Chapter 3. A Peircean View of Explanatory Reasoning
3.3. Two Problems for Explanationism
In the preceding section, I took for granted the idea that potential explanations can be ‘better’ than each other. However, it is important to recognise that the slogan “infer the best explanation” conceals an important distinction between two ways an explanatory hypothesis can be better than its competitors (Lipton 2004: 59-65). In one sense, a hypothesis can be better simply because we think it is more likely or closer to the truth than its competitors. For instance, we may be able to rule out, or show highly improbable, all plausible alternative explanations in light of our available evidence and accepted background theories. Here, the remaining hypothesis would be the likeliest available explanation, and in this sense the best. However, as defenders of explanationism such as Lipton (2004: 60-62) and Psillos (2002: 617) point out, if explanationism merely recommends inferring the hypothesis which we already think is most likely or closest to the truth, it would be a fairly uncontroversial but also rather uninteresting position.66 What motivated the interest in IBE is that the explanatory quality of the competing hypotheses is supposed to give us an independent, non-empirical criterion for choosing between hypotheses. It is this feature which (i) allows IBE to provide a solution to Quine and Harman’s worries about the empirical underdetermination of theories, (ii) makes it an
64 See also my discussion of the possibility of normative accounts of generative reasoning in Chapter 1,
Sections 1.4 and 1.5.
65 This section and the next draw on and expand the argument presented in Nyrup (2015).
66 As the “best of a bad lot” objection shows, even this inference is not completely uncontroversial,
depending on what is concluded on the basis of an IBE. See also Achinstein’s (1990) criticism of ‘the only game in town’ inferences.
account of the use of non-empirical criteria of scientific theory choice and (iii) would make IBE a means for supporting metaphysical theories. For the purposes of my discussion in the rest of this chapter, I will ignore the trivial interpretation of IBE.
The more interesting version of explanationism I examine here, then, brackets what we believe about the truth of the hypotheses and instead focuses on their explanatory qualities. Let us say that the explanatoriness of a hypothesis H consists in how satisfying
H would be qua explanation if it were true.67 This will generally depend on the number and quality of explanations that H would provide if it were true.68 Since the quality of an explanation is often taken to consist in how much understanding it provides, we might also say that the explanatoriness of H consists in the amount of additional understanding
H could potentially afford us. There are of course different accounts of explanation
(causal, unification, etc.), and these emphasise different theoretical virtues (simplicity, unification, coherence, elegance, quantitative precision, specifying a mechanism, etc.) as being characteristic of good explanations (Thagard 1978; Lycan 2002: 414-16; Lipton 2004: 122). Often, these criteria overlap with the theoretical virtues that Kuhn (1977) highlighted as important to theory choice. However, since the arguments of this chapter will not depend on any particular view of explanation or understanding, I will bracket these details and simply assume that it makes sense to distinguish between more and less explanatory hypotheses. Whichever accounts of these matters suffice for explanationism will be equally good for my argument.
67 ‘Explanatoriness’ here corresponds to what Lipton (2004: 59) calls ‘loveliness’.
68 This subjunctive formulation is also motivated by Lipton’s view (2004: 57-8) that only true hypotheses
can be genuine explanations. Some philosophers deny that explanation requires truth (e.g. van Fraassen 1980) or even hold that achieving understanding sometimes requires sacrificing truth (Cartwright 1983: ch. 2). For the purposes of this chapter I will follow most explanationists in assuming that successful explanation requires truth. Notice that in my decision-theoretic argument for why explanatoriness justifies pursuit (Section 3.4 below), if an explanatory hypothesis can be valuable even if it is false, this only strengthens the argument.
Given this focus, the core claim of explanationism is that having a high degree of explanatoriness can give us some additional reason for the truth of a hypothesis. Of course, explanationists do not claim that explanatoriness should trump all other considerations; there may be other independent empirical or non-explanatory theoretical reasons which tell against the truth of the most explanatory hypothesis (Lipton 2004: 61). Thus, they still hold that which hypothesis counts as “the best” explanation in the IBE inference schema is determined by the likeliness (or truth-closeness) of the hypotheses. What attenuates the charge of triviality here is the claim that explanatoriness can serve as a guide to the truth of a hypothesis, in addition to any other non-explanatory considerations. I will call this the truth-guidance claim.
The truth-guidance claim is what makes explanationism interesting but, when interpreted as a normative claim, it is also the source of some of the most pressing problems for explanationism. These were voiced already by Reichenbach (1938), in response to Nagel’s (1936: 508) observation that scientists often accept hypotheses on partly the basis of non-empirical “esthetic grounds”:69
It may be true that a physicist believes in his theory because he thinks it to satisfy esthetic standards; but I do not see any reason why we should believe in predictions which are based on esthetic arguments; or why a technician should do so. I do not see any relation between esthetic qualities and predictional qualities—and the latter are what a good theory must have. The beauty and harmony of a theory is a matter of taste; it should be easy to construct theories of an extreme beauty which are obviously false. … I cannot accept the esthetic argument as anything connected with the validity of scientific theories in an objective sense; i.e. as an argument which makes the acceptance of a theory justifiable (Reichenbach 1938a: 34-5, original emphasis).
69 Nagel raised this objection in a review of Reichenbach (1935c). See my discussion in Chapter 1, Section
Reichenbach here raises two distinct objections, both of which have been echoed by later critics of explanationism. First, explanatoriness (like Reichenbach’s “esthetic qualities”) seems too subjective to provide a plausible, objective guide to the truth—they are a “matter of taste” and therefore not objective enough to make a theory justifiable. This corresponds to what I, in Section 3.1, called the subjectivity problem. This is also the objection alluded to by Hacking’s (1984: 167) quip that IBEs are just inferences “from what makes our minds feel good”. Lipton (2004) similarly notices that if explanatoriness (like beauty) is merely “in the eye of the beholder” (143),70 then it becomes unclear how
it could provide a reliable guide to truths about the world.
Second, Reichenbach points out that there is no obvious logical or conceptual connection between the explanatoriness of a hypothesis and its truth. This is what I call the truth-connection problem. Why should the fact that a hypothesis would be a good explanation if it were true have any implications for whether it is in fact true? Just like there are many beautiful and harmonious theories which are clearly false, there are many false theories which would provide very good explanations if they were true. As Lipton notes, with a nod to Voltaire, “Why should we live in the loveliest of all possible worlds?” (2004: 144); to assume so seems worryingly close to a form of wishful thinking. Furthermore, this is not just an abstract logical possibility. As Duhem (1954) and later Laudan (1981) highlighted, the history of science contains many theories which, in their respective times, were regarded as excellent explanations of the same phenomenon. Since these explanations are mutually incompatible, most of them must be false (Cartwright 1983: 89-91).
70 Since the Irish writer Margaret Hungerford is thought to be the first to use the phrase “beauty is in the
To be clear, I do not regard these objections as knock-down arguments against explanationism. What they highlight is that the truth-guidance claim cannot simply be assumed. Explanationists need to give some positive argument for it. Of course, explanationists have proposed a number of solutions to these problems. In effect, these attempt to deny the premise of the problems by arguing that the criteria for explanatoriness are not completely arbitrary, and that explanatoriness can be a reliable or rational guide to the truth.
I will consider several arguments that explanationists have proposed for the latter claim in Sections 3.5 and 3.6. As I argue there, these face serious problems. First, I want to argue that for the Peircean view, these problems do not arise at all: even if explanatoriness is completely subjective and unconnected to the truth, it can still provide reasons for the pursuit of a hypotheses. Thus, the Peircean view side-steps the problems altogether.