By coincidence, my seminar on science and values covered Rudner’s Argument from Inductive Risk on the same day that Matt Brown posted an exchange about the Argument with Joyce Havstad. It’s taken me a couple of days to collect my thoughts.
Matt (quoting Heather Douglas at length) provides a version of the Argument according to which it applies to almost every determination to accept or reject an hypothesis. Joyce raises several worries. The strongest of them, as I understand it, is that ubiquitous inductive risk won’t sort cases which require critical discussion of values in practice from cases which don’t involve any contentious values. She writes, “I want to be able to identify and address those extremely risky decisions which are currently being made without proper consideration of ethical and social values, but which are in dire need of them.”
As an epistemic matter, however, it seems to me that the in-principle intrusion of values is as strong as Matt suggests (possibly stronger, but that’s another post). Whether we want to call the argument which brings out that point the Argument from Inductive Risk or not, it’s a point that should get made somewhere and under some heading. In my work, I call the claim that belief always indirectly reflects value commitments the James-Rudner-Douglas (JRD) thesis. Calling it a thesis rather than letting it be the conclusion of a named argument plays up the fact that it is a ubiquitous feature of our epistemic life rather than a specific move or inference.
Joyce suggests that it is a familiar point under another heading, writing that “I have a hard time distinguishing the supposed problem of inductive risk from the plain old problem of induction.” I’m a fan of reducing allegedly new problems to familiar ones, but I think there are several things that distinguish this from the old problem of induction.
One difference is the JRD thesis applies to observation reports as well as for the conclusions of inference. Joyce gives an example of counting nematodes. The problem of induction is just about general conclusions and does not hold for observation reports.
Another difference is that you can pose the problem of induction without recognizing the JRD thesis. As far as I can tell, the JRD thesis wasn’t posed until the late 19th century (by the eponymous William James).
So I think we should recognize the broader thesis as a matter of epistemological honesty. It doesn’t have serious consequences when the values in play aren’t significant ones, when everybody agrees that the costs and benefits to be reckoned with are small. However, the judgement that the costs and benefits are small is itself a value judgement. So the JRD thesis still applies.
Joyce’s question remains: How can we separate the cases where the JRD thesis matters for practice? Which cases are the ones in dire need of intervention?
The answer is that those cases are ones where the values in play are significant. That’s not very informative on its own, because there are several ways it can manifest:
- People disagree about the relevant costs and benefits in a way that underwrites different judgements.
- People disagree about the scale of the costs and benefits and so disagree about whether the matter is vital or trivial.
- Peoples’ epistemic judgements don’t appropriately reflect their value judgements.
- The value judgements people ought to be making would produce one of the above outcomes.
Maybe some of these are more important or ripe for intervention than others.
Comments on our exchange plus your post here raise the following question for me: What’s the difference between the AIR / JRD thesis and the argument from underdetermination for the value-ladenness of science?
Although “the argument from underdetermination” gets used to mean several different things, one version which was common 15 years ago went this way: Rationality or concern-for-truth are not enough to determine theory choice, so further criteria must be adopted. For closing this gap, there is no principled way to separate so-called epistemic criteria (like explanatory force or simplicity) from social or political criteria. So the social or political criteria are or could be legitimate ones for settling theory choice.
That argument makes space for any values whatsoever. In Douglas’ terminology, values can play a direct role.
The JRD thesis only concerns the benefits of possibly getting it right or the costs of possibly getting it wrong. Values play an indirect role.