Chapter 2. Reasoning about Pursuit: A Decision-Theoretic Approach
2.4. The Consequentialist Approach to Pursuit Worthiness
The acceptance of a theory concerns the question of what the world is like: whether the hypothesis true, or partially or approximately true, of the world (or the observable parts of the world). In contrast, decisions about pursuit concern which course of action to pursue (McKaughan 2008: 454; cf. Kapitan 1992, 1997). The normative basis for judgements about pursuit should therefore be construed in terms of practical rationality. More specifically, as Šešelja, Kosolosky and Straßer (2012: 53) point out, it is natural to construe pursuit worthiness in terms of a broadly consequentialist or goal-oriented conception of practical rationality. On this account, the pursuit worthiness of a hypothesis should be evaluated in terms of how well the expected consequences of doing so contributes to achieving the goals of research (whatever these are), compared to how one could have otherwise spent the available time and resources.
One reason for adopting this approach is that it is already implicit in most of the discussion reviewed above. It is clearly the basis for the ‘Economy of Research’ argument, which Weinberg and Peirce use to make that point that it is necessary to make judgements about pursuit worthiness. This is also the thinking underlying Franklin’s account of the fifth force hypothesis: after the hypothesis had been refuted, it was still
reasonable to carry out the experiments despite modest the expected epistemic gains, because the experiments were already set up and thus still cost effective. Finally, the arguments against there being normative standards for pursuit (distinct from those on acceptance), made by Feyeberabend/Lakatos and Kitcher, all assume that if there were normative constraints, these would be related to the expected outputs of pursuing the hypothesis. Feyerabend’s argument (in the radical interpretation) implicitly relies on this assumption to conclude from the premise that we cannot predict the future development of science that there are no normative constraints on pursuit. Similarly, Kitcher’s argument assumes that the pursuit worthiness of a hypothesis should be evaluated in terms of whether this is likely to result in more falsehoods or truths.
2.4.1. Drawing Distinctions
To start developing this into a more systematic account, it is useful to draw attention to some distinctions pointed out by Šešelja, Kosolosky and Straßer (2012). Saying schematically that it is rational for Y to pursue X if, and only if (or to the extent that), pursuing X contributes to achieving the goals Z,39 they distinguish different ways to instantiate the variables X, Y and Z. Each of these corresponds to a different kind of pursuit worthiness judgments.
First, we can distinguish between different agents, Y, for whom the judgment of pursuit worthiness is made. In particular, we can distinguish between whether we are making a community level judgement regarding what should be pursued by someone within a scientific field or whether the judgement concerns what an individual scientist
should pursue.40 As argued above (Section 2.2.2), both types of judgements are legitimate but should not be conflated.
Second, we should distinguish which type of item we take to be the object, X, of pursuit. In this thesis, I will mostly consider cases where X is some kind of theoretical representation, i.e. hypotheses, theories or models which purport to represent some target phenomenon. Here, the aim is usually to find out how well that hypothesis represents or at least predicts the relevant aspects of the target. In other cases, scientists are primarily pursuing some technological development, aiming to develop a specific kind of instrument or technique or to refine and calibrate an existing one. For example, part of the reason for continuing the gravitational experiments in the fifth force case, according to Franklin, was to improve experimental techniques.
Third, we can distinguish the type of goals, Z, that we are evaluating the situation in terms of. Two distinctions are relevant here. First, we should distinguish between (a) evaluating a case from the internal perspective, i.e. assuming the goals endorsed by the agent making a decision about pursuit, and (b) evaluating the case from the external perspective, i.e. using the goals we (the evaluators) think the agent ought to have. Second, we can distinguish between whether the goals refer narrowly to the epistemic or
intellectual goals—e.g. accepting theories that are close to the truth or obtaining
explanations of puzzling phenomena—or whether it includes a broader set of moral and political values and goals as well. Following Šešelja et al., I will refer to analyses taking into account only the former, narrow set of goals as concerned with epistemic pursuit worthiness and analyses taking into account the latter, broader set of goals as concerned
40 One can also draw more fine-grained distinctions between different levels, e.g. science as a whole,
with practical pursuit worthiness. I will mainly focus on epistemic pursuit worthiness in this thesis.41
2.4.2. Relevant Factors
On the consequentialist conception of pursuit worthiness, the kinds of considerations which can be used as reasons for or against a pursuit worthiness judgment can, in principle, include anything which is relevant to estimating the outcomes of pursuit. While these will presumably vary between contexts, some general suggestions can be made. As argued above, although the likeliness of a hypothesis has some relevance, it is not the only factor. Summarising Peirce’s view, McKaughan (2008: 457) concludes that the most important considerations in deciding whether to prioritise a hypothesis for pursuit are “factors like our time, resources, and value of the estimated payoff in comparison to other courses of action. … If we estimate that testing the hypothesis will be easy, of potential
interest, and informative, then we should give it a high priority”. Independently, Franklin
concludes from his case studies that “the decision to pursue an investigation seems to depend on a weighting of at least three factors: the interest of the hypothesis, its plausibility, and its ease of test” (1993b: 122). He also mentions factors to do with conserving resources, such as “recycling expertise” and continuity with already ongoing research programs (Franklin 1993b: 101).
In addition to these factors, Peirce emphasises that since “very rarely can we positively expect a hypothesis to prove entirely satisfactory”, it is important to consider the “effects upon other projects” of pursuing the hypothesis, that is, “we must always consider what will happen when the hypothesis breaks down” (CP7.220).
41 Exactly what the epistemic goals of science are differs between realists and anti-realists. I discuss the
What Peirce has in mind here are strategic considerations of how learning that a hypothesis is false can inform later stages of inquiry.42 He draws an analogy with playing twenty questions, where the most strategic yes/no questions are those that narrow down the field as much as possible regardless of what the answer will be. Similarly, it can in some cases be worth testing a hypothesis simply to “clear the field”. Peirce also argues that it can be worth testing a simpler hypothesis because this makes it easier to interpret how the results differ from the predictions of the hypothesis, thereby providing suggestions for how to formulate a better hypothesis. Finally, as Musgrave (1976) points out, we should also bear in mind that an experimental result which serves to falsify a hypothesis can sometimes constitute an interesting discovery in itself.
If reasoning about pursuit consists of comparing different candidates for pursuit in terms of these kinds of factors, it raises the question of how those factors should be weighed against each other. In practice this usually will be a matter of informed judgement. However, in order to clarify the underlying logic of these decisions, it can be useful to think of pursuit worthiness in terms of simplified, idealised decision-theoretic models. I will now develop such a model which captures many of the factors discussed above and which will be particularly useful for my purposes in this thesis.43
42 Recent interpretations have for this reason characterised Peircean abduction as a form of “strategic
reasoning” (Hintakka, 1998, Paavola 2004, Pietarinen and Belucci 2014). This type of strategic considerations will be particularly relevant for our discussion of diagnostic reasoning in Chapter 6.
43 The models developed here draw on and extend the model presented in Nyrup (2015). Decision-theoretic
models of pursuit worthiness have also previously been developed, although in a different direction to the one taken here, by Kukla (2001, 2010: ch. 1) and Harp and Khalifa (2015). Similar models have also been used to analyse decisions about whether to test uncertain diagnostic hypotheses (Pauker and Kassirer 1980); I discuss these further in Chapter 6.
2.5. Decision-Theoretic Models