Morality as a System of Categorical Modals
Morality as a System of Categorical Modals
We express our moral beliefs in sentences like these: “Murder is wrong”, “Stealing is wrong”, “Generosity is good”, and “Violence is bad”. What do they mean? Some have said they are equivalent to categorical imperatives, others suggest that hypothetical imperatives provide a better analysis; the concepts of duty or obligation are frequently brought in; the word “ought” is deployed. I want to suggest that the propositions in question are modal in character: the concepts of necessity and possibility can be used to analyze what moral statements mean. Thus “Murder is wrong” means “You must not murder”, and “Generosity is good” means “You must be generous”. The “must” here is specifically moral—it isn’t the “must” of logical necessity or metaphysical necessity or epistemic necessity. It isn’t that people can’t do otherwise than be generous, or that it is certain that people are generous. No, the thought is that one must morally be generous and not murder or steal. That is, there is a specifically moral “must” in our language (or our use of language). What is the evidence for this? We say things like, “You must visit your grandmother soon” or “I really must remember her birthday this year” or “We must do something about global warming”. We subscribe to such assertions as that promises must be kept, lies must not be told, gratitude must be shown, kindness must be practiced, justice must prevail. Some things we can do and not invite moral rebuke, but other things we can’t do without inviting moral censure. There is a contrast in our thinking between what it is possible to do morally and what it is necessary to do morally. These are perfectly good uses of “possible” and “necessary”, and they don’t mean what they do in non-moral contexts (modals are notoriously protean). They are entirely natural ways of expressing our moral attitudes. Thus moral language is modal language.
If so, we would expect modal logic to apply to moral language. One central rule of modal logic is that necessity implies possibility: what is necessary is possible. This rule has its counterpart in the well-known principle that “ought” implies “can”: you cannot be obliged to do what you can’t in fact do. If you must see your grandmother tomorrow, than it must be possible for you to see her. No one can say “I must cure cancer tomorrow” when everyone knows that this is not possible. As necessity implies possibility, so moral obligation implies the ability to carry out the action in question. Of course, “can” doesn’t imply “ought”, as “possible” doesn’t imply “necessary”. Thus the basic rules of modal logic apply to moral reasoning.
The modal analysis also comports with the sense we have that morality is binding—that it leaves us with no alternative but to do what is right. Morality restricts freedom; it cuts down our options. But so does modality generally: the number 2 has no option but to be even (it isn’t free to be odd), and epistemic necessity eliminates all alternative possibilities (I can’t not be in pain given my current epistemic situation). If I must repay that debt, then I have no alternative but to repay it—no moral alternative. I have no choice in the matter; I am not morally free to act otherwise. We are bound to do what morality requires of us, and moral necessity captures this idea.
Modality also captures the meaning of prudential statements, as in “I must eat less sugary food” or “I must exercise more frequently”. Again, this is a perfectly natural way to express our prudential attitudes. But the connection between morality and prudence is conceptually close: we can even view morality as a kind of extension of prudence to other people. If it is prudent for me to take care of my health, then it is moral of me to be concerned about other people’s health. The Golden Rule makes this connection explicit: do to others what you would want done to you, i.e. what you would do for yourself if you could. Thus I must do for others morally what I must do for myself prudentially—not harm them and promote their welfare. Given that a modal analysis of statements of prudence is plausible, so is a modal analysis of moral statements plausible. The same conceptual apparatus applies to both, i.e. modal concepts.
None of this is yet to say how these uses of “must” should be analyzed, and that will be open to debate depending on one’s view of the content of morality. I have only given a theory of the logical form of moral statements, taking the moral modals as primitive for this purpose. The canonical form of a moral proposition is, “Everyone must X”, where “X” ranges over action types (e.g. “Everyone must refrain from murder”). Moral sentences are a special type of modal sentence: that is their correct semantic analysis. If we were giving a truth-theory for moral sentences, we would first translate them into the corresponding modal paraphrase and then offer a Tarski-type theory of truth for these modal sentences (or some other type of formal semantics, according to preference). We might try to construct a Kripke-style modal semantics for moral discourse following these informal ideas.[1]
[1] The inspiration for this paper came from a road sign I saw the other day: “Right lane MUST turn right”.
I attained confidence in a similar conclusion a while ago, especially after reading your paper entitled ‘modal reality’, whose introductory remarks encourage us to notice how “our prior conception as to the constitution of modal reality should find its way into the semantics we propose” and that “we are obliged to take seriously whatever our favoured semantic theory pins on our serious modal assertions”.
Your enumerated ‘regulative canons of ontological commitment’ also help to keep our metaphysical ideas on their proper semantic rails.
Elsewhere (eg 2000, Ch 5) you encourage us to imagine doing without the concept(s) in question.
In 1997 (passim) you compare moral knowledge to modal knowledge (“moral knowledge is knowledge of ideals”)
We do talk in such ways and we mean what we say when we say such things.
I’ve come to regard moral knowledge as a kind of modal knowledge (knowledge of necessity tells us what must obtain, knowledge of possibility tells us what could obtain and knowledge of goodness tells us what ought to obtain). They seem like cognitive modes of property instantiation, without which we should be condemned to rely on inductive inferences from observations.
There is also perhaps a parallel with your ‘copula modifying’ idea. Nothing is actually good. It’s just that when we point to something and say “that is good”, the copula is tacitly modified (like a sort of error theory?).
I admire your erudition, Alan. You know my work better than I do!
I shouldn’t presume, though I often refer to you as ‘the philosophers’ philosopher’, I have much to thank you for.
I gather that Lewisian style, objectual ‘possible worlds’ might be regarded as something of a straw man, were it not for their formal success…one wonders why (apropos the present thread) it doesn’t occur to Lewis and his disciples to lump in ‘ideal worlds’. Then, it seems to me, the nonobjectual nature of modal and moral epistemology becomes harder to resist.
Also, might it be the case that the ‘naturalistic fallacy’ is just a symptom of taking the copula (in moral ascriptions) too literally?
The most adamant resistance to the realist conception I have encountered so far comes from ‘constructivists’ (whose views I find very hard to take).
Thanks.
It all stems from Quine: to be is to be the value of a variable. Thus there can be no reality (being) that doesn’t consist in existing objects. It may be that the uniform copula is responsible for ontological leveling.
Aye, yet reality hosts fiction.
Fiction, yes, but does it host fictional objects?
I think it does, in the sense that I can’t imagine (fiction) without imagining some object(s).
Maybe this all turns on the reality of property instantiation: my imagination instantiates real properties: Chewbacca (say) is a hairy biped who carries a ‘laser crossbow’ yet Chewbacca doesn’t exist and neither does the property of being a wookie, or being a ‘laser crossbow’, or being friends with Han Solo, etc. But being hairy, being a biped, being able to shoot with a crossbow, etc. are not unreal properties just because Chewbacca may instantiate them.
I suppose I am rolling with the idea that being does not necessarily entail existence. That ‘nothingness’ may host a kind of being sans existence. I’m trying, possibly in vain, to carve out a space (pace McGinn, 2000, Ch. 2) where real properties can have or lack existence and hence, for reality to come apart from being, courtesy of the instantiation relation.
If we accept that fictional objects lack existence, we have accepted that they differ from things that do exist. Thus in one sense they are not part of reality–Sherlock Holmes is not a real detective (but a fictional one). The rest is just terminological. Of course, non-existent things can have real properties–the property of being a detective is real enough. Fiction itself exists, i.e. novels, films, etc.
If being different from existent things affords a sense in which they are just not part of reality, then the terms ‘existent’ and ‘real’ may be exactly synonymous.
I’m still unsure that they are, although I concede the point that my confusion may be merely terminological.
Does the sentence ‘unicorns really lack existence’ necessarily express a tautology?
Or the sentence ‘in reality, unicorns don’t exist’?
Are those ‘really’ and ‘in reality’ locutions purely for emphasis? Is the second sentence just as tautological as if we said ‘fictional unicorns don’t exist’?
Yes, for emphasis. In these discussions “real” and “exists” are synonyms. The right thing to say is “Some objects of thought don’t exist/are not real”.
Got it, thank you.
You have outlined a very interesting schema whereby nonexistent entities are metaphysically impossible (though epistemically possible), merely possible entities are existent (though not actually so) and impossible objects are bifurcated into the existent kind (which “lack the possibility of actuality”) and the nonexistent sort.
Just to explain what prompted my (worthless) musings, in 2000, Ch. 2, you maintain an undogmatic attitude to this schema, leaving it open to challenge and potential modification, which prompted me to wonder whether it might be fruitful (just as in your actual/nonactual distinction and especially considering the uncontentious reality of properties) to work up a sense of reality that admits of nonexistent things, courtesy of instantiation. My guiding thought was that ‘existence’ and ‘reality’ could mean different things.
But I shall now abandon this train of thought, assured as I now am of its fruitlessness.
Thank you.
If only more people were as receptive to reasoned argument as you! These are very difficult questions.