Norms: do normative facts exist?
Position: Normative pluralism
The normativity of reasons
It has been common to operate a distinction between motivating reasons and normative reasons. The former are traditionally described as the reasons someone has to be in a certain state; the latter cover the reasons that “count in favour” of being in a certain state. In a nutshell, normativity is about the "ought": it is preoccupied with how things ought to be, rather than with how they are.
However, I am inclined to reject that distinction altogether, and to accept that reasons are essentially normative. First, looking at the (supposed) motivating vs normative difference, it seems to be re-litigating the rationality/reasons distinction from the last subsection. That Ed believed he was a bird qualifies (by most accounts) as a motivating “reason” for jumping off a building. But there is no normative reason for him to do so (nothing objectively “counts in favor” of that). Ultimately, I would argue that what some authors call motivating reasons simply evidence instances of an agent being rational. And what fits the definition of reasons that I have proposed is really a normative reason.
This conclusion can easily seem counter-intuitive, in virtue of the habit we have of reflecting on morality as the pure archetype of a normative system. Let’s come back to Peter, who jumped to save a child from a falling rock. There is no debate that one reason to jump was normative: Peter was morally compelled to save a child in danger. Now, imagine that someone else spotted the rock as it was about to fall, and quickly offered Peter one million dollars to save the child. It is a fact the offer was made, and we may want to say that counted for at least something in Peter’s quick thinking, but it seems awkward to put that reason under the normative category. Seeking retribution for a good action should indeed not be the bedrock of anyone’s moral code, but the point here is that normativity is much broader. The rule that, all else being equal, the promise of a prize increases, ever so slightly, someone’s bundle of reasons to make the correct move, seems like the kind of common-sense norm that guides human behavior, for better or worse.
To accept that is not to make a value judgement, though, like accepting some moral norms would be. To take another example, being hungry is a reason for me to initiate the action “grabbing my sandwich”. Again, the implicit norm that being hungry is good reason to grab any food around is common-sense, and not borne out of some intricate moral stance. If anything, it is grounded in biology, but can be counter-acted by a plethora of social norms. I regard it as a norm nonetheless, notwithstanding a difference of nature and degree with something like “Thou shalt not kill”.
A typology of norms
So, normative reasons are multifarious, but we can still try and provide some structure to their domain. I will outline three categories:
Deontic reasons are reasons to act. They are connected to notions of obligation: to say I am overwhelmingly justified in acting a certain way (and not at all justified to do nothing) is equivalent to saying that I must act that way. Moral or ethical norms are a natural fit for this category.
Evaluative reasons are reasons to feel or to wish. Aesthetic norms are built upon evaluation: to find something beautiful is to have reasons to feel pleasure, emotion, appreciation. Moral obligation can also be framed as an evaluation of situations: I may not be the agent but feel an action is good if I have reasons to wish someone would take it.
Epistemic reasons are reasons to believe. All subjective minds spin a web of beliefs. But those don’t exist in a vacuum: they connect to each other, but also to facts that justify them: what the agent has seen, has been told...
I’ve been using the term norm without a proper definition. Now I will state that any proposition that involves the reason relation \(\mathcal(R)\) is a norm. Further to that, I will single out a sub-set of those norms which I’ll call principles. Colloquially, it is quite clear what the difference is: principles, for instance moral principles, are “laws”, or “rules” with limitless purview, while regular norms can pertain to very specific situations. Let’s formalise things a little bit.
We can start again from our very simple example. The fact that Camille (a child) is in danger of being crushed by a rock (\(Dc\)) is a reason for Peter to take the action: try and save Camille (\(Spc\)). \[\mathcal{R}[[\lambda Dc] p (\textsf{Act}[Spc]\delta])\] This is:
agent-relative: about the agent’s judgment of his own (purported) actions.
special: about a given, specific agent (Peter).
particular: involving a specific object (Camille).
To make a norm agent-neutral instead of agent-relative, one needs to have the reason and the response be independent of the agent’s identity. For instance: the fact that Camille is in danger is good reason for Peter to wish that someone would save Camille. In other words, whomever may save Camille, Peter would garner normative support towards that scenario. \[\forall \, h \exists \delta>\delta_0 \ \left( \mathcal{R}[[\lambda Dc]p (\textsf{Des}[Shc]) \delta] \right)\]
To make a norm general instead of special, it needs to be true of any agent. For instance: the fact that Camille is in danger is good reason for anyone to wish that Peter saves her. \[\forall h \, \exists \delta>\delta_0 \left( \mathcal{R}[[\lambda Dc]h (\textsf{Des}[Spcl]) \delta] \right)\]
To make a norm universal instead of particular, the reason and response need to be rid of any proper name (i.e. any direct-reference, Kripkean acquaintance that is attached to a specific ordinary object) other than the agent. For instance: the fact that a human is in danger is good reason for Peter to try and save them: \[\forall h \, \exists \delta>\delta_0 \left( \mathcal{R}[[\lambda Dh]d (\textsf{Act}[Sph]) \delta] \right)\]
We shall define a principle as a norm that is agent-neutral, general and universal. For instance: the fact that a human is in danger is good reason for anyone to try and save them: \[\forall h \, \forall j \, \exists \delta>\delta_0 \left( \mathcal{R}[[\lambda Dh]j (\textsf{Act}[Sjh]) \delta] \right)\]
From a given principle, or, for that matter, from universal, general or agent-neutral norms, more specific norms can stem at all time. That’s one of the basics of rationality: to take one’s principles and apply them to actual situations. Which reasons, or types of reasons, can be derived from principles is naturally a key question for normative metaphysics. It determines the structure of the entire normative system.
But we’ve eschewed much more momentous questions. This axiology section started by introducing reasons as a common ground to a cluster of seemingly disparate concerns. Then the connection between reasons and normativity has been reinforced. But that’s still just talk, discourse, pixels, inkblots or soundwaves... Indeed, we may have talked a great deal about reasons and norms, but we have done so light-heartedly, as we might in daily conversation. It’s time to get a bit philosophical and to follow our usual line of questioning:
The semantic question: do normative sentences express truth-apt propositions?
The ontological question: do normative facts exist?
The metaphysical question: what grounds normative properties?
The epistemological question: how can we come to know normative facts?
I won’t answer these questions here unequivocally — there are many flavors of normative sentences that may fall in one category or the other. This subsection is then placed under the sign of normative pluralism. The next one, centered around value-based norms, will be more trenchant.
The semantics of norms
It is time to confront the foremost bogeyman in philosophy. Are we fooled by language? Remember that we started talking about action, rationality and normativity after acknowledging their importance as a domain of discourse. There are normative sentences, that’s an unassailable fact.
But we’ve also started to use the reason relation \(\mathcal{R}\) in a carefree way, writing propositions, implicitly admitting its existence in the world. It’s time to question these priors. We’re only spending time speaking about normativity because it is strange, unlike anything else we’ve built so far. Perhaps it’s just an illusion, then?
Let’s summon some examples of normative sentences that will be our guinea pigs:
Deontic: It is forbidden to swim in the river.
Evaluative: ’Penny Lane’ is a beautiful song.
Epistemic: You must believe what you see.
We know that these sentences exist, because they have actually been uttered. It is a natural impulse to automatically assume they express a proposition. Indeed, their syntactic structure is similar to non-normative sentences:
It is cold to swim in the river.
’Penny Lane’ is a pop song.
Alice did believe what she had seen.
Those sentences, when uttered, are reflective of a cognitive state. In short, to say that \(\phi\) is to express a belief in \(\phi\), that is, to signal that I am in the mental state \(\textsf{Bel}[\phi]\). Since \(\phi\) is a proposition, it is truth-apt: it can be either true or false. Is it the case for normative sentences?
To accept that is to espouse a position called normative cognitivism. If that is the branch we pick, we need to face more questions, at the very least the ontological one, which we’ll discuss in a moment. If we don’t accept normative cognitivism, that makes us (drum roll...) normative non-cognitivists. Sometimes theses opposite positions are called descriptivism and non-descriptivism.
Now is as good a time as any to say this: the answer to the semantic question (and to the ones that follow) need not be the same for all kinds of normative sentences! One may well be a moral non-cognitivist but an aesthetic cognitivist (for the sake of argument).
If non-cognitivists believe that normative sentences are not really about normative propositions, then what do they make of them? Well, the standard stance is to say they are disguised expressions of conative states: mental states that include desire, fear, appreciation... It is common to group them under the more neutral guise of "a pro- (or anti-) attitude" towards something.
Indeed, lots of our non-sentential utterances express conative states. Shouting "Boooooo, Novak!" in a stadium reflects a wish that the so-named sportsman lose. "Yes, he lost!" arguably doesn’t reflect a belief that he did lose (which is likely obvious in context) but mostly expresses the pleasure the speaker experiences.
Similarly, non-cognitivists include the following main groups:
emotivists would say that the impermissibility of swimming in the river is just an alternate way of "booing" that action, really expressing a strong dislike for it.
prescriptivists would rather say that sentence expresses a very strong wish or a direction, best rendered as an imperative: "don’t swim in the river!"
expressivists cover a more varied, and dare we say sophisticated, range of positions which boil down to a generalization of the above towards more generic pro- and anti- conative attitudes.
The ontology of norms
Do normative facts exist? For non-cognitivists, the question is moot: a fact is an obtaining proposition, so if there’s no normative proposition, there can be no normative fact. What of cognitivists? They accept that there are truth-apt propositions about normativity. Those can be true or false. If a normative proposition is true, the corresponding state of affairs obtains, and a corresponding fact exists. Accepting that makes us proponents of normative realism.
The complementary stance is normative anti-realism. Of course, non-cognitivism is a flavor of anti-realism. But is it possible to be a cognitivist and an anti-realist? It is, if one believes that all normative propositions are false. That is why the resulting position is often called error theory: it argues that all of our beliefs in normative propositions are misguided, as there is no fact of the matter.
That may seem like a peculiar position: what’s the use of an entire class of propositions that are uniformly false? Like other anti-realists, error theorists are struck by the queerness of (purported) normative properties; refusing the existence of normative facts avoids dealing with that. Unlike non-cognitivists however, they don’t deny that there’s genuine cognitive content behind our normative sentences that can play a similar explanatory role as "morals" do in folk psychology. The general conceit of error theory is actually not new in this work. It is very close to mathematical fictionalism or to our treatment of fictional entities proper, to such an extent that we could develop a sub-set of error theory we’d call normative fictionalism.
About fictional entities, we had established, in summary, that:
"Rastignac is ambitious" is false.
"Rastignac is a fictional character" is true.
"Rastignac is ambitious" is true in Balzac’s literary world.
"Rastignac" is an abstract object that encodes "being ambitious"
Normative discourse is not so much about objects than about properties, so it might be more useful to flip the perspective, as in this example:
"Romulus was a demigod" is false.
"’Demigod’ is a fictional kind" is true.
"Romulus was a demigod" is true in Roman myths.
"Being a demigod in Roman myths" is an abstract object that encodes "being exemplified by Romulus"
If we believe in mathematical fictionalism, we can similarly say that:
"\(2\) is even" is false.
"\(2\) is a mathematical object" is true.
"’Being even’ is a mathematical property" is true.
"\(2\) is even" is true in Peano number theory.
"\(2\)" is an abstract object that encodes "being even".
"Being even" is an abstract object that encodes "being exemplified by \(2\)".
Moving on to a normative sentence, we get:
"Stealing is wrong" is false.
"’Being wrong’ is a moral property" is true.
"Stealing is wrong" is true in the moral code \(\mathfrak{M}\).
"Being wrong in \(\mathfrak{M}\)" is an abstract object that encodes "being exemplified by stealing".
Let’s use a last example to shed a different light on this idea:
"Moving the pawn backwards is illegal" is false.
"’Being illegal’ is a deontic property" is true.
"Moving the pawn backwards is illegal" is true in the rules of chess
"Being illegal in the rules of chess" is an abstract object that encodes "being exemplified by ’moving the pawn backwards’".
We can see the close relation between fictionalism and formalism. Balzac’s books and Roman myths are like guidebooks to their universes: they make a large number of statements, from which we can ourselves extrapolate, using some form of metaphysical inference. Number theory and chess rules are an even clearer fit since they are explicitly axiomatic systems from which an infinity of theorems can be inferred. In this worldview, normative codes such as morals or aesthetics can follow either an informal, pragmatic, individual "code" made up of countless micro-rules, or a codified set of "tables of the Law".
The formalization itself is similar to what we’ve done for the aforementioned topics of mathematics and fiction. We can start by writing that stealing is wrong in a certain moral code: \[\mathfrak{M} \vDash W s\] Then the concept of being wrong (in the code) is just the abstract object that encodes all the things that being wrong exemplifies in the code: \[W_{\mathfrak{M}}=\imath y ( \kappa y \wedge \forall \mathcal{F} ( y\mathcal{F} \equiv \mathfrak{M} \vDash \mathcal{F} W )\] The code itself can be defined as the abstract object that encodes all propositional properties that can be inferred from its axioms: \[\mathfrak{M} = \imath x ( \mathbf{A}x \wedge \forall P ( xP \equiv \exists \phi ( P=[\lambda y \phi] \wedge \alpha_1,...\alpha_n \vdash_M \phi_T ) ))\]
We should now find it easier to relate to error theory. It is fairly consistent with our daily experiences to say that something is permissible based on a certain set of rules, or is beautiful based on the tenets of some aesthetic movement. The obvious drawback is that relativism is part of the package, and can quickly veer into nihilism. If moral or epistemic codes are inferred from arbitrary axiom schemas, then we are seemingly disarmed against people arguing that the Earth is flat, or that hurting children is a good thing...
The metaphysics of norms
This one only matters to realists. It’s about the place of normative things in our ontology. Anti-realists don’t believe there are normative facts, and a fortiori, don’t believe in normative things outside of the fictionalist theories. Realists however don’t have that luxury. Our ontology so far has made use of two procedures to generate properties and abstracta. The first one, the "natural route", starts from fundamental natural properties (originating in \(\mathfrak{ToPM}\)); even though it can make heavy use of abstraction via the comprehension principle, and even combine the outcome with mathematical properties, the end result is recognizable as an entity whose existence depends on the physico-mental world.
"Having an electric charge", a physical property, "feeling pain", a mental one, obviously fall into this category. More complex properties like "being a bicameral constitution" or "being a sonnet" are too, since they can be exemplified by objects that are themselves typified by ordinary things. The other route is purely abstract and is represented by the procedure we’ve just reminded ourselves of, that can lead to the sprouting of mathematical theories, fictional worlds or legal rulebooks.
Adopting the latter solution brings us back to anti-realism. The choice is then simply between the natural route, and an additional solution, one that is always at hand: the profligate route that simply posits new, independent entities.
The natural route leads to an attitude called normative naturalism. A minima, its central tenet is the idea that normative facts are grounded in natural facts. As we’ve seen when exploring inter-level metaphysics, there are many sub-cases. We’ll distinguish two here.
Reductive normative naturalism has it that normative properties are not just grounded in natural properties, they are natural properties, in the sense that the right bridge laws can reduce the former group to the latter. A very simple version would be a direct equivalence like "to be a good thing is to elicit joy" or "to be beautiful is to be symmetrical". Of course, a serious attempt in this direction would result in an ultra-Leibnizian stance, whereby goodness or beauty would be defined as fearsomely complex algebraic combinations of physico-mental primitives. In other words, we would need to argue that there are a “formula for good” and a "formula for beauty" with which an omniscient being could evaluate every normative proposition.
Synthetic normative naturalism says that normative properties are grounded in, but not reducible to natural ones. The relationship between the two would then occupy the space that’s broadly referred to as emergence. That is not a great fit for our own model, at least at first sight: the principle that properties, relations and abstract objects are built algebraically enshrines reduction as the primary type of grounding between properties and tends to conflate emergence with independence, which is the next position we’ll examine.
The profligate route eschews the idea of emergence indeed, to simply conclude that normative properties form a metaphysically primitive category, another category of being altogether. Because of the existence and independence of normative properties in that worldview, it is often called normative Platonism. That can be quite easy to accept in a religious or mystical context. Taking the example of monotheism, normative properties are commonly related to God (e.g. what is obligatory is what God wants), and since God and God’s "cognitive states" are already considered to be beyond the natural world, normative concepts have a cozy home to themselves already. It is also possible to accept this outside of religious settings: that would entail the addition of new "tables of the law" to the existing "tables of natural law", namely a normative complement to those causal and descriptive principles that govern our universe.
The epistemology of norms
This is a basic philosophical question, but also a profund one that has been at the core of belief systems of all eternity: how do we come to know what’s good or what’s bad? There’s not enough space here to delve into the complexities of this topic; suffice to say that the answers to previous questions will come to determine the realm of epistemic possibility.
Moral expressivists and formalists could readily consider normative epistemology to be moot since they defined normative judgments on the basis of actual mental states (conative ones for the expressivist, knowledge of a formal system for the fictionalist).
Reducive naturalists would rely in our times on evolutionary psychology: natural properties we come to see as synonyms for "good" must be related causally to some traits that’s been selected as advantageous for our species. Knowledge of "good behavior" is thus pre-programmed in subjective agents through their genetic heritage. Pure naturalists would perhaps suggest that a belief in universal values is a hallucination that self-conscious beings make up when confronted with the evidence of that behavioral bias.
As to Platonists themselves, they must contend with the same issue as realists of every obedience: if some norms are neither mental nor physical, then how can we physico-mental beings know of them?