Cooperation A Philosophical Study
In Cooperation, A Philosophical Study, Tuomela offers the first comprehensive philosophical theory of cooperation. He builds on such notions a collective and joint goals, mutual beliefs, collective commitments, acting together and acting collectively. The
- PDF / 58,555,671 Bytes
- 447 Pages / 481.89 x 691.654 pts Page_size
- 81 Downloads / 226 Views
PHILOSOPHICAL STUDIES SERIES VOLUME 82
Founded by Wilfrid S. Sellars and Keith Lehrer
Editor
Keith Lehrer, University ofArizona, Tucson Associate Editor
Stewart Cohen, Arizona State University, Tempe Board of Consulting Editors
Lynne Rudder Baker, University of Massachusetts at Amherst Radu Bogdan, Tulane University, New Orleans Allan Gibbard, University of Michigan Denise Meyerson, University of Cape Town Fran 0; and they are maximally conflicting if and only if corr = -1). In general, a given amount of covariance between the preferences of the participants can be achieved by means of several different patterns of interdependence (viz., different patterns of control over their own and their partner's action as well as of purely interactive control) between their actions. (See Chapters 8 and 9 for details.) High degree of correlation of preferences is in general a precondition of cooperation and an element facilitating it, making it both more stable and more flexible. Let me emphasize here that when speaking of preference-correlation I generally mean the correlation of outcome preferences in a dependenceinvolving cooperation situation. Notice that some agents might all like apples very much (similar and correlated preferences) but if, say, only one apple were available in a certain choice situation, the outcome-preferences ofthese agents would not correlate highly but would indeed be conflicting. Let me give the following example to illustrate preference-correlation. Consider a two person two-choice situation with persons A and B and their respective choice-alternatives aI' a2 and b l and b2 . As in the standard game-theoretical setting there will be four choice-combinations or joint outcomes, viz., a lb l, a lb2 , a2b l, a2 b2 , which both agents value from their own points of view. Suppose the first three satisfy the joint action X. If the agents value these three outcomes similarly in the sense of giving them the same preference ranking (and, if numerical values are given, the same numerical utilities), we say that the satisfaction conditions of X correlate perfectly. Conditions 3) and 3') - or rather the contents of the mutual beliefs - are equivalent given some assumptions (see Chapter 9). Let me here just briefly indicate why. Suppose we have perfect correlation in the above example, and assume that the combination alb l is valued most highly by A and B. Then, for example, if A chooses a l it is rational for B to help A to arrive at his highest score by choosing b l; whereas if A chooses a2 it is obviously rational for B to choose b l for X to be satisfied at all. Similar considerations of course apply to B. On the other hand, a maximal amount of helping (at least if understood in the above simplified setting) will require that the preferences be perfectly correlated. (Clause 3) does not strictly require maximal helping.) As to a richer setting of cooperation, the possibility of helping in the context of a fully cooperative joint action type means that every participant's part-performance or a required
