Non-Ideal Epistemology and Vices of Attention

ABSTRACT

McKenna’s critique (rather than criticisms) of idealized approaches to epistemology is an important contribution to the literature. In this brief discussion, I set out his main concerns about more idealized approaches, within and beyond social epistemology, before turning to some issues I think he neglects. I suggest that it’s important to pay attention to the prestige hierarchy in philosophy, and to how that hierarchy can serve ideological purposes. The greater prestige of more abstract approaches plays a role in determining what issues and what voices we attend to.

KEYWORDS:

• Social epistemology
• ideal theory
• prestige
• ideology
• attention

Non-Ideal Epistemology and Vices of Attention.

‘All models are wrong, but some are useful.’ George Box

The world is complex, and we are finite. To understand it, we must create models that abstract away from much of the detail. We must choose, first, which aspects of reality to attempt to model, and then – because those aspects are almost certainly causally downstream from many different variables, and we can’t include them all in our model – which of the many variables that might be relevant to focus on. If we’ve chosen well, the model will still be wrong, in multiple ways. We can’t achieve fidelity to reality; rather, we aim to construct models that are not importantly wrong (Box 1976).

Philosophers don’t usually think of themselves as constructing models of reality, but modelling is not a bad (ahem) model for much of what we do. We, too, aim to understand particular aspects of reality, and must make choices regarding what to include and what to exclude when we try to home in on the aspects we’re interested in. A thought experiment might be thought of as a sort of model. It’s common to complain that thought experiments are excessively simplified (why are those people on the track in the first place?), but simplification is a feature, not a bug. The simplification enables us to home in on the right aspects of reality, bracketing those we’re not (currently) interested in.

Thinking of philosophy as modelling is useful for my aim here, to assess Robin McKenna’s (2023) proposal that an epistemology for the real world will be less idealized than the epistemology that currently dominates so much discussion. Thinking of philosophy as modelling nicely highlights the significance of the choices we must make. Since all models are wrong, we must be alert to whether ours are importantly wrong. Whether a model is importantly wrong is relative to our aims. It’s in this light that we should assess epistemological models. ‘S knows that p’ epistemology abstracts away from the identity of the agent who knows, the proposition they believe, and the social relations in which they’re embedded, but for certain questions such abstractions are appropriate. Such a model may not be importantly wrong, relative to these questions. But it would be importantly wrong relative to others. A model illuminates some aspect of reality; by design, it leaves others in darkness. ‘S knows that p’ epistemology does a great job illuminating some aspects of knowledge and justification, but leaves the power relations between individuals and the social embeddedness of agents obscured.

Non-Ideal Epistemology (McKenna 2023) has as a principal aim arguing that it’s not enough for epistemology to go social; it must also construct models that pay attention to some of the details of agents’ identities and the contexts in which they find themselves. Debates over reductionism in the transmission of knowledge via testimony, for example, no more illuminate the particularities of agents’ identities than does Getteriology. Does that make the models that a more idealized social epistemology works with bad models?

I said above that whether a model is importantly wrong is relative to our aims. Box himself suggests something stronger: a model might be bad without being bad relative to our aims, because our aims might be bad ones: ‘It is inappropriate to be concerned about mice when there are tigers abroad’ (Box 1976, 792). But stated so flatly, that’s wrong. Even in areas where tigers are a threat, mice may not be insignificant at all – they can carry disease and threaten crops. Moreover, while practical benefits are an important factor to weigh in deciding how to allocate our attention, they’re not the only factor. For one thing, pursuing only the research that seems most likely to be beneficial is likely not the best strategy for reaping those benefits. We may be wrong about what questions matter most. Even if we’re right, and even if we should invest most intellectual energy in direct efforts to answer them, responses might benefit from input from people with different interests and different sets of expertise. As a great deal of work in social epistemology has emphasised, cognition is socially distributed, and work in any area might be better because intellectual energy has been invested elsewhere. Finally, knowledge is itself a great good, and we shouldn’t apologise for pursuing it.

So what’s wrong with ideal epistemology? McKenna is explicit that he’s not an opponent of ‘S knows that p’ approaches. He does not take issue with people pursuing fundamental questions concerning the nature of knowledge, for example. Nor does he object to idealization in social epistemology. As he recognizes, idealization may have benefits; indeed, the kind of formal modelling pursued by people like Bright, Zollman and O’Connor surely sheds unexpected light on practical problems (e.g. Zollman 2013; O’Connor, Bright, and Bruner 2019; Arvan, Bright, and Heesen forthcoming). While he insists, rightly, that ideal epistemology obscures certain phenomena that we would expect epistemology to explain, he also accepts that it is not incumbent on every epistemologist to focus on these issues. He accepts a central lesson of social epistemology: good cognition is distributed cognition.

Nevertheless, McKenna has three worries concerning idealized social epistemology.

First, it represents a misallocation of resources. The drunk man of the allegory looks for his lost keys under the lamppost and not where he lost them, because that’s where the light is best. Model building is the construction of the lamppost: a model illuminates a certain aspect of reality. That’s a problem when there are things we should attend to, but few lampposts to illuminate them. It’s not wrong to be interested in mice rather than tigers, but it is wrong for all of us to be interested exclusively in mice when there are tigers about. We live in a world characterized by gross inequality, oppression, and the deliberate distortion of truth. It would be grossly unfair to tigers to call these the tigers of human societies. In any case, they’re phenomena that demand attention.

Second, idealized social epistemology sometimes produces models that are importantly wrong for the purposes to which they’re, nevertheless, put. Whether a model is good or bad depends on the aims of the modeller. Idealized models of testimony transmission, such as those that feature in debates over reductionism, are perfectly appropriate for homing in on some issues, but these same models are importantly wrong when the aim is prescriptive. They’re ill-suited for guiding behavior. By abstracting away from certain messy realities, they make some issues more tractable, but leave obscure too many of the questions relevant to the prescriptions they’re taken to support. They’re good models of something else, and we go wrong when we forget that fact.

Third McKenna suggests that models that are less idealized, and therefore better models of the kinds of phenomena that he’s interested in, may nevertheless issue in prescriptions that are appropriate only for a world that is substantially more just and substantially less hostile than the actual world. In the actual world, the prescriptions they generate may be useless. They may even be worse than useless: a prescription that is apt for some contexts and some agents may make things worse for others or in other contexts. For example, the agent who pursues the genuine good (according to McKenna) of intellectual autonomy might end up less autonomous, given the actual power differentials and the way in which the epistemic environment is shot through with ambushes and traps. The models the social epistemologist produces may be good models of the very phenomena they’re concerned with, but bad models of those phenomena in this environment.

While I take issue with some details of McKenna’s criticisms, I agree with the main thrust of all three criticisms. While ideal epistemology, including ideal social epistemology, is a valuable enterprise, it’s plausible that it occupies too much of our intellectual attention. I’m very confident, moreover, that some prescriptions are generated on the basis of overly abstract models, and that some social epistemologists inappropriately generalize prescriptions that would serve only some agents in some contexts. What follows, then, is friendly criticism, if criticism it is. I aim to highlight two factors that McKenna doesn’t mention, or to which he doesn’t give sufficient weight, and which contribute to the costs of ideal epistemology. To reiterate: there’s nothing wrong with (and a lot right about) ideal epistemology. The costs I point to arise from features of the context in which it’s pursued, and may arise regardless of the intentions or even the conscientiousness of those who pursue it.

The first is the prestige of some areas of philosophy. Some areas of philosophy – feminist philosophy in particular – are still sometimes disparaged as ‘not real philosophy’ (Mackenzie 2019). Though these accusations are less prevalent today than they once were, it remains true that some areas of philosophy are much more prestigious than others. As De Cruz (2018) documents, the so-called ‘Lemmings’ subdisciplines of philosophy – philosophy of language, epistemology, and mind – are massively overrepresented in the most prestigious ‘generalist’ journals, like Mind and Noûs, relative to other areas of philosophy. Importantly, the prestige of epistemology attaches to its ideal form: in general, the more abstract an area of philosophy, the more likely it is to be prestigious, and the more applied the less likely to be prestigious. Non-ideal epistemology, which is less abstract, more applied and – worst of all! – often draws on relatively marginal subdisciplines, like feminist philosophy and philosophy of race, is very much less prestigious than ‘S knows that p’ epistemology.

While – again – there’s absolutely nothing wrong with pursuing the questions that preoccupy those in the Lemmings subdisciplines, the prestige of idealized approaches, and of the questions they illuminate, function to ensure that other issues and other approaches recede into the background. Just as a model focuses attention on certain issues and leaves others obscure, so the hierarchy of prestige in philosophy focuses attention on certain approaches. We can expect the prestige hierarchy to bias social epistemology, in the sense that more abstract approaches will not only get more attention, they will also more readily be drawn on for (often inappropriate) prescriptive uses than the approaches developed by non-ideal epistemologists. The prestige hierarchy may help to explain why, for instance, some epistemologists take their abstract models to be action-guiding, when they’re ill-designed for such purposes. It may also explain why they don’t properly consider whether their prescriptions generalize to marginalized individuals and polluted epistemic environments. Because we orient toward the prestigious, and away from the non-prestigious, we may overlook the lacunas in our own work.

The second factor I want to point to is the ideological function, or at any rate effects, of ideal theory. McKenna is explicitly indebted to Charles Mills and his work on ideal theory; in particular, he draws on Mills work on ideal theory as ideology (Mills 2017a). Mills points to several different ways in which ideal theory can serve an ideological function. McKenna notes one in particular: how ideal theory tends to represent the actual as a mere deviation from the ideal, and therefore not worth theorizing about. But he neglects another emphasised by Mills: the way in which ideal models can represent the interests of a narrow (but hugely overrepresented in philosophy) group of people. People who are marginalized within society more generally are also (and entirely non-coincidentally) marginalized within philosophy: their voices are less likely to be heard at all, or to be heard only in forums that are less prestigious. As a consequence, models may misrepresent the kinds of epistemic situations, and the power asymmetries, they encounter. Models of (say) testimony as an exchange between equals, and in a benign epistemic environment, do not merely idealize testimonial exchange, they present a picture of it that is at best useless when it comes to testimony in a context of epistemic injustice and power disparities.

Ideal theory is entirely legitimate, but its prestige has ideological effects. It allows the interests of certain groups to predominate and ensures that work, and people, that challenge them remain peripheral. It also might serve to obscure epistemic disparities and even help to support their maintenance. As Mills suggests elsewhere, ideology can perpetuate epistemic injustice, by ensuring that certain voices are not heard and that the privileged do not develop the concepts to understand the ways in which oppression impacts cognition (Mills 2017b). It may also underwrite the hermeneutical injustice of preventing members of oppressed groups from developing the concepts they need (see Brancazio and Levy 2024 for discussion). Mills argues that non-ideal theory is superior to ideal theory. Like McKenna, I disagree: which is better depends on our aims. But ideal theory does not deserve to dominate attention and prestige, and its doing so serves an ideological function.

I want to conclude by placing prestige and ideology within a social epistemological framework. Ideal theorists may see these criticisms as harsh and unfair.¹ Most, after all, are well-motivated people, pursuing worthwhile research projects. They don’t see themselves as serving narrow sectional interests. In this light, it’s important to get clear on how prestige can serve ideology, even while those who respond to and reinforce it behave rationally and without any malice or ill-will.

In recent work, Gardiner (2022) has identified the vice of misallocating attention. There are many issues that it is proper to be aware of, but where attention becomes vicious when it exceeds a certain degree. Many of the battles that constitute the culture wars might be understood in this sort of way. It might be fine to wonder whether drag shows are appropriate for children, but when that issue comes to be the centre of one’s political identity, something has gone badly wrong. Drawing heavily on Medina (2013), McKenna argues that agents can be responsible for the vices they manifest. It is therefore open to him to blame agents for the vices of misallocated attention. It is also open to those who accept both his view and Cassam’s (2018) claim that vices are distinguished from mere defects by reference to agential responsibility for them to maintain that philosophers who misallocate attention to ideal theory manifest a vice.

But as non-ideal theory teaches us, it’s important to be attentive to the details of the context in which a putative vice is manifested. While we often think of prestige bias as analogous to racial or gender bias, there are reasons to think it functions quite differently. The prestige of an individual, an institution, or an approach is (higher-order) evidence in its favor (Levy 2021, 2023). Prestige correlates with success, and the fact that something is prestigious is evidence that it works or that it’s worth attending to. Prestige can be both a vehicle of ideology and also function as evidence for those who respond to it. I don’t intend to take a stand on whether agents must be responsible, much less blameworthy, for a character trait or disposition in order for it to count as a vice (as far as I can tell, nothing turns on that question other than terminology). I do want to suggest, however, that there are good reasons to think that those philosophers who pursue ideal theory should not be blamed for misallocating attention. They respond to evidence, and they behave conscientiously.

This is all the more true because in this case the misallocation of attention is not vicious at the individual level. Someone centring their political identity around opposition to drag shows manifests an individual-level fault. But someone pursuing ideal theory may behave entirely appropriately, rationally and morally. It is not their allocation of attention that’s faulty, but the philosophical community’s. We can remedy this misallocation by ensuring that non-ideal theory is taken more seriously, granted more prestige, published in more and more prestigious venues, taught more widely, and so on. Individuals and the profession can pursue this goal in multiple ways, not all of which require reorienting one’s own research. We can pursue a more equitable allocation of attention at the community level; the level where it is most powerful and manifests most clearly. Among its other virtues, Robin McKenna’s work helps us towards this goal: it constitutes a powerful argument that non-ideal epistemology deserves much more prestige and (therefore) our attention.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

I am grateful to the John Templeton Foundation [grant #62631] and the Arts and Humanities Research Council [AH/W005077/1] for support.

Notes

1. Indeed, these criticisms make me uncomfortable. My own work is not clearly non-ideal, in any interesting sense. My appeal to the centrality of the division of cognitive labor may give rise to the impression that my work is squarely non-ideal, but I don’t appeal to it as a compensation for our limitations. In my view, agents with vastly more cognitive resources of every sort available to them would also have good reason to distribute cognitive labor. Perhaps godlike intelligences would not benefit from a diversity of perspectives, but I suspect enormously idealized versions of ourselves would. Given my focus, it’s absolutely fair to say I neglect power dynamics and differentials: as much as anyone else, I need a response of some sort to these criticisms.