Braithwaite (and Stanley) are only half right about the relationship between risk and credence

While there are differences between R. B. Braithwaite’s theory of “acting as-if” belief and Jason Stanley’s project, both share the idea that the higher the stakes are to an agent, the better evidence that agent needs before he may act as if (assert to know) that claim is true. The idea behind the more you care, the less you know has been with us for some time now. Even so, it is only half correct.

Stakes-sensitive intuitions are stoked, and nowadays tested, by examples of the following kind. You may act as if a vaccine against bird-flu is non-toxic to your chickens, but refuse to act as if the vaccine were non-toxic to your children. The idea is that since your children are more valuable to you than your chickens, and the stakes of losing a child are greater than losing a chicken, you’ll require greater evidence of non-toxicity for your children than you will for your chickens. And this seems right, so far as it goes.

But consider now an antibiotic which is the only treatment for an infection F that occurs in both chickens and children. The treatment has a fatality rate of 25% in both populations, but chickens almost always recover from F on their own whereas children nearly always succumb to F if untreated. Here you would have a reason to treat your children with the antibiotic but to refuse treating your chickens. That is, you would have a reason to act as if the antibiotic were non-toxic to your children but toxic to your chickens.

What is happening here is that the Braithwaitian view only considers the magnitude of risk, but what is really going on in such examples is that we consider a ratio of risk to reward. The standard intuition-pump examples do not vary gains.

So it isn’t true that ‘the more you care, the less you know’ because the disposition to act (or affirm that you know) is not a function of the magnitude of the stake. Instead, a disposition to act (or affirm that you know) is a function of the ratio of the amount risked to the amount gained.

Now, it may well be that Stanley’s view can survive translation out of Braithwaitian terms and into (Kyburgian) risk-reward terms without loss. But we shouldn’t continue thinking that magnitude of risk is what is driving these examples.


Braithwaite (and Stanley) are only half right about the relationship between risk and credence — 2 Comments

  1. An senior thesis I just advised goes through some cases like this. The student pointed out that it is not just risk and reward ratios, but also cost and benefits of acquiring evidence. Here is one of his cases of the latter. If you were weighed a month ago and it said “150 pounds” and you are standing next to a scale today, and your friend says “Do you know how much you weigh?” Whether or not you are inclined to answer “yes” can depend on whether or not you need to pay to use the the scale. If you need to pay $5, you are inclined to say “Yes I know, let’s go” whereas if it were free, you might say “No” and get on the scale just to see.

    The student’s example of the former is just varying Jason’s Hannah-and-her-wife bank cases to make it so that there is some cost to Hannah standing in line at the bank on a Friday (making her late to dinner with her wife perhaps), versus some low cost to not depositing the check if the bank happens to be closed on Satuday.

    The general idea is that knowledge is sensitive to both the practical consequences of both acting as if P were true and P were false, and acting as if you do not know that P and P happened to be true, as well as the benefits of acting as if P were true and P were true, and acting as if you do not know that P and P were false.

    I’m pretty sure Jason anticipated these issues (maybe DeRose too??), though I’m not up to speed on the most recent literature.

  2. Hi Barry, I would understand the cost and benefit example likewise in terms of risk and reward. I’d also mention that it is more natural to me to formulate a risk-reward ratio account of as-if belief in terms of a purely epistemic notion of evidence.

Leave a Reply

Your email address will not be published. Required fields are marked *