1     Reconfiguring the Realm

Philosophy valorizes truth. There may be practical or prudential or political reasons to accept a known falsehood. But there can, it is held, never be epistemically good reasons to do so. Nor can there be good reasons to accept modes of justification that are known not to be truth conducive. Such is the prevailing consensus. Although it seems reasonable, this stance has a fatal flaw. It cannot account for the epistemic standing of science: for science unabashedly relies on models, idealizations, and thought experiments that are known not to be true. Modern science is one of humanity’s greatest cognitive achievements. To think that this achievement is a fluke would be mad. So epistemology has the task of accounting for science’s success. A truth-centered, or veritistic, epistemology must treat models, idealizations, and thought experiments as mere heuristics, or forecast their disappearance with the advancement of scientific understanding. Neither approach is plausible. We should not cavalierly assume that the inaccuracy of models and idealizations constitutes an inadequacy; quite the opposite. I suggest that their divergence from truth or representational accuracy fosters their epistemic functioning. When effective, models and idealizations are, I contend, felicitous falsehoods. They are more than heuristics. They are ineliminable and epistemically valuable components of the understanding science supplies.

These are bold claims. In what follows I develop a holistic epistemology that accommodates the cognitive contributions of science. It acknowledges that tenable theories must be tethered to the phenomena they concern, but denies that truth is the sole acceptable tether. It maintains that felicitous falsehoods relate to their targets via exemplification, the relation of an example to the features it is an example of. It urges that exemplification provides at least as strong and stable a link to the phenomena as truth does. Although models, idealizations, and thought experiments are inaccurate—sometimes wildly inaccurate—they exemplify features they share with their targets and thereby afford epistemic access to aspects of their targets that are otherwise overshadowed or underemphasized. They are true enough. They advance understanding of the phenomena they bear on. To accommodate science, I advocate shifting epistemology’s focus from the knowledge of individual facts to the understanding of a broader range of phenomena. In science and other systematic inquiries, relatively comprehensive constellations of epistemic commitments stand or fall together.

Once truth is dethroned, truth-conduciveness falls as well. Many epistemologists hold that arguments and methods of inquiry are justified by being reliably truth conducive. If so, in deploying them and eschewing arguments and methods that are not truth conducive, we increase our prospects of epistemic success. Having sidelined truth, I can make no such claim. This raises an important question about epistemic normativity. If truth-conduciveness does not vindicate a method or form of argument, what does? I develop a deontological account of epistemic norms. I argue that acceptability turns on epistemic responsibility rather than on reliability. I provide criteria for objectivity that are satisfied by representations that respect deontological epistemic norms. Worries about subjectivity fade, but there is, even in principle, no guarantee that conclusions that are backed by objective reasons are true. What a tenable account does guarantee is that conclusions are reasonable and defensible in the epistemic circumstances.

The position that emerges not only secures the epistemic standing of science, it accords the same status to art. For the symbolic devices that underwrite the acceptability of felicitous falsehoods in science are also deployed in the arts. Thus, I urge, the arts embody and advance understanding if the sciences do. Throughout the book, discussions of the arts and the sciences intertwine. My position vindicates the oft expressed but seldom justified claim that, although works of art are not true or even truth apt, we learn something important from our encounters with art.

In places, I frame my position by criticizing ways others address the same issues. I make no attempt to provide a comprehensive survey of the available alternatives. I discuss other approaches only when and only to the extent that the juxtaposition illuminates the points I seek to make. Nor do my criticisms purport to be refutations. Like Quine (1961), I believe that we can hold fast to any contention come what may, so long as we are willing to make compensatory adjustments elsewhere in our system of thought. One can, for example, continue to insist that scientific theories are unacceptable unless true, or that to understand is to be able to explain, or that science is cognitive but art is not. But, I will urge, the cost of such tenacity is surprisingly high. It is high enough to make it reasonable to entertain my alternative.

The shape of the book is as follows: I begin by arguing that if epistemology is to accommodate science, it must weaken its commitments to truth and truth-conduciveness. After examining a variety of widely accepted scientific models and showing that standard veritistic tweaks, such as construing them as approximations, fail, I introduce the idea of a felicitous falsehood—an inaccurate representation whose inaccuracy does not undermine its epistemic function. I do not contend that we should believe felicitous falsehoods. Rather, building on L. Jonathan Cohen (1992), I maintain that we should accept them. This is a matter of being willing and able to use them as a basis for inference or action when our overarching interests are cognitive.

To accommodate the fruits of science and other systematic inquiry, I urge, epistemology should shift its center of gravity from knowledge to understanding—in particular, to what has been called objectual understanding, the understanding of a topic or subject matter (Kvanvig, 2003). Throughout the book, that is what I shall mean when I use the word ‘understanding’ without qualification. Understanding is not mirroring. It is neither a matter of believing a host of separately justified individual propositions nor a matter of believing a network of interconnected propositions that mirror the facts. To understand a topic involves knowing how to wield one’s commitments to further one’s epistemic ends. It involves being able to draw inferences, raise questions, frame potentially fruitful inquiries, and so forth. Since some of the commitments are likely to be felicitous falsehoods and others to be methodological or normative commitments that are not truth apt, understanding is not factive. It is not a type of knowledge; it does not consist exclusively or primarily in believing or accepting truths.

Because the commitments that comprise an understanding are mutually supportive, understanding must be construed holistically. I show how the mutual support of independently doubtful items can increase their credibility. Drawing on my previous work, I argue that an understanding of a topic consists in accepting a system of commitments in reflective equilibrium (Elgin, 1996). A network of commitments is in reflective equilibrium when each of its elements is reasonable in light of the others, and the network as a whole is as reasonable as any available alternative in light of our relevant previous commitments. Even if some components would be doubtful in isolation, collectively they constitute an interwoven tapestry of commitments that we can on reflection endorse.

What differentiates a felicitous falsehood from an infelicitous one? If truth and truth-conduciveness are sidelined, what replaces them? Rather than being epistemic reliabilists, I urge, we should be epistemic responsibilists. Acceptable epistemic norms are norms that would emerge from the deliberations of suitably idealized epistemic agents. Extrapolating from Kant (1981), I urge that such agents are legislating members of a realm of epistemic ends. The realm of epistemic ends is an idealization of real-life communities of inquiry that eliminates factors that distort actual deliberations. I identify and justify the constitution of the realm, the factors legislators can draw on, and the conditions on their deliberations. I argue that legislators must be, in a political sense, free and equal. They must be free to advance any hypothesis they consider viable and equally entitled to be heard. This does not mean, of course, that they must be equally intelligent or that their views are to be treated as equally cogent or equally worthy of acceptance. It only means that views are not to be rejected without a fair hearing. The political requirements on a realm of epistemic ends fix the framework within which the distinction between good and bad reasons, methods, strategies, and conclusions can be responsibly drawn.

I go on to argue that intellectual integrity requires that certain values—those that figure in or contribute to trustworthiness—are simultaneously moral and epistemic. Here again I focus on science. Since nature is enormously complicated, epistemic agents need to work together to achieve an understanding of it. Since even the best methods are not failsafe, the resources for identifying and correcting received errors must be readily available. Scientific misconduct undermines the enterprise by producing results that are untrustworthy. Because the value of trustworthiness is integral to science—because a scientist needs to trust the findings she builds on—I argue that in teaching what good science is and how good science is done, science education typically imbues novice scientists with certain moral/epistemic virtues. Other disciplines do the same for their practitioners.

As I’ve construed it, epistemic responsibilism may seem too weak to secure objectivity. Can a discipline claim to be objective if it recognizes that there is an ineliminable human element in whatever vindicates its claims? I argue that the sort of objectivity that epistemology should seek is not that delivered by the ‘view from nowhere’ (Nagel, 1986) or the ‘absolute conception of reality’ (Williams, 1978). It is an objectivity that protects against idiosyncrasy, bias, and chance. Drawing on van Fraassen (2008), I argue that it is often perspectival. I discuss two modes of objectivity. One is impersonal. We define magnitudes and develop ways of measuring them that omit the personal element entirely. Or anyway, they eliminate the personal or interpersonal element once the work of defining magnitudes and developing measurement procedures is done. The other is impartial. Although someone must be doing the assessing, it makes no difference who that (suitably trained) someone is. The results of both are trustworthy and would be endorsed by legislating members of a realm of epistemic ends.

Having argued for a holistic account of understanding and explicated the norms that understanding is subject to, I return to the question of what vindicates felicitous falsehoods. I urge that a falsehood or inaccurate nonpropositional representation is felicitous only if it exemplifies features it shares with the phenomena it bears on. By making those features salient, such a representation enables us to appreciate their significance for the phenomena. I explicate exemplification and show how it figures in ethics, mathematics, and dance.

I go on to argue that laboratory experiments, thought experiments, and works of fiction also advance understanding by means of exemplification. Because laboratory experiments involve pure samples tested in isolated, pristine environments, they distance themselves from the natural phenomena they bear on. They eliminate or control for irrelevant features in order to highlight relevant ones. Thought experiments involve further distancing. They contrive imaginative scenarios where situations that are not realizable in fact can be entertained and their implications investigated. This enables them to bring to the fore factors that are typically overshadowed. Literary fictions, I urge, are elaborate, extended thought experiments. In all three cases, delicacy of discrimination is needed to determine which features are exemplified and how they project back onto the facts.

Effective models exemplify features they share with their targets. They streamline, eliminating irrelevant factors so that features of interest stand out. This enables them to reveal occluded aspects of their targets and to disclose how seemingly disparate phenomena are alike. Evolutionary game theory, for example, shows that, from the perspective of natural selection, predator—prey relations and reciprocal altruism are the same sort of thing. Following Hughes (1996), I urge that a model represents its target as having the feature the model exemplifies. This, I argue, is analogous to what occurs in portraiture.

If legislating members of realms of epistemic ends set the constraints that bind them, how rigid are those constraints? What prospect is there for interdisciplinary understanding? I suggest that the restrictions set by different disciplines set strong presumptions but not rigid requirements. This allows for epistemic cross-fertilization. We can develop epistemic hybrids that bridge disciplinary divides.

Epistemology is haunted by the worry is that despite our best efforts we still might be wrong. Nothing in my position immunizes against this worry. The most I can say is that if a consideration is multiply tethered, it is relatively secure. That is far from a guarantee. The permanent possibility of error leads many to embrace fallibilism. This is not so straightforward as it might appear. The fallibilist is burdened by a version of Moore’s paradox (Moore, 1991). ‘I know that p, but I might be wrong that p’ has the air of giving one’s word in one breath and taking it back in the next. But if we repudiate fallibilism, and insist that we should dogmatically endorse a true conclusion that reaches the threshold for acceptability (wherever that threshold is set), we confront Kripke’s (2011a) dogmatism paradox: If I know that p, then p is true. In that case, any evidence that ~p is misleading. I ought therefore to disregard any evidence that ~p, since if I take it seriously, I am apt to lower my credence and lose my knowledge. That being so, when I know, I ought to be dogmatic. I ought to turn a deaf ear to evidence that tells against p. But to do so seems epistemically irresponsible.

Taken together, the two paradoxes strike me as serious problems for a theory of knowledge. They do not undermine the possibility of knowledge. Neither individually nor jointly do they point toward skepticism. But they undermine the value of knowledge. Concessive knowledge, which tolerates Moore’s paradox, and dogmatic knowledge, which deafens the knower to objections, both undermine the right to be sure. An epistemic agent would rightly be reluctant to accept testimony or to take a big risk on the basis of claims to knowledge that were accompanied by Moore’s paradoxical or Kripke’s paradoxical riders. But, I argue, fallibilism with respect to understanding does not face these difficulties. It is not a doctrine, but a stance—indeed, I urge, an epistemically valuable stance. Recognizing where and how we might be wrong discloses something significant about the fabric of our understanding. Possibilities of error, then, are not merely—and perhaps not mainly—vulnerabilities. They are avenues of insight into our epistemic situation.