I have picked up Cass Sunstein’s latest book on cost / benefit trade offs, and am enjoying it. But it seems to me that there is a fundamental problem here with the framing. The model being put forward is one in which we straight-forwardly calculate costs and benefits for any choice and then we make the right, informed and rational choice. Yet, we know that this model breaks down in two significant cases – and that is when the costs or the benefits become very large.
At that point, the probability is subsumed by the gravity of the cost or benefit and deemed unimportant. These decision spaces, let’s call them the ”rights”-space and ”risk”-space, are spaces where we navigate in a mostly rule-based fashion, and where deontological and kantian methods apply.
We will not calculate the benefit of sacrificing human lives, because people have a right to their own life and the individual benefit of that is vast. We will not calculate the cost of a nuclear break-down accurately because if it happens it has such a great potential cost. Even if the probability is miniscule, and the expected cost and benefit could be calculated well, we don’t. Rationality breaks down at the event horizon of these two decision singularities.
Now, you could argue that this is just a human thing, and that we need to get over it. Or you could say that this is a really interesting characteristic of decision space and study it. I find that far fewer take the second approach, and so expose an interesting trait: rationality blindness. A striving for rationality that leads to a blindness for human nature.
If we were to develop a philosophy of decisions, one things we would need to do is to show that not all decisions are the same. That there is a whole taxonomy of decisions that needs to explicated and examined and explored. As this example shows there are decisions that do not admit of a probability calculus in the normal way.
(Is this not Kahneman’s and Tversky’s project? No, it is in fact the opposite. Showing that the idea of decisional bias actually reveals a catalogue of different categories of decisions – not weaknesses in human rationality.)