2. The shape of Hobbes’s theory

I begin by registering three important constraints on any interpretation of Hobbes’s theory. Although Hobbes sets out the same basic account of evaluative language in all the relevant works (including Elements of Law, Anti-White, De Cive, and De Homine), to help fix ideas, I frame my remarks around the following well-known passage from Leviathan:

As Hobbes sees it, in our actual practice, we apply the predicate term ‘good’ to the objects of our appetites or desires, and ‘evil’ to the objects of our aversions (compare also EL vii.3; DCi xiv.17; AW xxx.24, xxxviii.5, xxxix.4). Whatever the official ideology of good and evil promulgated in the universities and pulpits, this is in fact how ordinary speakers use evaluative terms, and hence how such expressions get their meaning.Footnote ³ Moreover, since the appetites and aversions of one person differ from those of another, and since each individual’s appetites and aversions change over time, ‘good’ and ‘evil’ will be applied to different objects on different occasions of use (L vi.6, 80; L xv.40, 242; see also EL vii.3; DCi iii.31; DCi xiv.17; DH xiii.8). So as we look to their actual employment, we find that ‘good’ and ‘evil’ are used in a context- and speaker-relative fashion – that ‘good’ is in this sense an indexical term, ‘relative to person, place, and time’ (DH xi.4).Footnote ⁴

When Hobbes draws the corollary that nothing is ‘simply and absolutely’ good, he is repudiating the notion of an objective or intrinsic value independent of human appetites and aversions (L vi.7, 80–82; see also EL vii.3). Commentators have occasionally claimed to find an objective standard of value in Hobbes’s works – one that regards peace, the security of the commonwealth, or individual self-preservation as an objective good, whether or not we or the sovereign happen to desire any such thing (Gert Reference Gert, Hobbes and Gert1991, 13–16; Arp Reference Arp1999, 385).Footnote ⁵ But, in fact, the texts that might appear to endorse an objective or non-relational standard of value can all be accounted for as instances of the most important case of indexical and relational usage, as Hobbes himself makes plain in De Homine (DH xi.4), and, most vividly, in Anti-White:

Two other points about the basic structure of Hobbes’s theory will prove relevant. First, when Hobbes writes in L vi.7, 80–82 that the ‘words of Good [and] Evill […] are ever used with relation to the person that useth them, ’ he is often interpreted as suggesting that only the speaker’s appetites can set the relevant indexical context for the use of an evaluative term. But that would be a bizarre view. Can I not sensibly say that lemon tea would be good for my friend, but not perhaps for me? And, in fact, Hobbes recognizes that a speaker can also use evaluative terms in ways that shift the relevant indexical context to another person’s appetites. As we just saw, he remarks that in ordinary English usage, ‘[p]eople say […] that such and such is good or evil for themselves, for this person or that, or for the body politic’ (AW xxx.24, my emphasis). Or again: ‘what is it to say an action is good, but to say, it is as I would wish, or as another would have it, or according to the will of the state, that is to say, according to law?’ (OLN 14, my emphasis; compare also L xxv.3). So we can speak of something as good for someone else, perhaps placing ourselves in their shoes and imagining what would count as good from the perspective of someone with their appetites. This is a perfectly ordinary use of ‘good’ that Hobbes certainly allows.

Second, in L vi.7, 80–82, Hobbes states that a new ‘Rule of Good and Evill’ controlling the proper use of these terms takes effect in the transition from the state of nature to a commonwealth. Where there is no commonwealth, the appetites of private individuals govern the usage of evaluative expressions. ‘[I]n the condition of meer Nature, ’ as Hobbes puts it elsewhere, ‘private Appetite is the measure of Good, and Evill’ (L xv.40, 242). But once we are in a commonwealth, the appetites of the state (as represented by the sovereign) trump such private appetites in providing the standard for correct usage. At least on those questions of value where the sovereign has expressed an official opinion through law, the standards of private appetite are displaced by a new rule of good and evil: ‘Not the Appetite of Private men, but the Law, which is the Will and Appetite of the State, is the measure’ (L xlvi.32, 1090). Notice: Hobbes is not simply saying that we ought to defer to the will of the state, right or wrong, whenever there is controversy over the correct use of evaluative terms, but that the will of the state actually determines what the correct usage is – a point that commentators have not always appreciated, but which is corroborated by the various texts where Hobbes straightforwardly identifies good and evil in a commonwealth with the mandates and prohibitions of the law (L xlvi.11, 1058–1060; L xx.16, 316; L xxix.6, 502; DCi preface para. 8; DCi xxii.1; OLN 14). This is not something that he ever says of non-evaluative terms.Footnote ⁶ Interpretations of Hobbes on evaluative speech have often focused simply on the use indexed to private appetite, as if that were the only employment of these terms, or at any rate, the most fundamental. But any adequate interpretation must also be able to explain the bifurcated character of Hobbes’s theory, and accommodate those cases where the rule of good and evil is fixed not by private appetite but by the civil law.Footnote ⁷

3. The reductionist interpretation

According to the reductionist interpretation, Hobbes is claiming that evaluative judgments are equivalent to claims about human appetites or aversions, and that our evaluative interpretations of the world can be fully analyzed – reduced, reinterpreted – in these psychological terms. At first glance, values might seem difficult to fit into Hobbes’s strictly materialist picture of the world, but the reductive analysis would reveal that judgments about value are simply another way of putting claims about human desires and aversions (which are themselves simply so many material motions or impulses in the human body). In the simplest sort of case, an utterance of ‘X is good’ might perhaps be translated as ‘I desire X, ’ and ‘X is evil’ as ‘I am averse to X.’ But the reductive analysis could also be extended to accommodate cases where we shift the indexical context beyond the speaker’s own appetites and aversions. Depending on the context of use, a given utterance of ‘X is good’ might be translated as ‘You desire X, ’ or ‘She desires X, ’ or ‘The commonwealth as represented by the sovereign’s legal enactments desires X, ’ and so on.

There might seem to be decisive textual evidence in favor of the reductionist reading. Consider Hobbes’s blunt declaration that ‘Good, and Evill, are names that signifie our Appetites, and Aversions’ (L xv.40, 242). If Hobbes is using ‘signifie’ to mean ‘refer to, ’ then this passage and others like it (L iv.24, 62; DCi iii.31) tell us that we employ the evaluative expressions ‘good’ and ‘evil’ to refer to our appetites and aversions; and thus that when I say that ‘coffee is good, ’ in the simplest kind of case where I am speaking from the perspective of my own appetites, I will be saying something to the effect of ‘I have an appetite for coffee.’ Some version of the reductionist interpretation might then seem inescapable.

But this textual argument is a paper tiger. Hobbes’s relation of signification is not our semantic relation of reference, and a word might ‘signify’ something in his sense without thereby referring to it. In fact, Hobbes explicitly defines ‘sign’ and ‘signify’ simply in terms of a relation of reliable correlation between two kinds of phenomena, the sign and the signified, ‘as clouds are a sign of rain to come, and rain of clouds past’ (EL iv.9; see also AW xxx.13; DCo ii.2; L iii.8, 44). So clouds can signify rain; facial expressions and body language can signify one’s passions (L vi.56, 94); customary obedience to a person can signify his or her power (L xxxi.9–10, 562); and declarations of God’s greatness can signify our own pious reverence (EL xi.3; L xxxi.33, 568): all cases where Hobbes’s relation of signification cannot be understood as any sort of semantic relation of reference. Humans are able to intentionally introduce ‘conventional’ signs, including words, that through habitual use and regular correlation come to signify our thoughts and passions (DCo ii.2; see also L iv.3, 50). But Hobbes never says or implies that we can only understand such words as referring to the thoughts and passions that they signify. So when Hobbes says that ‘Good, and Evill, are names that signifie our Appetites, and Aversions, ’ he is saying that we use these words in reliable correlation with our appetites and aversions, and thus that our use of these words tends to display these passions. If I speak sincerely, then an observer should be able to infer my appetites and aversions from my evaluative pronouncements. That is something that each of our rival interpretations can easily accommodate. However, there is no need to suppose that Hobbes endorses the further claim that the expressions ‘good’ and ‘evil’ refer to our appetites and aversions any more than we need suppose that he holds that expressions such as ‘Hurrah!’ and ‘Get me a coffee!’ refer to the appetites and aversions that they might typically ‘signify’ – i.e. evince or display.Footnote ⁸

The problem with the reductionist interpretation is that it saddles Hobbes with an account of evaluative judgment that is both implausible in its own right and in conflict with his own wider systematic commitments. First, the account is unable to explain what people are disagreeing about when one asserts and another denies that something is good, at least in the case where the two speakers are using ‘good’ in a manner indexed to their own respective private appetites. According to the reductionist interpretation, the one speaker is simply declaring his own attraction toward that object, and the other his own indifference or aversion. Understood as reports or avowals of the speakers’ respective psychological attitudes, the two evaluative judgments are not at odds with one another: there is no disagreement. But this is implausible, and would be a particular problem for Hobbes, given that he holds that disagreements between private judgments about good and evil are not only possible, but a major source of the ‘Disputes, Controversies, and at last War’ that roils the state of nature (L xv.40, 242).

Second, according to the reductionist interpretation, Hobbes holds that evaluative judgments can be fully analyzed into statements about human desires and aversions. But as with any theory that attempts to reinterpret evaluative language in non-evaluative terms, the proposed analysis would seem to leave out the all-important normative force of the original evaluative expressions altogether. The claim that an object is worthy of desire is one thing; the claim that I desire that object seems to be another. And in calling something good, I am surely making the former claim, which carries with it an impersonal commendatory force that is missing from a simple factual report of my own desires (Peters Reference Peters1966, 152; Darwall Reference Darwall2000, 315–316; Martinich Reference Martinich2005, 58).

Third, the reductionist interpretation generates absurdities when we consider the bifurcated character of Hobbes’s theory. Recall that Hobbes characterizes the proper use of the terms ‘good’ and ‘evil’ disjunctively. When one is in the state of nature or when one is addressing an issue on which the civil law is silent, the indexical context for the use of an evaluative term will be the appetites or aversions of some particular private individual. On the other hand, where the civil law speaks, the indexical context for the use of evaluative terms switches to the appetites of the state as encoded in the law. So if the reductionist interpretation is to be adequate to Hobbes’s bifurcated theory, it seems that it must propose a similarly bifurcated reductive analysis: where the law is silent, an evaluative judgment will be equivalent to a proposition about private appetites; but once the law has spoken, an evaluative judgment will be equivalent to a proposition about the appetites of the state. In this way, evaluative expressions and judgments shed their old semantic content for new whenever the sovereign passes a relevant ruling in the civil law.Footnote ⁹ But this bifurcated semantics is quite incredible. It means, for instance, that seditious subjects who continue to use terms such as ‘good’ and ‘evil’ in ways that conflict with new laws are making not merely a political but a linguistic mistake. They are now simply misusing the relevant terms, and whenever a private individual states that a certain law mandates evil practices, their utterance will be oxymoronic, a contradiction in terms. The sheer implausibility of this result ought to give us pause, and it is worth noting that, while Hobbes is firm that there is something improper about subjects using evaluative terms in ways that conflict with the civil laws, he never says that such utterances involve linguistic errors or verbal contradictions, the sort of ‘Absurd, ’ ‘Insignificant, ’ and ‘Non-sense’ speech that he delights in exposing in expressions such as ‘incorporeall substance’ and ‘free-Will’ (L iv.21, 60; L v.5, 68). And, in fact, Hobbes cannot hold that a private individual’s evaluative pronouncements contradicting the official views of good and evil expressed through the laws are always absurd or nonsensical, for while he is clear that all laws are just simply in virtue of being laws, he is equally clear that they are not thereby also all good; and indeed, he himself explicitly condemns certain kinds of laws as ‘not good.’Footnote ¹⁰

4. The error theorist interpretation

According to the error theorist interpretation, Hobbes holds that although our evaluative judgments are in fact simply an expression of our subjective appetites and aversions, those judgments involve the mistaken attribution of an objective value to the things that we desire and an objective disvalue to the things to which we averse. On this reading, Hobbes is claiming that our value-laden interpretations of the world are involved in systematic error, and that all of our evaluative judgments are, strictly speaking, false. The objective and impersonal value or disvalue that we ascribe to things in characterizing them as good or evil is an illusion, if perhaps a natural or even psychologically inescapable one.

The error theorist interpretation plainly avoids the first two problems facing the reductionist reading. It can explain what two people are disagreeing about when the one asserts and the other denies that something is good: the one is claiming that that thing has an intrinsic value and is objectively worthy of desire; the other is denying this. And the error theorist interpretation also avoids saddling Hobbes with the implausible view that our evaluative expressions can be translated without loss into non-evaluative terms; instead, it credits Hobbes with the view that words such as ‘good’ and ‘evil’ involve an irreducible reference to an objective value or disvalue, albeit an objective value or disvalue that does not in fact exist. The error theorist interpretation handles the third (‘bifurcation’) problem rather less well, however. It is difficult to believe that evaluative terms are supposed to capture an objective and appetite-independent feature of the things that we call good and evil while also believing that the meaning of these terms is systematically bifurcated, governed in one context by a ‘Rule of Good and Evill’ reflecting private appetite and in another by a rule reflecting the appetites of the state – and indeed to believe not merely that we ought to defer to the state’s judgment regarding the proper usage of ‘good’ and ‘evil, ’ but that the state’s judgment actually determines what the proper usage is. This sort of systematically bifurcated ‘[r]ule’ or ‘measure’ of good and evil seems to wear its subjective and appetite-relative character on its sleeve.Footnote ¹¹

In any case, the texts simply do not support the view that Hobbes accepted an error theory regarding our practice of evaluative judgment. First, he never says that our general practice of evaluative judgment is involved in systematic error, nor does he ever say that we take ourselves to be describing some objective feature of reality when we make evaluative judgments. Second, Hobbes seems to regard at least some of our evaluative judgments as non-erroneous – perhaps most obviously, where they align with the laws of the commonwealth. Third, Hobbes holds not only that our use of evaluative terms ‘signifies’ – i.e. displays – our subjective appetites and aversions, but also that if we look to our everyday use of such expressions, we find that, at least some of the time, ordinary language users understand this fact quite well. ‘One must recognize, ’ he says, ‘that good and evil are names imposed on things to signify desire for or aversion from the thing so named’ (DCi iii.31, last emphasis mine). We can use these words precisely in order to put our appetites and aversions on display, and must therefore expect our audience to understand that these terms are being employed in a subjective and appetite-relative fashion, not as an attempt to characterize some objective and impersonal reality. Other texts confirm that we often employ these terms in a self-consciously subjective and appetite-relative way, both when we speak from the perspective of our own appetites, and when we speak on another’s behalf. ‘[W]hat is it to say an action is good, but to say, it is as I would wish, or as another would have it, or according to the will of the state, that is to say, according to law?’ (OLN 14, my emphases; see also AW xxx.24) ‘The common name for all things that are desired, insofar as they are desired, is good’ (DH xi.4, first emphasis mine). These passages are difficult to reconcile with any reading of Hobbes as holding that our practice of evaluative judgment presupposes an objective, impersonal, and appetite-independent system of value. And they seem to have been simply ignored by proponents of the error theorist reading.

What then is the positive case that advocates of the error theorist reading put for their interpretation? Tuck (Reference Tuck1989, 52–55) suggests that there is a telling structural analogy between Hobbes’s treatment of color on the one hand and his treatment of good and evil on the other. Darwall (Reference Darwall2000, 323–328) agrees, and notes that Hobbes himself explicitly registers a parallel between his account of color experience and his account of ‘the apparence, or sense of Good’ (L vi.11, 82). Given that Hobbes is plainly some sort of error theorist about color, Tuck and Darwall take the analogous treatment of the two cases to intimate some sort of error theory about value as well. First, consider the case of color. As Hobbes sees it, our everyday pre-philosophical talk and thought about color systematically mistake the subjective color sensations of our own visual field for an objective and mind-independent feature of external bodies (EL ii.9–10; L i.4, 22–24). Furthermore, in his analysis of color vision, Hobbes also draws a distinction between the underlying material basis of our color experience – a motion produced in the optic nerve and brain as a result of pressure on the retina (EL ii.7; EL ii.8; L i.4, 22–24; DCo xxv.2) – and the color experience itself, an ‘apparition’ or ‘phantasm’ that is how this material motion manifests itself in first-person consciousness (EL ii.4; DCo xxv.2). And very shortly after the notorious paragraph on good and evil L vi.7, 80–82, Hobbes goes out of his way to register the following correspondence between his treatment of color on the one hand and his treatment of good and evil on the other:

Moreover, since we apply the terms ‘good’ and ‘evil’ to the objects of our appetites and aversions, and since delight and trouble of mind – i.e. pleasure and displeasure – are how our appetites and aversions manifest themselves in first-person consciousness, Footnote ¹² Hobbes can write that

So we have the following parallel. As sensations of color are how certain material motions in the brain are experienced in first-person consciousness, sensations of pleasure and pain are how certain material motions around the heart are experienced in first-person consciousness. And as the former gives the appearance or sense as of an objective and mind-independent feature of external bodies – i.e. the color we falsely attribute to bodies in our everyday pre-philosophical judgments – the latter gives ‘the apparence, or sense of Good’ and ‘the apparence, or sense of Evill.’ For Tuck and Darwall, this structural similarity between the two cases shows that Hobbes is likely an error theorist about value as well as an error theorist about color. Hobbes holds that our natural, pre-philosophical judgments about color involve ‘the illusory projection’ of our own subjective states onto the objective world, and ‘it seems reasonable to interpret [Hobbes] as saying that value is in this respect just like color’ (Darwall Reference Darwall2000, 327–328; compare also Tuck Reference Tuck1989, 53, 55).

There is less to this analogy than it might appear, however. First, although Hobbes notes the parallel between color and value here at L vi.9, 82 and L vi.11, 82, he does not mention it in any other place, including the various texts before and after Leviathan in which he sets out the same basic account of evaluative language as a systematic expression of human appetites and aversions (EL vii.3; AW xxx.24; DCi iii.31; OLN 14; DH xi.4). It would be surprising if Hobbes saw the practice of evaluative judgment as presupposing a false belief in the objectivity of value, but neither thought fit to say this explicitly in any work nor to intimate it by way of an analogy in any text but Leviathan – and then only in a rather oblique fashion. For consider, second, that L vi.9, 82 and L vi.11, 82 do not say that color and value are alike in that our everyday thought and talk about each involve the illusory projection or presuppose the objectivity of the properties in question. All Hobbes says in L vi.9, 82 is that color and value are alike in that each involves a sensation – a visual sensation on the one hand, a sensation of pleasure or displeasure on the other – that is our first-person awareness of what is, at bottom, some sort of material motion in the body. That is the real point of the analogy in L vi.9, 82: not to suggest that value is a kind of secondary quality, but to emphasize that value, like color, is ultimately just a product of the workings of the human mind, and hence ultimately just a product of a complex system of matter in motion.Footnote ¹³ Third, Darwall (Reference Darwall2000) wants us to read Hobbes as saying that pleasure and displeasure are appearances whose content is as of an objective, mind-independent feature of the world – i.e. good and evil construed in realist terms as ‘categorical, subject-independent’ features of the objects of our appetites and aversions (328, see also 325–326). But it is difficult to imagine that we might take our subjective sensations of pleasure and displeasure to present us with mind-independent features of external objects in the way that we might naturally take our subjective color sensations to present us with mind-independent features of external objects. In fact, Hobbes himself explicitly contrasts color, which is naturally, if mistakenly, thought to be a mind-independent feature of external objects, with pleasure and pain, which are quite obviously subjective sensations in the mind; and he even offers a physiological hypothesis to explain this contrast between our natural interpretations of the two respective kinds of sensation. Colors and other sensory phantasms involve ‘a reaction and endeavor outwards in the organ of sense, caused by some endeavor inward from the object.’ The result is that a sensory phantasm, ‘by reason that the [bodily] endeavor is […] outwards, doth always appear as something situate without the organ’ (DCo xxv.2). In contrast,

So Hobbes has no truck with the view that our sensations of pleasure and displeasure present us with an appearance as of something beyond our own bodily responses and feelings, or that we project our desires and aversions, or their manifestation in consciousness as pleasures and displeasures, onto the world in the same way that we naturally project our color sensations onto external objects. Of course, this is not to deny that Hobbes holds that there is a characteristic phenomenology of pleasure and displeasure that tracks with our appetites and aversions, and hence with our dispositions to declare things good or evil. But there is no textual evidence that Hobbes takes evaluative judgment to involve the projection of this private phenomenology onto the world, and in fact, there is clear textual evidence against it.

5. The prescriptivist interpretation

There is no need to read Hobbes as treating our practice of evaluative judgment as any sort of attempt to describe or represent either some feature of the world (such as its objective value), or some feature of our own psychology (such as our appetites and aversions). In fact, we make better sense of both the texts and the philosophy if we take his view to be that evaluative judgment has a purely non-representational function – and in particular, a purely prescriptive function. On this view, when we call something ‘good’ or ‘evil, ’ the role of our utterance is to urge that a certain course of action be pursued or that a certain sort of practical stance be adopted. The specific illocutionary nature of the prescription will presumably vary with the indexical context of the evaluative utterance, as we shift from contexts fixed by the speaker’s own appetites (giving us commands, pleas, and petitions) to contexts fixed by the appetites of another person (giving us acts of counsel and advice), or to contexts fixed by the appetites of the state (giving us commands and one or another kind of exhortation). The authority of the prescription might also vary as we shift the intended audience from intimidated subordinate to unintimidated master, or to contexts of first-person deliberation and acts of self-counsel or self-exhortation. And the urgency and insistence of the prescription might vary with the strength of the appetites or aversions that are signified by the evaluative utterance. But for all this potential species-variation in the force of our evaluative utterances, in each kind of case, our utterance is an attempt to shape practice, and its role is not to represent the way the world is, but to legislate action within it.Footnote ¹⁴ Notice: I am not claiming simply that (for Hobbes) evaluative judgments have this particular non-representational function, but also that that they have no representational function. Evaluative judgments are purely prescriptive, and the alternative hybrid view that they involve some representational component alongside a prescriptive component fails quite as badly as the purely representationalist readings both (i) to account for the passages I will be citing in favor of the prescriptivist reading and (ii) to plausibly explain the systematic bifurcation of the rule of good and evil.Footnote ¹⁵

As we have seen, Hobbes holds that evaluative judgments ‘signifie’ our appetites and aversions, and that ordinary language users have an implicit understanding of this fact. On my proposed interpretation, this display of appetites and aversions is not the purpose of evaluative judgment, but simply a means or byproduct. The point can be illustrated with reference to Hobbes’s exhaustive four-part taxonomy of the various ‘uses of Speech’ in Leviathan. Here, Hobbes does not explicitly mention evaluative judgment, but it would presumably fall, not in the category of speech that registers the nature and causes of phenomena, nor in the category of speech that teaches this acquired knowledge to others, nor in the category of wit and wordplay, but in the category of speech that serves ‘to make known to others our wills, and purposes, that we may have the mutuall help of one another’ (L iv.3, 50). Both the means and the ends here square nicely with the prescriptivist interpretation. Evaluative speech puts our appetites on show (as indeed would any prescriptive utterance, regardless of whether or not it invoked evaluative language). But its purpose in so doing is to guide action.

Evaluative predicates operate grammatically like descriptive predicates, and so an evaluative judgment will typically take on the grammatical appearance of a descriptive judgment, as when I say that a certain cup of coffee is dark, hot, fresh, and good. My proposal is that Hobbes regards this similarity as merely skin deep, and that he holds that, the surface appearance of our language notwithstanding, our actual employment of evaluative terms shows that their function is not to describe but to prescribe. But could Hobbes really have endorsed a form of expressivism about this part of our language – i.e. the view that this kind of speech serves purely to express some non-representational attitude or state of mind, in spite of its superficially representational appearance? The proposal might seem anachronistic. At least, those who have gone looking for early proponents of expressivism have generally stopped short several generations before Hobbes, either with Hume’s account of causal and moral judgment, or with Berkeley’s instrumentalist handling of the language of scientific theory (Blackburn Reference Blackburn1993, 5; Flew Reference Flew and Todd1974).

But the charge of anachronism is misplaced. Hobbes is quite plainly an expressivist about some areas of discourse, and takes evident pride in exposing the purely non-representational function of certain kinds of speech whose representational appearances have led other philosophers astray. The clearest example is his treatment of pious language that might seem to describe the nature of the deity. Consider this from Anti-White:

The same expressivist treatment of pious speech about the nature of God appears in the Elements of Law, De Cive, and Leviathan (EL ix.3; DCi xv.14; L xxxi.28, 566; L xxxi.33 568). When we speak of God as acting in such and such a way, or as wishing for such and such a thing, or as being infinite, eternal, just, or holy, we are not describing anything – not the character or behavior of the deity nor the states of our own minds. Appearances to the contrary, we are not asserting propositions about the nature of God, but simply expressing (but not describing) our own awe and reverence. Hobbes sees nothing wrong with the devotional language itself, but he mocks to scorn literal-minded philosophers who are misled by the representational appearances and attempt to draw truth-evaluable propositions out of the devotee’s pious words (L xlvi.23, 1084–1086; compare also L xxxi.33, 568).Footnote ¹⁷

What is the case for thinking that Hobbes extends this sort of expressivist strategy to evaluative language, and, more specifically, that he regards evaluative language as having the particular non-representational function of prescribing actions and practical attitudes? We might begin with a distinction that Hobbes draws in De Corpore between two general kinds of speech:

Here Hobbes draws a distinction between the kinds of speech that ‘signify the desires and affections of men’ and the kind of speech that asserts truth-apt propositions. The latter form of speech is descriptive or representational in character: it is used to ‘affirm or deny, ’ and therefore ‘expresseth truth or falsity.’ In Hobbes’s implied contrast, the former kinds of speech must then serve one or another sort of non-representational function, as is clearly the case in each of Hobbes’s chosen examples: interrogations, prayers, promises, threats, and the rest. Now, presumably, Hobbes does not mean to deny that we might also on occasion use the representational, proposition-expressing form of speech in order to ‘signify [i.e. display] the desires and affections of men’ (as when I say, purely to report a fact, that my friend has a desire for lemon tea, or that I have a desire for coffee). His point is rather that the forms of speech that are distinguished peculiarly by the signification or display of our desires and affections have some different function from representation and the expression of truth-apt propositions. We also find this same basic distinction between the kinds of speech we use peculiarly to ‘signifie the Appetites, Aversions, and Passions of mans mind’ and the kind of speech we use to represent the world through truth-apt ‘affirmations’ in Leviathan (L iv.11, 54; L iv.23, 62). So if we recall Hobbes’s repeated insistence that ‘Good, and Evill, are names that signifie our Appetites and Aversions’ (L xv.40, 242; see also L vi.7, 80–82; DCi iii.31), we might then expect him to locate the vocabulary of good and evil within the former, non-representational species of language.

Two passages in Leviathan help confirm this picture. Each appears in the main body of chapter 6 (i.e. Hobbes’s official account of ‘the passions’ and ‘the Speeches by which they are expressed’), just a few pages away from our familiar proof text L vi.7, 80–82. So any reading of L vi.7, 80–82 and of Hobbes on evaluative language ought to be able to accommodate these two texts. Yet, each seems to stand in the way of the traditional representationalist readings.

First then, at L vi.59, 96, Hobbes tells us that ‘[t]he forme of Speech whereby men signifie their opinion of the Goodnesse of anything is praise,’ and various other passages confirm that he equates the labeling of a thing as ‘good’ with the act of praising it.Footnote ¹⁹ But what is it to praise something? According to the two traditional readings of Hobbes on evaluative language, it must be to represent it as having some property, to describe it in some way – either as being desired by someone (as with the reductionist interpretation), or as having an objective value that is independent of human desires and attitudes (as with the error theorist interpretation). But Hobbes does not regard praise as a matter of description, as the passive registration of putative fact, but rather as a speech act with some sort of non-representational function. Thus, in our passage from Anti-White on religious language, he clearly contrasts the ‘proclaiming’ of the ‘stat[e] of mind that govern[s] our wish to praise’ with representational speech ‘pertain[ing] […] to the explanation of philosophical truth’ (AW xxxv.16, quoted more fully above). Or, similarly, in Leviathan, he writes that ‘[i]n orations of Prayse […] the designe is not truth, but to Honour’ (L 8.6, 106). Passages such as these show Hobbes’s resistance to treating praise as any matter of description or representation, and indicate that he thinks of this form of speech – and hence that he thinks of our use of the predicate ‘good’ – not as directed toward truth and representation (or indeed to any hybrid role partly involving representation), but rather as having some sort of exclusively non-representational function.

My proposal is that that function is to prescribe actions and practical attitudes. The closest Hobbes comes to saying this in so many words is in the following important passage:

The passage begins with Hobbes allowing that any of the passions could be expressed by speech in the indicative mood, and he clearly implies that that these indicative utterances – unlike imperatives, optatives, and interrogatives – would be so many ‘affirmations.’ So we can signify the passions in this truth-apt form of speech, as when we simply report facts about our own desires and affections, or, presumably, about the desires and affections of others. However, Hobbes also asserts that when our passions gain expression in their own distinctive forms of speech – their ‘particular expressions by themselves’ – these forms of speech ‘are not affirmations.’ Among these distinctive passion-expressing forms of speech is ‘[t]he language of Desire, and Aversion, ’ which is ‘Imperative’ in character, and comprehends the various speech acts of command, prayer, and counsel. So the language of desire and aversion has a particular non-representational function. It serves to prescribe, and its role is, in one way or another, to guide actions and shape practice: to mandate, urge, recommend, exhort, and the rest. But the language of desire and aversion must comprehend the vocabulary of good and evil. At least, one could resist this only if one were prepared to say that Hobbes does not intend his expression ‘[t]he language of Desire, and Aversion’ to comprehend ‘names that signifie our Appetites, and Aversions’ – for on Hobbes’s account, ‘good’ and ‘evil’ just are such names (L xv.40, 242; see also L vi.7, 80–82; DCi iii.31; DH xi.4; OLN 14). Unless we are willing to make this rather desperate move, it seems that we must say that, for Hobbes, the vocabulary of good and evil, like the language of desire and aversion more generally, serves a prescriptive function; and further, that when we use this distinctive form of speech, our expressions ‘are not affirmations’ – and so are not representational, do not convey propositions, and do not ‘expresseth truth or falsity.’Footnote ²⁰

The prescriptivist interpretation of Hobbes on evaluative language is also attractive on systematic grounds. First, it avoids the two specific problems faced by the reductionist interpretation. It can explain what two speakers are arguing about when the one asserts and the other denies that something is good. In the standard case where the rule of good and evil is fixed by the speakers’ respective private appetites, the two will be in a practical conflict, with the one urging one course of action or practical stance, and the other resisting this in favor of a rival practical policy. Moreover, with the prescriptivist interpretation we also avoid attributing to Hobbes the implausible view that our evaluative vocabularies can be translated without loss into the non-evaluative terminology of appetites and aversions. The core function of our evaluative vocabularies is to prescribe, and so any proposed reduction of evaluative judgments to descriptive claims about appetites and aversions will leave out the essential ingredient altogether. (In this way, the prescriptivist interpretation also explains why deliberation about good and evil is, inescapably, practical deliberation.)Footnote ²¹

Now consider the way in which the prescriptivist interpretation meshes with what Hobbes has to say about the employment of evaluative terms under the civil laws. Recall that whenever the sovereign announces a new law, the ‘Rule of Good and Evill’ controlling the proper use of these terms follows accordingly, and any conflicting use of ‘good’ and ‘evil’ reflecting private appetite becomes improper. Both the reductionist and the error theorist readings of Hobbes on evaluative language struggle to explain this feature of his overall account – and indeed so long as evaluative language is regarded as representational in function, it seems that we will be left with an implausibly bifurcated semantics for evaluative terms, with such terms conveying one sort of semantic content when the law is silent, and an entirely different sort of semantic content once the law has spoken.

But with the prescriptivist interpretation, we avoid this Humpty Dumpty semantics and its paradoxical consequences, such as the result that subjects cannot criticize the laws without falling into verbal contradiction. Moreover, we have a natural explanation of why evaluative speech has its distinctive bifurcated employment while yet maintaining a unified underlying purpose. Whenever the sovereign enacts a law, it is just issuing a certain sort of command, and prescribing, in its capacity as representative for the overall body politic, that subjects behave in such and such a way. So the laws are naturally seen as the sovereign’s pronouncements about what is ‘good’ and what ‘evil, ’ given that the function of evaluative predicates is (on the proposed interpretation) just to prescribe this or that kind of action or behavior. Further, since all subjects have agreed to be collectively represented by the sovereign and ‘to submit their Wills, every one to his Will, and their Judgements, to his Judgment’ (L xvii.13, 260), the civil laws – i.e. the sovereign’s official pronouncements about what is ‘good’ and what ‘evil’ – have a particular authoritative status. That is why the new rule of good and evil encoded in the civil law makes any conflicting use of these terms improper. It is not that those who defy the state and continue to use the words ‘good’ and ‘evil’ in a manner reflecting private appetite have somehow managed to misunderstand the meaning of evaluative terms. Rather, their utterances are improper simply in that they are failing to align their own speech acts of command, counsel, and deliberative self-counsel – as expressed in their declarations about what is ‘good’ and what ‘evil’ – with the prescriptions of the law. Such uses of ‘good’ and ‘evil’ can remain perfectly meaningful and in good linguistic order, even if they are out of order politically.Footnote ²²

Finally, while this sort of lawless speech act will typically be a sign of criminality or sedition, the sovereign would be wise to allow some free expression of contrarian prescriptive speech, if only within a closed circle of counselors and advisors. Such counselors may have differing judgments from the sovereign and use evaluative language to urge different courses of action, though being bound to obedience they should speak only when and where permitted, and must accept any final decision by the sovereign as speaking for all. And here we can dissolve one last paradox, or putative paradox, in Hobbes’s philosophical system. There is often thought to be a kind of performative contradiction running through Hobbes’s writings, as he repeatedly asserts that we should give up the right of private judgment of good and evil to the state, and yet is himself so insistent in pressing his own particular private view of how the state should be ordered – including, as we have seen, his own private judgments about what laws are good and what ‘not good.’ But this is no contradiction at all on the current interpretation. So long as the state allows the philosopher to express his advice – as, in one way or another, the Caroline Crown, the Commonwealth, and the Protectorate each allowed Hobbes to circulate or publish his political recommendations – then the philosopher may criticize the laws while also insisting that they must be obeyed.