Bas van Fraassen once characterized the difference between English and Prussian conceptions of law this way: in the former, everything not explicitly prohibited is permissible; in the latter, everything no explicitly permitted is prohibited. Not likely to be quite accurate, but instructive nonetheless.
I was thinking about this idea with respect to theories of rationality and irrationality, in the context of some work I’ve been doing on the concept of mens rea. Suppose you prefer an English-style theory of rationality and you conceive of irrationality analogous to the way you conceive of criminal behavior: there is no crime without a guilty mind. Then we should expect legal theory on the concept of mens rea to be relevant to the nature of epistemic principles.
If so, the question is what level of mens rea involvement is requirement for an epistemic crime.
The analogy with the law will thus require us to distinguish between the mental element (mens rea) and the objective element (actus reus). For example, murder and mere homicide share the same actus reus but differ in terms of mens rea requirements. Mens rea requirements run, in order of strength, from the intentional to the known to the reckless to the negligent. Below these are strict liability laws that involve no mens rea requirement at all.
So: what is the epistemic analogue of actus reus? I’m not sure, but suppose we start with the suggestion that it is believing the false or the likely to be false. Doing so turns the initial focus on our theory away from irrational true beliefs and irrational likely-to-be-true beliefs, but that’s OK initially so long as we take a final step and generalize our results in the other area to these areas as well.
If we think of epistemic laws in terms of strict liability laws, we get the view that all false beliefs are irrational, or that all likely to be false beliefs are irrational. Why? Because not even questions of negligence or recklessness enter into the story of whether an epistemic crime has been committed. The central difference between negligence and recklessness concerns first whether the risks incurred were known to the agent and disregarded (recklessness) or whether such risks would have been known to a reasonable person in the situation in question even though they weren’t known to the agent (negligence).
Hard-core externalists might find such a position attractive, but I don’t and won’t bother to try to argue here for my distaste. The strongest requirements look equally puzzling, however. To believe that p intending to believe it in spite of its falsity or in spite of its strong likelihood of falsity are obviously irrational, but it is hard to hold out such a strong intentionality requirement on irrational belief. To believe that p knowing that it is false is hard to envision as possible, and to believe that p knowing that it is likely to be false hardly seems necessary for irrational belief (though, with some refinements, it may be sufficient for irrational belief).
So, if we are looking for minimal requirements on irrational belief, when the belief is false or likely to be false, it looks like either recklessness or negligence is all that’s left. The negligence standard, however, seems inadequately perspectival: suppose I don’t know of the risks of error and so don’t take them into account; how is it relevant that reasonable people, in general, would know? Of course, we can trivialize the requirement by quantifying over all reasonable people, and then the fact that I don’t know shows that I’m irrational (in virtue of this very lack of knowledge), and it’s a bit hard to argue that even irrational viewpoints must be allowed to help define the perspective from which rationality derives. But it isn’t a very plausible interpretation of the legal notion of a reasonable person to require quantification over all reasonable people, so the perspectival concern remains.
By elimination, we are left with the recklessness standard. Elimination arguments are generally quite weak, because they work only to the extent that all possibilities are covered by the original categories themselves, and here there is no special reason to think that the various levels of mens rea requirements are exhaustive in this way. But the exercise is still interesting and perhaps instructive: maybe you have to know and disregard the risk of error in order to be irrational. The hard work to be done, under this rubric, would be to say what levels of risk can’t be disregarded in this way and what the concept of risk involves–is it chance of error, subjective probability of error, etc.