMF view as I have reconstructed it. On the strength of TCM, I suggest that (1)-(4) are all analytic, in the sense that the each predicate in each identity statement expresses the same concept.
X.I: Summary of the connectedness model
TCM is a relational theory of concept identity which provides the conditions under which different individuals express the same concept or, internally, think the same thoughts. According to TCM, concepts acquire their semantic value from facts about agents in relation to the socolinguistic tradition in which they are embedded. Such facts concern what members of the tradition would conclude after reaching wide reflective equilibrium on a given topic, taking into account the inputs and standards of rationality respected by the tradition. Concepts are identical (have the same semantic value) if they meet both the "connectedness" and "congruence" constraints, which constitute the necessary and sufficient conditions for conceptual identity (Schroeter & Schroeter, 2014, p. 35). The former constraint requires that two concepts "are connected to each other via a shared representational tradition (constituted by apparent de jure sameness relations)" whilst the latter requires that "the understanding and historical context associated with them does not diverge so radically as to undermine a univocal interpretation" (Schroeter & Schroeter, 2014, p. 35). It is in this way that the semantic value of our terms (etc.) over time and between users is reliable, consistent, and unitary. This solves the core problem of providing a theory of concepts that accommodates disagreement; we want to say that two people can have different opinions about the very same topic, and this involves sharing the same concept. If concepts are individuated on a fine-grained individualistic basis, we cannot properly say they disagree insofar their differing opinions constitute the concept, and thus the topic, that each has in mind. But on the connectedness model, two people can use the same words (express the same thought-tokens, manifest the same concepts, address the same topic) and yet have substantially different opinions about it, as long as each is connected to the tradition such that if they were to reach the kind of reflective equilibrium described they would come to the same attitudes.
As an illustration, consider the word/concept 'douchebag'. Two people might associate different things with it, and yet we may suppose that under idealising constraints they would recognise that it denoted a particular social type: white, wealthy, heterosexual males with a tendency to colour-blindness, sexism, narcissism and selfishness (for instance) (Cohen, 2014). They determine this by assessing how we (the tradition) uses the word, and the particular interests at stake in doing so (for instance, they are not the same interests as those attaching to a 'miser' or an 'asshole', which do not serve as spears to attack the exact social type in question).
The traditional cognitive difference principle is also recognised as a necessary condition for concept identity which pertains to "the special case of co-conscious thoughts" (Schroeter & Schroeter, 2014, p. 7). The principle states that two concepts fail to be identical if they "don't seem from the perspective of the subject who entertains them to obviously and unquestionably pertain to the same topic" (Schroeter & Schroeter, 2014, p. 4). Given that the open question argument relies on the cognitive difference principle, TCM's concession to the traditional 'template view' might seem to endanger the analyticity of (4). But its bite is not fatal. Proposition (4) escapes the 'cognitive difference constraint' insofar as those who oppose its analyticity are not themselves entertaining the concept of '[the cashed-in version of DR]'. We might postulate that, were an agent to entertain both of the concepts involved in (4), it would not be open to doubt that they pertained to the same topic. It would follow from the meaning of 'right' that, if a certain action is right, then it is [cashed-in version of DR], because they are the same concept.
To explain this more fully, note that concept identity here involves the fact that the semantic content of some natural predicate N, and some moral predicate M are equivalent. My response to the OQA* is that this content-equivalence does not guarantee the transparency of 'M=N' to ordinary users of those predicates. Although it is an analytic, apriori truth that 'M=N', it is not transparent insofar as we have not reached the epistemic state required to discern it. In this regard, we might distinguish between naïve and mature epistemic states. In a mature epistemic state, an agent will have the concept 'right' ('M') as given by the 'mature tradition' (which will come out as articulable by some natural predicate 'N'). In a naïve epistemic state, an agent will not have 'M' in their conceptual command. It will therefore indeed seem, to them, open to doubt that 'M' is content-equivalent to some 'N'. We might always think that it is instead 'N*' that is equivalent to 'M', where 'M' is not merely our current approximation of 'M', but 'M' itself.
X.II: The argument from ideal disagreement
But the objection could be pressed that this simply shifts the burden. On the picture just given, 'M=N' is only known to be analytic by what we may characterise as 'ideal agents'. The question is, then, whether or not we are entitled to doubt that it is an analytic truth that
'M=N is indubitable by ideal moral agents'.
We might suppose that it is, given the alleged coherency of imagining a possible, conceptually competent, ideal moral agent who might doubt it by disagreeing with 'M=N'. The idea is that, even according to the ideal standards we (implicitly) appeal to in fixing the content of our moral terms, it is still an open question whether 'M=N'. That is, the moral concept 'M' leaves it open whether it is identical to one of a series of potential specifications of 'N', which may be denoted as 'N1', 'N2', and so on.
Two things hang on meeting this objection. Firstly, if it cannot be met, then the analytic nature of the EN I am endorsing is put in doubt. Secondly, and less obviously, its naturalism is put in doubt. To clarify matters, I refer to Bedke's "semantic entailment test" for ethical naturalism (Bedke, 2012).
Bedke argues that any EN must meet the 'semantic entailment' test. EN states that a purely natural ontology sufficies for giving us the truth conditions of normative descriptions of the world He calls this the "minimal normative reduction thesis" (Bedke, 2012, p. 113). The entailment test for this thesis involves whether facts about the natural ontology furnish "ideal agents" with "semantically-grounded entailments" to the normative truths (Bedke, 2012, p. 115). Ideal agents are defined in terms of their cognitive abilities, specifically their reasoning and information-processing abilities (p. 115). Normative concepts are defined as containing application conditions which are available to their users. If these application conditions are natural (either specifiable in natural terms, or pertaining to natural facts in the right way), then the entities meeting these application conditions are natural. That is, the extension of the concept is natural (the properties it picks out are natural). To test this, Bedke describes the following thought experiment: we imagine ideal agents possessing normative concepts and all the natural truths, and ask whether they can, on that basis alone, determine the extension of their normative concepts. If they can, then the properties referred to by the latter are natural.
It is worth noting Bedke's distinction semantic and metaphysical entailment. Although it may be the case that certain natural truths metaphysically entail certain normative truths (assuming supervenience of the normative on the natural), this is insufficient to prove a reduction, because it is compatible with a relation of "brute metaphysical necessitation" between two distinct ontologies (Bedke, 2012, p. 117). Semantic entailment, on the other hand, does ensure reduction. If all agents need to know to determine the normative truths are the natural facts, then nothing else is determining the extension of normative concepts.
Bedke argues that in fact natural truths do not semantically entail the normative truths. In this regard, he consider the case of the relational normative concept of 'being a reason for' and corresponding reason-truths (i.e., what considerations count as being reasons). Since all normative claims supposedly entail reason-truths, if reason-truths require irreducible normative properties, then so do all normative claims, including ethical ones, and SN is thereby refuted.
Bedke's objection can be seen as an elaboration on the open question argument. The key reason why he rejects semantic entailments from natural to normative truths is that he thinks his thought experiment allows for ideal disagreement: ideal reasoners in possession of normative concepts can disagree about the normative truths whilst agreeing on the natural truths. Such disagreement demonstrates the openness of an identification between natural and normative properties. The only way to get semantic entailment (and thus a true reduction) is to close the 'question'. This would be demonstrated by the absence of (ideal) disagreement.
Bedke's point may be understood as targeting the idea that 'ideal morality' expresses a single semantic value which fixes the truth of the statements in question. To return to the locution used above: if 'M=N' is contentious among ideal moral agents, then dissenters must be using 'M' in a different way from assenters, if we are to insist that M has a single semantic value.
Insisting on univocality is crucial to the present thesis. Given that meaning determines reference, if an extensional concept has a multivocal or ambiguous meaning, then it has an ambiguous reference. But this is incompatible with the naturalist thesis I wish to defend. As Streumer points out, if a moral concept may ascribe natural property N1, or N2, or… then the property ascribed by that moral property cannot be identical one of these natural properties (Streumer, 2010, pp. 8-9). Streumer calls this the "inconsistency objection", and he frames it against AMF. The first part of the objection brings out Bedke's point: ideal reasoners can disagree. According to AMF, the normative truths are determined by what agents maturely conclude. The problem is that it is reasonable to think that such mature conclusions may diverge. The alternative is to stipulate some criterion of rationality or reasonableness which ensures convergences, but this will itself be a normative property and trigger a vicious regress (Streumer, 2010, pp. 14-20). Given such divergence, then, we will have multiple normative truths pertaining to the same concepts – such as 'morally wrong'. Thus the normative properties they ascribe cannot be identical to any descriptive property; if property M is identical to property N, then there is no possible world in which property N is identical to property M*, where M* is not identical to M (transitivity of identity).
On the hybrid model I am considering, such semantic values in the case of normative concepts determine the normative truths, so the same kind of objection applies here just as it applies to Jackson's theory. The fact that the Schroeter's model has a story about semantic determination which makes it more likely that disagreement will not emerge – since the "rationalized interpretation" which determines the normative truths pertains to a single semantic tradition – does not rescue the hybrid model from Streumer's point. This is simply because there is no principled reason to exclude the possibility of divergence. Information about the tradition underdetermines conclusions about what it represents, in the final, even ideal analysis.
The 'argument from ideal disagreement' thus forms central objection to the view I am considering. It produces a dilemma: a) reject that the dissenting agent is wielding the same concept as us, or b) reject that such an agent is possible at all. I will take each horn in turn, concluding that the first – relativism – is preferable.
Relativism
The relativistic response concedes that it is possible for an agent to doubt the identity statement, where this agent is in a suitable position to doubt it because they have aposteriori access to the ideal morality which supposedly makes it true. But it is then claimed that this doubt or disagreement is only apparent because they are considering 'right' not as DR, but as DR* - where the latter is the property which fulfils the role specified by their independent moral system. That is, the agent's concept 'right*' is distinct from our concept 'right'.
If this kind of 'disagreement' with the proposed analysis is limited, then we can exclude the outliers and deny that fundamental moral disagreement (FMD) is going on. In other words, if there really is overwhelming agreement on what our moral terms mean (where this is relativised to a huge community, such as all Homo sapiens), then maybe it is indeed plausible to say that the disagreers are simply using terms in a different way. Horgan and Timmons call this option chauvinistic (Horgan & Timmons, 2010, pp. 9-10). However, I think it is unpalatable only when such disagreement is extensive. Extensive disagreement makes it look considerably less plausible to say that 'each has their own truth', given how entrenched commitment to the possibility of genuine disagreement – indeed, FMD – is in 'folk morality' (Zangwill, 2000, p. 285).
Forcing universality
The universalising response rejects the possibility of a competent user of moral terms who disagrees with the proposed analysis. The imagined possibility is one in which an ideal agent competent with the use of, say, 'wrongness' resists its attribution to 'prosecution of genocide against a particular group', even though the analysis of wrongness implies that such genocide is wrong. But this is clearly impossible. The idea us that such an agent would not even using their own, distinct concept of wrongness. The only possible concept of wrongness is ours; there are no concepts such as 'wrongness*'. All ideal agents will agree with the proposed analysis.
This reply would seem to either beg the question or imply that the proposal is revisionary, depending on how it is supported. In the first case, the reply is that our modal intuitions merely reflect limited insight. That is, although it seems like the meaning of a moral term permits ideal disputation, this is just because we have not properly grasped its meaning. The reason why it does seem possible – that we are wrong regarding the disagreement that our moral terms allow – is that we have simply not reached wide reflective equilibrium (WRE). Although moral disagreement among actual conceptually-competent agents is rife, all such disagreement occurs in the absence of WRE. All of this would dissolve were WRE reached - whereupon the individuals in question would realise the true semantic content of their moral terms and simply agree. Any intuitions to the contrary could then be dismissed as signalling failure to attain WRE.
In the absence of good reasons to think this is so, however, such a line of defence appears question-begging, as it simply reaches into the indeterminacy characteristic of the notion of WRE to fetch out its preferred answer. To clarify, this response claims that WRE would generate universal consensus (regarding the meaning of our moral terms). All dissent is explained by failure of reaching WRE. This looks like an optimistic, and question-begging, stipulation. It is difficult to refute, because it is hard to pin down criteria for when WRE has been reached, and thus to cite examples of coextensive dissent and WRE.
In the second case, the argument might be that although a moral concept may admit of ideal disputation, we can rule this out by normatively stipulating that our moral concept takes precedence. That is, such disputation should not be permitted, and any user who disagrees with the proposed analysis is thereby incompetent. All other moral systems are in error. The problem here is that it is unclear how to justify such a claim without appeal to other normative claims. Why are ideal disputants in error? Appeal must be made to some normative rule or principle which itself must be outside the circle of moral terms. But it is hard to see what it might be.
In light of these problems, I suggest the relativistic response is preferable. This conclusion might seem unsatisfying. The price for its reward – naturalism – is unclear, since it is determined by how much FMD there turns out to be, and this is left open to empirical investigation. (What I mean by 'FMD' is FMD as it is usually imagined, where fully informed people cannot come to a common moral judgement. But on this view, such people are not really disagreeing, since they are using different concepts.)
One final point in reply to the argument from ideal disagreement. Does my account address Matt Bedke's essential worry? It depends. Bedke worries that the application conditions of normative concepts are not naturalisable because when we (or cognitively idealised versions of us) wield such concepts, we cannot determine their reference on the basis of natural facts alone. But my account draws a line between merely 'wielding' a concept and being 'fully acquainted' with the content of that concept. Only in the latter case are semantic entailments apparent to those who wield the concept, and they involve as idealising agents in a way which decisively removes them from ordinary users. However the upshot remains that semantic entailments do obtain between the natural-cum-phenomenological facts and the normative facts, which yields naturalism.
XI: Reconciling AMF with TCM
I have defended ethical naturalism by appealing to the ideas of both AMF and TCM. While I offer no formal presentation of a systematic synthesis, I will refer to the combination of views I endorse here as a hybrid model. In this section, I address the worry that such a model is untenable from the outset in virtue of theoretical incompatibility between AMF and TCM. The feature of both views which most saliently unites them is the Peircean idea that truth (at least in the moral domain) is something like 'that which a community of ideal enquirers would conclude upon at the limit of their enquiry'. For AMF, this is constituted by 'mature folk morality', whereas for TCM, it is constituted by 'the rationalized interpretation of a semantic tradition'. However their theories concerning reference and the nature of concepts seem to differ problematically.
Firstly, it might seem that their accounts of reference are in conflict. This is important because while the more traditional account associated with AMF secures naturalism, it is the TCM account that secures analyticity (of proposition (4)). The problem first appears as the problem of co-reference, where TCM enters to ensure co-reference and analyticity. How do we co-refer when we use moral concepts in their first-order capacity (i.e., when making moral judgements)? If the ordinary case is one in which, as AMF argues, the descriptions we associate with the moral term do not match those associated with it by 'the mature tradition', then, in order to truly refer and therefore co-refer, it cannot be those approximate descriptions which are doing the work of fixing reference. Instead, we might refer because, as TCM suggests, we are suitably connected to a shared tradition and intend to so refer. But if this is our account of reference-fixing, then the naturalist conclusion I want to draw is in trouble. What fixes the reference of the concept 'right' (as spelled out by the mature tradition)? It is on the assumption that its descriptivist articulation (as in AMF) is reference-fixing that we get a naturalistic reduction of the property of rightness. TCM's account of reference-fixing does not secure naturalistic reduction.
In addressing this worry, note that the AMF-style analysis seems entirely compatible with the TCM account, which has it that the semantic value of our words (i.e., their truth-conditions) is determined by our "holistic rationalising interpretation" of the tradition as a whole, and it is open whether or not this interpretation turns out to be descriptivist a la AMF. Therefore we can distinguish two claims: i) what users of words refer to is what those words refer to according to the 'holistic rationalized interpretation' of the associated semantic tradition, and ii) what this 'interpretation' refers to is a matter of a collection of descriptions (a la AMF). The first is explicitly advanced by TCM, but the second is left open. Unless this mere possibility is shored up into an analytic truth as AMF proposes, users of terms will not know, apriori, that they refer to descriptive facts. Either way, there appears to be no irresolvable disagreement here.
Another apparent conflict arises when we consider how much semantic information we can obtain apriori. TCM proposes to address 'object-level questions' (such as "what is rightness?") in terms of 'metasemantic-level questions' (Schroeter & Schroeter, Forthcoming). This distinction corresponds to the distinction between (4) and (2) above. That is, we can ask the object-level question "what is rightness?", and the answer, I suggest, will be "[the cashed-in version of DR]". But according to TCM, this can only be addressed by first answering the metasemantic question "what does the concept of rightness mean?" The answer I am defending is something like 'that which plays the rightness role in mature folk morality', i.e., DR.
The Schroeters argue that this question is not a matter of purely apriori conceptual analysis, since it involves not only reflection on the concept, but on how the corresponding term is used by oneself and others in the associated tradition. In asking what a word means, we are broadening the scope of our inquiry. Whilst the object-level question may be a matter of one's internal understanding alone, the metasemantic question is, according to those oriented to a more social, later-Wittgensteinian semantics, about wider practices of usage, and these are not available to pure introspection. What a word means is a communal matter to do with language use; we find out the meaning of a word by discovering the content of the relevant mature folk theory, or semantic tradition as a whole (although these may not be fixed points). In this respect – how we address the aposteriori metasemantic question – both theories seem to be aligned.
Consider the test case used above, and by the Schroeters: the word 'water'. What does it mean? There are different positions, and TCM says that deciding between them is a matter of seeing which one best fits the representational tradition that uses the word. This amounts to inquiring into which interpretation best makes sense of the practice and its implicit interests. Some might give an account that privileges causal-explanatory interests, and end up with the interpretation that 'water is H2O'. Others might argue that practical everyday interests are dominant: 'water is the watery stuff'.
If we look at these two interpretations, we notice a parallel with the distinction descriptivists draw between the analysis of the conceptual role of 'water', and what the role is filled by. If 'the watery stuff' contains the idea that water must be a discrete natural kind, then the role will be filled by H2O. But if not, as the imagined proponents of the latter view suggest, then the role-filler may be a 'disjunctive property' describing the set of all things whatsoever which could (imagining other possible worlds, too) do all the things we practically expect water to do.
Looking at normative concepts, it is quite arguable that we should not take our practice concerning concepts like 'right' to be committed to the idea that it ascribes a natural kind. Still, it may be the case that DR fills the role, showing naturalism to be true. My point is just that TCM does not debar this conclusion – it is implicitly an option following from the supposition that, say, 'the watery role' analysis of 'water' is correct as a gloss on what the representational tradition represents about 'water'.
The upshot of these reflections is that TCM does not seem to be irreconcilably antagonistic to AMF. Rationalising interpretations appear to be statable using the particular cognitive-dispositional 'template-model' advanced by AMF – namely, a functionalist network. Given plausible generalisation of this logic to all moral terms and discourse, I have argued that an analytic ethical naturalism on the hybrid model is a defensible option. Moral facts/properties may be analytically equivalent to natural facts/properties.







Bibliography
Anscombe, E. (1958). Modern Moral Philosophy. Philosophy, 33(124).
Ayer, A. (1936). Critique Of Ethics And Theology Language, Truth, and Logic (pp. 104-126). London, England: Penguin.
Bedke, M. (2012). Against Normative Naturalism. Australasian Journal of Philosophy, 90(1), 111-129.
Blackburn, S. (1993). Essays in Quasi-Realism. Oxford, England: Oxford University Press.
Brink, D. (1989). Moral Realism and the Foundations of Ethics. Cambridge, England: Cambridge University Press.
Chappell, R. (2006). Against synthetic ethical naturalism. Retrieved from http://www.philosophyetc.net/2006/04/against-synthetic-ethical-naturalism.html
Cohen, M. M. (2014). Douchebag: The White Racial Slur We've All Been Waiting For. Retrieved from https://medium.com/human-parts/douchebag-the-white-racial-slur-weve-all-been-waiting-for-a2323002f85d
Copp, D. (2012). Normativity and reasons: five arguments from Parfit against normative naturalism. In S. Nuccetelli & G. Seay (Eds.), Ethical naturalism: current debates (pp. 24-57). Cambridge: New York: Cambridge University Press.
Cuneo, T., & Shafer-Landau, R. (2014). The moral fixed points: new directions for moral nonnaturalism. Philosophical Studies, 167. doi: 10.1007/s11098-013-0277-5
deVries, W. (2014). Wilfrid Sellars. In E. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Autumn 2014 ed.).
Dowell, J., & Sobel, D. (2013 (Draft publication)). Advice for Non-analytical Naturalists. In S. Kirchin (Ed.), Reading Parfit (forthcoming): Routledge.
Enoch, D. (2011). Precis of Taking Morality Seriously (Oxford University Press, 2011). Philosophical Studies, 168, 819-821. doi: DOI 10.1007/s11098-013-0215-6
Horgan, T., & Timmons, M. (2010). Analytical moral functionalism meets moral twin earth. In I. Ravenscroft (Ed.), Minds, Ethics, and Conditionals: Themes From the Philosophy of Frank Jackson (pp. 221-236). Oxford Scholarship Online: Oxford University Press. Retrieved from http://www.oxfordscholarship.com/view/10.1093/acprof:oso/9780199267989.001.0001/acprof-9780199267989-chapter-10.
Jackson, F. (1998). From metaphysics to ethics: a defence of conceptual analysis. Oxford: Oxford University Press.
Jackson, F. (2012). On ethical naturalism and the philosophy of language. In S. Nuccetelli & G. Seay (Eds.), Ethical Naturalism: Current Debates. Cambridge; New York: Cambridge University Press.
Jackson, F., & Pettit, P. (1995a). Moral Functionalis, Supervenience and Reductionism. The Philosophical Quarterly, 46(182), 82-86.
Jackson, F., & Pettit, P. (1995b). Moral Functionalism and Moral Motivation. The Philosophical Quarterly, 45(178), 20-40.
Joyce, R. (2001). The Myth of Morality. Cambridge, England: Cambridge University Press.
LaPorte, J. (2011). Rigid Designators. In E. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2011 ed.).
Mackie, J. (1977). Ethics: Inventing Right and Wrong. London, England: Penguin Group.
Moore, G. (1903). Principia Ethica. Cambridge, England: Cambridge University Press.
Murphy, M. (2011). The Natural Law Tradition in Ethics. In E. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Winter 2011 ed.).
Nuccetelli, S., & Seay, G. (2012). Does analytical moral naturalism rest on a mistake? In S. Nuccetelli & G. Seay (Eds.), Ethical Naturalism (pp. 131-143). Cambridge ; New York: Cambridge University Press.
Psillos, S. (2012). Causal descriptivism and the reference of theoretical terms. In A. Raftopoulos & P. Machamer (Eds.), Perception, Realism and the Problem of Reference (pp. 212-238): Cambridge University Press. doi: http://dx.doi.org/10.1017/CBO9780511979279
Schmitt, F. (1995). Naturalism. In J. Kim & E. Sosa (Eds.), Blackwell's Companion to Metaphysics (pp. 343-345). Oxford: Basil Blackwell.
Schroeter, L., & Schroeter, F. (2014). Normative concepts: a connectedness model. Philosophers' Imprint, 14(25), 1-26.
Schroeter, L., & Schroeter, F. (Forthcoming). Rationalizing self-interpretation: the epistemology of reference. In C. Daly (Ed.), Palgrave Handbook of Philosophical Methods. Basingstoke: Palgrave Macmillan.
Shafer-Landau, R. (2003). Ethical Non-naturalismMoral Realism: A Defence (pp. 55-80). Oxford Scholarship Online: Oxford Scholarship Online.
Street, S. (2013a). Normativity and Water: The Analogy and Its Limits.
Street, S. (2013b). Nothing "Really" Matters, but That's Not What Matters.
Streumer, B. (2010). Are Normative Properties Descriptive Properties? Philosophical Studies, 154(3), 325-348.
Streumer, B. (2013). Why There Really Are No Irreducibly Normative Properties. In D. Bakhurst, B. Hooker & M. O. Little (Eds.), Thinking about Reasons: Themes from the Philosophy of Jonathan Dancy. (pp. 310-336).
Williams, B. (1986). Ethics and the Limits of Philosophy. Abingdon, England: Routledge.
Zangwill, N. (2000). Against analytic moral functionalism. Ratio, 13(3), 275-286.