Chapter 8
Rationality
Overview: In this chapter I examine and critically assess two of the core assumptions of rational choice theory; that individuals are rational and self-interested. It identifies two ways in which rationality might be defined and defended. In the first a rational person is someone who's preference-ordering over bundles of goods and services is reflexive, complete, transitive and continuous. In the second a rational person is someone, who possesses optimal beliefs and acts in optimal ways given those beliefs and desires. Each of these definitions is critically appraised. The first involves making controversial claims about the content and structure of individual preferences and ties rational choice theory to an 'instrumentalist' conception of social science. The second is both descriptively inaccurate - people do not always hold optimal beliefs and do not always act in optimal ways - and theoretically self-defeating. It is self-defeating because if people do indeed act optimally then the costs of acting rationally are likely to be such that they will often be led to act in sub-optimal ways. By identifying those instances in which individuals will find it in their interests to act in optimal ways we can, however, pinpoint those circumstances in which rational choice explanations will prove effective. The result is a qualified defence of the assumption of rationality. In the final part of the chapter I go on to discuss the assumption of self-interest and, in doing so, extend this argument.
Introduction
In the opening chapter I stated that rational choice theorists employ an instrumental conception of rationality in which actions are judged as being rational to the extent that they constitute the best way of achieving some goal. I have subsequently said very little about the nature or status of this assumption; preferring, instead, to show how it has been 'cashed-out' in practice. But rationality is a controversial assumption to make in so far as most political scientists would argue that individuals operate with, at most, a 'bounded' rationality (see Box 1.4). The assumption of rationality is one which must not only be stated, but justified.
181
182   Rational Choice
Rationality 183
The most important claim I make in this chapter is that there a re tw0 very different ways in which an instrumental conception of rationality might be understood. The first, the 'axiomatic' approach, finds its roots in the psychology of behaviourism (Box 8.1) and the practice of neo-classical economics. It defines rationality in terms of a person's
Box 8.1   Psychological behaviourism
In the introductory chapter I noted the pioneering role, in the postwar years, of 'behaviourists' in responding to the challenge of developing the theoretical foundations of political science. Their work was inspired the arguments of behavioural psychologists like John'Watson (1914 1928) and-B.F. Skinner (1953, 1972)." Behavioural psychology Watson and Skinner included, may be understood as making three pi : cipat claims.
Firstly, that we cannot directly observe mental phenomena and that reference to them in explanations of individual behaviour must therefore be entirely inferential, non-falsifiable and unscientific. Behaviour.il psychologists do not deny the ontological reality of mental states such as fear and expectation; they do, however, maintain that such states have explanatory force. Secondly, that what Skinner calls 'mentalistic' explain tions - i.e. explanations whichjnvoke the existence of such mental srates -. are not only unscientific but redundant. They are redundant because they simply restate the facts of physical behaviour in a more obscure language. Behaviourists argue that mental states are displayed through and are synonymous with physical behaviour. When we say that somebody i.s ; hungry (mental state) and that they are eating (physical behaviour), 'a single set of facts [is being] described by the two statements' (Skinner, 1953:30). Thirdly, that it is possible to explain animal and human behaviour in terms of external physical stimuli, responses and behavioural reinforcements without reference being made to mental states.
Psychological behaviourism was an immensely influential theory "" within the social sciences from around the 1930s until the early 1970s. During this period, most economists came, for example, to accept the -behaviourist's argument that scientific explanations could not take a ■ mentalistic form. Yet psychological behaviourism, together with political science cousin, has now fallen upon harder times. The claim that mental states are displayed through and are synonymous with physical behaviour and that mentalistic explanations are therefore entirely redundant is now generally regarded as being misleading. Mental states. -may generate dispositions to behave in particular ways but they are not-"" the same as those dispositions. We can feel what we feel before we act upon those feelings and it is an important part of what it means to be human that we can sometimes resist the urge to behave in a particular way.
possession of a preference-ranking which satisfies certain logical criteria. The second, the 'optimizing' approach, draws more on everyday 'folk psychology'. It defines rationality in terms of a person's possession of optimal beliefs, and their selection of those actions which can best realize their desires given those beliefs. Which of these two approaches is the more defensible? Part of the answer to this question will depend upon what it is that satisfactory social science explanations are thought to involve; an issue addressed more carefully in the following chapter. Nevertheless, I argue here that the axiomatic approach, although often presented as making a minimal set of demands on what constitutes rational action, actually makes a number of controversial claims about the content and structure of individual preferences. I also argue that the axiomatic approach naturally lends itself to an 'instrumentalist' conception of social science and that this is, for rational choice theory, a potential hostage to fortune.
In the case of the optimizing approach I suggest that an unqualified assumption of rationality is self-defeating. If people are always rational in the sense of always selecting optimal actions, they will often act in sub-optimal ways. People do not acquire optimal beliefs or take optimal decisions by chance. In order to acquire optimal beliefs they must deliberate about what it is that they believe and in order to take optimal actions they must then deliberate about how they are going to act. This process is a costly one which rational people will seek to economise upon even though their doing so will routinely lead them to ^possess sub-optimal beliefs and to act in sub-optimal ways. But damning as this criticism might appear for rational choice theory; the optimising approach is not without value. For in certain situations it will be optimally rational for people to act in optimal ways and when this is the case rational choice explanations may prove successful. Part of the task rational choice theorists and their critics face is of identifying when and where the optimizing account is a defensible one.
The axiomatic approach
Individuals are called upon to and are blessed with the capacity to make choices. In our everyday lives we must choose whether to go to :Work or phone in sick, whether to save or spend, whether to buy apples or pears. In the political arena, politicians must choose whether to adopt those policies they believe will appeal to the median voter and ■chief executives must choose whether or not to invest in the pursuit of
184   Rational Choice
some rent. The choices people make can be said not only to reveal their preferences but to be constitutive of them. If a person must make a choice between A and B and chooses A, we can say that they preferred A (Box 8.2}. If we look at a series of choices a person makes between different bundles of goods and services, their choices constitute their preference-ordering.
Assume now that individuals' preference-orderings satisfy four conditions, (i) Reflexivity: this requires that any bundle is always as good as itself, (ii) Completeness: imagine there are just three bundles-of goods, A, B and C; a person's preference-ordering is complete if thf y either prefer one bundle to another (for example A > B) or are indiffci -ent between them (for example A = B). (iii) Transitivity: a person's preference-ordering is transitive if it is consistent; consistency requires thai '■■ if a person prefers A to B and B to C that they also prefer A to C. (iv) ^ Continuity: this requires that, given any two goods in a bundle, it'will always be possible to identify another bundle which that person is ; indifferent to by either fractionally increasing the amount of one:good in the bundle or reducing the amount in another.
If and when someone's preference-ordering satisfies these conditions: it can be represented by a utility function which assigns a number to each possible bundle of goods such that for any pair of bundles, A and B, when A is preferred to B, the utility associated with A is higher tha n ; that of B. In such cases it will be 'as if the individual, in making their choices, judged different bundles according to the utility they generated and always chose that bundle which maximized their utility ; (Hargreaves-Heap era/., 1992: 6-7).
We have so far talked about preferences, utility and choices. "What, though, of the subject of this chapter, rationality? The link here is quite : a simple one to make. A person is rational if they are instrumentally ; rational, they are instrumentally rational if they have a preference : ordering which is reflexive, complete, transitive and continuous, and if they have such a preference-ordering their rationality will manifestj itself in utility maximization.
There are many practical advantages for rational choice theorists in using the axiomatic approach. Most importantly, and as I will now go On to; argue, it allows them to sidestep potential objections to the; assumption of instrumental rationality and so to get on with the business of formulating theories:
Rationality 185
Box 8.2   Revealed preference theory
Say we want i.o know whether a person prefers apples to pears. We couid,sjmp{y ask them .what theyprefer, .but..thisis,notalways.a reliable strategy to pursue. At times people have an incentive to mislead others about what it is that they want. At other times people do not have sufficient reason to^^ wonld want iffie^^^^
are aware, the way in which people say they are going to vote in an election: cannot-ate
will actually vote. The alternative way forward here is therefore to s'lbotcat people?s=behaviOur as ^is^eveiile^hy^heit-'aetHaf'cho'iceSi-'-If-vre
want to know whether a person prefers apples to pears, we ought to
give that person a choice between apples and pears' to see which one -. they choose. If they choose apples rather than pears then they can be
described'as having 'demonstrated" arevealed preference for apples
over pears......
Revealed preference theory, which is sometimes invoked by rational , choice theorists, (Buchanan,,,1979; Buchanan and Hillock,,1962), was developed by an economist, Paul Samuclson, in the 1940s. Once applied
to eliminate all references to such terms as utility, indifference and pref--«er from their
analyses of market exchanges. This was significant because psychoiogi-.....sealhehayiouristsiarguedithatireli^
scientific.
The problem with revealed preference theory is that our behaviour (choices) does not always reveal an accurate or complete account of our
......preferences (1 lausman, 1992: 39-22). Consider the prisoner's dilemma
game first introduced in Chapter 5. In a one-shot game, instrumentally rational actors will 'defect' rather than 'cooperate' even though both are worse-off by doing so. If we want to apply game theory by saying that a group of particular people have found themselves caught in a ^prisoners? dilemma^we^ needs toT^e^abfc
ence-ranking in which each most prefers an outcome in which they are
rtheonly person ^ are the only person not to do so. But if we accept the basic tenets of s. revealed'preferene&tibeory: thenthe;^mytebehaviouriweRwill;see5and,be able to infer a preference from is people's decision to defect. But this does not tell us very much. In particular, it does not tell us that they defected because they were caught in a prisoner's dilemma in which taking the individually rational course of action nevertheless led to a . coUectiveiy sub-optimaloutcome. .....
1 In. order to justify the assumption of rationality within any particular theory, whether it is of party competition or budget acquisition*
186   Rational Choice
Rationality 187
it might be thought that rational choice theorists would have to establish that actors really do have reflexive, complete, transitive and continuous preference-orderings. Yet this is not the case Rational choice theorists can instead follow economists in maintaining that the assumption of rationality is axiomatic in the sense of having 'only to be stated to be recognized as obvious' (Robhins 1935: 78). Rationality, far from being a controversial assumption requiring a great deal of carefully justification, is entirely self-evident and in no need of further discussion.
2 Critics of rational choice theory routinely argue that people are not self-interested. Yet once it assumed that people have reflexive complete, transitive and continuous preference-orderings, their actions can be analysed and understood without having to make any assumptions about whether they are self-interested. Rational choice theorists need to assume that people will consistently choose one bundle of goods over another. They do not, however, need to make any assumptions about why people prefer one bundle of goods to another. The reason why people prefer one bundle of goods to another is, in a sense, entirely irrelevant to the practice of rational choice theory and so arguments about self-interest are entirely misplaced.
3 Once it is assumed that people are rational in the sense of having a preference-ordering which is reflexive, complete, transitive and continuous, rational choice theorists can analyse actors' behaviour on the assumption that it reveals their preferences. Rational choice theorists using the axiomatic approach do not therefore have to 'peer' inside the minds of the people whose actions they are trying to account for in order to discern their beliefs and desires. For a behaviourist this is important because attempts to peer inside and understand people's minds are doomed to failure. Whether or not this argument is thought plausible, the focus on behaviour is also important in so far as rational choice theorists find themselves having to
: account for the behaviour of a large number of actors. It would, if nothing else, be extremely time-consuming to have to peer inside the mit'ds of a large number of actors before conducting any research. The axiomatic approach absolves rational choice theorists from the need to do so.
4 Rational:choice theorists do not, in practice, study actors' choices in order to discern their preferences before then constructing theories;: they instead use models which make assumptions about what it is that actors prefer. Downs does not show that politicians prefer more
votes to fewer votes, he simply assumes that this is so. But how can we know whether the assumptions made in rational choice models tell us something about the preferences of actors in the 'real world'? Instrumentalists side-step this question by arguing that theories ought to be judged in terms of the accuracy of their predictions rather than the realism of their assumptions (Box 8.3). In so far as they commit themselves to such a position, rational choice theorists can thereby argue that debates about whether or not people are 'really' rational miss the methodological point. What counts is not whether people really are rational but whether rational choice theory can be used to predict outcomes and events.
In identifying the weaknesses of the axiomatic approach, I will concentrate upon the first and fourth of these arguments and postpone the discussion of self-interest until the end of the chapter. The first argument, it will be recalled, is that the attribution to people of reflexive, complete, transitive and continuous preference-orderings is simply axiomatic. Yet critics might argue that there is actually plenty of evidence to suggest that peoples' preferences do not always conform to such standards, (i) People do not always have complete preferences; they do not have complete preferences because they do not usually have preferences over goods they have not heard of, experiences they have not yet tried and ways of life they have not contemplated (Hollis, 1987:21}. (ii) People often have intransitive preferences; this is because they compare alternatives across different dimensions. Imagine a person who faces a choice between voting for parties A, B and C; they compare A and B in terms of their policies and prefer A, compare B and Gin terms of their leaders and prefer B, and compare C and A in terms of their probity and prefer C. Yet the result is a intransitive preference ranking of A > B, B > C and C > A. (iii) People do not always act to maximize their utility; they instead sometimes act out of a sense of duty or obligation (Sen, 1977,2002). Imagine that we face a choice between visiting an uninteresting relative and going to the beach. When describing our dilemma to someone else we might say that we would prefer to go to the beach but that we will visit our relative because we think this is what we should do. Yet if people are instrumentally rational in the way stipulated here, such behaviour cannot make any sense.
The fourth argument was that the assumption of rationality lends itself to an instrumental conception of science and that this reduces the pressure on rational choice theorists to justify the assumption of rationality. The first problem here is that many scientists, of both the
188   Rational Choice
Rationality 189
Box 8.3   Instrumentalism <:
Instrumentalism is the name given to the view that scientific theories do not make claims about how the world is which should be assess judged as being literallytrue or false. Instrumentalists argue that theories Ought to be viewed as instruments; that is as tools which we can use to understand the world (Rosenberg, 2000: 93-6). For this reason they should be judged in terms of the rigour and accuracy of their predictions -rather than the realism of their assumptions. Within the natural sciences, ^ instrumentalism is a venerable methodological1 tradition, bur withm the social sciences the instrumentalist position was not clearly artu    .. .; until the early-1950s. Milton "Friedman's" (1953) essay or-   ; ■. Methodology of Positive Economics' is, as Daniel Hausman (1992: 162) :i observes, 'by far the most influential methodological statement (within economics] of the century [and] the only essay on methodology that a ■■ i large number, perhaps even "a majority, of economists have evei :■!' This is significant because Friedman actually articulates and defends n particularly extreme version of instrumentalism.
Friedman starts his essay by asserting that''the only relevant icm of the validity of a hypothesis is comparison of its predictions with experience' (pp. 8-9). Only when two theories have equally good predictive records is it appropriate to compare them in terms of other criteria such as 'simplicity' or 'fruitfufness'. Turning to the question of whether it is ever appropriate to judge a theory in terms of the 'realism' or aci. i1 i-.ilv of its assumptions, Friedman argues that it is not. Indeed at one point he seems to suggest that significant theories, by which he appears to mean " -
those that generate non-obvious predictions, 'must be descriptively false in [theirj assumptions* (p. 14). In defending this view he relies upon the following example (pp. 19-20). Consider the position of a scientist trying to explain the density of leaves around a tree. Imagine he proceeds 1 by assuming that the leaves are positioned as if they had deliberately sought to maximize the amount of sunlight they received, as if they knew
" the physical laws determining the amount of sttnlightthey would receive in various positions, and as if they were capable of instantly and effort-lessly moving from one place to another. These assumptions can be used to construct a model capa ble of. accurately predicting how the leaves will fall around the tree. Does it matter that the assumptions are false?
■. Friedman argues that it does not. All that counts is prediction and so the fact that the leaves do not have the properties attributed to them is not
■■,'vitally relevant'(p..20),
Beyond these headline quotes, Friedman's argument is actually a
. great deal more subtle. At one point he accepts that scientists are entitled to judge whether a theory is likely to make accurate predictions by
. assessing the realism of its assumptions prior to formal testing.
. Elsewhere, he accepts that theories which have more realistic, assump-
. tions are more likely to. successfully predict a wider variety of phenomena. As far as I can see, the implication of these claims is that the realism
. of assumptions does matter in important ways. Yet such qualifications tend to get ignored. As he has been interpreted by several generations of'
. economists, Friedman demonstrates that theories, ought to be judged in terms of the accuracy of their predictions and not the realism of their assumptions, the assumption of rationality included.
natural and social variety, argue that we ought to be more intereste4(fe^:tl^ that rational choice theory has often been used to generate successful
explaining why something happened than predicting what;-#ti3.ii^i^^^ predictions (see De Mesquita, 2004). Critics might also argue that
happen. Successful prediction may sometimes be useful in so far as it Green and Shapiro do not understand the nature and requirements of
reassures us that our explanations are correct, but we should not successful prediction. Social scientists ought not to be in the business
assume that a model which successfully predicts some phenomena °£ making 'point' predictions about specific events but rather predic-
must therefore also constitute a successful explanation of it. ClicHeia^^^^ tions about what will happen if the world changes in particular ways
it may be, it is nevertheless true that correlation does not always mean (see Samuelson, 1972). Political scientists ought not to be predicting
causation. who will win the next election, but what will happen to support for
The second problem is that critics like Green and Shapiro {19Mr::i§^S^l tue incumbent government //there is an economic recession. I do not
6) maintain that tests of rational choice theory 'have either failed.^n^i^^i. want to reach any final judgements about Green and Shapiro's argu-
their own terms or garnered theoretical support for propositi(jrt^^^^^: merit here. The point I want to make is simply that //' Green and
that,1 on reflection, can only be characterised as banal'. Green and Shapiro are correct in arguing that rational choice theory has an
Shapiro's argument is, as I emphasized in the opening chapter, a unimpressive predictive record, instrumentalism becomes a rather
controversial one. Critics argue that they base their conclusions-oh a /unfortunate peg on which to hang a defence of the assumption of
selective and often dated review of the literature (Fiorina, 1996) and rationality.
190   Rational Choice
Rationality 191
The optimizing approach
Philosophers use the term folk psychology to refer to the conceptual scheme by which we predict and explain peoples' actions on the basis of the beliefs and desires we attribute to them (Cottrell, |99.5: 1.61) Folk psychology assumes that people are rational in the sense that they have reasons to believe what they believe; reasons to act in the way that they act given their beliefs and desires, and that their beliefs and desires actually cause them to act in the ways that they act through the creation of an intention (see Davidson, 1980). Now there is obviously more that could be said here about what ought to be counted as a reason and about the way in which reasons can cause actions. But at this point I simply want to observe that rationality, when defined in this way, might not seem so unreasonable an assumption to make; Most of us proceed in our day-to-day lives by assuming that people are rational and by successfully using this assumption to make predictions about how people will act (Dennett, 2002). This is not to say that people are always rational. People sometimes deceive themselves into believing something they know to be false but which they want to be true (Finagrette, 1989). At other times they succumb to arkasia or weakness:*' of the will (Elster, 2000). As anyone who has tried to quit smoking will know, having a reason to act in a particular way and wanting to act in that way does not necessarily guarantee acting in that way. But it is not difficult to imagine how behaviour of this sort might be presented as the pathological exception to a general rule of rationality.
Rational choice theorists do not, however, simply define rationality in this way. They equate rationality not simply with reason but with optimally (Elster, 1985,1986). When rational choice theorisis assume that people are rational they are not simply assuming that people have reasons to believe what they believe. They are assuming that their beliefs are the best possible beliefs they could have given the information available and that this is the reason why they believe what they believe. In a similar way, when rational choice theorists assume that people are rational they are not simply assuming that they have reasons to act in the ways that they act. They are assuming that their actions were the best possible actions they could have taken given their beliefs and desires.
v Understood in this way, rationality is a term we can use to describe particular beliefs and actions or, more generally, the people'whose* beliefs and actions they are. Rationality is, in a sense, an output. But this begs the following question. How is it people can come lo act
optimally? One possible answer to this question is provided by Amartya Sen (2002) and John Searle (2001). They argue that rationality is best understood not as an output but as a process. People, they argue, ought to be described as rational to the extent that they deliberate or reason about what it is that they ought to believe and how it is that they ought to act:
In the normal case of rational action, we have to presuppose that : the antecedent set of beliefs and desires is not causally sufficient to : determine the action. This is a presupposition of the process of : deliberation and is absolutely indispensable for the application of
■ rationality. We presuppose that there is a 'gap' between the 'causes' of the action in the form of the beliefs and desires and the 'effect' in
: the form of the action. This gap has a traditional name. It is called ::;  'freedom of the will'. In order to engage in rational decision-■-; making we have to presuppose free will. Indeed ... we have to presuppose free will in any rational activity whatever ... to see this
■ point you need only consider cases where there is no gap, where the belief and the desire really are causally sufficient. This is the case, for example, where the drug addict has an overpowering urge to take heroin; so, compulsively, he takes it. In such a case the belief and the desire are sufficient to determine the action, because the
■ addict cannot help himself. But that is hardly the model of ratio-nality. Such cases seem to be outside the scope of rationality.
.. (Searle, 2001: 13-14)
Searle and Sen present this distinctive account of what rationality entails because they want to show how people can acquire 'desire-independent reasons' for acting out of a sense of duty or obligation. I want to use their argument in a different way. It is their capacity for deliberative rationality which allows people to act in optimal ways. But people often regard the exercise of this capacity as costly in terms of both time, effort and, in the case of difficult decisions, mental anguish. As economists would put it, the exercise of deliberative rationality has a positive opportunity cost. Time spent deliberating about some belief is time that cannot be spent in other ways. Time spent deliberating about one issue is time that cannot be spent deliberating about another issue. The claim that deliberation is costly is certainly not true of all the decisions we deliberate about. There are some things we enjoy deliberating about and :some people would regard it as being a huge and perhaps unbearable cost if they did not have the opportunity to deliberate about anything.
192   Rational Choice
Rationality 193
But all that is being claimed here is that people do not enjoy deliberating about everything and that people do not want to deliberate all the time.
It is a fundamental assumption of economics that people economize on scarce resources. Given a limited amount of money, people will spend that money on the goods they most prefer. Given a limited amount of time, people will spend that time with the people they most want to spend it with. To say that people are economizers is not to say anything new. The assumption that people economize is contained within the assumption that they act in an optimal manner. But once it is recognized that the exercise of deliberative rationality is costly and that people will therefore economize upon it, we can see why people may not always act in an optimal way. People will only invest in the exercise of deliberative rationality up to that point where the marginal benefits of doing so are greater than the marginal costs. Beyond this point, people will not be acting optimally if they invest in the exercise of deliberation even if their doing so would allow them to act in a more optimal way. The problem with the claim made by rational choice theorists that people always hold optimal beliefs and always act in optimal ways is that it is self-defeating. For if people act optimally, they will sometimes act in sub-optimal ways.
In a moment I will consider the implications of this argument for the explanatory reach of rational choice theory. Before doing so I will, however, sketch the terms of a possible objection. I have argued that rational people will not always act in optimal ways because doing so will require a sub-optimal investment in deliberation. But this would seem to assume that people make optimal decisions about how much to invest in the exercise of their deliberative capacities. Yet it might be argued that it is precisely the issue of whether people make optimal decisions that is in dispute. So how might the claim that people make optimal decisions about how much to deliberate be defended? It might be argued that people make optimal decisions because they deliberate about how much they ought to deliberate, but this simply leaves us grappling with the same problem at a different level. For we then need to establish why people make optimal decisions about how much to deliberate (about how much to deliberate). If we then argue that people make optimal decisions about how much to deliberate about how much to deliberate (about how much to deliberate) we are simply creating an infinite regress.
The reach of rational choice
For many of its practitioners, a part of the appeal of rational choice theory lies in its promised universalism; its claim to be able to explain any and every aspect of political life (Ferejohn, 1991; Green and Sbapiro, 1994: 23-8). Now, in so far as rational choice theorists define instrumental rationality in terms of the axiomatic approach, this would indeed seem to entail a commitment to universalism. For if we assume that peoples' preference-orderings are reflexive, complete, transitive and continuous, and justify this assumption as being axiomatic, there are no obvious grounds for then arguing that people will act rationally at some times but not others. Yet, by contrast, it is one implication of my argument about the optimizing approach that rational choice theory cannot be used to explain any and every political action. Rational choice explanations which assume that people possess optimal beliefs and act in optimal ways given their beliefs and desires will only prove successful when an (optimally) rational person would act in (optimally) rational ways. Can we say anything about the circumstances in which this condition will be satisfied? Three very general propositions suggest themselves:
1 People will, all other things being equal, invest more time deliberat-:   ing about those issues they believe are important. Given the link
between the exercise of deliberation and the selection of optimal actions, we might therefore say, all other things being equal, that people will be more likely to act in a optimal way when the beliefs they must choose and the actions they must take have, important consequences.
2 As I have already noted, people do not always regard the exercise of their deliberative capacities as a cost. People enjoy thinking about some issues. People will invest more time deliberating about those issues they enjoy deliberating about. Given the link between the exercise of deliberation and the selection of optimal actions, we might therefore say, once again all other things being equal, that people will be more likely to act in an optimal way when deliberating about beliefs and actions they enjoy deliberating about.
3 Some decisions are more difficult to take than others either because a larger range of options have to be considered or because the relationship between options and outcomes is uncertain. So, for any given investment in the exercise of deliberation, an investment which, as we have seen, will depend upon the importance of the issue
194   Rational Choice
Rationality 195
and the costs of deliberating about it, it is more likely that people-will act in a optimal way when the choice is an easier one to make.
How does this bear upon the rational choice theories we have beeh examining? It would appear to me-that the first proposition is the least problematic for rational choice theory and the third the most problematic. The first proposition is the least problematic because in many of the cases we have been examining the choices actors must make would indeed appear to have important consequences for their welfare. It would not, for example, seem unreasonable to suggest that politicians will carefully deliberate about what electoral strategy their party should adopt or that bureaucrats will carefully deliberate about how large a budget to seek from their political sponsors. For in these cases it is fairly obvious how the selection of the right electoral or budget strategy can make a significant difference to a person's welfare. There are, however, exceptions here. It is not at all obvious why people will deliberate carefully about who to vote for or whether to contribute a small amount of money to some collective endeavour. In such cases, people may have reason to act expressively rather than instrumentally (see Box 7.2).
Turning now to the third proposition, rational choice models are of course simplifications. They present actors with unambiguous choices and clearly specify the relationship between those choices and eventual outcomes. It is not hard to see how actors in these models might succeed in maximizing their utility. In reality, political actors operate in a more complex and messy environment. They must routinely choose between a large number of courses of action and must often do so without knowing with any certainly what the pay-offs associated with each choice are. Consider, for example, Downs' model of party competition. Here, it is assumed that party leaders not only know what policies are available but what the levels of support for each of these policies are. This is not plausible. Even in an age of focus groups and private opinion polling, politicians cannot always know how a policy they are about to commit themselves to now will be regarded in the future. Politicians often lose elections not because they deliberately chose to adopt unpopular policies but because they failed to spot which the most popular policies were. To the extent that uncertainty is a pervasive and inescapable feature of political reality, the assumption that actors have the best possible beliefs and that their actions will be the best possible actions they could take given their beliefs and desires appears problematic.
What, finally, of the second proposition that people will deliberate more carefully about those issues they enjoy deliberating about? Here, there might be more comfort to be drawn for rational choice theory. For whilst different people no doubt enjoy deliberating about different things, it does seem conceivable that politicians, bureaucrats, interest-group leaders and other actors who have chosen a career in politics will have chosen that career partly because they enjoy deliberating about politics. It is also possible that those actors who have risen to positions of authority within political parties, bureaucracies and interest-groups, have done so partly because they are good at making the right decisions. Successful politicians confront as complex a political environment as you or I; they are, however, presumably better able to cope with it.
Self-interest
Having largely skirted around the subject so far, I will finish this chapter by saying something about the status of the assumption of self-interest within rational choice in general and the optimizing account of rationality in particular. In the opening chapter I presented self-interest as being one of the core assumptions of rational choice theory; yet I also noted that rational choice theorists do not need to assume self-interest. Laver and Shepsle's portfolio-allocation model, which we examined in Chapter 3, offers, in many respects, an exemplar of the potential of rational choice theory. Yet it explicitly assumes that politicians have a consummate commitment to particular policies. But it nevertheless remains the case that most rational choice theory, and most of the theories examined here, do assume that actors are self-interested. What though of the earlier argument that the axiomatic approach allows rational choice theorists to avoid having to make any
■ assumptions about whether people are self-interested? How can this be reconciled with my argument that self-interest is a core assumption of
r rational choice theory? The answer here lies in another of the arguments previously outlined. It is true that the use of the axiomatic
' approach means that rational choice theorists do not have to reach any judgements about why people prefer one bundle of goods to another. Yet, as I have also argued, rational choice theorists do not, in practice, ground their theories upon detailed examinations of actors' behaviour.
' They instead use models which make assumptions about actors' preferences and these models, by and large, assume that actors are self-interested and that their preferences reflect their self-interest.
196   Rational Choice
Rationality 197
So how might the assumption of self-interest best be defended? The most obvious argument here, and the one to which rational choice theorists do sometimes commit themselves, is that people really are self-interested (see Tullock, 1976: 5). I will consider this argument in a moment. Before doing so it is, however, worth emphasizing that there are a number of other ways in which the assumption might be defended which do not require us to maintain that people are necessarily se! I -interested:
1 Instrumentalism. As I have already noted in the case of rationality rational choice theorists can commit themselves to the instrumentalist position that theories ought to be judged in terms of the accuracy of their predictions rather than the realism of assumptions such as that of seif-interest. Having discussed this argument in the previous section, I will not say anything more about it here.
2 Protection from knaves. It might be argued that rational choice theory shows us how institutions and policies would workpeople were self-interested. Why would we want this information? Because it allows political scientists to design policies and institutions which will work if and when they are used by people a number of whom are likely to be self-interested. Rational choice theory can, in this way, be used to protect us from 'knaves' (Hume, 1741: 40) (see Pettit, 1998, for a more detailed discussion). Consider the theory of rent-seeking. We do not necessarily need to believe that every firm will ruthlessly pursue any opportunities to acquire rents in order to believe that we ought to design institutions which can minimize the incidence and costs of rent-seeking,
3 Economizing on virtue. In a similar vein it might be argued that whilst no one acts in a self-interested manner all of the time, that there is only a limited amount of altruism to go around. In order to economize on this precious resource it is therefore better, where possible, to design and rely upon institutions and policies which can reconcile the pursuit of self-interest with the achievement of the
.■..-■collective good (Brennan and Hamlin, 1995). Rational choice is a : useful theory because it allows political scientists to understand how best to economize on virtue.
4 Revealing hypocrisy. Politicians and other political actors routinely .claim that they are acting from the highest of possible motives: that
v they are driven purely by a desire to promote the public interest. No i: doubt such claims are sometimes correct. As I noted in the opening chapter, we live in a world in which not only do millions of ordinary
people volunteer to defend their country, give blood and donate money to charity, but in which politicians sometimes do go to j ail for their, beliefs. But no doubt such claims are also often self-serving cant. Rational choice theory can perform a useful function by showing how politicians' actions might be understood as an expression of their self-interest (see Weale, 1999: 103). For those on the political left who are often the most suspicious about the motives of politicians and others in political authority, rational choice may be a particularly useful analytical tool.
What, then, of the argument that people really are driven by self-interest? This does not seem particularly plausible. For in previous chapters we have seen: (i) that the outcome of coalition negotiations can be more effectively explained and predicted if it is assumed that politicians have consummate policy commitments; (ii) that people routinely act contrary to their self-interest in co-operating with each other in a one-shot prisoner's dilemma game; and (iii) that we can best account for the costs of rent-seeking by assuming that politicians have to provide a public interest 'cover' for their actions. This does not mean that the assumption of self-interest is beyond salvation. Rather than argue that everyone is self-1     interested all of the time, it might make more sense to argue, as we have already done in the case of rationality, that people are more likely to act in self-interested ways when the costs of not doing so are higher, j        Just such an argument has been offered by the philosopher Philip Pettit. Contrary to the standard assumptions of rational choice theory, he suggests that people routinely consider other people's interests when deciding how to act. But he also argues that if and when the results of their actions plunge them below their 'aspiration level' that they will then start to take more account of their own interests. Although self-interest does not always cause people to act in particular ways, it is nevertheless a 'standby cause' in the sense that it retains a 'vital pres-: ence that puts constraints on how .. . actual behaviour is likely to go' "   (1996: 275). What determines a person's aspiration level? Pettit suggests that it will usually be determined by their normative reference <        group. A person will start to take more account of their self-interest once their social and economic position is jeopardized relative to this group. To use Pettit's own example, a manual worker will not necessarily start to act in a more self-interested way simply because their position has deteriorated relative to that of a company director. They will, however, be led to consider their position if they are unable to maintain the same lifestyle as the colleagues with whom they identify.
198   Rational Choice
Rationality 199
'lo this argument I simply want to add one brief footnote relating to the impact of social norms. As I have already noted, norms tell people-how to act in particular situations and are not outcome-orientated. As I have also argued, the existence of norms cannot always be reduced to and explained in terms of people's long-term self-interest. Simply because we all benefit in the long-run from the existence of a norm requiring us to form a queue; we cannot thereby infer that the norm exists because it is in everybody's self-interest that such a norm exists? Yet once they have been established, norms do sometimes make it in people's self-interest to behave in particular ways because there are often social costs attached to breaking that norm. People do not simply join a queue because they think they ought to do so; they join a queue because they will be chastised for not doing so. This bears upon our previous argument because one set of norms in society govern the extent to which it is appropriate to act in a self-interested manner. Such norms dictate that it is, for example, not only inappropriate but reprehensible to behave in a self-interested manner with friends and family but entirely acceptable to do so when negotiating the price of a house sale. I have argued that people are more likely to act in a self-interested: manner when the costs of not doing so are higher. Yet if there exists a norm proscribing self-interested behaviour in a particular situation and if there are costs attached to violating this norm, it may be in a person's self-interest to act in another-regarding manner in order to adhere to that norm. In such cases, the higher the costs attached to violating the norm, the more likely it is that actors will, qua Pettit, find it in their interests to abide by it.
What norms are there regulating the pursuit of self-interest within the political arena? Within the kind of liberal democracies we have been examining in this book, it is, on the one hand, possible to discern a normative expectation that politicians and other actors ought to put; the public interest ahead of their own interests. Certainly politicians routinely castigate each other for failing to do precisely this and it is difficult to make sense of such attacks unless there is a norm proscribe ing such behaviour. Yet, at the same time, and as I noted in the opening: chapter, recent decades have seen the growth of a 'tabloid' political culture which maintains that politicians only care about getting elected, that bureaucrats are all lazy, and that local councillors are all corrupt. The: suspicion must be that we live in a world in which more; and more people expect political actors to behave in a purely self-interested manner, in which the costs of behaving in precisely this way are much lower, and in which the assumption of self-interest has become a
self-fulfilling prophecy. Rational choice theory is well suited to the analysis of such a world. Indeed by so zealously propagating the view that people are self-interested and that the pursuit of self-interest can sometimes be reconciled with the achievement of the public good, rational choice theory may, in some small way, have contributed to its emergence (see Box 6.3).