Robert Weston Siscoe (University of Notre Dame & University of Graz), "Being Rational Enough: Maximizing, Satisficing, and Degrees of Rationality"

Australasian Journal of Philosophy, 2023

By Robert Weston (‘Wes’) Siscoe

In his 1975 book Ignorance, Peter Unger introduces what he calls absolute terms, terms that express concepts which have an absolute limit. Unger’s example of choice is ‘flat’. It  is possible for one surface to be flatter than another, but there is also an upper limit of absolute, perfect flatness. Now according to Unger (1975: 54), the term ‘flat’ picks out this uppermost point on its underlying scale: “To say that something is flat is, so far as content goes, no different from saying that it is absolutely, or perfectly flat.” 

David Lewis accepted Unger’s view that ‘flat’ is an absolute term, but he also added his own spin on the view to accommodate cases where we call surfaces flat that are not perfectly flat. In his 1979 paper “Score-Keeping in a Language Game,” Lewis argued that there is a standard of precision that determines what counts as a bump or blemish. In some contexts, it is permissible to disregard small protrusions, while in others, the tiniest of irregularities will disqualify a surface from being flat. This makes it possible that there are contexts where it is appropriate to call a surface flat even when it departs from absolute flatness. 

A lot has happened since 1979. In contemporary linguistics, a whole theory about gradable adjectives has sprung up around the Unger-Lewis account of absolute terms. Linguists now distinguish between absolute gradable adjectives, what were previously Unger’s absolute terms, and relative gradable adjectives. Let’s call the former AAs and the latter RAs. 

RAs do not have a conceptual upper limit, and they exhibit considerably more variability in the regions they pick out on their underlying scales. Take `tall’, for example. There is no uppermost point on the scale of tallness. There might be a metaphysical limit, like the size of the universe or something like that, but it doesn’t seem like there is such a thing as being perfectly or absolutely tall. Furthermore, depending on the context ‘tall’ can pick out wildly different points on the underlying scale. Depending on whether we are talking about skyscrapers or schoolchildren, it takes vastly different heights to qualify as tall. 

My recent research has focused on applying the research on AAs and RAs to questions about rationality. On my view, when it comes to beliefs and actions, ‘rational’ is an absolute term. Suppose, for example, that you watched the weather report for the day and learned there’s a 50% chance of rain. If you don’t have any additional evidence, it seems clear that it is rational for you to suspend judgment on whether it is going to rain today instead of believing that it’s going to rain or believing that it’s not going to rain.

It also seems clear that suspending is the only rational attitude you could have because suspending is the perfectly rational response to your evidence. If you believe that it’s going to rain, or believe that it’s not going to rain, not only would both of these responses be less rational than suspending, they would be positively irrational. So to have a rational doxastic attitude, you must adopt the one that is most rational.

Maybe I am loading the dice here by only dealing in terms of belief, disbelief, and suspension. Things would surely be more complicated if I asked what credence you should adopt. Maybe it is perfectly rational to adopt a .5 credence that it will rain, and it may be less rational to adopt a credence of .49, but is it really irrational to have a credence of .49? 

Even though I agree that this case is less clear, I don’t think this calls into question the view that ‘rational’ is an AA. As Lewis argued, AAs come with a standard of precision. In many ordinary cases, the standard of precision is fairly forgiving, allowing that someone can have a rational doxastic attitude even though it’s not the most rational one they can adopt. But I think it is possible to create a context where .49 isn’t rational.  After all, if I make it clear that the news report is all the evidence that you have, and that you have no other reasons for not adopting a credence of .5, it does start to seem positively irrational that you are at .49. What reason could you have for adopting such a credence?

Maybe you’ve found this convincing. Maybe you haven’t. But whether or not ‘rational’ is an AA does have some interesting upshots for debates in action theory and epistemology.  Take, for example, the debate between maximizers and satisficers about practical rationality. Maximizers take it that it is only rational to choose the course of action that promises the most utility, while satisficers think that there is a utility threshold below the top of the scale that is still rational. 

If we suppose that ‘rational’ is an AA, it is possible to make sense of this debate. When we are using the strictest standard of precision, maximizers are correct that only the most rational course of action is rational. But satisficers are right that, when we are using more relaxed standards, suboptimal actions can count as rational as well. That’s at least what I conclude in my paper “Being Rational Enough” in the Australasian Journal of Philosophy.

But wait, there’s more! If ‘rational’ is an absolute term, then strictly speaking there is no rational supererogation. In order to believe or do what is rational, you still have to do what the standard of precision deems is most rational, even if that particular standard is fairly relaxed. Or so I argue in my “Rational Supererogation and Epistemic Permissivism” in Philosophical Studies. 

Furthermore, if rational doxastic attitudes are only those that are the most rational, then it becomes difficult to make sense of the formal epistemologist’s claim that rationality requires us to be probabilistically coherent. I take on this puzzling conflict between what ideal epistemology says is rational and what our ordinary judgments say is rational, in “Real and Ideal Rationality” in Philosophical Studies and “Does Being Rational Require Being Ideally Rational?” in Philosophical Topics. 

Finally, I use the distinction between AAs and RAs to distinguish between rationality and justification. I argue that ‘rational’ is an AA, but then I make the case that ‘justified’ is more like the RA ‘tall’, with an adjustable contextual threshold. After all, being justified in believing that it is raining doesn’t mean I have the maximum possible amount of justification.  That idea can be found in my paper “Belief, Rational and Justified” in Mind.

So there you have it. I think there is a case to be made that ‘rational’ is an absolute term, and I think that this insight has a lot of important upshots. Little work has been done on ‘rational’ as a gradable adjective, and it appears that depending on whether ‘rational’ is an RA or AA has a number of surprising consequences. So even if my view ends up being wrong, I think it is at least wrong in an interesting way. And maybe that’s what philosophers have been shooting for all along.