There’s no such thing as normative harmony
Psychophysical Harmony #5
This post is part of a series. See other parts of the series here:
The argument for psychophysical harmony that Cutter and Crummett present in their forthcoming paper appeals to normative harmony. Yet their account of normative harmony seems to rely on a realist conception of normativity. If it didn’t, then it’s unclear why they’d take error theory to be a possible objection to it. However, error theory isn’t the only viable basis for rejecting normative harmony, and they offer little in the way of a compelling response to antirealist objections to normative harmony. Not every paper has to respond to every possible criticism, and I don’t expect them to offer a comprehensive critique of every possible antirealist response.
Nevertheless, I would like to see them provide a more substantive response to antirealist objections to normative harmony. Here, I want to discuss a short exchange on the error theorist’s response to normative harmony that I believe illustrates this concern:
After describing normative error theory, Crummett states at ,53:01
First of all, that view seems wrong.
Green replies
It seems like it could just stop there. Like well, but that’s wrong.
Crummett replies:
I humbly suggest that view seems wrong.
(1) Underspecificity of seemings
People discussing philosophical topics often make claims that something “seems true” or “seems false,” or is intuitive, counterintuitive, and so on. Such remarks are rarely accompanied by explicit commentary on who they seem true or false to, who something is intuitive or counterintuitive to, and so on.
A given claim can’t seem true or false, or intuitive or counterintuitive (and so on) in principle, because the seemings, intuitions, and the like are features of the agents evaluating the claims, not the claims themselves. Claims that a given remark seems true could reflect how things seem to the person making this claim, but they could also reflect claims about how things do or would seem to other people.
If it’s the former, this is not especially informative. Simply because something seems true to you might provide you with some minimal, defeasible evidence that it’s true (a claim some would contest), but it provides much less evidence to anyone else. That it seems to some people that there are ghosts or magic powers is not good evidence to anyone else that there are such things.
If it’s the latter, such claims are empirical. In some cases, a claim can be trivial enough that others may not contest the claim. If someone says that it probably seems to most people that fire is hot, I wouldn’t argue the point. But philosophical claims are only interesting insofar as they say something interesting and potentially mistaken. The alternative is to simply presume one’s claims are true and dismiss anyone who thinks otherwise: hardly a philosophical attitude.
Here, Crummett and Green seem content to dismiss error theory for “seeming wrong.” There are a variety of questions we could ask about this:
What proportion of people likewise find that it seems wrong? What proportion of people find that it seems true? What proportion have no inclination one way or another?
What causal factors are associated with it seeming true, false, or neither?
Error theory doesn’t seem wrong to me for the same reasons it presumably seems wrong to Crummett or Green. I think it seems wrong because I think it relies on false semantic and/or empirical claims about ordinary normative discourse. Note that error theory is a conjunction of two claims:
(i) Normative claims attempt to describe stance-independent normative facts.
(ii) There are no stance-independent normative facts, so all such claims are false.
I endorse (ii), but not (i). Error theory “seems wrong” to me because I reject (i), but it likely seems false to Crummett and Green because they reject (ii).
This last observation brings me to my main objection:
(2) Antirealists have more options than just error theory
The main issue with Crummett’s characterization of possible ways to reply to the argument from normative harmony is that it’s a mistake to focus specifically on error theory. Note that error theory is the conjunction of a semantic thesis (normative claims refer to stance-independent normative facts) and a metaphysical thesis (there are no such facts), from which it follows that all such normative claims are false.
However, error theory isn’t the only option for an antirealist who rejects normative harmony. An antirealist can both deny that there are stance-independent normative facts, and reject error theory. Objections to error theory don’t necessarily apply to other forms of antirealism. If Crummett and Green are both moral realists, then they may maintain that all other forms of antirealism seem wrong as well, but note that, in that case, at least some of what Crummett says before rejecting error theory wouldn’t apply to such accounts.
According to Crummett, a normative error theorist would say that:
“there is no reason to pursue pleasure or avoid pain, really. In fact, there is no evidential support relation between the contents of my consciousness and [...] my behavior.”
First, a minor technical point. Normative error theorists are not required to deny that there are evidential support relations. Quite the contrary: proponents of error theory argue that evidential support relations are sufficient to account for their epistemic practices without appealing to categorically normative epistemic reasons. For instance, Cowie appeals to exactly this point as an objection to companions in guilt arguments. First, Cowie asks us to imagine a departure board which indicates that your train will leave at 8:25. We might take this to indicate that we have a normative reason to believe the train leaves at 8:25, and we might even take this to mean that we have a categorical normative reason. Yet as Cowie (2016) points out:
In the present dialectical context, however, this argument would be of little use. Moral error theorists are already committed to denying the existence of categorical reasons for action [...] (p. 117)
From this, it follows that:
The moral error theorist will simply deny that she possesses a categorically normative, epistemic reason to believe that her train leaves at 0825. She will deny this precisely because she denies the existence of categorical reasons. What she will not deny is that the departure board provides evidential support that the train will leave at 0825. But she will deny that this entails the existence of a categorically normative, epistemic reason for belief. (pp. 117-118, emphasis mine)
Error theorists can deny that evidential support relations entail or just are categorical epistemic reasons.
This isn’t the main issue, however. After pointing out that error theorists would say there’s no reason to pursue pleasure and avoid pain, and so on, Crummett goes on to state at ,52:45 that the normative error theorist would maintain that such examples of normative harmony
[...] are spurious because in fact none of this stuff I said is rationally appropriate … it’s not really rationally appropriate, it’s not rationally inappropriate, it’s just that that’s just a mistake [...]”
Crummett immediately follows this with the remark that such a view “seems wrong.”
Why does this view seem wrong? Given the context, it would appear that it seems wrong because the normative antirealist denies that there is no reason to avoid pain or pursue pleasure, that there are no evidential support relations, and so on.
Setting aside that normative error theorists might, in fact, say some of these things, if they appeal to (what they regard as) revisionary (perhaps reductive) or fictionalist accounts, there are two problems here.
First, as always, I suspect the reason why these views seem implausible may be due to ,normative entanglement. Error theory may seem wrong because people who are considering it mistakenly think that it has normative or practical implications that it doesn’t have. An error theorist who endorses the view that “we have no reason to avoid pain and pursue pleasure” is not thereby committed to the view that they’re making some kind of mistake if they avoid pain or pleasure pleasure. They are not thereby any less motivated to avoid pain or to pursue pleasure. They are not thereby saying or doing anything inconsistently if they express a dislike of pain (theirs or anyone else’s) and a preference for pleasure. A normative antirealist can act exactly like a normative realist. They simply don’t think there are facts that entail that they “ought to” do so that are true independent of their (or anyone else’s) stances.
People may balk at error theory because it entails that one thinks a sentence like “pain is bad” is false. But this is misleading, and it is misleading in large part because people are failing to disentangle what the error theorist is saying from what they aren’t.
All the error theorist is actually saying is that there are no stance-independent normative facts according to which pain is bad, and this is compressed into and represented by the statement “pain is bad.” Error theorists are not, as a consequence, committed to any of the aforementioned implications one might infer given this position. It is purely a view which holds that, given a particular semantic account of the meaning of certain propositional claims, such propositional claims are false. It does not require that the error theorist doesn’t care about pain, or whatever else one might infer about the normative or psychological characteristics of the error theorist. I suspect the mistaken presumption that it has such entailments drives much of the balking and incredulity and rejection of error theory.
However, even if we did reject error theory (and, after all, I do!), why is error theory presented as the only antirealist objection to normative harmony? There are a variety of antirealist positions that are also capable of dispensing with normative harmony. It seems strange to focus on one.
(3) Not a solution to every problem
Note that Crummett said “First of all…” but then never follows this up with anything else. This is unfortunate. When presenting an argument, one would hope that there’d be arguments against precisely those views which could resist the argument. Simply saying that the positions in question seem wrong doesn’t do much to achieve this.
Instead, by 53:18, we’re told that:
Even if you hold that view it doesn’t actually solve the problem because you still have cases of semantic harmony.
Sure, it doesn’t solve that problem, but that’s not a problem error theory could solve in principle because it concerns a different topic altogether. Of course it can’t resolve it, any more than the theory of evolution can solve questions in fundamental physics.
And we shouldn’t expect someone responding to the argument for psychophysical harmony to appeal to one philosophical stance to deal with normative and semantic harmony. One can simply appeal to distinct philosophical positions each of which addresses the relevant proposed form of harmony. In this case, any view according to which there are no psychophysical laws will do the job. This could be achieved by any account which denies that there are phenomenal states, physical states, or both: if one or both doesn’t exist, then there’s no way for either to correlate with the other.