Published in International Journal of Philosophical Studies 2 (1994), 306-321

(Note: This is a manuscript version. Please quote only the published version.)

IN SEARCH FOR THE COMMON MIND
A Critical Notice of Philip Pettit's The Common Mind

Raimo Tuomela
University of Helsinki

I INTRODUCTION

The philosophy of social science can still be regarded as a much less densely populated part of philosophy than most other fields in it. However, there are neighbours in which research is booming. Thus the philosophy of cognitive science (and philosophy of psychology, to use an older title) is very popular today and so is moral philosophy. Philip Pettit's new book (1993) is mainly on the philosophical problems of social science, but a substantial part of it is also devoted to philosophy of psychology and to moral and social philosophy. This book is a most welcome treatise and a major accomplishment in its field. It is written in the best tradition of analytical philosophy. Pettit is very skillful both at making central conceptual distinctions and at devising and presenting imaginative arguments for his theses. The book will be of great interest not only for researchers and students in the fields of philosophy it deals with but also - because of its avoidance of technical jargon - to readers in other fields. Even if accessible to a larger audience, the book is yet not elementary nor too easy. It is very rich both in the topics covered and in the detailed arguments presented and defended. Pettit is not afraid of the hardest philosophical problems; accordingly his readers will be guided on a very interesting trip in a tough philosophical terrain. Finally, I should say that the book is also a very informative and reliable guide to recent developments in philosophy of social science and can well be recommended as a textbook.
The present article will attempt to give a brief presentation of the main topics and ideas covered in the book. This will be accomplished in Section II (together with making some minor critical points). Sections III-VI are critical sections dealing with social holism (in Pettit's special sense), social regularities, as well as individualism, supervenience, and explanation. There some criticisms - both internal and external - against Pettit's views and arguments will be presented. At the same time some of the main topics of the book will be presented and discussed in more detail. Section VII concludes that while none of the criticisms presented is damaging, some one-sidedness in Pettit's approach can be seen. The view advocated in the book could and should have been enriched by giving a central role to "jointness-notions" such as joint action and joint intention relying in part on the notion of "we" and "our group".


II PETTIT'S MAIN IDEAS: A SURVEY

The book creates a systematic philosophical view of man and society. It consists of three interconnected parts: I Mind, II Mind and Society, and III Mind, Society, and Theory. Each part contains two chapters. In addition each part presents a lengthy - and very accurate and informative - preview of its contents. I will draw (sometimes verbatim) from these previews below. (Page numbers will all be to Pettit's book.)
The book is long - 365 tight pages. It is in part based on his recently published (and to-be-published) papers, often with the omission of some less central details. The first part consists of two chapters: 1. Thought, and 2. Intentionality. This part is concerned with some central problems in the philosophy of psychology. In the first chapter the intentional aboutness of such mental states as perception, belief, and desire is discussed. To be an intentional system, says Pettit, is to be a system that has beliefs and desires and the like, is to be exposed perceptually to a certain sort of environment in a way that makes sense, and makes sense non-accidentally, in belief-desire terms. This is an acceptable general characterization. Next, Pettit assumes that there are regularities characteristic of beliefs and desires, regularities that dictate both the effect of certain sorts of evidence on what beliefs and desires are maintained, and the effect of certain sorts of belief-desire profiles on what responses are evinced. Intentional agents interact with their environment under the control of intentional regularities, whereas non-intentional agents do not. Beliefs, desires, and the like are argued by Pettit to be higher-order states doing their causal work via the brain-states instantiating them. In this connection he presents the "program model" of explanation to help his defense of the mentioned view of mental states. Although I shall not here go into the matter, this view is both "externalist" and naturalist. Mental states are argued to supervene on brain states and whatever naturalistic features of the world are required: Assuming the absence of "non-actual" and "ghostly" stuff, if two possible worlds are naturalistically identical then they are also intentionally identical (p. 27). (I think one could here make a distinction between contingently and in some sense necessarily identical worlds; this is relevant in the consequent of the above statement.)
Chapter 2 is concerned with thinking intentional systems. To be a thinking intentional system two requirements must be met: a) the system must be capable of intentional ascent, viz. capable of having not only intentional states with contents but also have intentional states about such contents; b) it must be capable of following rules. Pettit wants to allow for thinking intentional systems without language and - in part therefore - formulates the notion of a rule of thought in terms of possible worlds. Propositions are thus regarded - in accordance with one standard conception - as functions from possible worlds to truth and falsity. To think is to try to conform to the rules which certain propositions represent. In this chapter, which is an innovative naturalistic account of the thinking man, Pettit gives a Wittgenstein-flavored "ethocentric" account of rule-following (including, by the way, a good account of the notion of favourable conditions). I will not here discuss its details, as the focus of this article is on the philosophical problems of the social sciences.
Part II contains two chapters: 3. For Individualism, against Collectivism, and 4. For Holism, against Atomism. What Pettit calls individualism is an ontological view according to which the autonomy (or "autarchy") of individual agents is not seriously curbed by higher-level aggregate and structural social factors. This is a "vertical" issue in Pettit's terminology. The other issue is "horizontal". Pettit's holism - as contrasted with atomism - is not a relation between individuals and higher level factors but between individuals themselves. Holism claims that individuals' thoughts are in part constituted by social relationships. The combination-view that Pettit defends in his book is "holistic individualism", viz. vertical individualism conjoined with horizontal holism. In Chapter 3 he defends individualism against various anti-individualist theses and views defended by some social scientists and philosophers. He regards as especially central (to individual autarchy) the question whether social-structural regularities of the Durkheimian kind (e.g., that unemployment leads to increase in criminality), which are "causally and logically discontinuous" with the intentional regularities of individual psychology, "override" or "outflank" intentional regularities (see Sections IV and VI below). Individualism emerges as a winner. The content and argument for Pettitian holism will be discussed in Section IV below.
The final part of the book contains two chapters: 5. Social Theory, and 6. Political Theory. Pettit says that he means "by social theory the discipline of explaining the doings of individuals in society and the patterns of social events" (p. 217). By political theory he means "the project of evaluating the different social structures that political activity enables us to contemplate as alternatives" (p. 217). The discussion in these chapters is in part meant to explore the significance of holistic individualism for those areas.
In Chapter 5 an "inference-theoretic" approach emphasizing the agent's own point of view is introduced. Here the discussion is sketchy and does not properly take into account the voluminous discussion available in literature (cf. the discussions emphasizing the "practical syllogism" and the like patterns of practical inference). This chapter also discusses social supervenience and its role in structural and historicist social explanation - see Sections V and VI below. Also rational choice theory is discussed here, and Pettit directs several criticisms against its limitations. All this is welcome. I would like here to put forth a couple of minor critical points concerning his treatment. First, Pettit claims that according to rational choice theory "agents are to a good extent, if not predominantly, self-regarding in their desires" and discusses the problem of accounting for other-regarding motivation (p. 265). But - although Pettit does not make a very strict claim here - let me note that neither rational choice theory nor its mother game theory makes the mentioned assumption (see, e.g., Hargreaves Heap et al., 1992). Nor is economic theory committed to self-regarding desires, even if in actual practice utilities perhaps normally are considered self-regarding. Secondly, Pettit introduces a distinction between action-dependent goods (goods one gets by grace of what the agent himself or others do) and attitude-dependent goods (goods one gets by grace of what the agents himself or others think). This is a good distinction, if it is understood widely enough. Pettit refers to the distinction also by speaking about a concern with economic gain and a concern with social acceptance. In addition to armchair considerations I would have liked to see some mention of empirical research in recent motivation psychology in this connection. How does, for instance, the claim by the so-called theory of achievement motivation fit in the picture? According to this relatively well-supported theory the main human motives are the motives for achievement, sociality, and (social) power. These motives do not directly fit into Pettit's classification (thus sociality and power each have aspects related to both action-dependent and attitude-dependent goods).
Chapter 6 is concerned with questions of value and justice. Pettit discusses what he calls the contract-centred, the value-centred, and the institution-centred approaches to political arrangements of various kinds. Pettit argues for a "republican" view of freedom involving the idea of the state's guaranteeing (negative) freedom. The discussion in this chapter provides a good, but perhaps too concise, evaluation of the current state of the art. Pettit defends an institution-centred view and tries to get rid of the involved problem of "who guards the guardians" in terms of intangible hands - norms which are obeyed more or less without costs. In this chapter Pettit applies his holism in the form of holistic social values and claims that atomists must adopt what might be termed value-holism (values which cannot be solipsistically realized). It is not clear, however, that an atomism in the sense of the negation of Pettit's holism of Chapter 3 is committed to solipsistically realizable values. This point should have been better argued for.
Chapter 6 abounds with arguments against developments within the other two approaches. I would like to say, however, that while criticisms of some details of other approaches are all right there is the danger of premature criticism. For instance, Pettit strongly (but correctly) criticises current economic bargaining models for failing to do this and that. It should be kept in mind that research about formal bargaining models is very recent and that therefore much better models are likely to emerge. (Pettit could have emphasized more that economic bargaining models basically concern rationality rather than justice or other moral notions and that some principles going beyond economic rationality will be needed.)
As said, the parts of the book are interconnected, Part III depending on Part II, which builds on the ideas developed in Part. Perhaps the most central key idea of the book is to view man as a thinking intentional system, constitutively bound to a social community (Pettit's holism!) via the rules followed in thinking.


III SOCIAL HOLISM

I will start my detailed critical comments by a brief discussion of Pettit's social holism - a doctrine to be kept apart from what more usually is meant by holism. Pettit defends "horizontal" holism - the constitutive dependence of human thinking on social relationships. What is more often called holism in the literature is a view about the relationship between agents and their properties and interrelations, on the one hand, and social macro-properties (social "wholes" like groups, group-properties, and social structures). This dispute Pettit calls the individualism-collectivism dispute. While this section will be concerned with social holism in Pettit's sense the rest of this paper is concerned with various aspects of the individualism-collectivism dispute. Thus Section IV is concerned with the nature of social regularities and laws in this respect, Section V discusses the assumption of supervenience claimed to ontologically connect the individualistic and collectivistic level. Section VI finally discusses the micro-macro connections in question from the point of view of explanation.
In medias res , Pettit's basic argument in defense of holism is the following, where holism is characterized by the conclusion of the argument (p. 181):
1. The interactive thesis. A human being can follow a rule only on the basis of interpersonal or intertemporal interaction.
2. The commonability thesis. The rules followed by a human thinker are commonable: they are rules that another can claim as a common possession.
3. A negative claim. If a human being follows a rule on the basis only of intertemporal interaction with herself, then that rule is not commonable.
Conclusion. The rules followed by a human thinker are not followed on the basis of such intraperson interaction alone; they must be followed on a basis involving interaction with others.

What does commonability mean? Pettit says this: "So what is the condition I envisage? Briefly, that one human being can knowledgeably identify, as such, the rules followed by another and identify them as rules that she can follow herself: in particular, that one human being can knowledgeably identify the propositions and propositional elements that another targets in her thought." He thus regards the mind as scrutable. (The word 'can' above must be understood in a liberal sense.) I shall here accept Pettit's Wittgensteinian condition of commonability (but with a reservation). The thesis is defensible on philosophical grounds in that it makes communication possible in principle. However, it is clearly an idealized condition. I think that no currently existing human language (or actual human system of thought, if you prefer) fully satisfies it - and yet we humans communicate rather well. Anyway, as Pettit convincingly argues on pp. 182 ff., commonability is necessary at least for some rules and, indeed, works as claimed, serving to make interpersonal rule-following feasible.
As to Pettit's above argument, he discusses and defends it at length in the book. I tend to regard its conclusion as right for some rules of thought but I still consider it to be at least conceptually possible that there be rules for which it does not hold (e.g. as a "Friday-type", it seems that I might follow the rule for moving my finger without there having to be interaction with others). My basic doubts have to do with premiss 3 of the argument, if we take this premiss to be a thesis true on conceptual grounds. Although Pettit does not in this context put the matter so, his constitutive holism understood as a view making thought constitutively dependent on social matters (p. 114) seems to presuppose that this premiss is true for conceptual (or at any rate, noncontingent and noncausal) reasons, at least in the case of some rules. The point of Pettit's argument under discussion is to proceed from the assumption of commonability to the exclusion of the possibility of mere intrapersonal following of rules (note: above 'intertemporal' must be read 'intertemporal and intrapersonal'). As indicated, it seems that a person might be able to follow a rule merely on the basis of intrapersonal interaction the rule yet being commonable. It would then be a conceptual possibility that a person invents a full-blown human language the rules of which he follows on a merely intrapersonal (but still objective) basis which yet is commonable. (The objective basis would include things in the environment marked as referents of words, objective memory traces in the brain, and so on. To the extent interpersonal elements are needed they can be just imagined to exist by the person.)
If indeed premiss 3 cannot be regarded as a conceptual truth (although I do not claim to really have shown this), what should we do? The commonability thesis is not perhaps by itself enough for Pettit's purposes. But if it can be assumed in addition that this thesis presupposes - or can separately be argued to presuppose - the existence of society, would not that be enough for Pettit's holism? For then our thoughts would logically presuppose social relations and be in part supervenient on them.
We can also approach the issue by arguing for the conclusion of the above argument in a more direct fashion - considering the possibility that only some rules need to be communable and indeed social in order for a language to be regarded as social. Thus the claim here is that at least some rules are such that they can only be followed on a basis involving interaction with others (Pettit in fact accepts this view on p. 182). This is a thesis which can be shown to be right on the basis of an example. The following will do: You and I jointly intend to play tennis with each other this afternoon. Having this joint intention involves following a rule which in turn conceptually and factually requires interaction between the two of us. If we jointly intend to open the window we must both follow the rule requiring us jointly to change the state of the world in which the window is not open into one in which it is open, the following of this rule presupposing that we correctly follow the rule specifying the meaning (or at least truth conditions) of the sentence of "The window is open".
How Pettit's holism relates to Wittgenstein's ideas about forms of life and to the questions of the understanding other cultures (forms of life) is a question which Pettit does not really discuss in the book. It seems that further assumptions need to be made to guarantee the possibility of understanding other cultures.

IV SOCIAL REGULARITIES

Pettit speaks of regularities in the following sense (p. 11): "When I speak of regularities, I mean lawlike as distinct from accidental regularities. I prefer to speak of regularities rather than laws, because the focus is on presumptive patterns in nature, not on the formulae in which such patterns are articulated by theories, and the word 'regularity' is the more suitable for this purpose; the word 'law' is used both of natural patterns and of theoretical formulae." Thus it seems that when Pettit speaks of "social-intentional" and "social-structural" regularities he is committed to mean by them something in part conceptual (if not linguistic), for he speaks of the "presumptive" patterns (in the above quotation) and also of the logical discontinuity of the social-structural regularities with intentional regularities. Maybe he wants to think of regularities in terms of possible worlds as some kind of functional connections between properties (which are characterized as functions from possible worlds to sets or - in the case of relations - Cartesian products of sets in those possible worlds). Whatever the precise ontic nature of regularities is - it is not so central for the present purposes - it is appropriate here to consider some examples of social-structural regularities: "Increased unemployment leads to a rise in crime", "Companies maximize expected returns", and "States decline in political influence as they fall behind in economic capacity". This kind of regularities seem to me pretty idealized (maybe they hold in some possible world but not in any current society in our actual world). The reader might want to see more refined candidates for social laws from contemporary social science. In fairness to Pettit it should be said that his examples are good enough for his arguments.
On p. 119 Pettit defines social regularities as regularities that involve social properties, and on p. 123 he defines social-structural regularities by reference to the feature that they are social regularities logically and causally discontinuous with intentional regularities. What is a social property? According to Pettit (p. 119), a property is social just in case "its realisation requires that a number of individuals evince intentional responses: they display certain attitudes or perform certain actions, at the same or different times". Pettit admits that predicates like 'goes to sleep at nightfall' accordingly express a social property when more than one person is involved. However, even if this property has collective exemplifications (and even if all of its exemplifications were collective) I find it hard to see much if anything social in such properties. On the other hand, Pettit regards 'is a judge' as expressing a social property (which it surely intuitively is) although it is clearly disqualified by the above criterion: 'is a judge' can be satisfied by one person, no matter how institutional the predicate is. Some sharpening of Pettit's criterion is accordingly called for. However, this issue does not seem central for Pettit's arguments concerning the nature of social-structural regularities. In this connection Pettit speaks of regularities connected with "office-holding", which are intuitively clear cases of structural regularities. Such regularities connected to social positions and offices can be regarded as paradigm cases of social-structural regularities.
A social-structural law is causally and logically discontinuous with intentional regularities. Let us consider logical discontinuity in more detail: "We can imagine two worlds that are identical in regard to the intentional regularities that obtain in each but which differ in regard to such (viz., social-structural) regularities. We can imagine a world that is exactly like ours in the intentional regularities that obtain there but which does not conform to the regularity linking unemployment with crime. Durkheim would not be impressed at the suggestion but the difference might simply be that in that world the weather is so good that the unemployed can spend their time on the beaches." Pettit's account involves the idea that the logical discontinuity (and lack of supervenience) is due to the different "boundary conditions" holding in these worlds; but were the same relevant boundary conditions to obtain in these worlds there would be no discontinuity - indeed, supervenience would hold. This account accordingly makes social-structural laws idealized - they must be implicitly relativized to some boundary conditions to have a chance of being true (they do not hold in all circumstances as a literal reading would suggest). Thus we know that unemployment does not increase criminality in all circumstances - but we also do not presently know precisely under which circumstances it does (and I even suspect that there is no way of specifying antecedent conditions so as to make it a true law.)
In Chapter 4 Pettit engages in a long discussion concerning the problem whether social-structural regularities "outflank" or "override" intentional regularities. I find his arguments acceptable on the whole, but there are still some peculiarities in this discussion. First, Pettit is concerned mainly with - what I would call - rules of thumb of the Durkheimian kind (e.g., that unemployment leads to increase in criminality). All talk about overriding and outflanking must obviously relate to a certain methodological situation in social science, for the social world is not inherently incoherent in the sense of admitting mutually inconsistent but true accounts. Such methodological issues accordingly do not directly concern what the social world is like (but rather how it now seems to be like to various theorists).
There are two central issues that I would like to emphasize in this connection (also Pettit considers them). The first is the issue of individualism and collectivism (in Pettit's vertical sense). Pettit discusses many features related to this problem, which basically have to do with current methodological problems -with current theories or regularities and the problems between their explanatory and other relationships concerning methodological betterness. I would say - and this is not a criticism of Pettit's basic view - that the main problems related to vertical collectivism are the following: Individualistically thinking theorists cannot make sense of talk about highly supraindividual entities and regard them as murky and spooky. This is often connected to such facts as that claims involving such talk are or seem to be untestable or at least have not been properly tested, the degree of empirical support for social laws accordingly being regarded by them as low.
The other central issue that I would like to emphasize - and this relates both to the vertical and horizontal individualism-collectivism holism debate - is the factual problem of how much the social context restricts an individual's thoughts and actions. What thus is needed is information - facts as well as true generalizations and theories - concerning how intentional agents' autonomy ("autarchy") possibly is diminished in social context. Not only macro-social (macro-economic, etc.) information is relevant here but also social-psychological information (e.g.,issues related to conformity and social influence). The more usual approach in philosophy of social science to the vertical individualism-collectivism problem is via putatively true or best-explaining collectivistic (or holist, in the standard jargon) social theories versus true "individualistic" (or "interrelationistic" or at any rate non-collectivistic) theories. When put in this way (cf. Tuomela, 1991, 1994), the debate is not about current methodological problems but concerns issues which are philosophical in a deeper sense.
It is accepted (or should be accepted) by all parties by now that social context affects individuals' behavior: an individual often behaves differently in the presence of other people or when under the influence of social factors than when alone or when not so influenced. What specific kinds of group effects (and what kinds of resulting "composition laws") are involved is a question to be solved by social science on a posteriori grounds. Of course, whether we are highly autonomous or not is of clear philosophical interest, but it is mainly a factual scientific problem to answer that question.
Accordingly, in the case of such problems as holism, reciprocity, revisability, dispensability, inevitability - discussed at length in the book - there accordingly is an important factual element. Holism in Pettit's "horizontal" sense of thoughts with logically inbuilt interaction is a philosophical issue, but various further questions of the nature of interactiveness are factual. Reciprocity concerns reciprocal causal relationship between social entities and individuals. This does depend somewhat on how the issue is conceptualized: e.g., groups as interacting individuals may have a causal effect on their members and other individuals - and vice versa - but there is not much here which is philosophically problematic: causation takes place basically at the individual level. But if groups are treated as ontic wholes with individuals as their parts then surely problems with causation arise. The revisability thesis says that structural regularities identified in social science or even in folk sociology challenge intentional psychology in the sense of pointing us towards unsuspected disturbing factors. But all parties can accept this, as Pettit emphasizes. The thesis about the dispensability of individual contributions to social life says that no individual makes indispensable contributions to social events: others would have acted instead if a certain individual would have failed. This is a factual problem - the evidence points against it (taken as a general thesis). But its truth is acceptable to all parties. The indispensability thesis that various features of social life limit the opportunities available to individual agents - they are structural constraints on feasible options - in such a way that it is inevitable that agents will act so as to sustain certain social constancies. I agree with Pettit that it is generally unsound - the issue is factual rather than philosophical.


V SUPERVENIENCE

Supervenience was concisely characterized in Section II. Consider now the following general thesis by Pettit (p. 148): "I hold that all non-physical regularities are supervenient on the actual physical regularities and other actual physical conditions. The claim is that we cannot imagine keeping the world unchanged in regard to relevant physical features, while altering the patterns that operate at chemical, biological, psychological, or structural levels. Fix those physical regularities and conditions and the other regularities will be fixed too." Those other actual physical conditions include certain boundary conditions (e.g., that certain physical constants have certain values). But this seems not to be enough - unless boundary conditions are understood in an extremely broad sense. We need to put in the supervenience basis also information about - what have been called - "composition laws". For example, in classical mechanics the three-body problem - as compared with the two-body problem - represents such a composition problem. In the case of biology physical and chemical laws by themselves do not generate life - special composition laws are needed to show under what circumstances living cells are generated (and we accordingly get in the supervenience base a chemically described basis for living cells). Analogously for the social case: a person acting alone in many cases behaves quite differently than when others (or so-and-so many others, either friends or strangers or superordinates, etc.) are present - sometimes he is more active and productive and sometimes less active (much research on this has been conducted in social psychology). A composition principle says how in a given situation change of this kind of social context affects attitudes and actions. Accordingly, I would like to suggest that supervenience should be formulated not as Pettit does but rather as follows: Given a level-structure of reality (be it, e.g., social, psychological, neural, biological, chemical, macro-physical, micro-physical), supervenience should always be claimed to hold only relative to the next lower level and also relative to what the composition principles have produced. Thus biological entities and properties would be taken to supervene on what the chemical level has produced, so to speak, relative to the chemical composition principles. Similarly macro-social entities and properties are taken to supervene on social psychological entities, properties and relations, viz. on what psychological laws in conjunction with social psychological principles have produced. Thus the social level would not supervene on merely the single-agent level and laws of intentional psychology (see my argument below).
My present suggestion also entails that it becomes problematic to defend the claim that, e.g., the social level supervenes on the physical level (and that, in general, supervenience is transitive). To vindicate this kind of supervenience we must take the entities and properties involved in the base to be those produced by the operation of the composition principles. However, were we to understand supervenience in a weak sense saying that there is no difference at the supervenient level without a difference at the base level, then the weaker view that might be suggested by Pettit's text would be right. But if the base level is assumed also to determine the objects and properties at the supervenient level - as he does assume, recall Section II - then we need to proceed in something like the above fashion.
However, although Pettit does not properly discuss the aforementioned issues related to composition laws I think his position after all can be regarded as compatible with what I have just been saying. He does speak of "boundary conditions" of laws and this suggests that what I have been arguing for above is consistent with his position. Given that, my criticism boils down to saying that a more explicit discussion of these issues would have been in place.
I shall now give my argument for the central role of the jointness-notions and we-concepts in social science. Pettit sees no central role for them, as he does not discuss them at all in his book. I could produce several different arguments, but here I will only give one argument, with the feature that Pettit seems committed to accepting it. Let us first assume (as will be seen, counterfactually) with Pettit that the social level is supervenient merely on the psychological level, consisting of intentional agents, their features, and the regularities of intentional psychology (cf. Pettit, p. 11, 148-153). Pettit says that structural social regularities supervene on intentional regularities, given certain boundary conditions (about which he says very little: are they or can they be social-structural?). Consider the idealised example - mentioned by Pettit - that companies maximize expected returns. For convenience, we here discuss the following simple reformulation of it: For all x and y, if x is a company and y is a "business situation", then as a matter of factual necessity x acts so as to maximize its expected returns. We start by noticing that a company can only act if some of its office-holders act. It can have a many-layered decision structure. But let us here concentrate on the simplest case in which there is some kind of board of decision makers who jointly make decisions for the company, to be carried out by the rest of the personnel. Those decisions in central cases concern issues that relate to the company's maximization of its expected returns. If the regularity envisaged is true, such decisions, when carried out, will indeed bring about the maximization of expected returns. What this entails is that the joint (and other) actions of some or all members of the social collective (viz. company) in question is both necessary and sufficient for bringing about the company's action. It is important that all this activity takes place in the right institutional context or "boundary conditions". But in any case, given those boundary conditions, the company's action is supervenient on its (relevant) members' joint actions and joint intentions, intentional joint action obviously being based on the latter. Suppose now that there had been no relevant joint plan-making and joint action (or other "we-involving" actions and attitudes). Then we might get group action but we do not yet get intentional group action (intentional action by the group "as a whole"). The source of the intentionality of the group's action lies in the members' joint intentions. Our present argument thus indicates that intentional group phenomena can be supervenient on members' actions (and relevant psychological states) only if the latter include relevant joint actions and joint intentions, expressible by a conatively used sentence "We will do ...". (For a more precise argument see Tuomela, 1989, and 1994.) The upshot here is that although Pettit's theory does not explicitly discuss "jointness-notions" he needs them for his programme and should indeed have given them a prominent place.


VI INDIVIDUALISM AND EXPLANATION

Let us now consider matters of explanation, especially social-structural explanation. Pettit's (and Jackson's) "program" model of explanation, assumed in the book, can be summarized as follows (p. 37):
A higher-order property programs for ("is causally relevant to) a certain result - say, the occurrence of an event E - just in case three conditions are fulfilled.
1. Any instantiation of the higher-order property non-causally involves the instantiation of certain properties - maybe these, maybe those - at a lower order.
2. The lower-order properties associated with instantiations of the higher-order, or at least most of them, are such as generally to produce an E-type event in the given circumstances.
3. The lower-order properties associated with the actual instantiation of the higher-order property do in fact produce E.

To illustrate the model, consider the physical example of a flask breaking when heated. "The molecular explanation of why the closed flask breaks directs us to a specific process involving a specific molecule and helps us to contrast what actually happens with other ways the flask might have broken. The explanation that mentions the boiling of the water contained in the flask abstracts from that detail and directs us to constancies that bind the actual world to more and more other possible worlds; it reveals that in practically all worlds where the water boils, however they differ at the more specific level, the flask cracks." (p. 256)
What happens in the social case? Pettit reconciliates social-structural and intentional explanation by means of the assumption of local causality: a higher-level factor is always causally relevant in virtue of the relevance of a lower-level factor (causing a same-level event), and in this sense a causally remote factor is always relevant in virtue of the relevance of a more proximate factor. (Note that the assumption of supervenience actually does the same job.) Thus the obtaining or occurring of a higher order factor would have to be reflected at the lower level. However, Pettit argues that we should not go down all the way. Thus, given the right boundary conditions, we should explain the rise in criminality by means of rise in unemployment. We should not explain it in terms of changes in certain bunches of hadrons and leptons. Why not? Pettit's answer is that our higher-level explanans is more informative than a lower-level explanans. It gives relevant comparative and - in view of the assumption of local causality - it also gives contrastive information (viz. the local cause that actually serves to make the difference). Comparative information is information which directs us to features that belong in common to the actual world and to more and more other possible worlds (p. 256).
What is new in this program model as compared with previous accounts and, especially, with the covering law idea of singular explanation? It is mainly the idea that there is a hierarchy or level-structure of properties involved in all cases of explanation. I am not sure we always need to take into account such a hierarchy - Pettit might accept that the explanatory property be at the same level as E (think, for instance, of the movement of a person's hand causing a knob to turn) and that there thus would be a standard, intra-level covering law. In general, the explanatory property need only be causally relevant to the event E: there need not be a strict (or even less strict) causal law - a covering law - involved. Thus in the case of the regularity connecting unemployment with criminality there is no inter-level causal law connecting unemployment with agents' thefts, not even when the climate is like that in Finland (recall Section IV). Thus - or so it seems at least - what we usually call causally relevant circumstances or initial conditions can qualify as explanatory factors in this model. But at least sometimes there is a causal law involved: for instance, in the case of psychology a certain want-belief constellation (say C) can be regarded as causally connected to action (say E) - although in virtue of a lower-level neural state. (This would presumably be a law at the psychological level - macro-level in this context - rather than at the physiological level, for it is brain-states as instantiating C that cause movements as instantiating E in the present account.)
When there is a causal law in the above sense the comparative information in question just concerns the type-type regularity connecting C and E; in other cases C might just be some kind of causally relevant circumstance. The point I want to make concerning the case of singular explanation is that comparative information in Pettit's sense is the information that type-type connections contain, while contrastive information concerns singular events and processses. Accordingly, we are here basically discussing singular explanation as contrasted with the explanation of regularities. However, a higher-order factor C might be explanatorily relevant to a lower level regularity connecting a factor C* and E, where C* is a property (such as the relevant neural state-type) corresponding to, and being involved because of, C (in part due to supervenience). However, mainly because of the problem of multiple realizability, there need not always be such a property C* although there are singular lower-level instantiations of C.
According to Pettit, explaining a singular event is providing causal information about its history or giving information on causally relevant properties (p. 255); and - as clause 3 of the definition of the program model makes clear - the causally relevant properties must also be operative, not only relevant at the type-level. But it seems that this account, in conjunction with the principle of local causality, would still drive one down to the micro-physical level, for that would always give the finest-grain information available. Pettit now argues that we should take into account the comparative information available in the situation. However, this not unproblematic, for it would often mean that we have to stay at the level of the higher-order programming factor C - at least for comparative information. This is because the program model and the assumption of supervenience do not entail that there is lower-level comparative information (such as C* above). On the other hand, for contrastive information we would at the same time have to go to a lower level. Pettit might say that this is what we may have to be satisfied with as far as philosophical assumptions go. But if science discovers lower-level type-type correlations (e.g. to the kind connecting C* and E lawfully), then so much the better. But what comparative information should we then use, having several possibilities?
I would like to recommend that Pettit's and Jackson's program model of explanation be systematically enriched by pragmatic, indeed question-theoretic, ideas. Thus we should make assumptions concerning what the explainee's interests are (or what his explanatory questions are) and what kind of explanatory answers he can understand (cf. Tuomela, 1980). This would help to say what the relevant and adequate pieces of information are in each context. Let me note that also Pettit suggests the need for pragmatic constraints on pp. 232-33.
Let me emphasize that in the social-structural case there are also grave problems concerning whether there on the whole are and can be true (viz. actual and not merely potential) structural explanations. As noted above, individualists tend to think that at least currently there seem to be no credible macro-social laws, viz. laws close enough to truth (so that one might have some nomological grounds for making predictions). In this sense, macro-social information is regarded by many - and not without reason - generally as second grade and the search for such social macro-laws as a waste of time and resources. This credibility problem together with a "fear" of spooky social wholes is what leads individualists rather to ask explanatory questions concerning a lower level - such as the jointness level advocated in Tuomela (1991, 1994).
Pettit discusses the "overriding" and the "outflanking" theses in connection with his account of explanation and supervenience (recall Section IV). The overriding thesis argues that structural regularities have a greater power than intentional regularities; the opponent denies it. The outflanking thesis says that the social-structural regularities outflank intentional in the sense of having been selected for in a survival-based way. With his assumption of supervenience (and the involved "causal fundamentalism") Pettit can claim that there can ultimately be no problem about "overriding" and "outflanking" explanations related to the vertical individualism dispute, because there cannot ultimately be a conflict. Given the principle of causal locality, individualism will win - at least as far as contrastive explanation goes. Because of his assumption of supervenience (entailing harmony between the various ontic levels in the world) and - or so I suppose - his emphasis on the role of comparative information he can adopt the position of "explanatory ecumenism" allowing all flowers (and hence macro-, micro-, and meso-explanations) to flourish.


VII CONCLUSION

My critical examination of Pettit's book - especially the parts of it concerned with the "common mind" - have produced some - relatively minor - internal criticisms. The most central of them are as follows: a) Pettit's argument for holism may require weakening and, as he himself notices, it attempts to prove more than the position really needs; b) the thesis of level-supervenience does not explicitly consider the problem of accounting for the operation of "composition" laws (however, this is mainly a question of leaving out of his explicit treatment something that would have deserved a detailied discussion); c) there is the problem of exposition that the various problems related to whether social or psychological explanations somehow conflict could have been presented from the beginning so that they are seen to be less like in-principle problems than the present discussion indicates; d) Pettit's ecumenical view of social structural explanation faces some problems - at least in case macro-social notions turn out not to be nomological, no true structural laws couched in their terms being capable of existing.
My main criticism against the book is external: the important "jointness-level" is missing. Surprisingly enough, the main discussion is conducted as a debate between the level of individual psychology and the social-structural level. But there are many reasons for emphasizing the importance of social-psychological notions, especially "jointness" and "we-ness". I presented a direct argument for showing that these notions are needed if Pettit's supervenience-view of the connection between the psychological and the structural level is to work. According to my thesis at least intentional group-level (or structural level) problems require jointness-notions for supervenience to hold. This critical notice is not the right place to discuss various other arguments there are for the central role of jointness-notions. However, it is consistent with what Pettit says that his psychological level contains jointness-phenomena, although he is silent on them. Thus he can consistently allow them in his supervenience basis for social phenomena and structures, and he can even consistently regard them as central in a finer account of the supervenience of the social on individualistic or - better - "interrelationistic" elements.
Even if a treatment and emphasis of the jointness- and we-ness-level - was regrettably found to be missing in Pettit's book that mainly affects the overall architecture of the book and the choice of problems regarded as important. What Pettit actually says and argues for in the book makes it an important and timely contribution to the philosophy of social science.


REFERENCES

Hargreaves Heap, S., Hollis, M., Lyons, B., Sugden, R., and Weale, A., 1992, The Theory of Choice: A Critical Guide , Blackwell, Oxford
Pettit, P., 1993, The Common Mind: An Essay on Psychology, Society, and Politics , Oxford UP, New York and Oxford
Tuomela, R., 1980, 'Explaining Explaining', Erkenntnis 15 , 211 - 243.
Tuomela, R., 1989, 'Actions by Collectives,' Philosophical Perspectives 3, Philosophy of Mind and Action Theory , 471-496
Tuomela, R., 1991, 'On Radical Conceptual Revolutions in Social Science,' Journal for General Philosophy of Science 22 , 303-320Tuomela, R., 1994, The Importance of Us: A Philosophical Study of Basic Social Notions , Stanford UP, Stanford (forthcoming)