W7: Original Draft

Moral Intuitions

Is it a problem for a normative ethical theory if it is in conflict with widely felt moral intuitions?

(January 2025): This was the first draft of my Week 7 essay, which I then edited to make move slower (as that was the main feedback that I’d gotten from my tutor for previous essays). It seems to have worked! But here’s the original, to illustrate the difference.

Here is one version of an argument (CONFLICT) that it is a problem for a normative ethical theory φ if it conflicts with widely felt moral intuitions (WFMI):

(1) It is a problem for φ if it is probably extensionally incorrect.¹

(2) φ is probably extensionally incorrect if it is in conflict with WFMI.²

(3) So, it is a problem for φ if it is in conflict with WFMI.

To defend some particular theory φ (e.g., Mill’s utilitarianism), one might accept CONFLICT but argue that φ doesn’t conflict with WFMI. A more general line is to attack (2) with an evolutionary debunking argument.³ If successful, this means CONFLICT fails to establish (3). However, this would not refute (3), since there might be other reasons to believe it. Here is a sketch of two:

(C1) It is a problem (COMPLIANCE) for φ if it is difficult to comply with.

(C2) φ is difficult to comply with if it is in strong conflict with WFMI.

(A1) It is a problem (ACCEPTANCE) for φ if it is difficult to accept.

(A2) φ is difficult to accept if it is in weak conflict with WFMI.

Note that these are two distinct problems: an implications of some φ might be easy to comply with but difficult to accept (e.g., ‘everything is permitted’) or vice versa (e.g., ‘we are morally obliged to give up meat’). However, as we’ll see, acceptance might be a special type of compliance.

Establishing (C1) and (A1)

On my usage, a positive theory describes how things are or could be, while a normative theory prescribes how things should be.⁴ So, a positive ethical theory may ‘describe our moral sense’,⁵ i.e., our ‘judgements in reflective equilibrium’,⁶ or perhaps ‘what we would be motivated to do if we were vividly aware of the relevant facts’.⁷ But this is not yet normative; a parallel normative theory would prescribe acting in accordance with that moral sense.⁸ Plausibly, a successful description only needs accuracy and simplicity. CONFLICT attacks the former. But I take it that a successful prescription needs, perhaps among other things, to be accepted and then complied with. It seems right to say: ‘The problem with φ is that nobody accepts it!’ (or similarly with ‘complies with’). Presumably, if φ recommends x, it’s a problem on φ’s own terms if people don’t accept that they should do x, or if people don’t end up doing x. So, on this understanding of normativity, it is a problem for a normative ethical theory if it will not get broad compliance, or if it will not get broad acceptance.

So, to establish (C1) and (A1), we must bridge from (a) φ is difficult to accept to (b) φ will not get broad acceptance (and same for compliance). This is fairly simple: by (a) I simply mean that by default φ probably will not be broadly accepted; i.e., it’s very likely by default that (b) will obtain. The relative weakness of this link means that the problems I raise are defeasible. Advocates of φ might need to spend quite a bit of effort promoting acceptance or compliance. This might range from ingraining patterns of thought to establishing social and political institutions that encourage acceptance or compliance.⁹

Establishing (C2) for strong or asymmetric conflict

By strong conflict between φ and WFMI, I mean some action x being obligatory under one but impermissible under the other. This reading comes from the intuition that two principles don’t really conflict if one can always comply with both simultaneously. Of course, if there are moral dilemmas (e.g., if on both φ and WFMI, Sartre’s pupil is obligated to defend France but also to stay home for his mother), then φ and WFMI are in strong conflict with themselves as well as each other; but this edge case supports the idea that φ is difficult (impossible) to comply with. More generally, I take it that people are strongly motivated toward acts which are obligatory under WFMI (e.g., calling an ambulance if a parent collapses), and strongly motivated against acts which are impermissible under WFMI (e.g., kicking a baby). Now, if φ strongly conflicts with WFMI on x, then either x is impermissible under φ but people are strongly motivated toward it, or x is obligatory under φ but people are strongly motivated against it. Either way, complying with φ requires overcoming this strong motivation from WFMI. This is difficult because, by default, I’m likely to act on my strong motivations, and very unlikely to act against them. This establishes (2) and thus COMPLIANCE for strong conflict.

Strong conflict is sufficient for COMPLIANCE, but not necessary. We can asymmetrically weaken our reading of ‘φ is in conflict with WFMI’ such that it is also satisfied by either some act y being obligatory under φ but strictly permissible under WFMI (e.g., donating a kidney), or else being impermissible under φ but strictly permissible under WFMI (e.g., consuming animal products). This reading is asymmetric because WFMI would not be in conflict with φ on y. This reading also sounds reasonable to my ear.¹⁰ Now, if φ asymmetrically conflicts with WFMI on y, then either y is obligatory under φ but people aren’t morally motivated toward it, or y is impermissible under φ but people aren’t morally motivated against it. I’m unlikely to always do (avoid) something if I don’t have any default reason to do (avoid) it. And sometimes, I’ll even have strong nonmoral reasons against doing (avoiding) it. Either way, complying with φ probably requires building sufficient motivation if it doesn’t come from elsewhere. So φ is difficult to comply with, so (2) and thus COMPLIANCE obtains for asymmetric conflict as well.

However, asymmetric conflict does seem necessary for COMPLIANCE. We can weaken our reading such that ‘φ is in conflict with WFMI’ is satisfied by some act z being strictly permissible under φ but either obligatory under WFMI (e.g., saving your child over a stranger) or impermissible under WFMI (e.g., defiling a corpse). We can’t establish (2) with this weak reading, since we comply with φ whether or not we do z (since it’s strictly permissible under φ).

Establishing (A2) for at least minimal conflict

ACCEPTANCE applies for weak conflict. We are strongly attached on the basis of WFMI to the claims that z is obligatory or that z is impermissible, and accepting φ requires rejecting these claims. By default, we are unlikely to reject these claims. So, φ is difficult to accept. This establishes one version of (4) and thus ACCEPTANCE.

We can push ACCEPTANCE further, but we need to be more careful. At first pass, accepting φ looks like a special case of complying with φ’s prescription to accept it. This assumes that every theory prescribes its own acceptance. However, some theories may be self-effacing;¹¹ that is, they might not prescribe their own acceptance, and might even prescribe their own rejection. So, let φ* be the beliefs that φ prescribes adopting. It might be that most normative ethical theories prescribe their own acceptance (i.e., φ* is identical to φ). But it seems plausible that some ethical theories do not have this feature. It might be permissible or even obligatory on some φ to retain WFMI (this includes rejecting the fact that WFMI are ultimately grounded in φ). In this case, where φ* and WFMI are identical, it seems like an abuse of language to say that φ and WFMI are in any sort of conflict. But so long as this isn’t the case, there is some minimal amount of conflict between φ and WFMI. Then, we can argue for (2) as above, based on the difficultly of rejecting intuitions to which we are strongly attached. Thus, ACCEPTANCE is a problem for any φ which even minimally conflicts with WFMI.

Further issues

Here is one worry about ACCEPTANCE. We might think, on the basis of something like pessimistic induction, that WFMI are surely wrong. Some (perhaps many) WFMI of the past were wrong by our lights today, so the WFMI of today will likely be wrong by the lights of future generations. One obvious candidate for such a WFMI is the permissiblity of eating meat. And it seems right that the WFMI of future generations will be more considered than ours, and thus more likely to be correct. Thus, WFMI are probably wrong about some things. But to avoid ACCEPTANCE entirely, a normative ethical theory must prescribe its own self-effacement in favor of our WFMI; and so such a normative ethical theory is probably wrong about some things. It also seems unable to correct our current behaviors, which might be something we want from a normative ethical theory. These problems might be much worse than ACCEPTANCE. So, ACCEPTANCE is a good problem to have.

I think the spirit of that worry is correct. But ACCEPTANCE is not a good problem to have. To address this, we must recognize that both the problems I’ve raised come in degrees. The degrees are not determined by what I’ve called the strength of conflict between φ and WFMI. Rather, they are determined by what I’ll call intensity: WFMI may range anywhere from very mildly felt to very intensely felt. If φ conflicts with intensely held WFMI, then the arguments for the two problems I’ve raised become stronger, and they become bigger problems. Note that strength and intensity are two separate considerations: if φ says that it’s strictly permissible to be indifferent between your child and a stranger, then this is a relatively weak (asymmetric) but very intense conflict with WFMI. So, COMPLIANCE and ACCEPTANCE are more serious problems when they are in especially intense conflict with WFMI, and relatively trivial problems (perhaps even good problems to have) when they are in only mild conflict with WFMI. Even if they are serious problems for φ, they aren’t automatically good grounds to think that φ is false. But as I’ve discussed above, they are still problems for the normative, prescriptive success of φ.

Recap

It is a problem for a normative ethical theory φ if it is in conflict with widely felt moral intuitions. One argument to this effect is CONFLICT. We can situate the debate about intuition and the methodology of normative ethics here. But even if CONFLICT is undercut by that debate, there might be independent lines toward its conclusion. I advance two of them: COMPLIANCE and ACCEPTANCE. Here are revised versions of the arguments. (Of course, read ‘difficult’ as ‘difficult for most people’ and ‘we’ as ‘most people’.)

(C1.1) It is a problem for φ if it will not get broad compliance.

(C1.2) φ will not get broad compliance if it is difficult to comply with.

(C2.1) φ is difficult to comply with if we have must overcome or create strong motivations to do so.

(C2.2) We must overcome strong motivationsto comply with φ if it conflicts strongly with WFMI, and create strong motivations to comply with φ if it conflicts asymmetrically with WFMI.

(COMPLIANCE) It is a problem for φ if it conflicts strongly or asymmetrically with WFMI.

(A1.1) It is a problem for φ if φ* will not get broad acceptance.

(A1.2) φ* will not get broad acceptance if it is difficult to accept.

(A2.1) φ* is difficult to accept if we must reject strong attachments to do so.

(A2.2) We must reject strong attachments to accept φ* if φ conflicts even minimally with WFMI.

(ACCEPTANCE) It is a problem for φ if it conflicts even minimally with WFMI.

References

• Parfit, Derek, and Samuel Scheffler. On What Matters. Volume Two. Oxford: Oxford University Press, 2011.
• Parfit, Derek. Reasons and Persons. Oxford: Clarendon Press, 1986.
• Rawls, John. A Theory of Justice. Revised edition. Cambridge, Massachusetts: The Belknap Press of Harvard University Press, 1999.
• Sandberg, Joakim, and Niklas Juth. “Ethics and Intuitions: A Reply to Singer.” The Journal of Ethics 15, no. 3 (2011): 209–226.
• Singer, Peter. “Ethics and Intuitions.” The Journal of Ethics 9, no. 3/4 (2005): 331–352.

Footnotes

1. i.e., wrong about what things are right/wrong, good/bad, etc. This might not automatically disqualify φ; e.g., it might still be useful (cf. Newtonian mechanics as a physical theory). ↩

2. e.g., extensional correctness might be determined by moral judgements in reflective equilibrium, which strongly-held WFMI are likely to persist through. ↩

3. If an EDA isn’t selective enough to undercut only actual moral intuitions, it risks undermining morality (or even human reasoning) in general; this might mean that advocates of EDAs are companions in guilt. See Singer for an EDA which Sandberg & Juth argue isn’t selective enough. ↩

4. See Part Six of Parfit for a defense of this usage, i.e., that natural theories are never normative; also, what I call prescriptions are more specifically external reasons. ↩

5. Rawls, p. 41 ↩

6. Rawls, p. 43 ↩

7. Attributed to Darwall in ‘The Unimportance of Internal Reasons’ in On What Matters. ↩

8. See ‘Substantive Subjective Theories’ in On What Matters. ↩

9. See the last chapter of Kagan, which is dedicated to dealing with the compliance problem. ↩

10. e.g., in set theory, ‘GCH is in conflict with ZF on AC’ vs. ‘ZF is in conflict with GCH on AC’. ↩

11. See ‘How S Might Be Self-Effacing’ in Reasons and Persons. ↩