Stephen Finlay.  Confusion of Tongues: A Theory Of Normative Language (2014)

In a paragraph

Normative words such as ‘good’ and ‘ought’ have consistent end-relational semantics.  Practical reason and morality therefore are about assessing means and ends.  Absolutist and emotivist features of moral language can be explained by conversational pragmatics.

Key points

• Normative language, in particular the words ‘good’, ‘ought’, and ‘reason’, are shown by linguistic evidence to have a unifying end-relational semantics in which they refer to probabilistic relations in which things stand to contextually relevant ‘ends.’

• While normative language has an underlying end-relational logic, moral and deliberative uses of normative language can have features of “practicality” and “objectivity”, but these can be explained as arising from conversational pragmatics.

• Good is the most basic normative adjective.It is relational, an incomplete predicate.  ‘It’s good for e if p, [given b]’ means that p makes e more probable, pr(e|p & b) > pr(e|~p & b). 

• Ought is the most basic normative verb. It is a modal verb with normative and probabilistic meanings.  It has a near-synonym ‘should’, and an absolute form with ‘must’, ‘have to’.  What ought to done is what most increases the probability of the end, or the best, most reliable means.

• A reason is an explanation why something is more good. Good is the primary concept.

• Semantics and pragmatics: what is said v. what is communicated by the saying of it. Communication is powerful from combining the two, but can cause philosophical confusions. We communicate much more than we assert.

• “Quasi-expressivist” and “quasi-absolutist” explanations through pragmatics of the seemingly expressivist and absolutist features of normative speech and thought. Normative semantics are purely descriptivist and evaluative. Strong arguments against expressivism (Blackburn) and primitivism (Parfit).

• When we conclude that we ought to do something, we are not deciding to do this thing, but coming to have a normative belief, though our decisions to act may be based on such beliefs. A normative belief is part of deliberation.

• Ends can be multiple, vague and fluid.

• To speak about morality (unrelativized and in the singular) is to speak as if there were only one uncontroversial morality. This employs the rhetorical device of moralism, pragmatically expressing the expectation that one’s audience shares these attitudes.

• For p to be good for its own sake is for p to be good for p. We say ‘intrinsically bad’ rather than ‘bad for its own sake’.

• Under the end-relational theory no properties or facts are normative per se (or absolutely), but only relative to agents or motivated perspectives. A Humean model with a priority of desire.

• The puzzles of metaethics are largely the result of philosophers’ failure to understand these pragmatics, a metaethical “confusion of tongues”. The hypothesis is that people have mistaken theories about the meanings of their own words as concepts are often opaque and their analysis can be difficult.

Comments

I believe this book is invaluable to understanding normativity, practical reason and morality.  Normativity may have seemed mysterious, but Finlay, through his painstaking linguistic analysis, shows that it is all about making judgments against criteria.  It follows that practical reason is about considering means and ends and that morality has a similar nature.  As well as showing that normative words have an end-relational semantics, he also explains expressivist and absolutist features of moral language in terms of conversational pragmatics. He thereby shows that normativity, practical reason and morality have straightforward natures, and that rival metaethical views should be dismissed.

When I first read the book it was  a revelation.  I chose to write my Philosophy MA dissertation to discuss and develop Finlay’s ideas.

On re-reading the book and my dissertation a couple of years later, I remain convinced of the importance of Finlay’s ideas.  Some points on which I differ from Finlay:

• I prefer ‘criterion-relational’ to ‘end-relational’ to avoid implying consequentalism.

• While Finlay sees ‘ultimate ends’ as a matter of desire, I see ‘ultimate ends’ as generally capable of evaluation against other criteria; in addition, at least some valanced experience has an objective character.

• I think an important additional argument why normativity is deeply end-relational is that for any normative judgement it is always possible to ask ‘for what end.’ It sems impossible to make judgements without criteria.

I wish Finlay’s ideas were better known and the implications further developed.  I am hopeful this will contribute to a better understanding of practical reason and remove the confusions caused by rival metaethical theories.

Links 

My blog includes my full dissertation and an abridged version which consider and develop Finlay’s end-relational approach. 

Confusion of Tongues is available on Amazon UK. Chapters 1, 2, 5 and 9 are particularly valuable and some of the linguistic analysis can be skipped.

Stephen Finlay’s papers are all available on his website.

There is a decent interview with Stephen Finlay on the Scientific Sense Podcast, despite the limitations of the podcast host. 

NOTES

1 Introduction

This book is about the meaning and use of normative language, in particular the words ‘good’, ‘ought’, and ‘reason’, which are of central interest to moral and practical philosophy. In the first half of the book I take a linguistic approach, arguing that the evidence from various ordinary uses consistently points toward a unifying semantics for normative words as end-relational.  According to this theory, normative words refer to probabilistic relations in which things stand to particular “ends” or potential states of affairs that vary from context to context. In the second half, I address the distinctive features of peculiarly moral and deliberative uses of this language in speech and thought.  These are broadly classifiable as forms of either “practicality” or “objectivity”, and largely comprise the central problems of metaethics. I demonstrate that the end-relational theory accommodates and explains these features systematically by appeal to basic principles of conversational pragmatics,

Tower of Babel talking past each other. A failure to understand our own use of normative language.

From a naturalistic perspective, human beings are simply animals motivated by contingent desires for varying ends, which inevitably bring us into conflict. Normative speech and thought has been the object of a grand philosophical ambition to elevate humanity above “mere” animality.  Call the various ways the normative is independent from individuals’ subjective attitudes its objectivity. This objectivity has been viewed by many philosophers as raising us from the natural to the divine. Call the special connection of the normative to motivation its practicality.  “The Moral Problem”: how to reconcile the objectivity of the normative with its practicality.

Do we desire things because they are good, or are they good because we desire them? Ultimately supports the priority of desire.

The puzzles of metaethics arise largely from failure to understand the meaning and use of our own normative language.

I shall apply an analytic method, seeking a metaphysical analysis of normative facts, properties, and relations by means of a conceptual analysis of the meanings of the normative words.

Analytic reduction to non-natural properties.  Analysis to normative no better than circular with regard to the central puzzles of metaethics, which concern the nature of the normative as such.

Concepts shouldn’t be regarded as additional entities between us and the world, but rather as properties, etc. as conceived of; to “have the concept” aunt is just to conceive of the relation.

An enquiry into the meanings of words as fixed by the linguistic conventions of a community, a semantic analysis.

I assume the following model. Linguistically, words compose in structures of grammatical syntax to form sentences. Conceptually, concepts compose in structures of logical syntax to form thoughts. Metaphysically, objects, properties, and relations compose in logical structures to form states of affairs. To call a state of affairs a “fact” is to represent it as actual rather than merely possible, while to call it a “proposition” is to represent it merely as possible, or “proposed”. Sentences in different grammatical moods (declarative, interrogative, imperative, etc.) are used by semantic convention to advance propositions with different illocutionary forces, such as to assert, enquire, prescribe, and so forth. 

Paradox of Analysis. Ordinary competence is a form of tacit knowledge or know-how: a disposition or ability to follow a rule R that functions unconsciously and typically doesn’t rise to the level of explicit knowledge or theory.

We may not be aware even on reflection that we apply an end-relational logic so propose a less subjective method of proof: to demonstrate that the end-relational theory provides the best explanation of our linguistic intuitions. 

Parfit: “normativity is either an illusion, or involves irreducibly normative facts”. Primitivist’s presupposition of the reductive project is that the semantic function of normative language is descriptive.

Normative language is indeed characteristically used for nondescriptive purposes,

Semantic externalism. Putnam: “meaning ain’t in the head.” Russell: “While my own opinions as to ethics do not satisfy me, other people’s satisfy me still less”.

End relational theory has simplicity and conservatism with regard to multiple dimensions—metaphysics, logic, psychology, epistemology, ethics, and especially linguistics—and vindicating our general confidence in our ability to have normative and moral knowledge.

This may all sound too good to be true. How could a maximally simple and conservative account be correct, when the puzzles of metaethics have resisted resolution.  The solutions have been hidden from view by the richness of the pragmatics of normative language. 

Close relatives of the end-relational theory, Ziff 1960, Harman 1975, 1996, Mackie 1977, Wong 1984, Railton 1986, and Copp 1997.

2.  A Good Word to Start With

Our most general normative adjective, it expresses the simplest normative concept. Whereas denials of its analyzability have largely been motivated by features exemplified by peculiarly moral uses, ‘good’ is used all the time in mundane ways. These ordinary uses don’t so plausibly attribute a primitive, unanalyzable property or merely express an attitude. Even many who are skeptical of analytic reduction of the moral ‘good’ happily allow it has another ordinary meaning susceptible to analysis. 

A common logical form underlying the diverse grammar. A unifying semantics that reductively analyzes the conceptual meaning of ‘good’ in terms of comparisons of probability.

Conceptual competence enables us to apply the word good in these different ways.

Indexical words like ‘I.’ But ‘good’ is a logically incomplete words like ‘tall’. Relational properties. Qualified with adjunct propositional phrases.

Gradable adjective.  ‘Good’ is a member of this family of incomplete predicates, various kinds of goodness are relational properties. Implications for the OQA, providing a first illustration of the pragmatic origin of metaethical puzzles.

Two different kinds of inverse, ‘bad’ ‘poor’.  Our default hypothesis should be that ‘good’ has just one ordinary, flexible meaning as an incomplete predicate.

Nine different grammatical forms.

Explications.  Chocolate is good [for people][to eat.]  ‘n is good for s to φ.’ 

Subject movement: the object of the verb ‘to φ’ moves up from its usual position after the verb to the subject position before the main verb, ‘is’.

Paul Ziff:  Patient-relativity.  ‘Good’ sentences are fundamentally end relational, and patient- and interest-relativity are merely special cases of this.

‘it is good for e if p’, = ‘it is good for sp ’s φe -ing, for sa to φ m n.’

‘Good K’, grammatically “attributive” adjective, attached to a noun phrase ‘K’ to form a complex predicate ‘good K’ rather than a “predicative” adjective,

Geach: ‘good’ has a unified semantics as logically attributive.  Every meaningful use of ‘good’ must be understood as completed by some intended noun phrase. Are for their “telos” or function.  ‘Person’ and ‘human being’ don’t seem to be functional nouns.

While functionalist attributivism only works for a limited range of data, the end-relational semantics isn’t similarly restricted.

Good for e is approximately being promotive of e, p increases the probability that e.

‘It’s good for e if p’ means that p increases the probability of e. A modal comparison, between the actual probability of e and some counterfactual probability of e,

greater than it otherwise would be. Greater if p than if not-p,

‘it’s good for e if p’ means that pr(e|p) > pr(e|~p).

 ‘It’s good for e if p, [given b]’ means that pr(e|p & b) > pr(e|~p & b). 

Some kind of balancing of the various relevant ways of not-p;

Natural equilibrium point of value neutrality.  Gradability.  It’s good if something exists.  Just in case it’s better that it exists.

Objective probability (or “chance”) v subjective probability (or “rational credence”).  Objective probability is based on the state of the world at a time, while subjective probability is based on (and so relative to) sets of evidence or beliefs about the state of the world.  Possibility-space.

Are the fundamental bearers of value ordinary objects or states of affairs?  Good-making properties. The greater value of p over q is entirely due to properties of the hammer. “Goodness”

3.  The Probable Meaning of ‘Ought’

Whereas ‘good’ is our most general normative adjective, ‘ought’ is our most general normative verb.

Denying the analysability of ‘ought’: “no definition can break out of the normative circle, a circle of ought-like terms.” “Hume’s Law”

So-called categorical imperatives like ‘Children ought to respect their elders.’

Kant, Wittgenstein, and J. L. Mackie all accepted that instrumental ‘ought’ sentences express “natural” facts but thought moral or categorical uses of ‘ought’ have an irreducibly “nonnatural” meaning. 

Detaching problems and four unsatisfactory answers.

Modal verbs can be analysed as quantifying over possibilities

All possibilities: must, have to.  Strong necessity

In some possibilities: may, ought, should. Weak necessity.

Had to normative. has to have epistemic

‘if you want biscuits, [then this will be relevant information:] there are some on the table.’

Synonym ‘should’ has the same ambiguity between normative and non-normative senses. It should rain.

Can be relativized to any background,

The means that ought to be taken relative to an end e is just that which most increases the probability of the end, or the most reliable means. ‘Ought’ is equivalent to ‘best’.

“Pure” instrumental conditionals, concerning a single end or d

Kratzer, ordering source,

Distinction between ‘must’ and ‘ought’ as that between only and best. 

The “hypothetical imperative”…prescribes the fittest means to any end that we have determined to aim at.

Single reading is both end-relational and normative.

Assuming an implicit ‘in order that…’ ‘[in order to make people laugh,] jokes ought to be funny’

A “control” verb, expressing a normative relation O(s, φ) between an agent and an action, akin to having most reason

Ought-to-do is just a special case of ought-to-be.

Deliberative set

4.  Explaining Reasons

The relation of favoring, that is allegedly primitive. 

‘Explanation’ means roughly information that reveals.

A reason and more reason. Count noun.

More Reason: ‘There’s more reason to φ than to ψ’ means there’s an explanation why it’s more good (better) to φ than to ψ. 

The singular noun is explained as concerning a single (though possibly complex) explanation,

Something is good in some way and to some degree just in case it has some reason in its favour, and is best and therefore also what one ought to do just in case it has most reason in its favour.

‘good’ and ‘ought’ express more fundamental normative concepts.

“holism of reasons”: part of the reason.

Explain (pragmatically) the prevalence of both Humean and non-Humean intuitions here. 

Talk about explanations that isn’t explicitly relativized to a subject can reasonably be presumed to assume a factual background, which is a step toward explaining why ‘r is a reason, for s to φ’ strongly prefers a more objective reading. 

Belief aims at truth. Referring to belief generally makes salient an end of truth. Aims approximately at believing that p if and only if p is true.

Reason for s to believe that p’

Explanation for s of why it increases the probability that s thereby believes that p iff p is true, if s believes that p.

‘That the prison fence is electrified is a reason for Arthur to believe that the escape plan will fail’.  The fact of the fence’s being electrified is an explanation for Arthur, relative to a background b, why his believing the plan will fail increases the probability that he thereby believes the plan will fail iff in fact the plan will fail.

Normative reasons are roughly things that agents are supposed to take into consideration in deliberating. When all is going well, therefore, agents are motivated as a result of being aware of their normative reasons.

Perverse reasons aren’t problematic at all given the end-relational semantics for ‘good’.  Good for some perverse end (“evil, be thou my good!”) Inverted commas: supposed or alleged reason.

5.  Pragmatics and Practicality

My case for the end-relational semantics is that it’s a maximally simple and unifying theory that accommodates the truth conditions of a wide variety of ordinary sentences.  But so far I’ve ignored apparently inconvenient evidence from particularly the moral and deliberative uses of this language.

Practicality—the especially close connection with motivation which seems to defy all descriptivist and especially reductive analysis.

Our semantics accommodates and explains them all by appeal to the pragmatics of how we use normative language in context, pursuing our desired ends, yielding what I will call “quasi-expressivist” and “quasi-absolutist” solutions.  End-relational theory predicts these features and provides the best explanation of our practices of normative speech and thought.

Semantics and pragmatics now widely accepted. Against Wittgensteinian dictum that meaning is use. Distinction between sentence meaning and speaker meaning. Eg sarcastic: ‘That was really smart.’ Meaning is conventional use. Whereas “semantic information is encoded in the sentence, pragmatic information is generated…by the act of uttering the sentence.”  Semantics concerns the information or signification associated by convention with words and sentences themselves, while pragmatics concerns the information or signification generated by the fact that a speaker utters a particular sentence in a particular context.  “Context” includes common ground.

A system of communication that can be factored into distinct semantic and pragmatic inputs is simpler than one that assigns all the variability to the semantics itself, and is therefore easier for people to learn, use, and interpret. Rather than learning a set of complicated, contextually variant rules for each word in our vocabulary, we need only learn simpler, general rules for each word, and combine this semantic competence with our comprehension of the systematic ways in which meaning interacts with context. 

Since Paul Grice in the 1960s, the distinction between semantics and pragmatics has proven highly useful for solving philosophical puzzles.  The supposedly problematic behaviour of normative words in certain uses is systematic and amenable to a principled pragmatic explanation.  Pragmatically indicated, pragmatically expresses.  Pragmatics are ubiquitous, playing a role in any successful communication whatsoever.

Even for a completely explicit and unambiguous sentence that semantically determines a single proposition p, an audience relies on the speaker being literal, sincere and reliable.  Ellipsis. Communicate much more than they say or assert.  What is said v. what is communicated by the saying of it.  Introspection finds similar patterns of ellipsis in thought as in speech.

Speaking is an instance of motivated behaviour.  “Each use of language has an ulterior purpose.” Speech is a means.  Instrumental Principle of Conversation: speak in a way that is best for “conversational end”

Instrumental Law of Pragmatics: Speakers always speak in the way they believe best for their conversational ends. ‘Five is a bigger number than three.’ You would naturally be bewildered. ‘What’s your point?’  Grice suggests “Cooperative Principle” but not every speaker or utterance aims at cooperation, or shared conversational ends.

Practicality. Expressivists. Downfall of all purely descriptivist semantics. Primitivists. 

Motivational Internalism.  Motivation or motivational attitude, like a desire or intention. (Interpersonal Influence).  “Put pressure on choice and action” (Practical Illocutionary Force). Endorse, recommend or prescribe. Imperatival force. 

The simple and conservative Humean psychological model of motivation draws a fundamental distinction between beliefs, as attitudes that aim at reflecting the world, and desires, as attitudes that aim at shaping it.

“Separate existences.”  Normative language as merely normally motivating. Normative facts are simply those that necessarily ought to motivate us, regardless of whether they actually do.

The strong practicality requirement is generally motivated by narrow focus on particular uses of words like ‘good’, ‘ought’, and ‘reason’, which overlooked or ignored a wide range of other ordinary uses

‘Ought’ and ‘reason’ have entirely non-normative uses, which clearly lack the features of practicality.

Simpler, unifying semantics identifying a common meaning across these uses, in which case these words can’t themselves be tied by semantic conventions to motivation or practical force.

The features of practicality are also only generally evident for the use of unrelativized normative sentences, and generally disappear when words like ‘good’, ‘ought’, and ‘reason’ are explicitly relativized.

Adultery is good for destroying marriages’ ‘Adultery is good.’a “good liar”

To say that a tree has “good roots” is to express the intention to have roots like that for the hypothetical circumstance of being a tree.

The relativized uses of normative words are (in Wittgenstein’s words) “not how ethics uses them”. 

Reluctant to embrace such a radically disunified semantics, 

The embedded or nonassertoric use of normative sentences that prompts the original charge of pragmatic fallacy against expressivism.

Embedded, nonassertoric uses of normative sentences appear to be counterexamples to a strong practicality requirement. 

Motivational externalism. ‘you ought to φ, of course. But don’t do that! Come ψ with me instead—it’s much more fun!’ In inverted commas to mean approximately what other people call “good”  2009 survey by Philpapers.org finds 35% of philosophers accepting internalism and 30% accepting externalism (about specifically moral judgments)

For anybody who desires that e, facts and properties involving what increases the probability of e will have a practical role.  “probability is the very guide of life”,

‘This hurts, and it’s good’, suggesting a perverse and unusual liking of pain rather than a disliking of goodness.

Remarkably close connection between the unrelativized use of ‘good’ and positive motivation

The mere use of unrelativized normative sentences is poor evidence of an absolutist semantics, in view of the ubiquity of ellipsis.

These intuitions are really due to our pragmatic competence responding to the paucity of contextual information. 

Because we haven’t been given any other clue, we naturally assume that the implicit end is of the kind salient by default: an end desired by the speaker of the utterance.

The end-relational theory accommodates many of expressivism’s central claims, despite being a purely descriptive semantics. By asserting unrelativized normative sentences, speakers do typically express their motivational attitudes without describing them, and perform practical speech acts like endorsements and recommendations. The theory achieves these results while rejecting expressivism’s essential claim that this is an element of the semantic conventions for this language, attributing it rather to the pragmatics of omitting the relevant end.

“quasi-expressivist” explain the seemingly expressivist features of normative speech and thought with purely descriptivist resources. Preserves a unified semantics based on other, ordinary uses of these words.  Our theory accounts for practicality with just conservative, Humean psychological resources, explains the special action-guiding character of normative facts.

(pure) expressivism. an uncompromising rejection of the most popular horn of the Euthyphro dilemma: that we desire or are motivated toward things because we judge them to be good.

Gibbard’s plans. Parfit. When we conclude that we ought to do something, we are not deciding to do this thing, but coming to have a normative belief. Though our decisions to act are often based on such beliefs, these decisions are not the same thing as coming to have these beliefs.

The implausibly radical idea that normative thought is simply a matter of arbitrary choice. 

Although our theory agrees that deliberation is ultimately grounded in or motivated by contingent desires, it identifies the content of normative thought with facts and properties, which justify the burdens of deliberation and function to guide our decisions and cause corresponding motivation. 

We generally choose and are motivated toward things because we believe them to be good or what we ought to choose; as Butler observed, probability is the very guide of life.

The end-relational theory therefore enjoys the intuitive advantages claimed by primitivists with regard to the deliberative role of normative concepts.  The common-sense view that our practice of giving normative advice (in circumstances of cooperation) generally treats others as autonomous, “rational” agents, seeking to influence them by sharing information they can recognize for themselves as justifying action, without any psychological manipulation.

Because our account of practicality is pragmatic rather than semantic, unlike expressivism it is able to explain why the use of normative language is practical “to just the extent that it is”.

6.  Multiple Ends

Understood sufficiently well. Sometimes an end is difficult to describe. A speaker may just be confident that some relevantly desired ends are at stake, without having any particularly in mind. Reasons for asserting unrelativized normative sentences with obscure ends. 

Best not merely in relation to some desired end, but “all things considered”.

Allegedly, if ‘ought’, ‘good’, and ‘ reason’ were always relativized to single ends, they would be unable to fulfill their essential functions in practical speech and thought,

Extension of the Humean psychological model is that desires aren’t all equal: they vary in strength, generating preferences

It’s precisely the omission of the end that pragmatically communicates the essential information that the speaker believes the end to be relevantly p referred, and thereby generates the recommendatory force.

Decision theory. Maximizes expected utility. Expected preference-satisfaction.  Contextual Preference: ends they are most motivated toward after intrinsic desire is balanced against credence. Total outcome the agent is most motivated to pursue.

Not only does a probabilistic semantics appear redundant once probability is assigned this pragmatic role of selecting ends, but it may interfere with its proper functioning. 

The fluidity of pragmatics. Conversational salience can shift rapidly in conversation.

Although the end-relational semantics relativizes normative language to individual ends, on the basis of our pragmatic principles it accommodates the various ways that normative speech and thought are sensitive to multiple ends. 

The end-relational theory may even accommodate more data than rival theories, as it offers pragmatic solutions to several general puzzles.

7. Categorical and Final

If we “give up on absolute moral facts,” we’ll be left “without any normative vocabulary whatsoever,”

This assumption has been incorporated in the basic, conventional, meanings of moral terms.

Three main absolutist challenges to relational theories can be distinguished: from Categoricity, Finality, and Disagreement.

Pragmatic, “quasi-absolutist” solutions

“the categorical imperative would be one which represented an action as objectively necessary in itself, without reference to another end”

Grammatical definition of a categorical use as one without explicit relativization to an end.

‘Still, you ought to stop.’ 

Dora’s claim can be interpreted as about what Frank ought to do in order to preserve his neighbor’s house.

‘England expects that every man will do his duty’ reflects the rhetorical device we know as “moralism”

speaking as if the prescribed proposition were factual

Although there are many moralities of many different societies, and important divergences also within societies, to speak about morality (unrelativized and in the singular) is to speak as if there were only one salient—and so uncontroversial—morality or set of preferences. This can now be recognized as itself employing the rhetorical device of moralism, pragmatically expressing endorsement of that morality and the moral expectation that one’s audience shares these attitudes.

Lacking a sophisticated theoretical understanding of our implicit pragmatic competence, our own quasi-absolutist moral practice is most naturally interpreted upon reflection as genuinely absolutist, making reference to objectively prescriptive facts. 31 our simple, unifying semantics therefore accommodates and explains 

This characteristic feature of moral speech and thought, perhaps even better than either primitivism or expressivism can; without postulating any lexical ambiguities or special moral meanings, countenancing mysteriously prescriptive facts or properties, or ascribing systematic errors to ordinary moral p

for p to be good “for its own sake” is simply for p to be good for p

While i believe there is such a reading of “good for its own sake” which trivially applies to everything, this isn’t the usual sense that interests moral philosophers, which applies only to a few things, nontrivially. 

People apparently find ‘bad for its own sake’ to be an intuitively peculiar construction, while having no such qualms about ‘intrinsically bad’

Quasi-absolutist account of final value

8. A Disagreeable Problem

It is a matter of long-standing controversy whether deliberation or first-personal practical reasoning is just a special case of theoretical reasoning aimed at reaching true belief, concerning a distinctively practical subject, or an irreducibly practical kind of reasoning aimed at reaching an intention directly rather than a belief.

An agent therefore has no better means from her own point of view for promoting her end through deliberation than by identifying what she ought to do relative to the fullest information available.

The end-relational theory supports a pragmatic, quasi-expressivist account of normative disagreement that accommodates intuitions across a wide range of different contexts involving differing information.

9.  Conclusion

My primary case has been that the end-relational theory is a maximally simple and conservative explanation across all relevant dimensions as a whole: linguistically, metaphysically, psychologically, epistemologically, and ethically. I’ll conclude by briefly addressing its commitments and implications for each of these dimensions. 

The theory is also linguistically conservative, especially in comparison to expressivism, which looks like an attempt to fit a square peg into a round hole. 

Normative sentences are naturally evaluated as “true” or “false” like all uncontroversially descriptive sentences, and unlike uncontroversial imperatives (e.g., ‘Do what I say!’) and pure expressives or performatives like ‘hello’ and ‘ouch’. Normative language is descriptive on its face.

Expressivists typically offer one principal reason for rejecting pure descriptivism: that only an expressivist semantics can accommodate the practicality of this language. But the end-relational semantics systematically accommodates and explains this practicality pragmatically.

The pragmatic fallacy of confusing what these words are commonly used to do with what they mean.

‘Normative’ is a philosopher’s term of art of recent origin.

In deliberation agents don’t generally attend to their desires and preferences, but rather attend from them, to relevant end-relational information. Under the end-relational theory no properties or facts are normative per se (or absolutely), but only relative to agents or motivated perspectives.

An important element of truth in the objectivist answer to the Euthyphro Question that gives normativity priority over desire: we are indeed motivated toward things because we judge them to be good, what we ought/have most reason to do, etc. The deeper truth in the subjectivist answer that gives desire priority over normativity: we judge things to be good, simpliciter, because of what we more fundamentally desire. It supports the priority of desire.

Normative speech and thought therefore appear to be just manifestations of our human, all too human nature as intelligent animals pursuing the varying and incompatible ends we contingently happen to care about.

End-relational semantics are broadly consistent with alternative psychological models that would give priority to normativity. Alternative view: to desire that p in the relevant sense isn’t simply to have  a noncognitive attitude or motivation toward p, but to either judge or perceive  that p has some more basic normative property and be motivated toward it on  that basis. It is hard to distinguish between instrumental ends and intrinsic ends.

Concept of importance, or of what matters. Relativization to ends that are important, rather than to those that are merely desired or preferred.  To make either a moral or an all-things-considered normative claim, like ‘one ought always to maximize happiness’, is fundamentally to express a contingent and subjective preference. Claims of basic moral principles are little more than coercive expressions of preference. The end-relational theory justifies skepticism toward normative ethical theory about “first principles”. 

The puzzles of metaethics are largely the result of philosophers’ failure to understand these pragmatics, a metaethical “confusion of tongues”.  Concepts are often opaque and their analysis can be difficult, so the simpler and more plausible hypothesis is that some people just have mistaken theories about the meanings of their own words and the nature of their own practices of normative speech and thought.