INTRODUCING SOCIAL ACTION AND COOPERATION

CHAPTER 1 INTRODUCING SOCIAL ACTION AND COOPERATION I WHAT IS COOPERATION? I.1 It is a commonplace to say that human beings are social and are disposed to cooperate. We have learned from biology and ethology that such factors as kin-altruism and reciprocal altruism can ground cooperative behavior in animals. Analogously, we think that people are disposed to behave cooperatively at least with respect to their kin and, perhaps by a kind of extension, friends and other close group members and, in the case of moral people, with all human beings. (We could term this friendship or we-ness cooperation.) We also think that people often cooperate with strangers in terms of reciprocal exchange in business and related contexts ( exchange cooperation ). While a general, biologically based disposition to cooperate can perhaps be seen to exist in human beings, it is not that easy to specify under what conditions people actually cooperate rather than defect, act competitively, selfishly, or even aggressively. These latter kinds of behavior are all in their different ways opposite to cooperation, and people seem also to be disposed to such behaviors. It is surely of interest to investigate deeper the nature of cooperation and especially the conditions and circumstances that make it feasible for people to cooperate. It can also be argued that it is a necessary feature of human beings qua thinkers and agents conceived in terms of the conceptual framework of agency that they are social and to some extent cooperative. At least this sociality assumption is a general presupposition underlying any person s thinking and action on the whole, although in actual practice this presumption may be retracted from on particular occasions. A central argument for this sociality view goes in terms of the assumption that human beings conceived as thinking and acting persons necessarily are language users. As language necessarily is based on shared meanings and shared uses, we arrive at the sociality view, or its presupposition version, of human beings. One of the central ideas of this book is that collective reasons, primarily ones related to common goals and practices or to morality (e.g., to the require- 1

2 CHAPTER 1 ment of being helpful and cooperative), are in many cases required to account for cooperation. Collective reasons, being opposed to selfish and self-centered reasons, seem primarily to be due to education and related environmental factors or are based on or relate to what institutional authorities (be they persons, bodies of persons, or norms) require and expect to be realized. Needless to say, there are lots of institutional practices which are cooperative or have significant cooperative elements think, e.g., of the practices related to teaching, business, religion, and science. Collective reasons for cooperation can be thought to underlie these institutional practices, and the same goes for cooperation on a larger scale, especially international cooperation in its various forms. In the present philosophical work, my aim is to give a well-grounded and informative answer to the problems of what the basic features of cooperation are and what kinds of cooperation there are. Using the conceptual resources of this book, I will investigate under which circumstances and with what kinds of motivational grounds it is rational or useful (etc.) to cooperate. In addition, the problem of the possibility of achieving and maintaining social order will be discussed from a philosopher s point of view. These problems can be and have been addressed by social scientists, biologists, and game-theorists; and my account will attempt to take into account these discussions. Cooperation is a truly interdisciplinary topic and one with great importance for practical life. In this introductory chapter I will start with some general remarks on cooperation and then proceed to a preliminary discussion of social action with cooperative features. I will illustrate cooperation in terms of examples and introduce some technical notions, but specific analyses and discussions will be deferred to later chapters. My approach to cooperation is based on a philosophical theory of social action. It is argued that cooperative acting together forms the core of full-blown cooperative action. The term g-cooperation will be used for this kind of cooperation (or, more precisely, for cooperation based on a shared collective goal of a strong kind): it is cooperation based on the group-mode (or, synonymously, we-mode, to be discussed in Chapter 2). This view will be called the collective goal theory of cooperation. In addition, one may speak of cooperation also in weaker senses. Thus, for instance, cooperation in collective action dilemmas such as in Prisoner s Dilemma type of situations is shown to be weak cooperation to be called i-cooperation (and to be analyzed in terms of the notion of compatible coaction ). This latter kind of cooperation is based on the participants private or I-mode preferences and goals. A goal-based account of i-cooperation will also be given in the book. What has been said of cooperation in philosophical and scientific and other

INTRODUCING SOCIAL ACTION AND COOPERATION 3 relevant literature? While there is no need here to give a fully satisfactory answer to this question, a reader may still be interested in some kind of overview, even if partial and incomplete. I will below consider the matter briefly and leave a more extensive discussion to the Appendix of this chapter (and to some later chapters of the book). I.2 Let me begin with some general remarks on cooperation research. Broadly speaking, we can divide the literature into theoretical and experimental and related empirical literature. Among theoretical writings we can count dictionary definitions of the term cooperation and related words, philosophical (e.g., moral philosophical) and mathematical (especially game-theoretical) studies concerned with the concept, nature, and kinds of cooperation, as well as theories of the causes and other conditions of cooperation as well as of the evolution of cooperation. As to experimental literature, I primarily have in mind the vast literature dealing with experiments on cooperation in situations of collective action dilemma. There are also field studies in biology (ethology) concerning cooperation between animals. The survey in the Appendix to this chapter will give some sense to the main theoretical and empirical approaches and problems involved in cooperation research. It is useful at this stage to consider what dictionaries tell us about cooperation. Webster s Third New International Dictionary says this about cooperate : 1. To act or work with another or others to a common end; 2. to act together; 3. to associate with another or others for mutual often economic benefit. Collins Cobuild Dictionary defines cooperate thus: 1. If people cooperate, they work or act together for a purpose. 2. If you cooperate, you help willingly when they ask you for your help. Notions of i) acting or working together and ii) a common or the same end or purpose are central to definitions of this kind (cf. also the Appendix). It is thus not surprising that in various theories and views about cooperation, such as in the theory to be developed in this book, these elements will play a crucial role. This book concentrates on intentional cooperation of human agents and seeks to clarify the content and variety of such cooperation. Thus it concentrates on relevant conceptual and philosophical what-questions such as What is cooperation?, What conceptual means are most adequate for studying intentional cooperation?, What are the basic kinds of cooperation?, What is rational cooperation?. It also discusses questions such as Why is it rational to cooperate in such and such a context? and What reasons for cooperation

4 CHAPTER 1 are there in general?. As to the ontological aspects of cooperative action, the theory to be created is compatible with a variety of different views, and this matter will not be much discussed in the present book (for my own views, see, e.g., Tuomela, 1977, 1984, 1985b, 1995). The causal and evolutionary reasons for cooperation are the proper object of study of various empirical sciences, and I will only survey some of the relevant issues and try to see to it that the book is informed from the point of view of empirical research. As indicated, one of the central theses of the book is that there are two different kinds of cooperation, full-blown cooperation, upon analysis termed group-mode cooperation (for short, g-cooperation), and cooperation as coaction or coordination, termed individual-mode cooperation (i-cooperation). The first, a common-sense dictionary notion, bases cooperation on collectively accepted goals (collective goals in a strong sense) while the second operates at best with shared private (viz., purely personal) goals. Goal notions will be analyzed in Chapter 2. The early chapters (especially Chapters 2 7) of this book concentrate on the g-cooperation and the later chapters on i-cooperation. In these later chapters, a broadly game-theoretical set-up will be mostly used and strategic dependence becomes a central notion. Both kinds of cooperation are important and worthwhile objects of study in their own ways. While most current empirical studies concern cooperation in the coaction sense, this book emphasizes the full-blown notion of cooperation. In game theory there are in a sense parallel developments. Thus, cooperative game theory is related to my theory of full-blown cooperation. While the precise connections will be established later in the book, a central claim will be that current game theory is not capable of giving an adequate account of cooperation based on a collective goal. In the later chapters (especially Chapters 7, 10 12) rational cooperation will be concentrated on; the discussion is connected to both noncooperative and cooperative game theory and indeed makes use of some game-theoretical tools and results. II INTRODUCING THE BASIC ELEMENTS OF COOPERATION II.1 It is a platitude that cooperation is collective activity: we speak of two or more agents cooperating in order to achieve their ends or their shared collective end. One always cooperates in some collective context, yet we can speak of a single agent s action being cooperative as long as it is somehow based on a collective end (or some kind of jointness, such as processual jointness in activity) to be achieved by it. I will speak of a shared collective goal a state

INTRODUCING SOCIAL ACTION AND COOPERATION 5 goal or process goal in such cases. In addition, we can also speak of somebody s being cooperative in the sense of his having a cooperative or willing attitude towards some collectively endorsed goal or activity and towards the participants in such activity. As cooperation obviously relates to collective action with suitable social features, it is useful to start with some general remarks about the latter. Cooperative action can then be marked on this broad map. In this book, the conceptual framework of agency is adopted. It regards human beings as thinking, feeling, and acting persons who are morally responsible for their actions, who cooperate, argue cognitively with each other, construct and maintain social institutions (think, e.g., of institutions like the state, school, money, marriage). 1 Most of our actions, assumed characterizable in terms of the entrenched action notions in our culture, take place in a social context and accordingly are factually and typically also conceptually dependent on the existence of other persons and their relevant actions, on social institutions, conventions, practices, or the like. Thus, the action of drawing money from one s bank account conceptually depends on the banking institution. Saluting conceptually depends on the greeting practice and may depend on factual causal coordination (think of a marching army unit saluting a general). Acting for a social reason is indeed very typical. An example of such causal dependence (not built into the concept of the action in question) would be a boy s picking flowers in order to give them to his girlfriend. Often the causal dependence is of the causal-intentional kind in that it involves the use of language and communication. When a person buys something from a store, he normally communicatively interacts with a salesperson. An example which does not require communicative dependence would be walking in the street: people causally (yet intentionally) interact with other persons by trying not to bump into them. We intuitively think that some social action is cooperative while some is not. For instance, carrying a table jointly or singing a duet together seem to be unproblematic paradigm cases of cooperation whereas quarreling is noncooperative. What about playing a game of tennis? Is walking in a crowded street cooperative if the people intend to avoid bumping into each other? How about each of us lighting candles in the evening of the Independence Day? A philosopher wishes to know in more detail what is involved in examples such as these. What kinds of elements are or must be involved in cooperation, and how weak can cooperation be? II.2

6 CHAPTER 1 We can approach cooperation from the point of view of collective action and try to find weaker and stronger kinds of collective action. Among these kinds of collective actions will be actions which are cooperative to different degrees. To make this latter kind of assessment, relevant criteria for cooperativity must be available. The notion of a collective social action in its broadest sense is obviously a rather complex affair. It can be argued that it is action performed for a shared social reason (see Tuomela and Bonnevier-Tuomela, 1997). The notion of a shared social reason can furthermore be argued to amount to sharing a weattitude. Ideally, a person has a we-attitude (for example, a desire) relative to his group if and only if he or she i) has (or shares) this attitude, ii) believes the group members have it, and also iii) believes that there is a mutual belief that the members have this attitude (see Chapter 2 for further discussion). A shared we-attitude is a shared social reason for which the agents perform their actions constituting the (intentional) collective social action in question. For instance, people may share the we-goal of putting lit candles in their windows on Independence Day: each agent s (or, perhaps family s) goal is to light candles and each such agent believes that the others in the community also share this goal (and, let us also assume, share it in part because the others have that goal) and that this fact of sharing is mutual knowledge. The people then satisfy their shared we-goals by lighting the candles and putting them in their windows. They then act collectively in part for the social reason that all the others put lit candles in their windows. Acting on the shared social reason will yield some collective harmony (in goals and actions) and it can be regarded as weakly cooperative. In general, this holds for analogous cases as long as the participants goals and preferences in the collective social action are not incompatible (or, more generally, are not negatively correlated). While excluding major conflict, collective social action still need not always deserve to be called cooperation as such, for the participants need not be connected in the right way. In the Independence Day celebration case there is enough connection to give us a case of i-cooperation, for the agents in this type of case take into account what the others are doing and, in principle, are prepared to adjust their acts to each others actions (e.g., if the time of lighting the candles is changed due to some sudden mutually known fact requiring social attention). However, there are resembling cases in which there is not enough dependence. Suppose it is a custom in Finland to go bathing in a sauna every Saturday afternoon. Even if it were mutual knowledge among Finns that everyone thus goes to sauna this would not yet create cooperation even in a weak sense. But

INTRODUCING SOCIAL ACTION AND COOPERATION 7 if their reason for going to sauna would in part be that the others also go (so that they also would not go unless the others go), we have a weak case of cooperation, viz., i-cooperation. Thus people merely acting to reach the same kind of goal, perhaps mutually believing that they do so, does not suffice for even i-cooperation. The above observations also suggest that collective action in the sense of action involving several persons as such is not a very interesting notion for many purposes. Only when the action is based on social reasons roughly, what the others want, expect and do do we get hold of action that can serve as a building block of a theory of cooperation. Intentional cooperative action, qua based on a shared collective goal, is necessarily social, for it involves taking into account other people (the participants in cooperation) as part of one s reasons for acting (cf. Tuomela and Bonnevier-Tuomela, 1997, for discussion). This creates the kind of dependence required by cooperation. Notice, however, that the dependence may be either physical or mental. A case of physical dependence is involved in the example of people trying not to bump into each other when walking in the street. In the mental dependence case e.g., the Independence Day celebration the actions in question may be performable independently of each other, although the participants act for a social, dependence-creating reason. Acting together involves sociality in the relatively strong sense that such action must be based on joint intention or shared collective goal. This makes any case of acting together cooperative at least to the extent that the persons are collectively committed to making true a certain state of affairs. What acting together in this general sense precisely involves will be clarified in Chapter 3, where three kinds of acting together will be distinguished. In the strongest sense of acting together we require acting on a joint, agreed-upon plan. This I will call proper joint action. Examples of such joint action typically are singing a duet, playing tennis, or walking together. This kind of joint action is collective social action in its most central sense, acting as a team. The joint action has a cooperative element in that the participants are jointly committed to acting together and to relying on the other participants performing their own parts. (Note that also joint refraining from an action can be cooperation: For example, we may jointly refrain from opening the door when the doorbell rings.) Cases of joint action with an inbuilt element of conflict, such as in playing a game of tennis, are to a considerable degree cooperative in contrast to cases of pure conflict such as being involved in a fight for one s life. Let us now consider cooperative joint action a special case of cooperative collective action in some detail (for fuller discussion, see Chapters 3 and

8 CHAPTER 1 4). It was said that such intentional joint action must be based on a relevant agreed upon joint intention (or, generally, plan of action ). For example, you and I may form the joint intention to paint a house together and we may agree upon some means of doing it. Thus, we may agree that you paint the front while my part consists of painting the back of the house. We also agree upon which paint to use and suchlike things. Carrying out our joint intention, we come to paint the house jointly. This kind of cooperative intentional joint action can be characterized more generally as follows, assuming a joint action to be an action divisible into single-agent parts which generate a purported outcome: The participants have formed a joint plan for a joint action; the plan is taken to involve a relevant joint intention, entailing for each participant the intention to perform her part of the joint action. Each participant is assumed to believe (and rely on the fact) that the various conditions of the success of the joint action will be fulfilled at least with some probability, and she must also believe that this is mutually believed by the participants. In general, the performance of a joint action can be regarded as agreement-based if the plan has been accepted by the participants and if they have communicated their acceptances appropriately to the others so that a joint commitment to perform the joint action has come about. This shared plan to perform a joint action gives a kind of cooperative base for a joint action and both a quasi-moral and an epistemic basis for the participants to trust (rely on) that the others will perform their parts and not let them down. It is important to notice that the participants preference structures in a joint action can be fully cooperative (cf. carrying a table) in the sense of being highly correlated or can be to some extent opposed (cf. chess, selling and buying). This is a feature of cooperation in the sense of preference correlation; it will be emphasized when the underlying motivation and rationality conditions of cooperation are discussed. We can speak of given preferences (preferences giv ) concerning the joint or collective activity in question. These are the preferences the participants have prior to action and prior to their possibly having considered the situation of cooperation in strategic terms. These given preferences can be need- and interest-reflecting, but need not be rational or considered preferences. 2 They may also reflect objective payoffs such as money or other quantifiable and transferable objective goods. Depending on the case at hand, they can be either natural or institutional and highly culturedependent. This last distinction need not be regarded as a dichotomy, and it is not a very clear one either. It corresponds roughly to a similar distinction concerning joint actions a joint action can be physical (e.g., carrying a table jointly) or it can involve a conventional or normative element such as transfer

INTRODUCING SOCIAL ACTION AND COOPERATION 9 of rights. As to the latter kind of cases, think, e.g., of toasting a national victory together (based on a social convention), making a business deal (legal transfer of rights), getting married (legal creation of rights and duties). When cooperating in a normatively defined situation the participants accept, or are assumed to accept, the goals, tasks, and parts defined by the norms in question. Typically, there are many cooperative elements (such as corresponding and interlocking part-related preferences) in norm-governed situations. The correspondence between the preferences in question thus will be at least in part normatively determined. Normative determination is taken to mean that the correspondence is defined by means of an agreement or a social norm (either a rule-norm or a proper social norm ) or a normative mutual expectation. Actual cooperative activity will take place on the basis of the participants final preferences (or preferences fin ), which so to speak by definition take all the situationally relevant considerations into account. One may participate in cooperation either willingly or only reluctantly. In the former case we can say that the participant acts with a cooperative attitude lacking in the latter (cf. the kind of cooperative activity involved in an Italian strike type of situation performed with an unwilling attitude). Part of what is meant by a cooperative attitude is that a participant with such an attitude is supposedly disposed to transform his relevant situational preferences in a cooperative way, e.g., to transform his given preferences into final ones so as to take the participants joint reward in the situation into account. This joint reward can concern both the collective end or ends in question and the means-activities related to its achievement. A standard example of a person acting for a cooperative attitude would be a moral person loving one s neighbor as himself and acting accordingly. However, a cooperative attitude need not in general be based on altruism. Collective social action can of course be cooperative also in a stronger, fullblown sense involving a suitable collective goal without yet being based on a shared plan (agreement). However, acting together must always be present (as will be argued in Chapters 2 and 4). Acting together, AT for short, can either be based on a shared plan (AT p ), mutual belief (AT mb ), or just plain belief (AT r, rudimentary acting together). These kinds of acting together will be discussed in Chapter 3. Consider people cooperating to keep the streets clean. Whether a singular example should correctly be classified as a g-cooperation or i-cooperation by definition depends on what kinds of goals the participants have in their minds. Determining this may often be difficult and will rely on observing people s actions, especially their helping behavior, and asking them relevant questions

10 CHAPTER 1 about their goals. I wish to emphasize that there are many varieties of i-cooperation and some kinds come close to g-cooperation. Thus, for instance, in the case of structured groups there are operative members for making decisions and forming joint intentions for the group as well as for carrying out such intentions. The rest of the group members are supposed to cooperate at least to some degree. Thus, some of the members may intend to contribute to the group end without really having that end and some others only tacitly accept the group end and the group action leading to it without even intending to contribute. At least the first possibility here qualifies as i-cooperation. II.3 There are broader issues of cooperation in society overall to which the developments in this book are relevant. In Rawls (1993) a theory of political philosophy is created in which society is viewed as a fair system of social cooperation between free and equal persons viewed as fully cooperating members of society over a complete life (Rawls, 1993, p. 9; also cf. pp. 300 301). In his idealized system justice as fairness is a shared common end (in my terminology: a shared collective goal) which serves to make society-wide institutional cooperation full-blown cooperation in the sense of the present book (see the discussion in Chapter 13). In addition to Rawlsian kind of liberal theory, communitarian and republican accounts of society rely on shared collective goals and cooperation (cf., e.g., material and structural equality, freedom as non-domination). Therefore, the theory of this book should help to clarify and even fortify these kinds of theories (cf., e.g., Pettit, 1997, for a discussion; also cf. the points made in Tuomela, 1995, Chapter 10). Society consists of social institutions and social institutions overall or in many cases rely on cooperation. For example, they provide and build on collectively rational solutions to collective action dilemmas, and such solutions require the existence of cooperative action patterns. Fairness is a feature typically present in cooperation, and the notion of just cooperation is of course important in moral and political theory (cf. Rawls, 1993, Moulin, 1995, and Roemer, 1996). However, there can clearly be also unfair and unjust cooperation. In this book, the political and other moral reasons (such as the moral requirement of being helpful to others) will not be in the forefront. Our discussion leads us to the following self-explaining diagram:

INTRODUCING SOCIAL ACTION AND COOPERATION 11 DIAGRAM 1. Let me still make a remark on my terminology related to cooperation. The terms g-cooperation, referring to full-blown cooperation, and i-cooperation, referring to cooperation as coaction, will be generally used in this book for a central dichotomy which derives mainly from a division concerning the goals (thus intention contents) the participants are striving to achieve. COOPERATION g-cooperation (based on a shared collective goal) i-cooperation (based on compatible private goals) institutional non-institutional institutional non-institutional

12 CHAPTER 1 III THESES ON COOPERATION The view defended in this book is that when human beings g-cooperate (viz., intentionally cooperate in the full sense) they are acting together intentionally in the pursuit of an intended collective or joint goal, whether or not they at the same time also are acting in the pursuit of their intended private (viz., merely personal) goals. The former is a we-mode goal (g-goal, properly collective goal) and the latter are I-mode goals (i-goals, private goals). Acting together involves sharing in an action in a we-sense ( we are doing this together ). There must be jointness or togetherness in full-blown cooperation at least in the sense that the participants intend to act together and base their action on the expectation that others will participate at least with some likelihood. The agents must minimally share the joint action (of the AT kind) with regard to which they cooperate. In accordance with what has just been said, g-cooperation must involve a jointness-aspect. I will call this shared jointness the agents shared (proximate) collective goal assumed to satisfy what will be called the Collectivity Condition (see Chapter 2). The collectivity condition involves that, because of the participants collectively accepting the goal as their collective goal, it is satisfied for a participant if and only if it is satisfied for the others. The joint thing shared here can be a joint or collective action defined by reference to a state for instance, cooking a saucepan of pea soup or be the activity itself (the agents might love acting together, independently of what the activity results in). Cooperation in its fullest sense requires in addition that the participants participate willingly (as opposed to reluctantly). My theory will emphasize the presence of a collective goal (end, purpose), a cooperative attitude, and some degree of mental and behavioral dependence generated by the collective goal and (possibly) by the cooperative attitude. Furthermore, the commonality of action-related preferences and interests will be argued to facilitate cooperation. The presence of a collective goal in the strong sense meant here is required of g-cooperation in its full sense, but it is not required of weaker forms of cooperation, viz., i-cooperation. The latter corresponds at least roughly to what some authors mean by speaking of cooperation as coordination (cf. Bicchieri, 1993). But it also can include cases of cooperating as a group member when the group goal or intention is not fully endorsed and when there accordingly is not collective commitment to the group goal. To be more specific, all intentional cooperation in the full sense must involve, firstly, intention-based commitment to a collective goal or plan. (In the special

INTRODUCING SOCIAL ACTION AND COOPERATION 13 case of plan-based joint action, the collective goal will be called a joint goal.) Secondly, if the participants interests or preferences, qua participants of the action, are strongly correlated, this fact makes the nature of the action situation cooperative (by making helping desired and rational). This contrasts with the case where the interests of the participants are opposed. Thirdly, the cooperative nature of an action is enhanced if the participants act with a cooperative and helping attitude perhaps deriving in part from their viewing the cooperative situation as fair concerning the expected burdens and rewards. It is one of the main tasks of this book to give a more precise and illuminating account of these three components of cooperation. In the case of i-cooperation the last two elements are central even if a shared collective goal is not required. While the developments in the chapters to come mostly deal with specific problems related to cooperation, some general theses will also be defended. I will next present a survey of most of them. First, the central thesis for cooperation in the full sense is: Basic Thesis of Cooperation: Two or more actors cooperate in the full sense if and only if they share a collective (or joint) goal and act together to achieve the goal. This thesis is an approximate statement of what cooperation in a full sense is argued to involve. If it is acceptable, cooperation can obviously be called g- cooperation, viz., cooperation in the group mode involving a shared collective goal; and I have already spoken as if this basic thesis were true. The present thesis must be understood broadly enough to be compatible with the claim that not all cooperation needs be acting towards a collective end-state. This is because there can be full-blown cooperation which only involves shared activity, a collective action-goal, but does not purport to lead to a shared collective end or purpose at all. Note, too, that cooperation in the sense of the Basic Thesis of Cooperation does not require an agreed-upon joint plan (cf. Chapters 4 6 and 10 12). The notion of collective action, viz., acting together, in the Basic Thesis of Cooperation will be clarified later in Chapter 3. It is weaker than the notion of agreement-based joint action. The Basic Thesis of Cooperation will be discussed especially in Chapters 3 and 4. It will be shown to precisely what kind of activity it applies, and the requirement of the presence of a shared collective goal a notion to be clarified in great detail in Chapter 2 is defended in Chapter 4. The theory of full cooperation of this book can in view of the Basic Thesis of Cooperation be called a collective goal theory of cooperation. As to i-cooperation (viz., cooperation as coaction), it is meant to cover coordination-achieving cooperation based on private goals. By i-cooperation

14 CHAPTER 1 (coaction) I accordingly mean interaction with compatible private goals and with the intention of satisfying one s goal by means-actions which do not conflict with others attempts to achieve their goals. Somewhat more precisely I require of such compatible coaction: 1) there must be compatible goals, viz., goals which can be satisfied in the situation without the kind of conflict preventing the others to satisfy their goals; 2) there must be at least an action-intention to avoid satisfying those goals by means-actions which strongly conflict with others attempts to reach their goals; 3) the actors are dependent in that action situation on each other s action, viz., they have to take the others actions into account in attempting to achieve their goals in an optimal way; 4) the goals may (but need not) be of the same type; they may also be shared; 5) the agents must have beliefs about the participants goals being compatible and about their intending to avoid satisfying them by means-actions conflicting with the others goals (or at least they must think the latter kind of cooperativeness will probably exist); 6) each agent intends to achieve his goal and believes he can do it in that context at least with some probability without coming in conflict with the other persons attempts to satisfy their goals. (See Tuomela and Bonnevier-Tuomela, 1997.) We thus arrive at the following schema explicating the most general notion of (compatible) coaction: (CO) Agents A 1 and A 2 coact compatibly in a situation S relative to their I-mode goals G 1 and G 2 if and only if 1) their respective primary goals (viz., action-goals) in S, i.e., types of states or actions, G 1 and G 2, which relate to the same field of mutually dependent actions in S, are compatible in the sense of being satisfiable without making it impossible for the other agent to satisfy her goal; 2) a) A 1 intends to achieve G 1 without means-actions conflicting with A 2 s attempts to satisfy his goal and believes that he can achieve it at least with some probability in that context although his relevant G 1 -related actions are dependent on A 2 s relevant G 2 -related actions, and he acts successfully so as to achieve G 1 ; and b) analogously for A 2 : 3) a) A 1 believes that 1) and 2), and b) analogously for A 2. By i-cooperation I thus simply mean compatible coaction in the present sense. I-mode goal in (CO) means private goal; see Chapter 2 for a fuller discussion. (CO) can be argued to be in conflict which the mentioned Collectivity

INTRODUCING SOCIAL ACTION AND COOPERATION 15 Condition assuming that it entails that a participant s goal can be achieved by acting alone (but with the social and other environment not preventing goalsatisfaction) rather than by acting together. (A resembling, weaker condition with a contingently true equivalence may still hold, where the equivalence is not true on the ground that the participants accept it as true.) (CO) by itself is rather weak but it does exclude cases of conflict. Although it deals with cases of dependence, the participants are required not to engage in conflict-involving behavior on purpose. Let us consider a simple example to illustrate (CO) in the case of a coordination situation. Agents A 1 and A 2 have as their shared private goal to meet (thinly conceived: to arrive at the same place). Let the means actions in the case of both agents be going to the station (s 1, s 2, respectively) and going to the church (c 1, c 2 ). Their actions leading them to meet (viz., the pairs s 1, s 2 and c 1, c 2 ) achieve coordination and lead to the satisfaction of their goals. The two other action pairs do not satisfy their goals. There is no conflict in this situation. Both agents are assumed to act individually ( privately ) rationally to successfully achieve their goals (cf. the Reward Thesis below). This entails that they must have coordinated successfully (viz., selected one of the action pairs leading to coordination). Accordingly, (CO) entails that the coordination problem in question has been coordinatively (and thus cooperatively) solved. Conflict can be introduced into the situation by assuming that the agents have different preferences concerning where to meet, although they still prefer to meet rather than not. Here the conflict is not disturbingly big, and this modified coordination case (a Battle of the Sexes situation) qualifies as i-cooperation in the sense explicated by (CO). Competitive cases and zero-sum cases with strongly conflicting means actions do not belong to coaction in our present sense. (CO) can accordingly be regarded as a general schema for compatible coaction, which can be strengthened into coaction towards a shared divided goal such that G 1 and G 2 will both amount to the same goal G, and to acting together with a shared collective goal. Next we consider some further central theses of the book, which are here presented without fuller discussion and justification (see especially Chapter 9). They basically apply both to g-cooperation and i-cooperation although the formulations are geared to the former case (cf. Chapters 9 and 12): Commonality Thesis: Ceteris paribus, the more commonality of interest (preferences) there is in a situation, the more likely cooperation is to be initiated and carried out successfully and speaking of rational cooperative situations to give the expected individual rewards from cooperation, understanding this to mean rewards from acting together (relative to not acting so).

16 CHAPTER 1 The preferences here concern all the action-outcome combinations possible in the situation of interaction in question (see Chapter 9). That cooperation is successful in the sense of the Commonality Thesis has to do with improvements related to a) the selection of a shared goal and the means of reaching it, b) the stability of the commitment to the collective goal-directed action, c) the opportunities to help the other participants succeed in their part-performances, and d) flexibility concerning change in a collective goal when called for. The Commonality Thesis can as such be regarded as a general empirical thesis claiming that similar want-based preferences tend to motivate and explain cooperative action. This issue and the rationality of helping under various conditions related to preference correlation will be discussed in detail in Chapter 9. It will be emphasized that cooperation may be, and often is, governed by normative factors (see Chapter 6). In particular, in the case of agreementbased joint action it is rational to give help concerning the coming about and maintenance of joint action, as long as helping is not too costly. These factors can be perfectly correlated, because of the nature of agreement making. Accordingly, in cases where cooperative action is not based on social norms or rules, the participants agreement-making can create perfectly correlated preferences concerning the content of the agreement, when this concerns what cooperative activity to pursue and how. One can also enter an agreement to compete (e.g., to play tennis or to run a race). In these rule-governed cases the agreement creates perfectly correlated preferences concerning only the contextual elements of the cooperative activity serving to make it possible (viz., concerning the so-called joint action opportunities). In a normatively governed cooperative activity the preferences (qua preferences normatively belonging to the action) may even be fully normatively, e.g., institutionally, determined. One has to keep the notions of a cooperative situation and the actual action of cooperation analytically separate. Both notions will be investigated in this book. If one chooses to characterize cooperation in terms of helping as is often done in this book one can and should distinguish the features of an action situation which make helping rational (and otherwise feasible) from actual helping behavior itself. The preferences the participants have on entering a situation of potential cooperation are given preferences (preferences giv ) while the preferences at which they arrive in their discussions, bargaining, or otherwise, in a particular situation and which they finally act on are final (preferences fin ). Here is a rather obvious claim concerning their relationship: Closeness of Given and Final Preferences Thesis: Ceteris paribus, the closer (and the higher) an actor s given and final preferences, viz., prefer-

INTRODUCING SOCIAL ACTION AND COOPERATION 17 ences giv and preferences fin, are, the more likely he is to cooperate rationally in the long run (in a sense respecting his given preferences). Closeness is to be measured in terms of utilities defined in terms of the respective preferences. Generally, cooperation is expected to be individually rewarding. This expectation is a relatively minimal rationality feature related to cooperating participants. This leads to the following thesis:

18 CHAPTER 1 Reward Thesis: Ceteris paribus, all intentionally, knowingly, and reflectively undertaken cooperation by normal ( normally rational ) and normally acting human agents is expected by the participants to be more rewarding to them than non-cooperation at least in circumstances favorable to carrying out the activity. This thesis is understood to be true in virtue of our common-sense ideas about motivation to cooperate rationally, even if reward-expectation need not be regarded as a conceptual feature of cooperation. Intentionally and knowingly undertaken cooperation here means, roughly, a case of cooperative activity in which the people know what is going on, viz., that they possess the concept of cooperation to an adequate extent and correctly apply it to the situation at hand. For example, when a person asks somebody to join him in performing a task, the latter must of course understand what is requested and what is required of him if he is to perform his part of the joint action. In the Reward Thesis, the reward in question is that obtained from the collective cooperative activity itself as compared with not cooperating. The reward attained by achieving the collective goal involved in cooperation is a separate matter. In view of the Reward Thesis, all conceptually and informationally adequate cooperation is at least minimally rational in the sense of involving an individual reward expectation. The normality (or normal rationality ) assumption excludes such people as small children and the mentally ill; unreflectively acting normal agents will also be excluded. (The notion of favorable circumstances will be discussed in Chapter 3.) Note, however, that the Reward Thesis is compatible with the existence of cooperation without reward-expectation: people can cooperate less than fully intentionally or without adequate knowledge of the nature of the situation simply by adopting a joint (or collective) goal whose joint achievement need not actually be rewarding for the agents nor believed to be rewarding, even in favorable conditions. This could be called the Plain Cooperation Thesis. We can also note here that, accordingly, people may cooperate towards a collective goal without the goal being i) Pareto optimal ( non-improvable ) or ii) an equilibrium, relative to the agents private preferences in the situation. (A goal may be Pareto optimal without being an equilibrium, and conversely, as collective action dilemmas show see Chapter 10.) Some refinements of the above claims about cooperation will be discussed in Chapters 9 and 12. In Chapter 11 the following thesis related to the reasons for cooperation will be defended, especially in the case of rational cooperation: Motivation Thesis: One may cooperate for one s private reasons (which are allowed to be selfish or other-regarding as well as short-term or long-term)

INTRODUCING SOCIAL ACTION AND COOPERATION 19 or for one s collective reasons; these reasons may be in conflict with each other (serving to create collective action dilemmas). There are situations of social interaction in which acting for a collective reason rationally furthers also the private interests (preferences, goals) of the participants. In some such situations individual interests cannot be satisfied to a maximal or optimal degree (relative to the possibilities inherent in the situation) or cannot even be satisfied to any degree at all without acting for a collective reason. These situations (also the latter kind of situations) include a) cases with no conflict between the participants private interests or between their private interests and collective interests (cf. for example jointly carrying a table and instances of pure coordination) and b) collective action dilemmas, viz., instances involving a conflict between private and collective interests. As said, the theory of g-cooperation to be developed in this book is a collective goal theory with emphasis on dependence be the dependence due to the (non-social) environment or to the participants joint creation in some sense. The account of i-cooperation will also rely heavily on dependence. Behavioral dependence is required only in a minimal sense in the general case the sense that any collective activity requires it. However, much of the book is concerned with situations of strategic interaction in which a kind of actional dependence is a central element. The dependence relations in question will technically be regarded as dependencies between the preferences (or utilities) of the interacting agents in the first place. The feature of dependence obviously relates to issues of social power. To what extent successful cooperation depends on various kinds of situational social control possibilities (power) will be investigated in the book. What was said at the end of Section II gives rise to the following thesis on institutional cooperation: 3 Institutional Thesis: Cooperative structures are central for the existence and maintenance of social institutions and, hence, society. The discussion especially in Chapters 6 and 13 is relevant to this thesis, which is not truistic as also conflicts of some suitable kind may be argued to be important. IV SURVEY OF THE CONTENTS OF THE BOOK The following is a concise chapter-by-chapter description of the contents of the book following the present introductory chapter: In Chapter 2, Collective and Joint Goals, the notions of collective and joint goals are discussed as important conceptual elements for the theory to be

20 CHAPTER 1 developed in the book. Various kinds of collective goals are discussed and a crucial condition, the Collectivity Condition, is introduced for collective goals. Another related principle (the criterion (ET*)) for distinguishing participants group-attitudes from their privately held personal attitudes is also introduced and discussed. Joint goals are the strongest goals that plan-based joint actions require. They are plan-based (or agreement-based) in a specific sense clarified by the so-called Bulletin Board view. Chapter 3, Cooperative Joint Action, argues that cooperative joint action, the prime case of g-cooperation, is based on the possibility of helping and on a willing or cooperative attitude towards the joint action and its participants; this also generally makes it feasible for the participants to trust each other s effort to participate. The possibility of helping presupposes highly correlated preferences either the given preferences built into the action situation itself or the final or effective, transformed preferences the participants qua cooperative agents have acquired. This chapter also gives a detailed characterization of various kinds of acting together (AT) and relates cooperation to them. In Chapter 4, Cooperation and Collective Goals, it is argued that g- cooperation, be it joint action or not, is based on the participants acceptance of a collective goal (one satisfying the Collectivity Condition). The requirement of the presence of a collective goal for cooperation is argued for in several ways. This chapter also discusses weaker kinds of cooperation, such as informationally incomplete cooperation, unilateral cooperation, and cooperation as coaction. Chapter 5, Cooperation, Practical Reasoning, and Communication, discusses the various patterns of practical reasoning pertaining to g-cooperation and i-cooperation. Communicative action is discussed in analogous ways and is similarly divided into g-communication and i-communication. Chapter 6, Cooperation in an Institutional Setting, examines various kinds of social norms and discusses the central elements of social institutions in the context of cooperation. It is argued that normative considerations have the effect of creating correlated institutional preferences contributing to cooperation. Normative authorities may, of course, also affect cooperation. This chapter also emphasizes the role of collective social construction. An important, structural part of social reality is in a sense constructed by collective acceptance, a weakly cooperative activity. This phrase, used here as a technical term, is to be understood broadly to include collective construction, acceptance and maintenance. This account is related to the problematic reflexive idea that a collective goal is one the members of the collective in question have collectively accepted as such.