Unpacking Non-Cognitive Values in Scientific Practice — Epoche C2
The choice a funder actually makes When a research council must decide between a two-year study costing one unit of budget and an eight-year study costing five, it applies a criterion — expected knowledge per unit of money, or per year elapsed — that appears nowhere in any account of what makes a hypothesis well confirmed. That criterion is the subject of this essay. The claim to be defended is that resource economy of this kind, the demand for cost per result and speed of result, is not a neutral housekeeping constraint on inquiry but a value with a systematic direction; that its direction is set by the natural timescales of the objects studied rather than by anything about the researchers; and that it operates at a stage of the scientific process which Helen Longino's account of objectivity, for good reasons of its own, does not reach. One framing in the earlier version of this essay has to go first. It presented its thesis as a challenge to "the common belief" that Longino ties objectivity to critical discourse and community diversity. That is not a common belief to be corrected; it is Longino's stated position, argued for at length. What follows is therefore an extension of her framework at a point where she does not apply it, not a correction of a misreading of her. What Longino established, and where its scope ends The argument of Science as Social Knowledge (1990) turns on a single structural observation about evidence. Whether a body of data e counts as evidence for a hypothesis h is never settled by e and h alone: it is settled only relative to background assumptions that connect the two — assumptions about what the instrument measures, about which populations are relevantly similar, about what would count as the same phenomenon recurring. Those background assumptions are not themselves fixed by the data they license. Longino's conclusion is that contextual values — social, political, personal commitments external to the content of the theory — can therefore enter the evidential relation without any breach of logic and without anyone lying, because they can shape the assumptions that make the relation hold. From this she draws the constructive part, which is what her reputation rests on. If values cannot be excluded from the individual's reasoning, objectivity has to be relocated to the community, and it becomes a matter of degree: a community is objective to the extent that its background assumptions are exposed to criticism from people who do not share them. The 1990 book gives four conditions for such a community: recognised avenues for criticism of evidence, methods and assumptions; standards that critics can publicly invoke; responsiveness of the community to the criticism it receives; and intellectual authority shared among qualified practitioners. The Fate of Knowledge (2002) revises the list, most importantly by replacing bare responsiveness with uptake — the community must demonstrably change in response to criticism, not merely tolerate it — and by tempering the equality condition, since authority cannot be equal across the qualified and the unqualified. The distinction that does the classificatory work in Longino is between constitutive values, internal to the aims of inquiry, and contextual values, belonging to its social setting. Now comes the limit. Every one of the four criteria is a criterion on criticism : they govern what happens to hypotheses, methods and assumptions once those are in play. None of them says anything about which hypotheses come to be in play. A community can satisfy all four conditions perfectly with respect to the proposals in front of it and never once be in a position to criticise the absence of the proposals that were never written, because there is nothing on the table to criticise. That gap is where the present argument sits. Disambiguating "economy": two values wearing one word The earlier version of this essay classified efficiency and economy as non-cognitive values. That needs splitting, because "economy" names two quite different things and only one of them is non-cognitive. The first is theoretical parsimony. On Kuhn's list of the characteristics by which theories are appraised — accuracy, internal and external consistency, scope, simplicity, fruitfulness — simplicity sits squarely among the cognitive criteria, and Kuhn's point in the 1977 essay is not that these are illegitimate but that they are values rather than algorithms: they admit of different weightings and different readings by different competent scientists, and so they underdetermine choice without ceasing to be reasons. Ernan McMullin, arguing directly against the constructivist use of that underdetermination, coined the term epistemic values for the same family and defended them as truth-conducive rather than merely conventional. Parsimony, on either account, belongs to the cognitive side. The second is resource economy: money, instrument hours, researcher-years, time to a reportable result. Nothing on Kuhn's or McMullin's list corresponds to it, and no one has argued that cheapness is truth-conducive. It is this second sense that the essay is about, and from here it will be called resource economy to keep it apart from parsimony. Heather Douglas's Science, Policy, and the Value-Free Ideal (2009) supplies the vocabulary for saying where such a value may legitimately act. She distinguishes a direct role, in which a value functions as a reason in its own right for accepting a claim, from an indirect role, in which it sets how much evidence is enough given the costs of being wrong — the inductive-risk role. Her thesis is that the direct role is illegitimate at the point of accepting or rejecting a hypothesis but entirely legitimate earlier, in the choice of what to study and how. Resource economy is a textbook case of a value acting directly at exactly that earlier stage. On Douglas's scheme it is therefore permitted — and because it is permitted, it is unexamined. The p