Towards Cognitive Moral Quasi-Realism

There is a long-standing discussion concerning the nature of moral discourse. Multiple views range from realism—according to which moral discourse is closer to scientific discourse than to fictional discourse—to anti-realism—according to which moral discourse is rather closer to fictional discourse. In this paper, I want to motivate a novel anti-realist account. On this view, there are no moral properties or truths, neither mind-independent nor mind-dependent ones (i.e., anti-realism). However, moral cognition results from the use of higher order cognitive abilities with enough resources to grant moral discourse with all the features of a realist talk (i.e., cognitive quasi-realism). I defend this view based on empirical evidence on human moral development and by showing that the resulting account can meet the demands of robust moral realism. The paper concludes by placing the proposed view within the metaethical landscape by comparing it against other forms of anti-realism, most significantly against expressivism.


Introduction
Understanding moral discourse is a challenging task.Under the assumption that assertions involving moral terms-i.e., 'good', 'bad', 'wrong', and 'right'-express some or other content several questions arise.What kind of content is that?Is it a representational one?If so, what kind of representational is it?Is it truth-evaluable?Can it be true?Is it like those representations associated with perception, or are they more like those associated with pretense and make-believe?
Ref. [1] has offered a defense of the view that moral discourse is about "response-independent, non-natural, irreducibly normative truths, perfectly universal and objective ones, that when successful in our normative inquiries we discover rather than create or construct" ( [1], p. 21).If this is so, then moral claims such as Lying is wrong are representational, and objectively truth-evaluable claims.Such claims may describe perfectly universal, objective truths, like the truth I describe by asserting That a table is made out of wood.This is the view of robust moral realism.
Anti-realist views oppose such robust realism.In its most radical version, moral anti-realism claims that there are no moral truths in no interesting sense, either as mind-independent or as mind-dependent truths.There are several different versions of such anti-realism.In this paper, I want to use Enoch's challenge against moral anti-realism to motivate a novel anti-realist account of moral discourse.Like other anti-realist accounts, this theory takes moral terms to be semantically empty.Unlike any other anti-realist account, this theory takes moral terms to have a genuine, truth-evaluable albeit non-semantic content.This content is understood as surrogate content, and is determined as the product of a non-linguistic, domain-general, cognitive mechanism closely associated to what cognitive psychologists have dubbed the "Theory of Mind Mechanism".By offering surrogate contents for moral terms, the theory paves the way for an account of moral discourse that meets the demands of even the most staunch moral realist (such as [1]).The resulting view is what I dub "cognitive moral quasi-realism".I start by presenting Enoch's account in order to get a clear view of the desiderata that any theory of moral mental states should meet according to the robust realist.In Section 3, I discuss empirical studies concerning moral development suggesting that moral cognition, pretense abilities and the development of the Theory of Mind Mechanism (ToMM) are closely related.This evidence will allow me to present the cognitive anti-realist view I want to defend in Section 4. In Section 5, I show how the resulting view meets Enoch's robust-realism demands, while properly accounting for the empirical evidence presented in Section 3. The paper concludes by classifying cognitive moral anti-realism as a form of quasi-realism and comparing it against its anti-realist competitors.

Robust Moral Realism
Enoch has devised an argument purported to show that robust normative truths (RNTs) are indispensable, just like electrons and protons.Enoch accepts [2] challenge according to which for an entity to be accepted in our ontology it must be indispensable: we must have sufficient reason to believe in its existence.Harman, however, thinks that indispensability runs only in one direction: the explanatory way.We have sufficient reason to believe in the existence of protons and electrons, for example, because we need them in order to explain the relevant data (e.g., certain scientific experiments).Protons and electrons are explanatorily indispensable; hence, we are justified in believing that they exist.
Unlike protons and electrons, however, RNTs are not explanatorily indispensable.As [2] shows, the relevant data concerning morality is best explained by appealing to facts (or truths) such as psychological, sociological, or cultural ones, that do not include RNTs.Enoch seems to concur.However, he has an alternative reply.Maybe Harman is correct and RNTs are not explanatorily indispensable, but this is not the only form of indispensability that may give us sufficient reason to believe in the existence of a given kind of entity.
According to Enoch, if we look at what happens when we deliberate and decide what to do we shall find that RNTs are indispensable.Deliberating what to do is an elite project.It is a project that we cannot fail to engage.Furthermore, without robust normative truths RNTs, we would (it appears) undermine whatever reason we had to engage in deliberating about what to do.Enoch presents three different features of practical deliberation that speak on behalf of robust moral realism.These features have to do with the phenomenology of deliberating.Like [1], I believe that any theory of moral discourse should account for each one of them: A.
Discovery: when S asks herself what to do, she assumes that such question has an answer, that some answers are better than others, and that S doesn't create the best answer but discovers it.B.
Non-arbitrary: when deciding what to do, S is not arbitrarily picking, without reasons, the best answer.On the contrary, she deliberates which answer to pick based on reasons.C.
Commitment: as a consequence of (A) and (B), S is committed to there being reasons (normative) for deliberating.
If we accept the existence of RNTs, we thereby have accounted for items (A) to (C).What is discovered through deliberation is a robust normative truth; the process is not arbitrary because it depends on what these truths are; and the commitment on the existence of reasons for deliberating is justified (and satisfied) by the very existence of them.I believe, however, that there is more to the phenomenology of deliberation that any proper account should explain.Here's a more inclusive list and one that, I hope, is fully acceptable by the realist: D. Argument: deliberations can take the form of arguments, things that include premises and conclusions, some of which can be valid and others invalid.E. Meaningfulness: as a consequence of (D) deliberations must be meaningful, they must have some content.

F.
Truth-evaluable: when deliberating, S argues as if the claims made where truth-evaluable, as if some or other premise in the relevant arguments may be considered to be true or false.
The inclusion of item F deserves a cautionary note.It is weak enough for it to be accepted by both realists and anti-realists.However, it may seem too weak for some realists, as they may prefer to make a stronger demand, e.g., truth and not merely truth-evaluability.It is controversial, however, whether this stronger realist demand can be included among the desiderata that any account should meet.Anti-realists could easily complain that it sets the stage in favor of the realist.It seems best, hence, to keep the weak demand as it appears in F.
With this more inclusive list, I believe we have a better understanding of the phenomenology of deliberation.In the following section, I will briefly discuss a few empirical studies from the psychology of moral development.This will help complete our list of the explananda of deliberation by adding three more (empirically based) demands.In Section 4, I will present an anti-realist account of moral mental states in terms of what I call "cognitive depictions", and, in Section 5, I will show how this theory accounts for desiderata A to F.

Moral Development
Since [3] and more recently [4], moral development has been at the center of human cognitive development.The amount of research in cognitive psychology has increased substantially in the last three decades (see [5]), allowing us to have a more substantial insight into the nature of human moral reasoning.There are three salient features of relevance for the metaethical debate: the structure and form of moral development; its close relationship with ToMM; and the variety of moral views across different cultures.Each one of these features constitutes one more explananda in our list.

Stage-Like Development
Moral development-unlike perceptual development-is not a one-off, strictly linear, upward kind of development.It follows several different stages, with forwards and backwards shifts.There seems to be no constant moral standing through development, yet there seems to be a common developmental thread: moral reasoning becomes more and more complex through the life span.
Humans get involved in moral forms of reasoning at the infant stage.Between three and five years of age, the reasoning is rather simple: it is self-interested and concerned with the avoidance of self-harm.At around age 7, children continue to be concerned still endorse this concern, which appears to be limited only to directly accessible acts (see [6]).There is no evidence of an understanding of fairness and reciprocity, and children turn out to be rather unable to consider the needs of more than one person when reasoning at this young age.
Through development, children develop a more subtle and substantial understanding of notions such as retributive justice, benevolence, reciprocity, and equality.This development allows children to reason their way through more complex situations involving, for example, indirect forms of harm (see [7]) at ages 12 to 14.A further step in moral development involves the understanding of social conventions, which itself involves a complex understanding of the human mind and the ascription of mental attitudes (see [5,8]).Finally, adult moral reasoning distinctively involves complex forms of reasoning involving several interested parties and even counterfactual forms of reasoning.
An interesting feature of moral development is its U-shaped pattern.Concerning cases of indirect harm 8-year-olds and 16-year-olds offer similar judgments that are completely opposed to those of 13-year-olds.When asked whether it is wrong not letting a person know that they have dropped money and keeping it for oneself (see [5]), 8-year-olds and 16-year-olds take it to be wrong, whereas 13-year-olds consider it acceptable.There is, however, an interesting difference between 8, 13, and 16-year-olds.Young children have a simple line of instrumental reasoning: they seem to assume there is a direct connection between the money and the person who dropped it, so the bystander must simply help out.At age 13, the subjects have a more complex understanding of social contexts, realizing that strange cases, such as the one in the example, give way to moral ambiguities.
For example, they realize the observer is not responsible for the loss of the money, since it did not occur because of her actions.Furthermore, 13-year-olds realize that "in the absence of an observer, the money would have been lost in any case".It takes a further step in development for 16-year-olds to resolve such ambiguities by concluding that "the act of observing rendered the bystander obligated to return the money" ( [5], pp.426-427).
This constitutes one more explananda: why does moral development and, with it, moral discourse, observe this peculiar stage-like, U-shaped developmental pattern?To put it more specifically, if there really are robust normative truths (e.g., the bystander is obligated to return the money), why do subjects first acknowledge the truth, then reject it, in order to acknowledge it again?If the goal of moral development is to apprehend such robust moral truths, isn't it useless (if not senseless) to follow such developmental lines?

Closeness with ToMM
There is a widespread agreement in cognitive psychology concerning the existence of something like a cognitive mechanism, the so-called "Theory of Mind Mechanism" (ToMM) that allows competent subjects to appreciate people's mental states, such as beliefs and desires.The general developmental line of ToMM is well known, with a crucial step appearing at some point between three and five years of age, when children realize that subjects may have false beliefs.Since [9], there is general agreement concerning ToMM and its use of something like a decoupling mechanism that allows subjects to use, manipulate and interpret representations (e.g., perceptual representations) in order to create representations of mental states (see Section 3).This ability to manipulate representations is acutely important for the understanding of false belief, as it allows the subject to form a representation as if it were a truthful representation of what is the case while knowing it is not.It is also well known that ToMM plays a crucial role in moral development: to understand moral responsibility, children must understand that people have beliefs and desires upon which they act.
Aside from this substantial relation, recent studies have shown a more substantial connection between ToMM and moral reasoning.Ref. [8], for example, argues that reasoning about obligation and permission is central to ToMM.One important feature of ToMM is that it allows subjects to take different perspectives concerning a given situation.Closely related to this, ordinary reasoning about obligation and permission is intertwined with volition.Children easily realize that "an actor can only (sensibly) be obliged or permitted to do something he or she can carry out volitionally [...].It is because permissions allow volitional, intentional acts that they can be exercised or not, at the discretion of the actor" ( [8], p. 112).Thus, reasoning about obligation and permission makes little sense without reasoning about belief and desire.Similarly, considering different perspectives or possible plans of action, both for oneself as for others, makes little sense without considering the relevant obligations and permission.Furthermore, not only do they seem to be conceptually intertwined, but also developmentally.Ref. [8] shows that there is substantial cross-cultural evidence that "belief-desire reasoning and obligation reasoning are early and rapidly achieved, [and] they demonstrably interrelate in early life" ( [8], p. 113).
In a more recent study, Ref. [10] points at another aspect of the relation between ToMM and moral reasoning involving the ability to use, manipulate, and interpret representations previously associated with ToMM.Lane and colleagues begin by underscoring the relevance of both an understanding of mental states as well as of emotional reactions, which play an important role in moral reasoning: for example, when deciding whether to do something that may affect different subject's interests, competent subjects usually consider both, the underlying beliefs and desires as well as the emotional states (i.e., whether they would be sad or happy about this or that outcome) of the people involved.
The goal of [10] is to show that ToMM and emotion understanding, more particularly false belief and apparent emotion understanding, interact in order to substantiate moral development.However, I want to focus on another interesting insight of their study.The evidence suggests a parallelism between the understanding of mental states and the understanding of emotions, both of which seem to be triggered by the more general ability to form and consider different perspectives by manipulating mental representations.Just as understanding false belief is central to ToMM, understanding apparent emotions is central to emotional understanding.In both cases, competent subjects must be able to form representations as if they were representing what is the case while at the same time know they are not.It is not surprising, then, that false belief understanding and apparent emotion understanding appear pretty much at the same developmental stage: between three and five years of age.
These results give place to yet another element of the phenomenology of moral deliberation: accounting for the intricate interrelations between moral reasoning and the development of ToMM.Why does competent moral reasoning depend on the development of a general cognitive ability to form, manipulate and interpret representations that allows subjects to understand false belief and apparent emotions as well as higher-order representations, and mental states such as belief, desire and, of course, pretense?

Varieties
It is a well-known fact that so called "moral values" differ across human groups and or cultures.Concerning the psychology of moral development, there is a debate as to whether this variety has an important impact in moral development: is moral development a matter of internalizing a given set of social norms?Or is there substantive agreement across cultures?Some "culture" theorists [8] like to give social norms a crucial role; other "emotion" theorists [11] argue that moral thinking is fundamentally emotional and not so substantially rational.What seems clear, though, is that moral values, the very examples of what would count as moral truths, vary substantially across cultures while general forms of moral reasoning (e.g., involving harm and obligation) are more common and widespread.
This rather simple issue constitutes another problematic aspect of moral development and moral deliberation: to account for this particular kind of variety and commonality of moral representations.If moral truths are in fact robust, response-independent, and universal: why do competent subjects across cultures agree upon only some of them?In addition, why is that agreement concerned with what appear to be general forms of moral reasoning rather than particular moral claims or truths?
This brief review of the empirical evidence delivers three more items to our list of explanatory demands for an account of moral discourse.G. Developmental line: Moral development is stage-like and follows a U-shaped pattern.Moral reasoning begins early in life (from three years of age on) as a simple self-interested consideration involving harm, and develops into ever more complex forms of reasoning involving perspectives, permissions, and appearances.H. ToMM closeness: ToMM and moral reasoning develop early and rapidly, and both benefit from general representational abilities that allow the subject to take different perspectives and give place to an understanding of false belief and apparent emotion.I.
Varieties: Moral claims (i.e., alleged moral "truths") vary across cultures, while general forms of moral reasoning are shared.

Cognitive Moral Anti-Realism
Cognitive moral anti-realism is a view about moral reasoning and moral discourse.According to this view, the following two theses are true: Cognitive: moral reasoning and, thus, moral discourse, is made possible by means of higher-order cognitive abilities.These abilities allow subjects to form and express contentful and truth-evaluable mental states associated with meaningful concepts and representations that may be guided by inferential, pragmatic, and other kinds of rules.
Anti-Realism: moral terms and concepts are empty, they do not denote any object or property.Moral discourse does not describe any mind-independent (or even mind-dependent) object or property.It is, strictly speaking, devoid of truth-values.
These theses result from the acceptance of a cognitive account of moral terms as empty names.In this section, I will present, develop and defend this latter view.To do so, I will show, first, that the empirical evidence on pretense and empty names strongly suggests substantial developmental parallels between moral development and pretense development, thus motivating an understanding of moral discourse on the basis of our understanding of pretense.This will, I hope, clarify the way in which the evidence in Section 3 motivates the proposed account.

Moral Development and Pretense: A Parallel Developmental Course
There are several interesting similarities between the use of empty names and the use of moral terms, as there are between pretense abilities and moral cognition.Cognitively speaking, they both depend on the development of what has come to be known as the Theory of Mind Mechanism (ToMM).ToMM has been associated with different functions.The most well known one is that of providing the subject with enough resources to ascribe representational mental states (i.e., beliefs and desires) to their conspecifics, thereby achieving a better understanding of their behavior.
The relations between ToMM and the use of empty names are rich.Thanks to several different studies of human pretense and its development (see [9,12,13], as well as [14], among others), it is now generally accepted that ToMM plays a central role in our ability to pretend and get involved in games of make-believe, some of which involve the use of empty names.Ref. [9] famously offered an account of the workings of ToMM in terms of a decoupler mechanism (for more on this, see Section 4.2).
Compellingly enough, there are also interesting relations between moral development and the development of ToMM.It is not just the case that moral development depends on the understanding of belief and desire.It also depends on the subject's ability to distinguish between reality and appearance (e.g., between real and pretended emotions).Furthermore, as Ref. [10] shows, there is a significant correlation between performance in false-belief tasks and the performance in tasks where subjects are asked to evaluate the goodness or badness of characters.Thus, moral cognition seems to require very similar resources to those involved in the use of empty names: belief-desire understanding, pretense understanding, false-belief understanding, and the resulting understanding of higher-order thoughts (e.g., thoughts about beliefs). 1 Aside from the use of common cognitive resources and the presence of a common developmental timeline, there is a further striking similarity between pretense and the use of moral terms.Both fictional uses of 'Hamlet' and moral uses of terms like 'good' may be said to have different referents (i.e., different objects or different properties) while still having one and the same meaning or interpretation.Different actors may be used in different contexts as surrogate referents for 'Hamlet' in the representation Hamlet is courageous without thereby changing the meaning or truth conditions of such representation.Similarly, different properties, such as desirability, utility, or pleasantness, may be used in different contexts as surrogate properties for 'good' in the representation Feeding others is good without thereby changing the meaning or truth conditions of the representation.I will clarify what is meant here by 'surrogate' referent or property in the following subsection.For now, let it be enough to underscore the referential flexibility of both fictional names and moral terms.This flexibility shows not that empty names have different referents, as they in fact have none.Rather, it shows that when trying to meaningfully interpret representations using fictional empty names subjects assign surrogate referents instead of referents: i.e., objects that purport to represent the referent of the empty names.What I want to suggest is that something similar goes on with moral terms.On this view, moral terms turn out to be empty terms.When trying to meaningfully interpret representations involving moral terms, subjects assign surrogate contents: i.e., properties that purport to represent the referent of the moral term.
Together, these striking cognitive similarities between the evidence presented in Section 3 on moral development, moral reasoning and discourse, and pretense development offer motivation for the view I hereby propose, namely, that moral terms are more properly understood as fictional names: terms that have no semantically assigned content and, hence, do not refer to any existing object or property.This, of course, does not mean that moral terms (or empty names) are absolutely meaningless, contentless, or in any sense useless.In what follows, I will describe the details of the cognitive account of moral.This will include some semantic, pragmatic and cognitive details, as well as an account of how such empty moral terms can be adequately used to report and assert mental states, to deliberate, and, ultimately, to guide one's life.

Decoupler, Cognitive Depictions and Moral Discourse
In ref [15], I argued for a view according to which empty names are semantically sterile.They have no semantically determined content, precisely because they are empty.As a result, the sentences in which they are used also lack a semantically determined content and, thus, have no truth-value and express no proposition whatsoever.Fortunately, language use also involves a host of cognitively general resources allowing speakers to convey information in a more flexible manner.
Thus, on this view, empty names are semantically sterile but cognitively fruitful.Speakers use them to convey information, to express beliefs, to ascribe beliefs, to deliberate, and to argue.For this to happen, empty names must have some kind of associated content, yet not a semantically determined one.This content is fixed by means of other cognitive resources: the relevant content is provided by the decoupling mechanism.Let me develop the details of the account by extending it to moral terms.To consider moral terms as empty names is to accept the following theses: Semantic: In virtue of semantics, all predicative uses of moral terms do not refer to any property nor do they contribute any descriptive information to the content of utterances in which they are used.
Pragmatics: All predicative uses communicate a non-proffered piece of information: that there is a property that is being referred to or described.
Cognitivism: Predicative uses of moral terms are intelligible not only in virtue of their semantics and pragmatics but mainly in virtue of the cognitive resources they recruit, such as the decoupler.
Flexibility: Two attitudes, the content of which is cognitively determined, may have the same associated representation (a cognitive depiction) even if the associated representations have been assigned different surrogate objects or properties as part of their interpretation.

Depictions:
The relevant contents associated with the use of moral terms are given by cognitive depictions (CDs) that result from the decoupler's assignment of a surrogate property as the referent of the term.Now, what are CDs?What is the decoupler?As I mentioned before, the decoupler has been postulated (see [9]) to be part of our cognitive machinery, allowing us to produce the representations needed in both pretense and ToMM.The most convenient way to understand the decoupler is to see it as a machine that delivers representations, different kinds of them for different purposes.The decoupler allows the subject to use, modify, and create new representations for the benefit of the task at hand (e.g., ascribe beliefs, pretend, etc.).The decoupler has four important features.First, it works independently from general cognition and, hence, it is available for different tasks.Second, it offers representations that can be put to use with different mental attitudes (i.e., beliefs, desires, pretense, etc.).Third, the decoupler offers a way to manipulate and create representations and, as such, it is free to determine the referential, truth, and existential conditions of the representations it delivers.In other words, the decoupler freely selects an object or property to be the surrogate referent or property of the relevant representation, for which properties are to be associated with it for the representation to be true, and whether the representation in question is to have existential commitments or not.Fourth, given this interpretative autonomy, the decoupler is free to substitute surrogate referents or properties without modifying the representation to which they are assigned (e.g., 'Hamlet is courageous' remains the same representation whether it is assigned Daniel Day Lewis or someone else as a surrogate referent for 'Hamlet'.
These features, cognitive independence, attitude generality, interpretative autonomy, and flexibility are essential to the decoupler.Because it is cognitively independent, the decoupler can work when other mechanisms, such as the linguistic system, cannot.Thus, the decoupler can interpret representations that are semantically emptied.The attitude generality of the decoupler is evidenced by the fact that subjects may seriously believe the representations it delivers.LeVerrier, for example, seriously believed that Vulcan perturbed the orbit of Uranus, and so he aimed his telescope at finding Vulcan.However, given that 'Vulcan' is semantically sterile, the representation endorsed by LeVerrier must have been interpreted by non-semantic means, e.g., by the decoupler.The interpretive flexibility of the decoupler goes hand in hand with its cognitive independence.This is partly because the decoupler can freely assign objects and properties as surrogate referents of singular and general terms and it can be put to work when other mechanisms cannot.The decoupler has no restrictions when it comes to making meaningful a representation that is otherwise not comprehensible.Finally, given that the decoupler achieves meaningfulness by means of surrogates, it can maintain representational consistency even when there is no "real" object or property to be associated and, thus, many different objects or properties are equally well suited to work as a surrogate.
The general goal of the decoupler is to produce adequate representations for a given task.These representations are what I dub "Cognitive Depictions" or CDs.To understand what a CD is, it is important to have a clear understanding of how they are produced.To produce a CD, the decoupler must deploy three basic functions: copying, manipulating, and interpreting representations.To copy a representation is to produce a copy of it while divorcing this copy from its previous semantic values.This process is similar to that of quoting a sentence, whereby both its meaning and truth-value are suspended.The resulting representation is then manipulated by using further information available from memory.During manipulation, for example, certain rules of inference may change (either by suspending them or by adding them).This explains how, after undergoing the third and final process of interpretation, the resulting representation may easily be part of an inferential process. 2 Finally, the decoupler interprets the manipulated representation by fixing an assignment of reference, truth, and existential conditions.This assignment is given according to the context.It is not the same to find a surrogate referent for 'Hamlet' as to find a surrogate property for 'good'.The decoupler is meant to have access to central memory, and so it is capable of using general knowledge or even information previously gained through perception in order to fix an interpretation (see [15], for more details).
Cognitive Depictions, or CDs, are the product of the decoupler, which, in turn, is triggered whenever a referent is not available either in perception or in memory.CDs are made up of manipulated, decoupled, and interpreted representations.CDs have an interpretation, but this is not a set of sentences or descriptions: it is an arbitrary assignment of referential, truth-and existential conditions.This assignment can be as arbitrary as needed by the relevant context.CDs are not propositions; they are not made up of logical space.They do not represent possibilities but, as I said, representations of possibilities.However, precisely in virtue of this, they do divide the space of possibilities. 2 This is central for moral discourse and practical deliberation, just as much as it is for games of make-believe, where subjects must be able to follow the rules of the relevant pretense.
It is difficult to overstate the relevance of this latter feature.CDs are not propositions, this means they are not the kind of content that can be semantically determined to be the linguistic meaning of any sentence of any natural language.Given that the interpretation of a CD is the result of an arbitrary assignment of surrogates to fit certain referential, truth, and existential conditions, there are no substantive relations of synonymy between a CD and a proposition, or between a sentence expressing a CD and a sentence expressing a proposition.Thus, a given CD cannot be considered to be synonymous with any proposition.If, for example, the decoupler interprets 'Hamlet is courageous' by assigning Daniel Day Lewis as a surrogate for 'Hamlet', it simply does not follow that this CD is synonymous with the proposition expressed by the sentence 'Daniel Day Lewis is courageous'.Similarly, if the decoupler interprets a representation involving the moral term 'good' by assigning, say, PLEASANTNESS as a surrogate meaning for 'good', it does not follow that the resulting CD will be in any sense equivalent to a proposition involving the property of PLEASANTNESS.
Aside from its arbitrary origins qua representations, there is one further reason why CDs cannot be made equivalent to any semantically determined proposition.As I said already, CDs are the product of representational manipulation by the decoupler.It is of utmost importance that this manipulation be purely representational and not conceptual.For the former to be the case, the arbitrary assignment of meanings for the representation must modify or fix the interpretation involved and only such representation.For a conceptual manipulation to take place, the decoupler should have access to the linguistic system, more specifically to the lexicon, and be capable of modifying its contents.Conceptual manipulation involves some gerrymandering of lexical concepts.This latter kind of manipulation is theoretically and empirically excluded since Ref. [9]'s initial proposal.When a 5-year-old uses a banana to pretend that it is a telephone, she does not thereby modify her concept BANANA by including the feature DEVICE FOR COMMUNICATION.Neither does the child modify her concept TELEPHONE by including the feature EDIBLE.All the child needs to do is to use the banana as a surrogate referent for the representation 'Telephone' while pretending that this telephone is ringing.If conceptual gerrymandering were needed for representational manipulation, fictional imaginings and games of make-believe would be almost impossible, as they would result in a substantial loss of general knowledge.Merely pretending that bananas are telephones would falsify almost all our beliefs about fruits and communication devices.
Thus, representational manipulation without conceptual gerrymandering must be possible.In other words, it must be possible to arbitrarily assign referential, truth, and existential conditions to a representation without any semantic commitments, i.e., without thereby fixing or changing the semantic values of the expressions involved in the representation.This is precisely what CDs are meant to do.They are manipulated representations made up of surrogate referents that do no constitute semantic referents.Thus, for example, the fact that 'This telephone' has as a surrogate referent this or that banana does not imply that it is, at the same time, the referent of the demonstrative expression 'This telephone'.Briefly put, representational manipulation without conceptual gerrymandering is possible because surrogate referents, whether objects or properties, are not referents, and surrogate interpretations are not meanings, they merely help us understand representations that cannot be parsed under the semantic restrictions of the linguistic system.
The best way to describe the nature of a CD as content is in terms of the sentences it satisfies.Consider a proposition-for example, the proposition that Jon killed the cow by stabbing it this proposition satisfies the normal sentence (1): (1) Jon killed the cow by stabbing it.
There is, at least on the view I defend, no proposition satisfying the normal sentence (2): (2) To kill a cow by stabbing it is wrong.However, Ref. ( 2) is clearly a meaningful, truth-evaluable, sentence.The view I defend claims that normal competent speakers interpret sentences such as (2) by cognitive means (i.e., by using something like the decoupler) and assigning a surrogate property in place of the alleged property referred to by 'wrong'.To assign a surrogate is to take a property (a perceived one, a mental representation, an image or what have you) and use it as if it where the property referred to.The assignment of surrogates is clarified by the distinction between representations of possibilities and representations of representations of possibilities.Sentence (1), by having the proposition it has as content, represents a given possibility, namely, the possibility that Jon kills a cow by stabbing it.Sentence (2), however, has no proposition as content, for it makes use of an empty moral term.It represents no possibilities.However, it can be interpreted as representing a representation of a possibility.We can illustrate this by considering surrogate sentences, such as (3): (3) To kill a cow by stabbing it is wrong s .Unlike (2), sentence (3) can by satisfied, not by a proposition but certainly by a CD: one that assigns a surrogate property as the referential condition for 'wrong'.As a result, (3) is not strictly speaking true or false.However, it can be interpreted as true or false depending on its assignment of referential, truth, and existential conditions.To illustrate, suppose that the decoupler uses the property of UNPLEASANTNESS as a surrogate referent for the predicate 'wrong'.The result is a CD that satisfies ( 3), yet one that is not equivalent (or synonymous) to the proposition semantically expressed by (3)*.
(3)* To kill a cow by stabbing it is unpleasant.
Any competent speaker can seriously believe the CD satisfying (3), with UNPLEASANTNESS as a surrogate property, without thereby being committed to metaethical hedonism, or the idea that the moral 'good' and 'bad' are defined as what is pleasant and unpleasant.Surrogate referents are not similar, much less equivalent, to semantically determined referents or denotations.They result from an arbitrary assignment of referential conditions by a non-linguistic cognitive system.For the purposes of the CD, and the decoupler producing it, whose object or property it uses as a surrogate is much less important than the fact that it is so used.The same CD, satisfying the same sentence (3), would result if the decoupler had used the property of UNHAPPINESS instead of UNPLEASANTNESS.

Truth-Evaluability, Meaningfulness, and Inference
Thus, normal sentences using moral terms do not have a semantically determined content and, thus, do not have a truth-value.However, this does not entail that they cannot be evaluated for their truth.Ordinary sentences using empty names such as (4) do not have a semantically determined content and, hence, do not have a truth-value.However, this does not entail that, when interpreted, it cannot be evaluated for its truth: (4) Hamlet killed the King.
When properly interpreted (i.e., as conveying a CD), (4) will be taken to convey something like (5), which, in turn, can be either true or false depending on the relevant theatrical or literary piece.
(5) Hamlet s killed the King.So (4) has no truth-value, but its cognitive interpretation in terms of (5) does.That makes (4) truth-evaluable in terms of (5), even though (4) strictly speaking has no truth-value.Similarly, sentences involving moral terms such as (2) have no truth-value, but they are truth-evaluable in terms of their cognitive interpretations, such as (3).
(2) To kill a cow by stabbing it is wrong.
(3) To kill a cow by stabbing it is wrong s.This explains why sentences using moral terms can be truth-evaluable.However, there is more.Just like sentences using moral terms can be interpreted as having content (i.e., a CD), satisfying a given prop sentence using moral terms, mental states that are properly described by using moral terms can also be interpreted as being directed towards some or other CD.Thus, in this view, sentences (or mental states) using moral terms (or concepts) can be meaningful and convey non-trivial information.
Furthermore, it also explains how such sentences (or mental states) can be involved in inferential processes.According to this view, sentences using moral terms are interpreted as representing a given CD, CDs are themselves products of the decoupler.For this to be done (see [9]), the decoupler must have access to the information stored in memory, information that may include general rules of inference and concepts.Thus, the CDs delivered by the decoupler may simply inherit, as part of their interpretation, general rules of inference used by the speaker.For example, it may include the general rule according to which if A kills B then B is dead or, even more generally, if all As are Bs and this x is an A, then this x is a B. This explains how a sentence such as (3) can be part of the following inferential process: P1.To kill a cow by stabbing it is wrong s .P2. Jon killed the cow by stabbing it.C. Jon did something wrong s .

Moral vs. Fictional Discourse
I have been comparing moral terms with empty names and, specifically, with fictional empty names.There might be, however, a very important difference between fictional and moral discourse.Fictional claims, it seems (see [16]), are self-warranting: if Shakespeare claims that Hamlet is courageous, then ipso facto, it is true that Hamlet is courageous.However, this is not the case with moral discourse.Killing cows by stabbing them is not wrong just because I say so.Fortunately, the self-warranting character of at least some fictional claims is not something that is owed to their use of empty terms.
There are serious uses of empty terms that are, clearly, not self-warranting.When Le Verrier claimed, as we may well suppose he did, that Vulcan perturbed the orbit of Uranus it was not ipso facto true that Vulcan perturbed the orbit of Uranus.It required some more empirical research to realize that, in fact, what Le Verrier claimed was false.Something similar to this seems to be going on with moral claims.Moral claims are not accompanied by presuppositions of self-warrant, such as those that are part of fictional contexts. 3The fact that there can be conflicts (e.g., imaginative resistance; see [17]) associated with moral claims within fictional discourse supports this view.

Quasi-Realism
Let me now go back to our list of desiderata.

A.
Discovery: A subject S may discover what is the correct answer to her moral questions.B.
Non-arbitrary: S may decide what to do by deliberating, not by arbitrarily picking, without reasons, the best course of action.C.
Commitment: as a consequence of (A) and (B), S is committed to the existence of reasons for deliberating.D.
Argument: deliberations can be valid or invalid arguments, constituted by premises and conclusions.E.
Meaningfulness: as a consequence of (D), deliberations must be meaningful, and they must have content.F.
Truth-evaluable: S argues as if the premises presented where truth-evaluable. 3 There are several questions concerning moral epistemology that I will not consider here: how are moral claims and arguments justified?Are there fundamental moral claims that are just accepted without further discussion?Should we be foundationalists, coherentists, contextualists, or reliabilists about moral discourse?All these are interesting questions, but still questions that cannot be addressed here.

G.
Developmental line: Moral reasoning starts as a simple self-interested consideration, and develops by involving perspectives, permissions, and appearances.H.
ToMM closeness: ToMM and moral reasoning develop early and rapidly.Both benefit from perspective-taking abilities and the understanding of false belief and apparent emotion.I.
Varieties: Moral "truths" vary across cultures, while general forms of moral reasoning tend to be uniform.
In what follows, I will show how the account of moral terms as empty terms that I just offered can help meet all these desiderata.This will constitute a positive argument on behalf of cognitive moral anti-realism, the view that moral discourse and deliberation is made possible by means of higher-order cognitive resources-such as representational decoupling.In the following section, I will further describe the view by comparing it with other anti-realist views in metaethics.
A: Moral Discovery Suppose you are deliberating whether to ride your bicycle or drive your car to work.You ponder different reasons on behalf of each option.Riding the bicycle will improve your health but will get you tired, driving will be more comfortable but will contribute to air pollution.The deliberation goes on until you reach a conclusion: you should ride your bicycle to work.This conclusion is, according to the alleged phenomenology, something you discover after deliberating.It is not something you created, it is something you found out as you pondered several different reasons and reached a conclusion.
All this is consistent with the view I am defending.The fact that the contents involved in the process of deliberation are cognitive depictions (CDs) and not propositions does not affect in any way their discoverability.CDs are just as real objects as propositions, mental representations, chairs, and buildings.They are, as I said before, manipulated perceptual representations.They can be interpreted as observing as many conceptual and inferential rules as may be needed for the purpose at hand and, thus, they can be part of inferential processes.
Discovering that one should do P is done in virtue of discovering that doing P is the conclusion of the deliberative process.For that, we need deliberation.There is no way around it.The content of should do P can simply be given by a CD, like the one satisfying (6).Of course, this does not in any sense imply that the subject involved in moral deliberation knows or is aware of the fact that it is CDs and not propositions that she is working with.The decoupler is meant to work at a sub-personal level.It is triggered whenever a term lacks a referent (object or property) either in perception or in memory.The contents offered by the decoupler are, thus, not the product of any decision (arbitrary or not) made by the subject.

B: Moral non-arbitrariness
According to the phenomenology, subjects involved in deliberations pick non-arbitrarily because they pick based on reasons.I take it that reasons here are what play the role of the premises, conclusions and sub-conclusions in the deliberative processes.Cognitive depictions (CDs) can very well play that role.They are meaningful, may have associated inferential rules, and can be assigned truth conditions.Nothing else is needed for them to be reasons for deliberating.
Suppose that you concluded (6) based on reasons (7) and (8).Then, accepting (6) will not have been arbitrary, and your reasons will have also been given by CDs, not propositions or robust normative truths.(7) Riding my bicycle to work is good s because it consumes fat and saves gas.(8) Riding my bicycle to work is good s because it delivers more oxygen to my brain and helps me work better.
The fact that it is some or other CD that satisfies the sentences expressing the reasons given in the process of deliberation, and the fact that CDs are the product of the decoupler which, in turn, is itself part of the cognitive apparatus of the subject, does not make those reasons arbitrary.What matters is whether (7) or ( 8) are accepted as reasons in the process of deliberation, and this does not depend in any relevant sense on whether they are satisfied by some or other decoupled representations.
Whether or not (7) is accepted in the process of deliberation depends on two conditions.First, it depends on whether (7) conveys something that is either already accepted or that appears to be a consequence of what is already accepted in the context.Second, when none of this is the case the acceptance of ( 7) in a deliberative context depends on whether there are further reasons (i.e., claims and arguments) to accept it.

C: Moral commitment
It seems then that deliberating subjects are committed to the existence of reasons for deliberation.However, that is, again, not problematic for the cognitive account.There are reasons constituting the deliberating process, that is, the relevant contents (i.e., CDs) of the relevant moral sentences (such as 6 to 8) that the subjects present as part of their deliberation.Subjects are committed to the existence of reasons, and they can safely do so because there are such reasons: those accepted CDs expressed by sentences such as ( 7) and ( 8) that are argued for in the process of deliberation.
I have already said why we must distinguish between the fact in virtue of which those CDs exist and the fact in virtue of which those CDs are reasons within a deliberative process.The former has something to do with our biological and cognitive endowment: we are animals capable of processing, manipulating, and interpreting representations in different orders of hierarchy.The latter has to do with the fact that the relevant representations are accepted as part of a deliberative process.The former accounts for the ontology needed to satisfy the commitment of subjects to the existence of reasons.The latter accounts for the apparent requirement that such reasons be non-arbitrary, non-response-dependent, and, ultimately, objective (as we shall see).D-F: Arguments, meaningfulness and truth I have already said something about how moral terms can be used in arguments and deliberations even if we take them to be empty terms.The account of their meaningfulness in terms of CDs allows us to understand how they can be truth-evaluable as well, even if, strictly speaking, they have no truth-value.The central idea here is that when speakers benefit from the use of empty terms by cognitively interpreting the representations at hand they do not thereby throw themselves into meaninglessness.There is meaning; it is a CD.This meaning can be understood as satisfying a prop sentence and this sentence can be arbitrarily assigned both a meaning and a truth-value.
To clarify this, I distinguished between a structure that represents a possibility (e.g., a proposition) and one that represents a representation of a possibility.We cannot represent, directly, talk about the possibility of something being good or bad because, in this view, 'good' and 'bad' are empty terms.However, we can talk about a representation of the possibility of something being good or bad by assigning a prop to 'good' and 'bad'.However, assigning a prop is not any kind of magic trick; it is merely a form of pretense, a sub personal one, of course.
This prompts the question: if sentences with surrogate content literally have a full blown content, why can't they be literally true or false?The answer should be clear by now.Moral claims purport to be about the world, and this is why they have no literal content and cannot strictly speaking be true.However, they are interpreted as having some, cognitively determined, surrogate content, and this is why we interpret them as meaningful and truth-evaluable.Speakers use the only content moral claims can have as a surrogate of the content they intend them to have.The former is available and latter is not forthcoming.Moral claims cannot be literally true or false, but they can be used as if they were.This use is not to be considered mere pretense in the sense that speakers consciously use empty terms and pretend not to.This use involves serious, non-pretense-like attitudes, as any other serious use of terms.It should be clear that to consider moral terms as empty is not to consider them as fictional and, much less, to consider their competent use as part of fiction-making. 4  G-I: Development, ToMM, and varieties Understanding moral terms as empty terms from a cognitive perspective offers an account of moral discourse with two important features: it is sensitive to developmental and cognitive timetables and it is open to cultural and group differences.
Central to the account of moral terms as empty terms is the thesis that moral claims are interpreted by being assigned a CD by the decoupler.The decoupler in turn has been hypothesized to be a cognitive mechanism associated with ToMM, which is itself a cognitive capacity that develops in infancy.As I said in Section 2, there is evidence that moral cognition is incomplete at age 3, with further developments at age 5 and more complex forms of reasoning in adolescence and adulthood.This developmental timeline is difficult to account for if you are a robust moral realist.If there are universal, objective, response independent, moral facts out there, why does moral cognition take so much time to elaborate?Why isn't it more like perception, which is clearly a form of human cognition about response independent facts out there?Why does it involve forward and backward steps?Why does it strangely follow a developmental timeline that is so close to that of ToMM and pretense capacities such as the appearance-reality distinction?All of these questions, however, have a natural answer within the cognitive theory I am offering.Moral cognition, just like pretense and ToMM, heavily depends on the maturation and use of the decoupler and other higher-order forms of reasoning.From this perspective, the peculiar timeline of moral cognition is a rather expected outcome.
Another important problem for robust moral realism is to account for the variety of moral views across different human groups.If there were objective, response-independent, moral facts out there, one should expect at least some kind of homogeneity of moral views across cultures and groups.However, the evidence shows exactly the opposite.Not only is there moral disagreement (see the objection from disagreement in [20]; see also [21]), but type-level differences concerning moral development across cultures.This is prima facie at odds with the existence of robust normative truths.As a realist, one may want to underscore the existence of shared general views and dub them "moral principles".However, one will have a rough time explaining why competent subjects fail to draw common, homogenous, inferences from such "principles" if such particular moral claims should be equally truthful.
This particular variety, however, is compatible with the cognitive view I have been developing in this paper.On the one hand, there is no problem with disagreement, heterogeneity, and cross-cultural differences concerning moral views because, strictly speaking, there are no normative truths.Normative views and normative standards depend on how each group puts to use its cognitively endowed capacities for cognitive interpretation, representation, manipulation and argumentation.It is not a surprise that different standards end up being accepted as a result of these capacities by different groups at different times in different locations.On the other hand, some shared, general, forms of moral reasoning are to be expected as a result of the shared cognitive abilities upon which moral development is based.4 Indeed, fiction and empty terms are two rather orthogonal issues.There can be fiction with non-empty terms as well as serious scientific discourse with empty ones (e.g., phlogiston).Furthermore, even empty fictional terms that are known to be empty may be used in serious non-fictional ways, as evidenced by the following passage of [18] review of [19] book on Hamlet: Why would Arabs choose Hamlet, a character frequently taken to embody a quintessentially Western deliberative rationality, to define their collective identity and articulate their political aspirations?When Arabs adapt, rewrite, comment on, and otherwise invoke Shakespeare, they are not automatically responding to a former colonizer; instead, they are addressing an array of more parochial domestic issues.Given Litvin's central premise, her impressive study is concerned not only with telling the fascinating story of Arabs' engagement with Hamlet in the mid-to late twentieth century, but also with adopting a theoretical vocabulary to make sense of Shakespearean appropriation more generally, especially in situations where ordinary postcolonial explanations are inappropriate or misleading [18].
I have shown how the cognitive account of moral terms as empty terms meets the desiderata A to I.This constitutes a strong positive argument for cognitive moral anti-realism.In the following and final section, I will further describe the view by comparing it against prominent anti-realist views in metaethics.I will underscore the comparative advantages of the cognitive account against its competitors, as it is uniquely well suited to understand moral discourse as realist talk while keeping an anti-realist ontology.Because of this, the resulting account is more properly understood as a form of quasi-realism (see [22,23]).

Cognitive Quasi-Realism
I have presented the cognitive account of moral discourse as a form of anti-realism, but I have said little about the kind of anti-realism it is meant to be.Aside from its capacity to meet the demands of robust realism (A to F), as well as the psychological evidence on moral development (G to I), the account exhibits further advantages when compared against other anti-realist views about moral discourse.
Like all anti-realist positions, the cognitive account rejects the existence of mind-independent moral truths.Furthermore, it also rejects the existence of mind-dependent moral truths.All we have is mind-dependent, manipulated and interpreted representations that make use of real non-moral properties as surrogates.Thus, the view differs from constructivist, idealist and non-objectivist views, all of which accept a form of mind-dependent moral realism.Thus, the cognitive account is in better standing, as it has no commitment to the existence of moral properties or moral truths of any kind.
The literature includes two major groups of theories that, like the cognitive account, fully deny the existence of moral truths, both as mind-independent or mind-dependent.These are, first, the so-called "non-cognitivist" views-which deny that moral discourse aims at truth-and, second, error theories-which accept that moral discourse aims at truth, but fails systematically to secure it.Cognitive anti-realism differs importantly from both these views.It understands moral discourse as truth-directed.In addition, unlike error theories, moral discourse plays a central role in human cognition and is, thus, not a systematic failure.Because of these features, cognitive moral anti-realism avoids the objections presented against these traditional anti-realist accounts.
On the one hand, cognitive moral anti-realism understands moral discourse as aiming at truth, in virtue of aiming at the correctness and adequacy of the moral representations that are cognitively produced (see Section 4).In so doing, it is properly understood as a cognitivist view.On the other hand, the view takes moral discourse to be capable of securing that which it aims at, i.e., the correctness and adequacy of the moral representations that purport to be about moral truths by means of surrogate contents (i.e., by means of cognitive depictions).In so doing, the view is not properly understood as an error theory.Error theorists consider moral discourse to be a mistake, something we are better off without.That is not how cognitive moral anti-realism understands moral discourse.It is, rather, a central part of human cognitive development, perhaps even one we cannot do without, even if it does not, strictly speaking, achieve any truth.
Cognitive moral anti-realism is, thus, a very unique view on moral discourse and reasoning.It is an anti-realist view as close to moral realism as possible.In this sense, it may be appropriate to consider it a form of what [22] has dubbed "quasi-realism", since it endorses an anti-realist moral ontology together with a cognitive account of moral cognition with enough resources to grant moral discourse the status of a realist talk.
This results in important theoretical advantages, as cognitive moral quasi-realism appears to be consistent with a certain non-robust understanding of moral objectivity and universality.I will explain this in detail in what follows.

No Subjectivism
On the view I am trying to present, moral discourse is not a "load of bunk" (see [21]).There is surrogate content associated with moral claims, it is truth-evaluable, and it can be genuinely believed in.
A third form of moral anti-realism is what [21] labels "subjectivism".In this view, moral claims are in the business of aiming at truth (against non-cognitivism), and there can be genuine moral beliefs (against error-theory) because there are genuine moral facts, yet these facts are considered to be mind-dependent.This is, again, not the view I am defending.Unlike subjectivism, cognitive quasi-realism rejects the existence of genuine moral facts (even mind-dependent ones).In this view, moral terms are genuinely-i.e., completely-empty, which is tantamount to assuming that there are no moral facts of which we talk about when we use moral terms, not even mind-dependent ones.Cognitive depictions help the human mind have surrogate contents when no genuine ones are available.It does not create new kinds of things-e.g., moral properties-to become part of reality.

No Relativism
Another thesis that cognitive quasi-realism does not endorse is that of moral relativism.According to moral relativism, the truth or falsity of moral claims, or their justification, is not absolute or universal, but is relative to the traditions, convictions, or practices of a group of persons (see [24]).This view seems to entail, for example, that the abortion debate cannot be resolved since both pro-life and pro-choice groups are equally right and equally well justified, since moral truth and justification is relative to their groups.Moral relativism seems prima facie unacceptable.
Cognitive quasi-realism takes moral claims to be strictly speaking empty.As a consequence, the view takes moral claims to, again strictly speaking, lack any truth-value.Thus, the view entails that there are no absolute or universal moral truths or justification.However, this does not imply that moral truths (or justification), in the non-strict sense that they can be talked about, are relative to groups, traditions, convictions or practices.
Of course, cognitive quasi-realism claims that there is a non-strict sense in which moral discourse can be said to be true, just as much as realist talk.Can there be moral relativism in this sense?Whether a given moral claim is true (or justified), in the cognitive quasi-realist view, depends on which general moral claims (with their surrogate contents) are taken as principles, which particular moral claims (with their surrogate contents) are taken to be true, which inferential rules are observed, and which further observations are considered as relevant evidence (with or without surrogate contents).
That said, there is nothing special about the principles, claims, and evidence accepted in a given context or a given moral dispute.They all are subject to questioning, criticism and argumentation, precisely because there are no moral facts that may be appealed to as conclusive.This suggests that moral inquiry and deliberation is open-ended, as it in fact seems to be.However, as seen with respect to moral development (see Section 3.3), even though moral values and claims differ across human groups, there are still some general "principles" that are fairly uniform across human groups.These uniform views should be enough for proper moral deliberation to take place.
A proper analysis of these features, and how they interact, in moral discourse is of course needed.Pending such analysis, it seems appropriate to assume that competent moral deliberation must be at least consistent and based on argumentation.In addition, there is no single consistent way of arguing that delivers anything close to moral relativism.Thus, fortunately, cognitive quasi-realism does not have the unwelcome result that, for example, both pro-life and pro-choice groups are equally right, in the non-strict sense.Their views are not both true and both equally justified given a consistent set of moral principles, moral claims, inferential rules, etc.Furthermore, none of the views can merely claim to be based on objective, universal, self-evident, unquestionable moral truths, for there are none.Cognitive quasi-realism does not give place to moral relativism.On the contrary, it gives place to moral argumentation and deliberation.

Objectivity and Universality
It is a well-known fact about moral discourse that subjects across cultures consider that there are certain moral truths to be universal.Cross cultural studies (see [25]), for example, show that "children across cultural groups and social classes have been found to treat moral transgressions, such as unprovoked harm, as wrong regardless of the presence or absence of rules, and have viewed the wrongness of such moral transgressions as holding universally for children in other cultures or settings, and not just for their own group" ([5], p. 422).
This universality or, if you prefer, objectivity of certain moral claims has traditionally been one of the strongholds of moral realism: it is certainly something we do not want to get rid of and commonly believe to be unaccountable unless there are non-subjective, mind-independent, moral facts.I believe, however, that the traditional view has conflated two different things: the alleged objectivity of moral universal truths and the fact that speakers take certain moral claims to be objective or universally true.It is certainly true that, by accepting the existence of the former, we can account for the latter.However, to think that it is the former that is in need of explanation is to beg the question against the anti-realist.The phenomena in need of explanation is the latter, not the former.Why is it that, cross culturally, humans agree in taking certain moral truths to be objectively or universally true?
The cognitive account I am striving to present is, in principle, capable of offering an explanation of such phenomena.Although a fully developed answer is needed, I will limit myself to sketching such an explanation.First, it is important to underscore a rather obvious fact: the so called "objective" moral truths are general ones-e.g., it is wrong to inflict unjustified pain.Second, as such, they play the role of general inference rules helping speakers draw more particular or specific conclusions.This suggests a way for explaining the phenomena.Humans across cultures agree in taking certain moral truths to be objectively or universally true because these principled "truths" mark the pinnacles of moral development which, like other sorts of cognitive development, follows a common developmental progression for all human subjects (see Section 2).
Moral development follows a stage-like path, with different reasoning paradigms at different ages.It is not before three to five years of age that children are able to accept the existence of obligations, while keeping a self-interested form of reasoning.By age 8, children start considering harm as wrong (as opposed to there being an obligation not to harm others), while being able to consider the interest of others.From approximately age 12 and onwards, humans consider morality a matter of social convention and it is taken to be a sign of mature, adult-like moral reasoning, to consider morality to be a kind of principled reasoning.
The account I am offering is tailor-made to fit this developmental progression.It does not need to postulate the existence of robust moral truths to explain why humans take certain general moral truths to be objectively or universally true.All it needs to do is endorse a proper developmental account of moral reasoning.Just like ToMM develops, moral reasoning does-according to [8,10], they codevelop-and they both (on the cognitive view at least) substantially rely on the cognitive flexibility offered by something like the decoupler mechanism.

Cognitive vs. Expressivist Quasi-Realism
Ethical expressivism is perhaps the most well-known and most influential form of moral quasi-realism.Unlike cognitive quasi-realism, expressivism does not take moral discourse to have surrogate contents but rather to play a different communicative function than that of scientific discourse.According to expressivism, moral discourse is not in the business of describing how things are or may be, but rather in the business of expressing the speaker's attitudes towards a certain outcome.
Whereas cognitive quasi-realism understands (2) in terms of (3), expressivism does it in terms of (3 e ): (2) To kill a cow by stabbing it is wrong.
(3) To kill a cow by stabbing it is wrong s .
(3 e ) Expression of disapproval of killing a cow by stabbing it.
Moral claims turn out to be expressions of attitudes (e.g., approval or disapproval) according to the expressivist.This generates important problems when it comes to explaining how the content of moral claims embeds into more complex sentences-this is the so-called Frege-Geach problem.
A well-known instance of this general problem is the negation problem.Expressivism has a problem with negation because it cannot explain why (2) and (2-) are inconsistent statements: (2) To kill a cow by stabbing it is wrong.Intuitively, (3 e1 ) is the negation of (3 e ), yet it is not the expressivist analysis of (2-), but rather that of (2-1 ).
(2-1 ) To kill a cow by stabbing it is right.
Briefly put, expressivism has no clear way to explain how complex moral statements get their contents determined by their parts.There are, of course, several different expressivist proposals aimed at solving this problem (see [26] for a recent one).I do not intend to argue against such proposals.To defend cognitive quasi-realism against expressivism, it is enough to show that no embedding problem arises within the view-it has no problems with negations, attitudes, conditionals, etc.
Cognitive quasi-realism has no embedding problems because the proposed analysis of moral statements does not alter either the syntactic or the logical structure of the statement, and the interpretation of the statements in terms of cognitive depictions is meant to observe all the inference rules that may be needed, including those of classical logic.All that cognitive moral quasi-realism does is offer a surrogate content to terms (and concepts) that lack any genuine one.Once this is done, all compositional rules, whether syntactic, semantic or logical, may apply.

Conclusions
Cognitive moral quasi-realism exhibits a mix of explanatory power and parsimony.If what I said in Sections 4 and 5 is correct, then the account has at least as much explanatory power as even the most robust metanormative realism.However, it is certainly less ontologically committed.It does not postulate the existence of metaphysically special normative facts, truths, or reasons.All that is needed is a great deal of cognitive resources, such as those postulated by current cognitive psychology in order to account for human cognition.More specifically, it postulates the existence of mental representations and of decoupling abilities, together with inferential and argumentative abilities.I believe that none of these resources imply any extra cost for our theory.We already need them in cognitive psychology even if we don't use them in philosophy.
The resulting view also seems to excel against its anti-realist rivals.It avoids the negative consequences of error theory without falling into subjectivism or relativism, and it even appears to be consistent with an explanation of why human subjects genuinely believe there are certain objective moral principles.Finally, the proposed view appears to be much less problematic than ethical expressivism, and yet it has the same (if not more) explanatory power.
I believe these are good reasons to consider cognitive moral quasi-realism a serious theoretical alternative when it comes to understanding moral discourse and moral reasoning.

1
These relations between ToMM and moral development strongly suggest that, unlike what robust moral realists claim, moral cognition and human thinking about physical objects are importantly distinct.If moral discourse is to reflect what goes on with moral cognition, it should not be a surprise if moral discourse turns out to be substantially different from scientific discourse.

( 6 )
Riding my bicycle to work is the right s thing to do.

( 2 -
) To kill a cow by stabbing it is not wrong.According to expressivism, (2) turns into (3 e ), yet it is unclear which among (3 e1 )-(3 e3 ) is the negation of (3 e ): (3 e ) Expression of disapproval of killing a cow by stabbing it.(3 e1 ) Expression of approval of killing a cow by stabbing it.(3 e2 ) Expression of disapproval of not killing a cow by stabbing it.(3 e3 ) No expression of disapproval of killing a cow by stabbing it.