2. Moral and non-moral targets

Matheson and Milam (Reference Matheson, Milam and Timmons2022) attempt to show that non-moral blame is never justified, in that it is never a morally permissible response to a non-moral failure. They agree with the realist that people do exhibit blaming reactions in response to seemingly non-moral failures and note that our existing practices of non-moral blaming have several similarities to our moral blaming practices. Given this, the challenge is to identify a disanalogy between the two. Their approach is to show that, since there is ‘a conceptual connection between blameworthiness and obligation such that one is blameworthy only if one fails to meet an obligation’ (204), there are no non-moral obligations. Their method for achieving this in turn is to argue that all cases of putative non-moral blame are actually either moral blame, or unfitting.Footnote ³

Their first line of argument for thinking that many instances of non-moral blame are, in fact, moral is to note that ‘when challenged about the permissibility of their non-moral blame, blamers often fall back on different reasoning’ (p. 205). For example, when asked why a missed goal warrants athletic blame, a blamer might point to things such as the player’s huge salary, the fact that he should have practiced more or that his shot demonstrates a lack of care, all of which Matheson and Milam think are moral grounds.

The efficacy of this line of reasoning is dubious because most people are not adept at distinguishing questions of the circumstances that are sufficient for blame to be fitting with questions of what justifies blaming in those circumstances (indeed, Matheson and Milam do not make this distinction either). In response to ‘Why are you blaming him?’, the reply ‘Because he missed the shot!’ is a perfectly intelligible and fitting answer. What justifies blaming poor performances more generally is another question altogether, and most people are not particularly well-versed in the range of different options one might consider when trying to ground different types of obligations or blaming responses.Footnote ⁴ What would be needed is for the sceptic to show that no such grounds are available, and that would require engaging with questions of the foundations of normativity. (We shall return to this question in §5.) Consequently, this line of argument does not succeed against the epistemic blame realist. That a moral consideration in favour of blaming can be identified as present doesn’t thereby show that that consideration is what is doing all the work.

Teasing out what is doing the work is attempted in Matheson and Milam’s second line of argument, which is that ‘our attitudes seem to change when we strip away the moral features of a particular case’ (p. 205). If we consider cases in which someone makes a poor artwork, or engages in shoddy reasoning, but in which nothing significant turns on this particular performance, Matheson and Milam think that it seems harder to sustain an intuition that the agent is blameworthy.

However, this method threatens to bottom out in a mere trading of intuitions, which we can see by imagining arguing with someone who was convinced that only non-moral blame was ever justified. (Perhaps they had been inspired by Socrates’ argument that no-one ever knowingly does evil.) Suppose you encountered a community of philosophers who were convinced that all justifiable blame only ever targeted epistemic, aesthetic, or athletic (etc.) failures. There are plenty of such considerations to go around, e.g. the rationality of committing a murder often involves hasty reasoning, and people find dead bodies to be aesthetically grotesque. Suppose now you tried to convince them that their worldview was mistaken, and that blame actually targets moral failings, not non-moral ones. They would worry that your assessments were being confounded; that while you took yourself to be blaming moral failings, you were actually responding to these other types of failings. Given this, they would reason that to test your theory, we would need to consider cases involving purely moral failings with no non-moral considerations present whatsoever.

I suspect there would not be many cases both sides could agree meet this criteria, and of those that remained, it wouldn’t be difficult for these philosophers to report an intuition that blame didn’t seem especially warranted. Of course, we would find their position ludicrous, and the considerations that they pointed to as ‘really doing the work’ to be mistaken. But this shows such a line of argument doesn’t do much to advance the debate, so much as merely shore up one’s pre-existing position. Because the non-moral blame realist takes both non-moral and moral blame to be justified, the method of trying to fully ‘control’ for confounding factors and then restricting our attention to a very narrow range of cases in which it appears uncostly to report a particular intuition is something that only the sceptic can propose. Showing why this method is suspect requires seeing how it looks when made by someone who rejects a type of grounds for blameworthiness that we accept, which our imagined community of moral-blame-insensitive philosophers helps illuminate.

If we want to make some dialectical progress in establishing what considerations are doing the work, a viable method would be to reflect on our actual practices, or find some independent consideration from which to establish some kind of burden-of-proof. But these both seem to favour the non-moral blame realist: as Matheson and Milam note, people do seem to blame others for non-moral failings, and more importantly frequently take themselves to be doing so. The non-moral blame realist thus seems to better capture the phenomenology of our blaming interactions. When we blame an athlete for, say, missing the goal, we are blaming him for his poor performance. When the athlete then apologises, they don’t apologise for some kind of wrongdoing or moral transgression, they apologise for missing the goal. This is enough to mean that the sceptic’s interpretation of our practices, then, is somewhat revisionist.Footnote ⁵

Because appealing to intuitions about the blameworthiness of agents in particular cases (or the blameworthy-making features of such agents) threatens to result in a stalemate between the epistemic blame realist and sceptic, an alternative approach is needed. This is what Smartt (2024) attempt to provide, presenting a number of arguments against epistemic blame realism, which it will be useful to consider individually.

3. Force and frequency

Smartt’s first argument is that attempts to explain what force is provided by epistemic blame end up with an implausible picture of our practices of epistemic criticism. Here he focuses on Brown’s (Reference Brown2017, Reference Brown2020) belief-desire account of epistemic blame, itself built upon Sher’s (2005) account of moral blame. Brown takes blame to consist in a belief that the agent has believed badly, a desire that they had not believed badly, and a range of behavioural and affective dispositions which this belief-desire pair then manifests. When I blame someone for having inconsistent beliefs without a good excuse, for instance, I believe that they have believed badly, I desire that they had not done this, and this results in a range of dispositions (e.g. disappointment and reduction of trust) which count as blame. Smartt’s target, however, is broader than this. In the moral blame literature, a large number of accounts take blame to involve some kind of negative emotional reaction towards wrongdoing. Since it is plausible that we experience a range of negative emotional reactions towards agents who violate epistemic norms (qua epistemic norms) without an excuse, it is plausible that this is a distinct type of blame. My own (2021) account of epistemic blame similarly takes certain emotional responses (e.g. disappointment, frustration, non-retributive anger) to be forms of epistemic blame, and so any objections to emotion-based accounts will undermine this account too.

Smartt’s objection concerns how well claims about our epistemic blaming responses map onto our interactions with putatively epistemically blameworthy agents. He argues ‘The emotion-based account predicts that when we learn about an agent’s epistemic norm violation—say, they have probabilistically incoherent credences, or believe something on insufficient evidence, or form a belief via an invalid inference—we’re disposed to feel frustrated, angry, upset, or feel other negatively valanced emotions towards them’ (p. 1816). But, he notes, most epistemic norm violations don’t generate these kinds of emotions. And in those cases where negative emotions do seem appropriate, factors beyond epistemic blame can explain our reactions.

Epistemic blame realists can respond to this challenge, but let’s first distinguish two questions that seem to be being run together: why epistemic blame seems, on the whole, less frequent than moral blame, and why certain instances of epistemic blame seem less fitting than moral blame.Footnote ⁶ Accounting for the reduced frequency of epistemic blame is simple for realists: in those circumstances where we don’t blame, we didn’t care about those particular violations (or at least, didn’t care enough to experience any blame). These accounts of epistemic blame don’t predict anything about how frequently we will experience blame upon learning of a norm violation, and nor do they need to. The same obtains for moral norm violations: I might fail to experience moral blame in response to hearing about the latest political scandal on the news, but that doesn’t mean that the perpetrators aren’t morally blameworthy, or even that I don’t judge them morally blameworthy. It’s just that, for whatever reason, I don’t particularly care about them; perhaps I am disenchanted with politics. If it ever came to arguing that moral blame exists, what would matter is that there are plenty of cases where I do, in fact, experience moral blame, and that I have a principled reason for thinking that people can be morally blameworthy in those cases. A similar move can thus be made concerning the frequency of epistemic blame.

When we turn to examining the blame that people do feel (and its fittingness), Smartt thinks there are problematic disanalogies between moral and epistemic blame, in that only the former involve a victim and strong negative emotions. But it seems to me that if anything, the analogy holds: given there’s a difference in whether there’s a victim, we should expect some difference in the content and affective strength of moral and epistemic blame, and this is precisely what we observe. There’s a difference in how strongly we experience the reactions, but it isn’t prima facie problematic as there are also differences between kinds of moral norms and how we respond to violations of those. Some moral norm violations are so trivial that they also don’t generate much resentment, but this isn’t a problem for thinking of the emotional reactions that do occur as a form of moral blame.

When we try to explain away the difference in strength between moral and epistemic blame by pointing to the lack of a victim, Smartt worries that only we end up with another disanalogy, namely that moral norms communicate or signal information about victims, whereas epistemic norms don’t, despite epistemic blame realists sometimes motivating their account by appealing to the communicative elements of blame. But they need not take it to be an essential feature of blame that it involves concern for a victim. Why not think that blame targets and communicates disapproval of norm violations, and that moral norm violations often involve victims while epistemic norms don’t? Identifying that there is a disanalogy here between moral and epistemic blame doesn’t point to any problem, unless we should independently think that blame needs to communicate information about victims, or unless we should think that blame doesn’t target norm violations.Footnote ⁷

We can weaken Smartt’s framing by noting that many moral norm violations simply have no victim, and yet still warrant moral blame. Perhaps the wrongdoing was an act that conflicted with some abstract value that generates substantive duties, or violated something like the categorical imperative. Perhaps the agent failed to pull their weight in a collective action problem, but success with the problem was over-determined. Perhaps the agent had no effect on the victim, but acted with poor quality of will, such as sadistic voyeur. Perhaps in acting wrongly, the ‘victim’ was actually benefitted and was grateful that the agent acted wrongly!

In response to worries that we don’t need our account of blame to involve communicating concern for victims, Smartt notes that the case for epistemic blame seems strongest when the norm violation does have negative effects on other agents or their informational environment. But arguably it only seems strongest insofar as we lack an account of epistemic blame’s frequency and fittingness. Once this is on the table, there is little left to explain, and proponents will say the case for epistemic blame realism seems stronger. We already saw we can account for why people don’t frequently experience epistemic blame by saying they don’t care about those particular violations. We can likewise account for the cases Smartt thinks are strongest by saying they happen to be the instances in which people care more about the violations in question. It is indeed a legitimate question why we care about these cases more. Epistemic blame proponents can accept that such cases feature moral blame too, or that some epistemic norm violations are more important than others. Sceptics, in turn, might argue that moral considerations and moral blame are doing all the work, but epistemic blame realists are aware that there are multiple factors here potentially affecting our intuitions and have provided careful analyses of thought experiments to make their case (e.g. Piovarchy, Reference Piovarchy2021, p. 792–793; Brown, Reference Brown2020 pp. 2–3, 12–13; Boult, 2021, p. 521; 2023; p. 819; Goldberg, 2017, §4). As we saw with Matheson and Milam, if sceptics want to do more than report a different set of intuitions, they need to either engage with the realists’ treatment of these cases or hang their argument elsewhere.

Smartt’s first objection, then, doesn’t necessarily stick. However, the epistemic blame sceptic might nevertheless be on the road to their goal, because Brown’s account is known to face a number of objections. In particular, it seems one can desire that others not believe badly and have this manifest in certain ways without being blamed, one can blame without having the desire that someone not have believed badly, and there are some components of Sher’s account of moral blame that cannot be transplanted to epistemic blame (Boult, Reference Boult2023, §2; Piovarchy, Reference Piovarchy2021, pp. 797–799). If there is a way to also undermine competing accounts of epistemic blame, namely those from Boult and myself, the sceptic might succeed yet. Fortunately, Smartt has more objections.

4. Co-ordination and redundancy

Smartt’s next argument is that much of the work that epistemic blame is supposed to do is already carried out by epistemic evaluation, which plays two important roles. First, evaluation enables Deference. In saying ‘Bel knows whether the Prime Minister is in Sydney’, I mark Bel as a trustworthy epistemic surrogate, and thus worthy of being deferred to by others in the epistemic community. Second, evaluation provides Compliance Pressure. When I mark Bel’s belief as irrational, I not only mark her as untrustworthy, I pressure her to comply with the norms of the community. This, in turn, enables co-ordination, as their performance improves.

Smartt argues this particularly poses a problem for Boult’s (Reference Boult2021a, Reference Boult2021b, Reference Boult2023) account of epistemic blame, which is built off of Scanlon’s (Reference Scanlon2008) account of moral blame. The details of Boult’s account – which have several moving parts – do not actually matter for understanding and engaging with this objection.Footnote ⁸ All that matters is that Boult takes epistemic blame to aim at ‘promoting epistemic goods, like believing truly and avoiding believing falsely’ (2021, p. 5).

Smartt (2024, fn. 10) also has my (2021) account in his sights with this objection. On my account (itself building on Vargas’s (Reference Vargas2013) account of moral blame), epistemic blame starts off as an external motivator (pressure) to comply with epistemic norms, but gradually, over time, causes agents to internalise those norms. Epistemic blame helps to cultivate a distinctively valuable kind of agency, one that is responsive to epistemic norms. The difference between moral and epistemic blame’s natures is accounted for by the fact that the norms we are trying to sensitise other agents to in blaming to are quite different in character, with epistemic norms notably lacking any victims, but both rely on blame’s unpleasant features to achieve their goal.

Smartt’s complaint is that since epistemic evaluation already enables deference and provides compliance pressure, then epistemic blame is not adding anything. ‘To the extent that epistemic blame includes commitments to roles like Deference and Compliance Pressure, it fails to be all that different to epistemic evaluation’ (p. 1819–1820). This particularly means that the accounts of epistemic blame from Boult and myself fail to provide any required force or sting, commonly taken to be a desideratum on a theory of blame.Footnote ⁹

Smartt’s argument, I believe, is too hasty in its reasoning. Showing that evaluations can, in principle, play this role to some degree is not the same as showing that they alone do, in practice, play that role sufficiently.Footnote ¹⁰ We can see this by going back to the moral case. Moral evaluations also provide compliance pressure. Claiming that someone did the wrong thing, or is evil, or callous, or negligent, exerts some form of social pressure. But it would clearly be too hasty to infer that as a result of this, moral blame does not add anything beyond what is already provided by moral evaluations. Responding to a wrongdoer by simply evaluating them seems like a poor method for making them understand the significance of their wrongdoing and motivating them to apologise. It is one thing to show that epistemic evaluation provides pressure, but another to show that it provides pressure to the degree and in the same manner as existing accounts of epistemic blame.

This reply seems even stronger when made on behalf of the agency-cultivation account, which gives a more explicit role to compliance pressure. To be sure, there are many factors which can improve the chances that an agent will internalise norms. But we cannot infer from this that all factors work equally well in all contexts, which is what would need to be shown for charges of redundancy to stick.

Additionally, showing that epistemic blame’s role is already played by something else is not the same as showing that epistemic blame does not exist. To get to the latter, Smartt argues that we should prefer epistemic blame scepticism on grounds of parsimony, avoiding introducing idle wheels that do the same work as existing ones. But there is an important ambiguity here in what it means to ‘do the same work’. Epistemic evaluation and epistemic blame might ‘do the same work’ in the sense of both potentially being able to achieve the same end, given our collective interest in designing systems and practices that result in agents complying with epistemic norms. But parsimony is usually invoked in explanations or models which are aiming to account for our observations, and we observe some important differences in how people evaluate and how they blame, which count against thinking that evaluation alone can do all the work. Smartt elides a distinction in how it is that Deference produces coordination across an epistemic community. Evaluations primarily seem to be for the benefit of others; when I call someone irrational, I am letting others know not to rely on them. But when I engage in epistemic blame, modifying my relationship, the primary change is my willingness to rely on them. That these can come apart is clear, because I can sincerely call and evaluate someone as irrational, and yet still subsequently rely on them without experiencing any blame. Alternatively, I can epistemically blame someone, and yet refrain from making any public evaluations of them whatsoever. If epistemic blame is an idle mechanism, then Smartt needs to invoke some other mechanism to explain why these responses appear to be different, or show they are the same response simply described in different ways.

That evaluations of blameworthiness (whether moral or epistemic) do not seem reducible to evaluations of reliability is evidenced by the fact that we can intelligibly say (and observe others saying) things like ‘he did the wrong thing (believed badly; is too irrational to be relied upon), and so I’m not going to rely on him anymore, but I don’t blame him for it’ (cf. Coates and Tognazzini, Reference Coates, Tognazzini, Justin Coates and Tognazzini2013, p. 9–10).Footnote ¹¹ If Smartt is right that epistemic blame doesn’t add anything to evaluations, then such locutions become rather puzzling, which is another observation that would need to be accounted for. Conversely, reporting that one blames another agent, having already identified them as irrational, seems to make one guilty of redundancy. Worse still, if – as I pointed out above – Smartt’s point generalises to moral blame, then people who take themselves to be engaging in blame (understood as something distinct from a mere assessment of reliability) seem to be confused about what they are experiencing.

The potential redundancy in claims like ‘He’s irrational and I blame him for it’ also suggests a way that Smartt may be hoisted by his own petard. Recall that the purpose of evaluation is to mark different kinds of sources according to different criteria. On Dogramaci’s (Reference Dogramaci2012) picture of evaluations, which Smartt endorses, different evaluations thus serve different functions. If we notice that there is a distinct kind of evaluation, then it is probably serving some kind of function that isn’t being fulfilled by other kinds of evaluations. But this is precisely the kinds of grounds that should make us think that ‘epistemic blameworthiness’ is a distinct property which it is useful to keep track of!

Smartt’s second objection, then, is unsuccessful against either Boult’s account or my own. However, his third argument is much more promising.

5. How many kinds of blame are there?

A final objection from Smartt is that allowing there to be a distinct, ‘epistemic’ type of blame opens the door to there being too many other kinds of blame. Here he charges Brown’s approach with being too permissive: its structure allows for too many other kinds of blame that might stem from analogous belief–desire pairs, such as prudential blame (experienced when others violate prudential norms), or aesthetic blame in response to violations of aesthetic norms. Indeed, it can be hard to see why we should not allow all manner of different kinds of blame; if there can be ‘office furniture norms’, why not allow there to be ‘office furniture blame’ directed at office furniture ‘norm violations’?

This objection is stronger than I think Smartt presents, because it could generalise to other accounts of epistemic blame. Boult’s account depends on there being a distinct kind of relationship between agents according to some normative ideal. It certainly seems that athletes on sports teams have a kind of relationship with one another as teammates, and they definitely blame one another for poor sporting performances. One can even blame one’s rivals, if they break the rules, or even give up too easily. There could, in principle, be some kind of normatively guided aesthetic relationship between agents in the art world, for instance, comprising a set of intentions and expectations about how agents will behave with regard to some aesthetic standards. Perhaps some artists rely on one another to meet a certain set of aesthetic standards, and modification of this relationship on the basis of perceived faultiness in how one conducts themselves could be considered to be a form of blame? If Boult takes epistemic blame to be concerned with a modification of trust that someone will live up to the standards of the epistemic relationship, why not allow agents to have a modification of trust that they will live up to other kinds of relationship standards too?

The agency-cultivation account is also not immune to this objection. I’ve previously argued that epistemic blame is justified by its ability to cultivate a distinctively valuable kind of epistemic agency. But we need not think that epistemic and moral agency are the only valuable varieties. It is valuable for people to be better athletes, artists, philosophers, chess players, dancers, politicians, etc. To be sure, one can get better at each of these by improving their epistemic agency, but improving epistemic agency alone is far from sufficient. Each of these fields has distinct standards, methods of improvement, cultures, roles, and internal set of goods, and agents can experience blame in any of them. Why not think there are also different types of blame present here then? If my account were also unable to answer this, Smartt’s scepticism might succeed after all.Footnote ¹²

If moral blame exclusively targets moral failings, it seems we need to either claim that blame in other domains is always unfitting (because not moral), that blame in other domains actually targets some hidden moral properties, or that there are distinctively non-moral kinds of blame. None of these options seem appealing. As we’ve seen, saying that blame is always unfitting outside of the moral domain would make our theory somewhat revisionist given our actual practices of blaming people who believe badly and warrant an explanation for why blame is only ever justified towards moral wrongs or bads. Saying that there are hidden moral reasons doesn’t seem to quite fit the phenomenology of our blame, and some examples (e.g. the aesthetic domain) seem not particularly moral. But allowing for distinctive classes of blame, as the epistemic blame proponents would like, risks opening the door to all manner of types of blame.

I believe the third option has the most promise, because the unappealing consequence does not follow: we can allow for some non-moral types of blame without risking all manner of types to propagate. Here is how: there are moral norms, there are epistemic norms, but there is no distinct class of office furniture norms, or at least, not any that matter. ‘Norms’ in the sense we are concerned with are not merely potential kinds of reasons, or standards that one could theoretically point to relative to some domain. They are authoritative standards. They are the kind of thing that we collectively recognise as important and uphold as communities.

There is a tendency to sometimes think of epistemic norms as simply being what would characterise good reasoning and states (‘don’t hold inconsistent beliefs’), or to think of them in reference to some ideal theory, e.g. what a perfect Bayesian would believe. But this is not the relevant sense when it comes to thinking about blameworthiness and types of blame. Norms need to be the kind of thing that we normatively expect agents to follow, the failure of which licenses negative responses given the circumstances that the agent was in. An analogy will help. Suppose that consequentialism is, in fact, the correct normative theory. This alone does not entail that agents who fail to maximise good consequences are thereby blameworthy. Consequentialists regularly made a distinction between what is wrong and what is blameworthy as a defence against charges of demandingness (e.g. Morris, Reference Morris2017). Returning to the epistemic domain, we need to take care to not move from noticing that a certain belief or inference violates some epistemic standard without an excuse to inferring that the agent is thereby an appropriate target of blame. Their violation needs to be the kind of thing that we expect people to live up to, and which reflects badly on them as an agent, not merely reflects on them as a less than perfect being.

Returning to Smartt’s objection: it is not enough to merely observe that there could be some other set of non-moral and non-epistemic standards, or that some people think that there are some sets of standards which people can fail to comply with. One needs to show that these are the kinds of things that we expect people to recognise and treat as authoritative and think reflect badly on them independent of any moral or epistemic considerations that warrant experiencing the kinds of reactions we associate with blame.

One might ask why moral and epistemic norms have this authority, but others do not. It is worth noting that this question is essentially asking about the foundations of normativity, and it seems an unfair standard to insist that epistemic blame realism should be considered tenable only if realists have first solved that issue. Admitting epistemic norms have a bona fide authority (Grimm, Reference Grimm, Haddock, Millar and Pritchard2009) does not compel us to admit that other kinds of merely possible norms thereby are actually authoritative too, and thus nor are we compelled to say that there are other kinds of blame.

Appealing to distinct types of norms doesn’t particularly help Brown. Given she justifies her account with reference to what blame is (a belief-desire pair), the role that epistemic blame plays in some epistemological debates, and instances of people blaming others for believing badly, there is nothing here to rule out people having other kinds of belief-desire pairs and blaming others for other kinds of violations. Nor does it seem to help Boult: there is no inherent feature of moral/epistemic relationships and moral/epistemic norms as he understands them to limit other kinds of relationships allowing for other kinds of blame. But this answer does dovetail particularly well with the agency-cultivation account in Piovarchy (Reference Piovarchy2021). Though I argued there that agency cultivation is the goal of blame, this is not justified for the agent’s sake, but for the sake of the community. In later work, I’ve been more explicit about this (Piovarchy, Reference Piovarchy2023; Reference Piovarchyms), and clarified that I now think it’s more accurate to think of blame’s function as upholding valuable norms. This is a minor extension: while I originally argued that we blame people to cultivate their agency, the kind of agency we want is one that is responsive to authoritative moral and epistemic considerations; that is, one which complies with moral and epistemic norms.

This story is compatible with what is called social epistemic instrumentalism, which has been developed by others in a number of places (Dyke, Reference Dyke2021; Fleisher, Reference Fleisher2025; Hannon & Woodard, Reference Hannon and Woodardforthcoming; cf. Chrisman, Reference Chrisman2022; Boult, Reference Boultforthcoming). The basic idea is that moral and epistemic norms are particularly important for the functioning of communities, given how reliant we are on each other for our well-being. We need people to treat us well, and to be reliable sources of information, and agreed-upon standards help maintain the systems that provide these benefits. When we blame someone, we are saying that they failed to comply with the norm, and that they need to improve their behaviour. We are signalling that the norm matters, and communicating that we will not tolerate continued violations, thereby providing an incentive for them to comply, and helping them learn what the contours of the norm are. Once they have become competent with the norm and internalised it, they are then able to help maintain valuable forms of both moral and epistemic co-operation (cf. Henderson & Graham, Reference Henderson, Graham, Fricker, Pedersen, Henderson and Graham2019). They not only treat others well and avoid engaging in destructive behaviour, they also act as a reliable source and conduit of information, whether that be on matters they have first-hand experience with or regarding information they have heard from others.

Note this explanation makes sense of our ambivalence or disagreement about aesthetic and athletic blame: our uncertainty stems from whether we take the norms in those fields to be authoritative. Sports teams need to uphold certain standards in order to function well. If athletes and fans think that sporting excellence is a genuinely valuable activity worth cultivating among professional athletes, it makes sense that the former would blame poor performers who could reasonably be expected to do better. In contrast, while many artists care about aesthetic standards and expect others to uphold them, philistines like myself can find it difficult to see what the fuss is about and accordingly won’t take such norms to be authoritative.Footnote ¹³

This does not provide a full account of when norms are valuable or authoritative, or under what conditions other sets of norms that we find alien could come to have genuine importance. But epistemic blame proponents do not need a comprehensive account of the foundations of normativity in order for epistemic blame realism to be viable, any more than moral blame theorists need to first demonstrate that moral realism is true. It is enough to have made clear what the possibility of there being types of blame will depend upon, and provided a plausible answer at that. Blame functions to uphold norms, moral and epistemic norms are genuinely authoritative, so there are distinctly moral and epistemic forms of blame.Footnote ¹⁴ Other merely potential types of norms don’t serve any especially valuable function in communities, nor are they recognised to, so it makes sense that there won’t be corresponding types of blame. The connection between norms and blame, however, only fits well with the agency-cultivation account, which also happens to be the only account of epistemic blame capable of avoiding each of Smartt’s objections.Footnote ¹⁵ The epistemic blame realist, then, lives to epistemically blame another day.