Rutgers University professor Larry Temkin argues that certain deeply-held and widely-accepted assumptions about the nature of the good and practical reasoning are incompatible with each other.
The following is a summary of Larry Temkin's talk at the Legatum Institute, in which he discussed his new book Rethinking the Good, Moral Ideals and the Nature of Practical Reasoning. The audio recording of his talk is embedded below.
My talk’s aim is to introduce a few key issues and problems raised in my recent book Rethinking the Good: Moral Ideals and the Nature of Practical Reasoning. In doing this, I present several examples, or in some cases impossibility results, showing that certain deeply-held and widely-accepted assumptions about the nature of the good and practical reasoning are incompatible with each other.
Most people accept that “all things considered better” than is a transitive relation. This view, which I shall call the Axiom of Transitivity, holds that for any three alternatives, A, B, and C, if all things considered, A is better than B, and B is better than C, then, all things considered, A is better than C. The Axiom of Transitivity is one of the key premises of Expected Utility Theory, which is the key theory underlying game theory, decision theory, and much of modern economics. It also implicitly underlies many practical decisions that we make in everyday life.
“Certain deeply-held and widely-accepted assumptions about the nature of the good and practical reasoning are incompatible with each other.”
Most people also accept a principle I call the First Standard View: Trade-offs between Quality and Number are Sometimes Desirable, as well as a principle I call the Second Standard View: Trade-offs between Quality and Number are Sometimes Undesirable Even When Vast Numbers are at Stake. I point out that there are certain kinds of cases where the First Standard View seems deeply right, and relevant for assessing how two outcomes compare, all things considered, and other kinds of cases where the Second Standard View seems deeply right, and relevant for assessing how two outcomes compare, all things considered. I then present a Spectrum Argument, which involves a range of cases from 1 to n, such that the First Standard View seems relevant for comparing each of the adjacent cases along the spectrum (k and k + 1, for each k), while the Second Standard View seems relevant for comparing the cases at opposite ends of the spectrum (e.g. 1 or 2 versus n or n – 1). I then show that, together, the First and Second Standard views are incompatible with the Axiom of Transitivity. Thus, one faces the unpalatable options of rejecting the First Standard View, the Second Standard View, or the Axiom of Transitivity; none of which, I suggest, will be easy to do.
For many, forsaking the Axiom of Transitivity is not even an option, as they believe that the Axiom of Transitivity is true either analytically, that is, in virtue of the meanings of the words “all things considered better than,” or as a matter of the logic of better than. I shall claim that this position is mistaken. I shall note that there are two rival ways of understanding the good or ideals. On one, the Internal Aspects View, how good an outcome is depends solely on the internal features of that outcome. On another, the Essentially Comparative View, how good an outcome is sometimes depends on the alternative(s) with which that outcome is compared. I show that each of these views is deeply plausible, but that only the first is compatible with the Axiom of Transitivity. I further suggest that some of the ideals that people are most wedded to in assessing the comparative goodness of outcomes—including the Pareto Principle, plausible versions of maximin and utility, and a position I call the Narrow Person-affecting View (which assesses outcomes, in part, by focusing on how the particular people in those outcomes are affected for better or worse)—are essentially comparative. Thus, retaining the Axiom of Transitivity will require abandoning some of the ideals to which people are most committed.
Two other widely shared views are that, in assessing the goodness of outcomes, one should be neutral with respect to people, places, and times, and that one should accept various weak Pareto-like dominance principles of the following form: if one outcome would be better than another at every point in time, or at every point in space, or for every person who ever lives, then it must be better all things considered. I shall show that this constellation of views is incompatible, that, for example, A might be better than B at every moment in time, and yet B might be better than A for every person. In this case, I will argue, one should favor B over A on narrow person-affecting grounds; but, as just noted, such a view reflects an essentially comparative view of ideals that threatens the Axiom of Transitivity.
"Many deeply plausible views that people hold are incompatible. Giving any of them up will have far-reaching practical and theoretical implications."
Finally, it is widely recognized that Expected Utility Theory’s completeness assumption is dubious, that is, that some alternatives are only roughly comparable in the sense that they are on a par, but neither is better than the other nor are they exactly equally as good. But, unfortunately, the implications of incompleteness have not always been fully appreciated. Time permitting, I aim to show that once one abandons completeness, one must also abandon at least one other principle that seems deeply plausible and central to our understanding of practical reasoning, such as a Pareto-like dominance principle, the Axiom of Transitivity, a Reflection Principle (which holds, roughly, that if, on reflection, I know that at some point in the future I’ll have more knowledge than I currently have, and I now know that given that future knowledge it will be reasonable to assess two prospects in a certain way, then it is now reasonable for me to assess the two prospects in that way), a First Principle of Equivalence (which holds, roughly, that for any two prospects A and B, if for every possible outcome, w¬i, that might arise with a given probability, pi, if A is chosen, the same outcome, w¬i, might arise with the same probability, pi, if B is chosen, and vice versa, then prospects A and B are equally good (that is, V(A) = V(B)), or a State by State Comparison Principle (which holds, roughly, that for any two prospects A and B, if the value of A’s outcome stands in a particular comparative relation, R, to the value of B’s outcome for each possible state of nature, then prospect A stands in relation R to B).
The upshot of my talk is, I’m afraid, rather unsettling. I am confident that many deeply plausible views that people hold are incompatible, and that some of them will ultimately have to be rejected. But I believe that giving any of them up will have far-reaching practical and theoretical implications for our understanding of the good, moral ideals, and the nature of practical reasoning.
--- Larry Temkin
is professor of philosophy at Rutgers University, specialising in ethics and political philosophy. He graduated number one with a B.A. Honours Degree from the University of Wisconsin/Madison (1975), and earned his Ph.D. in Philosophy from Princeton (1983). He has received fellowships from the Danforth Foundation, the Andrew Mellon Foundation, the National Humanities Center, Harvard University's Center for Ethics, All Souls College, Oxford University, and the National Institutes of Health.