Consider the following commonsensical description. I am fully aware if I drink alcohol during my meal, my forthcoming piano concert will be impaired. I do prefer to play piano better rather than worse and I am not a compulsive alcoholic. Nonetheless, when time comes, I go straight ahead and have the wine. The situation is common, frequently experienced by many, if not with alcohol with some other temptation. In some moral views it is even a sin to fall into temptation. The morals do not interest me here and neither they interest Bratman. The problem is how to model the case in terms of what the agent is doing. If anything, what we would like to know is whether there a rational way to understand what it is to resist the temptation. On a simplistic line one may take, the only model is very straightforward. My actions are determined by punctual choices that reveal my preferences at the time of the choice. All one is allowed to conclude is that I do prefer at dinner time to drink over my (future) better musical performance. This appears to be a preference reversal, since, seen from afar (earlier in time that is) I would always claim that I rank good playing ability above wine intake. Rational strategies may diverge on what is the best tactics to overcome or resist temptation. Bratmans approach is in comparison and contrasted with George Ainslies. His proposal seems to be related to what is usually called an evidentialist line in Newcombs problem. If I take a choice on a single night to be a good predictor of choices about the same problem, I would have some reason to make the first choice against drinking. Note that I have to think that many such choices are at stake and that each one of them is not causally relevant to any of the following ones. Bratman very tightly draws the analogy with the one-box strategy in Newcombs like situation. The choice of one box is evidentially but never causally relevant to the outcome. Given the conditions of the problem ones actions are temporally irrelevant. It is important to be clear about what is needed here. It needs to be shown that my choice to abstain on day 1 is not merely evidence of an underlying tendency to make similar choices in the future but also a cause of such future choices. (p. 47) Bratmans proposal to scotch the preference reversal is to impress on his readers how more rational it is to consider that intentions are planning intentions, and so even choices turn out to be, contrary to appearances choices between long term plans and not single-shot cases (do I drink or not right now?)
I have my doubts on the practicality of this recipe if intended as psychological aide. That said there is much to be commended in the proposal to keep the focus of rationality on the long-term commitments it entails.
I close with two remarks. One is for the reader: do not read this book without having some acquaintance with Bratmans, Intention, Plans, and Practical Reason. The second is for myself: all of these theories of action present cases of an interesting mix in which prescriptive and descriptive elements are hard to extricate one from another; is rationality something like beauty or is it a property of particular cognitive systems like us?
. English (and other languages) is ambiguous on plans. One may have contingency plans, in the sense of specific means to be adopted, for total warfare against Albania and no intention of carrying them out. If, on the other hand, I plan to go to Rome, I may have no plan about how to do it (walking, flying on carpets, etc.) but intending to carry the plan out. Bratman is interested in the second sense of the word.
. Viz. Sec. 3 of the book.
.I assume familiarity with the problem Newcomb invented and R. Nozick discussed on several occasions (the last in his Nature of Rationality book.) Briefly and w/out epicycles: a good or very good predictor of my behavior will present me tomorrow with two boxes, one of them will contain X pounds of gold, the other will contain either 1000X pounds of gold or nothing. The predictor will fill or not the second box depending only on what she predicts I will do.
Shall I take one or both boxes? If the prediction is that I take both boxes the second box will be empty, and vice versa. What is the rational strategy? Note that my choices take place after the predictor did exactly what she wants to do and there is no trick.
Adrian Palma was educated in the UK, Italy, Germany and France before doing his Ph.D. in philosophy at Indiana University. He is currently a researcher at the Center for Research in Applied Epistemology (CREA) in the École Polytechnique, Paris, and is also at the University of Technology in Compiegne, France. His main interests lie in the intersection between philosophy of language and the philosophy of mind.