Some Of Our Books


« CFP: The ninth annual Felician Conference on Ethics and Public Affairs | Main | Coming Featured Philosophers »

January 06, 2015


Feed You can follow this conversation by subscribing to the comment feed for this post.

For relevant literature on the issue you raise, see Alex Barber's "Science's immunity to moral refutation" in AJP (

Thanks Alex - extremely helpful!

In the example you give, is my belief, that I ought to slow down, canonically normative? I would have thought that a canonically normative belief would be more along the lines of, I ought not speed to such extent that it is likely people will die as a result. The former but not the latter seems to be epistemically liberal in the sense you mention.

Hi Mert

Yes - I was thinking of the belief that I ought to slow down as a normative belief, full stop. Equally well, the belief that I ought not to speed to such extent that it is likely people will die as a result.

What your example seems to illuminate is the distinction between mixed and pure normative beliefs as Mark Schroeder suggests. Mixed statements have both normative and descriptive content (for example, part of the content of the belief that I ought to slow down might be a representation of the situation in which I am in) whereas it seems like there are also pure normative claims that have only normative content. It might be true that these latter beliefs do not admit both normative and other beliefs with ordinary naturalist contents as evidence and defeaters. It seems like they admit only other normative beliefs as evidence and defeaters, but not the other beliefs with ordinary purely naturalist contents. For my purposes, this would still be fine as it would still mean that there is an asymmetry in the epistemic norms.

Hi Jussi,

I'm having a difficult time seeing what the issue is (not trying to be difficult!) Can't we account for the asymmetry this way: the normative considerations don't have any bearing on the truth or probability of the contents of the normative beliefs but the non-normative beliefs do have bearing on the truth or probability of the contents of the normative beliefs (but, really, only because of background beliefs that connect the two)?

Hi Clayton,

well - I didn't even try to specify any "issue" here. I was just interested above in whether the epistemic asymmetry exists, which you don't seem to contest, and nothing else.

There probably will be many different accounts of why the asymmetry exists. There seems to be something wrong with the beginning of the one you give - it should probably read "the normative considerations don't have any bearing on truth or probability of the content of *non-normative* beliefs". Note that this explanation makes use of metaethically controversial notions such as contents of normative beliefs and their truth. People who are not keen to use these notions in their metaethical views would give alternative accounts.

For what it's worth, I am more interested in the consequences of the asymmetry rather than its causes. I've been looking at debates in philosophy of mind concerning phenomenal concepts. One argument against the reduction of phenomenal concepts to other physicalist concepts is that these concepts are governed by different epistemic norms and the identity of concepts in part depends on such norms. I've been thinking about how this argument would apply in metaethics against the new forms of reductive analytic naturalism.

Hi Jussi,

Is this a counterexample? I hear that you've been telling people looking for cars that they morally oughtn't buy a Marauder. I know that you're well-informed about cars and morality, so I believe that I morally oughtn't buy a Marauder. Recognizing that the usual reason why buying a particular brand of car is wrong is low fuel economy, I infer (correctly) that the Marauder has low fuel economy.

Isn't that a case of a normative/moral belief providing evidence for a nonnormative proposition? You might object that in this scenario it's the fact you've been making these judgments, rather than the proposition you've been asserting, that is my evidence. But your testimony is only evidence on the assumption that it's reliable, so the evidential path seems to go through the moral fact.

Hi Jussi,

I'm a little skeptical that there will be straightforward asymmetries, since I'm a holist about belief justification. This leads to a few thoughts about your examples:

In the case of normative beliefs evidentially supporting non-normative beliefs: I might be sure I'm driving in an 80 km/h zone and suspect I'm driving at about 100 km/h without being able to check my speedometer to be sure. Then I pass an apparent speed trap. The police cruiser does not pull out after me. So I think, "I guess I'm not doing anything wrong; so I am probably going closer to 80." (Of course, you could bring in the police officer's belief about norms instead of a normative belief itself, but I don't see why this would be obligatory here.)

Inferences like this are ubiquitous. Suppose I'm watching philosopher A and philosopher B argue, and A makes a remark that seems so cutting that my immediate inference is that A aims to hurt B's feelings with it. This inference might be undermined if I think A is a kind person or a very respectful interlocutor. Then I might wonder if maybe A and B have a history of teasing each other with similar remarks, so that there is a context for the cutting remark that makes it less likely to hurt, or if there was a slip of the tongue, etc. My non-normative belief about intentions thus gets reasonably pushed around by my beliefs about normative facts about A. (If you like, make the belief about A more blatantly normative: A wouldn't do something so flagrantly wrong here.)

Likewise, one might think, "well, since it's pretty clearly wrong to do X to my client, X is also probably forbidden by my professional association's code of conduct," even if one does not remember that part of the code of conduct. Again a normative belief is pushing around a non-normative one (a belief about what's written in a particular document).

Notice that what goes for the straightforwardly normative goes for the not normative but evaluative (if you think there's such a category). Take biological health and biological deformity. One might not think that beliefs about what one is seeing under a microscope ought, rationally, to be pushed around by beliefs about what is biologically healthy or in good condition. But it's reasonable to do so sometimes. I believe I'm looking at the first sample, and I seem to see very unusual clumping of cells. This clumping, if real, strongly implies that the cells are diseased. That might reasonably push against my belief that I really have the first sample under my microscope, and against my belief that I'm seeing genuine clumping (maybe the slide was prepared in a way that generated clumping as an artifact). Why would this be reasonable? Well, maybe sample 1 comes from a randomly selected 18 year old US college student, and the disease in question is very uncommon among such.

In all these cases, of course, there are beliefs that are not themselves normative that sit in the background, or come to the foreground, that are obligatory for these inferences to be reasonable: beliefs about the expected rate of norm violations and satisfactions in the relevant context, or about expected human responses to detected violations, or the like. But then, for beliefs about observations of natural phenomena to reasonably influence other beliefs about natural phenomena, there often need to be such bridging beliefs too. So again there's no special asymmetry that I can see.

Final thought: I didn't touch on purely normative beliefs, but there are several ways to think about them. If one is, e.g., a cultural relativist, such beliefs will be nicely integrated into one's larger web of belief. If one is a very strong realist, one typically holds that pure normative beliefs are like pure mathematical beliefs - which are not normative but not beholden to one's ordinary empirical beliefs either. In both cases, there's no asymmetry. Maybe there's a kind of metaethics that does commit one to asymmetry - would that be a mark for or against such a metaethical theory?

Hi Steve

that's an interesting case. Few thoughts:

1. I think I can live with few individual cases. All I need is that there is an asymmetry in the types of evidence that *typically* or *canonically* are relevant for when normative and other beliefs can be held with warrant. This would still be true even if in some rare and perhaps unusual cases normative beliefs provide warrant for ordinary beliefs.

2. It also seems to me that what is doing a lot of work in your example is the auxiliary premise belief that 'if I morally ought not to buy a particular brand of car, then that car has a low fuel economy'. Now, I accept that in the light of such bridge-principles normative beliefs can be evidence for other ordinary beliefs (given that I can use modus ponens). However, I think that I can say that what is relevant for my purposes here are the differences in the basic epistemic norms that entitle us to move from one atomic belief to another without the bridge-principles. And, it doesn't seem like your case works without the bridge-principle.

3. It also seems to be that perhaps there is a greater difference on the defeater side. Suppose that I have ordinary, non-normative evidence that Marauder has a good fuel economy. It doesn't seem like in this situation this belief can be undermined by my belief that I morally oughtn't buy a Marauder - even when I think that this is usually this would be because cars it is wrong to buy have a low fuel economy.

Hi Tim

thanks for the really good cases. I am sympathetic to holism too but I was thinking that asymmetries in the epistemic norms would be compatible with forms of holism (it's not that holists need to accept that anything can be evidence for anything).

About your cases: in the first it seems to me that 'I guess I am not doing anything wrong' means I am not violating any norms police care about. The latter version of the second case and the third case of the code of conduct seem best to me. I need to think about these cases more, but initially my reaction to these cases are both 2. and 3. in my reply to Steve.

Jussi, what are some examples of basic epistemic norms that entitle us to move from one atomic belief to another without bridge principles? Is the "I ought to slow down" example supposed to be an example of naturalistic evidence supporting a normative belief without bridge principles?

Hi Jamie

good question. I naively thought that it would be. The idea was that you could believe that you ought to slow down with warrant merely on the grounds that you believe that you are doing 80mph even if you didn't hold a principle of the sort [you ought to slow down if you are doing 80mph]. The idea is that you can form justifiedly normative beliefs on the basis of reasons even if you don't subscribe to any general bridge-principles (as you can also ordinary non-normative beliefs).

Now I am less certain about this after Tim's illuminating comment. He seems to be right in his examples that in some cases we do form ordinary non-normative beliefs on the basis of normative beliefs. It seems to me that in these cases there are back-ground principles involved - generalisations about what kind people tend to do, what professional code's conduct tend to be like morally speaking, why buying certain cars tend to be wrong and the like.

But, if I insist that such back-ground beliefs are always at work in the cases where normative beliefs are evidence for the non-normative beliefs, to get an asymmetry I would need to be able to argue that the bridging back-ground principles are not necessarily present in cases where ordinary non-normative beliefs are evidence for one another. For this purpose, I would need good examples of the basic norms you asking examples of. I am starting to suspect that defending such norms inside the non-normative realm and insisting that they must be present when we move from normative beliefs to non-normative beliefs with warrant will be harder than I initially thought.

Jussi, your asymmetry might not sit well with a Bayesian view of evidence, or confirmation. On this view, X confirms Y if Pr(Y|X) > Pr(Y). But this makes confirmation symmetric. That is, X confirms Y if and only if Y confirms X.


I apologize for introducing my own jargon here, but it seems to me that this puzzle might be cleared up by distinguishing between two different kinds of normative beliefs: beliefs about what I call pure normative facts and beliefs about mixed normative facts. (I make this distinction in Chapter 2 of Being Realistic about Reasons.) Facts about what I have reason to do depend on my situation, hence are "mixed" but the fact that if someone were to be in my situation he would have such reasons is a pure normative fact, which does not depend on non-normative facts. It is mixed normative beliefs for which both normative and non-normative beliefs can be evidence or defeaters. This is not true for pure normative beliefs, which are related in these ways only to other normative beliefs.

Hi Jussi and Tim,

I was also thinking about Tim's distinction between pure and mixed normative facts. But I wonder if Tim is right that it is only mixed normative beliefs for which both normative and non-normative beliefs can be evidence or defeaters. If we are aiming, not only for a "narrow," but also a "wide" reflective equilibrium, we might think about our pure normative beliefs partly in terms of their natural history. Though I would probably agree that no non-normative belief (or its content) could be a direct reason in favor of a pure normative belief, it seems plausible to think that a non-normative belief (or its content) could cast some amount of doubt on a pure normative belief. I myself am a little skeptical towards challenges against non-consequentialist pure normative beliefs based on the sort of considerations that Josh Greene and Peter Singer and others find convincing. But I think it at least makes sense to think - as they do - that the kinds of considerations they appeal to can cast doubt on certain pure normative beliefs.

Hi Campbell,

thanks. I'm not sure what to say about Bayesianism. It's true that this theory describes what kind of combinations of beliefs are rational - it sets certain rational restrictions. You are also right that these constraints themselves might not set any asymmetries in evidence. I wonder, however, whether the Bayesian standards capture all the epistemic norms there are. If not, then Bayesianism too might leave room for the type of asymmetries I have been suggesting. One way to think about this is that the Bayesian norms seem to me to be state requirements whereas I was thinking of process requirements (in the terminology of the Broome/Kolodny debate).


thanks - that distinction is very helpful and I did rely on it in my earlier response to one of the commentators. It still supports the asymmetry (and in a way makes it even stronger). The initial claim was that ordinary, non-normative beliefs only admit other similar beliefs as evidence and defeaters. If mixed normative beliefs admit both normative and non-normative beliefs as evidence and defeaters and pure normative beliefs only other normative beliefs, then the epistemic norms that govern holding normative beliefs with warrant and the norms that govern other beliefs would still be significantly different. This would suggest that normative concepts are not reducible to naturalist concepts given that the norms that govern their application are different.

This still seems right to me but I struggle with the response from Tim Schroeder above.


thanks. I can live with that. It's ok for me if pure normative beliefs admit both kind of defeaters (and perhaps even warrant).


For what it's worth, I'm convinced there is an asymmetry between reasoning about normative and non-normative matters, but I'm not sure I would draw it in the same way.

In what sense is "It is wrong to impose a risk on others" evidence for "I ought to slow down?" I would think that evidence for "I ought to slow down" would be propositions like "I am driving so fast that I'm imposing a risk on others." The (normative) fact that "It is wrong to impose a risk on others" is maybe the explanation for why "I am driving so fast that I'm imposing a risk on others" counts as evidence for "I ought to slow down." But just because X explains why Y is evidence for Z, that doesn't mean that X is evidence for Z. Right?

Here is another difference between reasoning about normative propositions and reasoning about non-normative ones that might be fodder for your project:

Generally, when someone is justified in believing some normative proposition, their reason for believing it is the reason it is true. E.g. I believe that it is wrong for me drive recklessly. I believe this because driving recklessly would endanger lives. Since my reason for believing it is also the reason it is true, I'm justified in my belief.

Contrast this with the case where I believe it is wrong to drive recklessly because my grandmother said so. That my grandmother said so isn't the reason it is wrong. Though I might have (some kind of) evidence in this case, seeing as how she is pretty reliable on moral mattes, there is something defective about my justification.

This phenomenon doesn't hold in non-normative cases. Right now I believe it is 20 degrees outside because my thermometer says so. It seems like I'm perfectly justified in this belief, even though my reason for believing that it is 20 degrees (i.e. the thermometer says so) is not the reason why it is 20 degrees.

Thanks for the post, Jussi. My dissertation is in this area and the thread is full of helpful ideas.

The comments to this entry are closed.

Ethics at PEA Soup

PPE at PEA Soup

Like PEA Soup

Search PEA Soup


  • Unless otherwise indicated, the views expressed in any given post reflect the opinion of only that individual who posted the particular entry or comment.