True Enough

Catherine Elgin’s True Enough is an impressively wide-ranging, meticulously argued, thought-provoking book. Its thesis is that ‘to accommodate the fruits of science […] epistemology should shift its center of gravity from knowledge to understanding’ (p. 3). For Elgin, an epistemology that aspires to illuminate how science is practiced, and why it manages to be one of the most successful epistemic practices, should be grounded in the concepts of felicitous falsehood, responsible inquiry, and understanding, and should weaken its commitment to truth, truth-conduciveness, and knowledge.

Elgin develops a sustained argument in support of this thesis over the fourteen chapters of True Enough. The point of departure is the observation that scientists routinely devise and use models they know not to be true, or rely on methods they know not to be fully truth-conducive. Such models and methods are ‘true enough’ (Chapter 2); and because of, rather than despite, their inaccuracy, they often deliver understanding of reality. Where the understanding they supply is not a type of knowledge or a type of true belief, their acceptability is grounded in a set of deontological epistemic norms for responsible, historically and socially situated inquiry (Chapters 3–5). These norms specify the moral and epistemic obligations that come with being a scientist. In particular, they set the boundaries of intellectually integrity, and govern trustworthy scientific inquiry that can promote genuine understanding (Chapters 6–8). Inaccurate models supply understanding only if they are suitably tethered to reality by exemplifying it. Exemplification allows felicitous falsehoods to bring to light features they share with reality, displaying the significance of these features for scientists’ epistemic aims and practical purposes (Chapters 9–12). The upshot is that any epistemology consonant with the practices of science should acknowledge there is more to final epistemic value than just accuracy and true belief, and that there is more to epistemic justification than truth-conduciveness.

Let us now consider this argument in more detail. After a helpful overview in Chapter 1 of the distinct threads from epistemology, philosophy of science, ethics, aesthetics, history and art that the rest of the book elegantly weaves together, Chapter 2 introduces the notion of ‘felicitous falsehood’. This is defined as ‘an inaccurate representation whose inaccuracy does not undermine its epistemic function’ (p. 3), and plays a key role in Elgin’s argument. In particular, according to Elgin, scientists’ systematic and successful usage of felicitous falsehoods shows that veritism is false, where ‘veritism’ refers to ‘any position that takes truth to be necessary for epistemic acceptability […] and [holds] truth-conduciveness is the appropriate standard of assessment for epistemic policies, practices, and their products’ (p. 10).

Felicitous falsehoods feature in practices like curve smoothing, ceteris paribus laws, thought experiments, and idealized modelling; familiar examples include Einstein’s elevator thought experiment, the Lotka–Volterra model of predatory–prey interaction, and Monte Carlo methods in statistics and machine learning. If these and other scientific representations are not true—and many scientific methods are not fully truth-conducive—despite successfully playing a variety of epistemically valuable functions in scientific practice, then what’s epistemically valuable in science is not exhausted by accuracy or true belief, and the central concept of epistemic justification need not be truth-conduciveness.

One potential point of confusion here concerns the shape of the dialectic between Elgin and the veritist. In some places, Elgin associates veritism with truth-conduciveness in a way that might lead some readers to conflate veritism and reliabilism. But veritism is a thesis about epistemic value that should be kept distinct from theories of epistemic justification like reliabilism. Veritism is a ‘forward-looking’ view that ascribes epistemic value to practices and rules as a function of the patterns of truth-value attainment they issue. Reliabilism is a ‘backward-looking’ view that ascribes epistemic justification to token epistemic states as a function of events upstream from those states (Goldman [2015]). Veritism is detachable from reliabilism; reliabilism does not entail veritism, as reliabilism by itself does not entail any claim about epistemic worth. So, some readers may wonder why—even if felicitous falsehoods show veritism is false—what makes a felicitous falsehood in science justified (or worthy of acceptance) is not its connection to a process that conduces towards the promotion of true belief and avoidance of false belief.

Elgin addresses this and other questions concerning epistemic normativity in Chapters 3–5, which are the densest in the book and of special interest to ‘traditional’ epistemologists. Chapter 3 argues that the paradigm of epistemic value in many scientific practices should be construed as ‘objectual understanding’. This is defined as ‘having a suitable grasp or take on a topic’ (p. 38). More specifically, objectual understanding is knowing how ‘to make certain inferences and eschew others, to perform certain actions and refrain from others [when one’s ends are cognitive], to engage in and endorse certain forms of higher-order evaluation and criticism and avoid and repudiate others’ (p. 56). The target of objectual understanding is a domain of inquiry or a topic, not individual events. And its vehicle is an account, which is a comprehensive, systematically linked body of information, cognitive commitments, methods, models, and theories.

Because scientific accounts that supply understanding often include falsehoods, objectual understanding is non-factive. And yet, Elgin maintains that understanding is grounded in fact. She calls ‘the required connection between a comprehensive, coherent account and the facts it bears on an understanding’s tether’ (p. 45). I will get back to the notion of a tether in a moment; for now, it is important to clarify that because objectual understanding is non-factive, its value cannot be grounded only in truth. The value of objectual understanding lies primarily in the fact that it is essentially an epistemic achievement or success, and achievements are arguably valuable for their own sake. That’s how felicitous falsehoods embedded in an account can acquire epistemic value: they contribute to the achievement of objectual understanding.

Chapters 4 and 5 flesh out the conditions under which understanding constitutes a success, and establish epistemic norms for intellectual integrity and responsible inquiry. Elgin’s general approach is holistic: she seeks to provide answers to questions of epistemic value, success, and justification at the level of the cognitive commitments governing an account as a whole and not at the level of its individual constituents. Taking this approach, she argues that epistemic success consists in ‘improving on the commitments we currently hold, where improvement itself must be measured by current standards’ (p. 67). The kind of improvement she talks about is produced by a community of ‘free and equal’ inquirers engaged in a deliberative process, which consists in ‘rejecting, correcting, revising, and augmenting’ the elements of an account until they display ‘tenability’ and stand in a ‘reflective equilibrium’ (p. 66).

But when are an account’s commitments or epistemic practices justified or worthy of acceptance? For Elgin, a commitment is worthy of acceptance just in case it would be reflectively endorsed by the legislating members of a realm of epistemic ends. This is what Elgin calls ‘the epistemic imperative’ (p. 105), which grounds a set of norms that specify the practical and epistemic obligations members of an epistemic community have towards each other, and determines when a falsehood in science is justified or worthy of acceptance.

While the account of deontological epistemic norms in Chapters 4 and 5 is fairly abstract, and some philosophers of science might fail to see how exactly it bears on scientific practice, Chapter 6 illuminates their functions, discussing how their violation can play out in practice. The central idea is that trust is one essential character of science, where different members of scientific communities depend on each other and cooperate to improve our understanding of reality. While this idea shows that understanding has both epistemic and moral dimensions, violations of deontological epistemic norms can constitute cases of scientific misconduct such as fabrication of data, plagiarism, and statistical analyses aimed only at p-hacking, where scientists breach obligations they have in virtue of their membership of a certain epistemic community. Scientific integrity, Elgin suggests, can be promoted by education that may lead students to internalize the epistemic and moral norms grounded in the epistemic imperative.

Though Elgin does not articulate this line, her account of epistemic norms might fruitfully be applied to the institutional arrangements of scientific communities. As the failures in the replicability of published results in fields like social psychology illustrate, the incentive structure and social institutions governing scientific communities can lead to problematic research practices. Practices such as hypothesizing after the results are known (HARKing), low statistical power, p-hacking, publication bias, and lack of data sharing can usher in non-replicable findings and erode trust in science. Remedies here should be structural, and involve a negotiation of epistemic, sociological, and economic considerations (for a recent proposal of how this negotiation can inform the distribution of replication efforts, see Romero [2018]). How exactly Elgin’s deontological norms can inform the design of new scientific institutions remains an important question for future research.

Chapters 7 and 8 expand on the discussion of trustworthiness, relating it to the objectivity of felicitous falsehoods in science and aesthetics. The account of epistemic norms Elgin defends countenances a form of objectivity that is ‘procedural and perspectival’. Perspectivism is a reaction against a ‘God’s eye view’, according to which there is exactly one true and complete representation of reality. Although perspectivism is committed to the idea that human inquiry is situated inquiry ‘that represents how things appear from a particular point of view’ (p. 155), this does not mean perspectival, felicitously false representations are not objective. Like trust, objectivity is grounded in norms for responsible inquiry. It is gained through the usage of publicly available and defeasible procedures of inquiry that ‘have been devised, calibrated, and validated to satisfy the standards of a realm of epistemic ends’ (p. 160).

To articulate this notion of objectivity, Elgin draws an analogy between science and aesthetics. Debates about the value and interpretation of artwork involve epistemic communities with shared commitments and standards, just as they do in science. And again like science, such standards are devised, tested, and certified so as to promote trustworthy judgement. Some aesthetic disagreements may appear intractable in a way that sets aesthetics apart from science. But reasons play similar roles in both science and aesthetics: they ‘highlight features and display their significance. In neither case ought the reasons purport to exclude alternative viable perspectives’ (p. 180). Practices of reason giving, reason accepting, and disagreement can thus secure objectivity in both science and aesthetics.

Aesthetics and the arts figure prominently in Chapters 9–12 too, where Elgin contends that works of art deploy the same devices as felicitous falsehoods in science. The claim here is that both artwork and felicitous falsehoods in science enhance our understanding, by distorting certain features, abstracting away from irrelevant confounding complexities, and making salient selected features of a target phenomenon they exemplify. The key notion is exemplification: ‘When an item serves as a sample or example, it exemplifies: it functions as a symbol that makes reference to some of the properties, patterns, or relations it instantiates’ (p. 184). Exemplification is a species of tether to reality that can acquaint inquirers ‘with features of things they had been unaware of or whose significance escaped their notice’ (p. 203). By exemplifying particular features or patterns, scientific practices like modelling, thought experiments, and mathematical proofs can afford epistemic access to those features and patterns scientists might otherwise miss. Likewise, people can learn something significant about features and patterns of reality from dances and literary fictions exemplifying those features. With their close examination of scientific modelling and its intriguing relationship with aesthetics, Chapters 9–12 will be the most interesting to philosophers of science working on scientific representation and, particularly, on the role of fiction in science.

Chapters 13 and 14 take up two remaining issues. Chapter 13 examines interdisciplinary work with a focus on the relationship between historical sciences and art. If individual disciplines are constituted by distinctive methodological and theoretical commitments, then how can these commitments change and be managed? For example, ‘are the commitments of art and history too disparate for a single work to satisfy the demands of both?’ (p. 273).

Elgin argues that the framing commitments of historical sciences can be flexibly negotiated, so that works of art can afford genuine historical understanding. Unpacking this idea in the light of Maya Lin’s Vietnam Memorial in Washington DC, Elgin explains that this negotiation ultimately rests on a decision by historians concerning how their research is used. While the key role of ‘usage’ coheres with the account of felicitous falsehoods and understanding that Elgin develops in the previous chapters, more attention could have been drawn to the ways usage of an interdisciplinary account is often limited by institutional and cognitive constraints—for example, by peer-review and criteria for tenure promotion, by differences in evidential standards and epistemic values, and by the opacity of certain specialist concepts (for a discussion of challenges to interdisciplinary work in the environmental sciences, see MacLeod and Nagatsu [2018]).

Chapter 14 concludes True Enough by examining the role of error in science. It argues for a fallibilist theory of understanding, where acknowledging the fallibility of our scientific achievements prompts a community of responsible scientists to gauge the depth and limits of their understanding of a topic: ‘Only someone who understands a good deal about a topic has the resources to make a significant mistake’ (p. 301). The vulnerability of scientific achievements to error is thus a strength, contributing both to the trustworthiness of science and to the understanding of reality it can deliver.

From this outline, it should be clear that Elgin’s True Enough is a philosophical tour de force of incredible breadth. Although she could have paid closer attention to the institutional and social structures that constrain scientists’ usage of felicitous falsehoods, Elgin makes significant contributions to the growing literature on scientific modelling and the nature of understanding. Her original account of the aims, methods, and trustworthiness of scientific inquiry is informed by, and engages with, many ideas and questions from traditional disciplines in philosophy such as epistemology, ethics, and aesthetics. In doing so, True Enough is a pellucid exemplar of how philosophy of science can avoid the risk of isolationism.