THE ECONOMICS OF RATIONALITY 223

events, the interpretive economist investigates social patterns that define the conceptual frameworks within which agents make their plans.

The inclusion of the agents’ intersubjective context in economic theorizing clearly distinguishes the hermeneutical-based, interpretive approach from positivist methodology. In addition, the interpretive view of rationality sheds new light on the criticism that RE theory is based on an unrealistic assumption of rationality. The argument that ‘real’ decision-makers are not rational is based on the unwarranted assumption that the RE view of rationality is justifiable. This view of rationality cannot be consistently defended when the process of justifying the agents’ beliefs includes an examination of positivist methodology. As the preceding analysis of foundational and coherence epistemology indicates, these theories can at best provide only a partial defence for the claim that agents in RE models are rational. In addition, philosophical hermeneutics solves many of the problems of justification and truth that economists face, but it cannot be applied to agents in RE models.

From an interpretive perspective, agents in economic models could be considered rational if they were more like ‘real’ economic agents. Thus, the epistemological and ontological analysis presented in this chapter complements the alternatives suggested by psychologists and others who find RE theory to be unrealistic. It is not ‘real’ agents that are irrational but, instead, the agents in RE models. Rather than indicating why real agents are not rational, the analysis of actual economic decision-making should be seen as an alternative which promises a view of rationality and of economic theory that is more justifiable than that of rational expectations.

ACKNOWLEDGEMENTS

I wish gratefully to acknowledge the valuable advice of Don Lavoie and Lynn Newbill, although I assume responsibility for any errors.

NOTES

1For a general introduction to RE models, see Attfield, Demery and Duck (1985), Havrilesky (1985), Klamer (1984), and Sheffrin (1983).

2RE theorists do not argue that their statistical approach is an attempt to compensate for imperfections such as ‘lapses in closure and completeness’. Their definition of ‘imperfect’ knowledge would therefore be somewhat different from the description given by Brodbeck (1962).

3For example, see Hogarth and Reder (1987).

4While most philosophers seem to agree that both truth and justification are required for knowledge, there are differences of opinion regarding the exact relation between these two requirements. For example, see Chisholm (1986), Cohen (1983), and Gill (1985).

224 ALTERNATIVE VIEWS OF HERMENEUTICS

5But, as Firth points out, the normative aspect is not in the sense of an ethical question to which someone would reply: ‘“Because you will be happier (or more loyal to your friends) if you believe these statements rather than those”’ (Firth 1964, p. 549).

6For a more detailed discussion on epistemic justifications see Alston (1985).

7While the justified true belief (JTB) account enjoys widespread popularity among philosophers, it has, for the last two decades, been roundly criticized for failing to account adequately for knowledge. Beginning with Gettier’s article (Gettier 1963), philosophers have constructed hypothetical examples showing that a person may hold a justified true belief, and yet it would be false to say that this person knows what they justifiably believe. Despite this weakness, I will use the JTB definition of knowledge for two reasons: first, no consensus has developed among philosophers that a superior alternative exists; and second, I do not think the Gettier-type problems affect the critique of justification offered in this essay.

8This description of truth is based on the well-known ‘correspondence theory of truth’ described later in this chapter.

9It may appear that agents could offer a different defence that would eliminate the requirement that the economist be rational. We could imagine, for example, that agents have justified their beliefs by conducting their own tests to see if any systematic errors result from following the model’s decision-rules. However, independent tests such as these do not eliminate the need for the economist to be a rational agent. Only if the economist has taken into account all relevant information about both the economy and possible testing procedures can we be sure that an enterprising agent will not find some remaining systematic error. If the economist has not eliminated the error from the model’s equations then the agents in the model are not rational. In other words, for agents in the model to be rational the economist must himself be a rational agent.

10The regress would terminate if the economist could claim that it was only important that he be justified in his belief rather than actually providing a justification. This view reflects the difference between justifying and being justified: a person’s belief could be justified even though they could not defend it themselves. However, this approach merely shifts the burden of justification. It must in principle be possible to justify the belief. The epistemic question of justification occurs precisely because we want to know that we are justified.

11For an overview to issues related to this problem see Alston (1985), Moser (1985), and Simson (1986). Although I refer to this as the problem of infinite regress, I agree with Alston that, in practice, justifications often become circular rather than regressing in a linear sequence. However, I think that referring to the justifications as circular obscures the root problem—which is avoiding the infinite regress. Circularity is but one of several unsuccessful solutions to that problem.

12It should be noted, however, that philosophers such as Annis reject the regress argument on the grounds that it ‘involves confusing the epistemological question of when a belief is justified with the meta-epistemological question of how we know or can justify a theory of justification’ (Annis 1982, p. 54). I do not find this view convincing. If the only restraint on what justifies beliefs is the existence of some theory of justification, arbitrary theories could be concocted to justify almost any belief. Such a result is clearly at odds with the purpose of deriving epistemological theories of justification.

THE ECONOMICS OF RATIONALITY 225

13Recent advocates include Firth (1964), Chisholm (1980), Pollock (1974), Alston (1976c), and Van Cleve (1979). Also see Stiffler (1984) and references for additional approaches to foundationalism.

With few exceptions philosophers have rejected scepticism and instead have tried to address the regress problem directly. There are three generally acknowledged responses to the regress problem. First, it might be possible to accept the infinite regress of justifications but deny the charge that it precludes epistemic justification. Second, there may exist ‘foundational’ or ‘basic’ beliefs which do not receive epistemic justification from other beliefs but are nevertheless justifiable and are the basis on which all other beliefs derive their justification. Finally, the regress might not be a linear progression but instead a circular structure of beliefs or systems of beliefs that are used to justify themselves.

14The requirement that the foundation of knowledge must be certain has been associated with a ‘strong version’ of foundationalism by some philosophers. For examples of weak or ‘relativized’ foundationalism, see Annis (1977), Alston (1976a, 1976c), and Van Cleve (1979). Others who argue that basic beliefs in foundationalism can be fallible include Alston (1976a, 1976c) and Pastin (1975).

15See, for example, Chisholm (1980) and Pastin (1978).

16See, for example, Goodman (1978), Lehrer (1974), Putnam (1981), Rorty (1979), and Will (1977).

17For an overview to problems related to foundationalism and perception, see Alston (1986).

18This similarity is not surprising because, as Hornstein (1982) has observed, positivist epistemic theories are often sophisticated versions of foundationalism.

19Muth also relies on the ‘as if’ argument (Muth 1961).

20It is perhaps worth noting that the assumptions identified by Friedman are not basic beliefs because they do not confer justification onto other beliefs. Rather, their justification depends on beliefs about the model’s predictive ability which, in turn, is dependent on the givens of empirical observation and the methods of modeltesting.

21See, for example, McCloskey (1985).

22In addition, Quine (1953), Aune (1967), Harman (1973), Lehrer (1974), Sellars (1963, 1973, 1979), and Blanshard (1939) have developed different versions of the coherence theory. Other related proposals include systems epistemology and contextualism. On systems theory see Seidler (1979) and references. On contextualism see Annis (1978) and Simson (1986).

23It is important to note that the coherence standard involves more than a logical consistency between beliefs. Even foundationalist theories are ‘coherent’ in this sense. The difference between the two theories depends less on the absence or presence of a ‘coherence’ criterion and more on whether or not the standard is based on the coherence of the entire set of beliefs.

24The roles of steady states and ordered wholeness are described in Laszlo (1972). In his systems theory of cognition Laszlo refers to the concept of ordered wholeness and to an adaptive process that is similar to the one described by Lucas (Laszlo 1972, p. 120).

25Other philosophers consider this a straw-man argument because all the leading advocates of coherence agree that some input from the world is necessary. See Brandt (1985).

226 ALTERNATIVE VIEWS OF HERMENEUTICS

26 In defence of his coherence theory Lehrer also allows for inputs which are ‘immediate and noninferential’ but he states that these beliefs must be justified because of their coherence with ‘acceptance systems’. Leherer’s theory of justification seems to include some beliefs that can be ‘immediately justified’ (Lehrer 1986, p. 21). However, his statements about the relationship between immediately justified beliefs and the system of beliefs appear to be inconsistent. While Lehrer explains that the coherence theory is compatible with the claim ‘that there are some beliefs that are justified in themselves without argument’, he also states that ‘even immediate knowledge depends on coherence’ (Lehrer 1986, pp. 21 and 22).

27The indeterminacy problem is independent of the first objection that the coherence standard does not allow for empirical inputs. Since the standard refers to internal coherence with all beliefs, allowing inputs would not prevent some systems from being equally coherent while remaining inconsistent with each other.

28Another alternative suggested by Moser is to replace ‘positive’ criteria such as those listed in the text with ‘negative’ criteria which determine whether a belief will reduce a system’s coherence (Moser 1984).

29However, at least one advocate of the coherence approach claims that coherence is universally recognizable as being the standard of ‘rationality’. Larmore states:

Coherence, then, underlies what I described as the contextualist justification of epistemological and theoretical standards…. [R]ationality enjoins that we strive for overall and mutual coherence, and thus any new standard must cohere with our more significant and steadfast beliefs… This norm of coherence constitutes a sense of rationality that, I again wish to insist, has been universally shared.

(Larmore 1986, p. 158, my italics)

Unfortunately, Larmore does not explicitly define this norm nor does he provide convincing evidence that such a norm is universal.

30On the need to allow for inconsistent beliefs in coherence theories see Foley (1979).

31Rescher identifies these epistemic beliefs as ‘principles that are no less crucial to the make-up of the system than the theses it accommodates’ (Rescher 1979, p. 5). BonJour refers to them as ‘laws and principles which underlie the various subsystems of beliefs and provide a significant degree of inferential connection between [beliefs]’ (BonJour 1985, p. 97).

32In contrast to Rescher’s attempt to justify the coherence standard, BonJour claims that even if the coherence standard is ambiguous this cannot be counted as an argument in favour of foundationalism. If the meaning of coherence is vague, then this creates a problem for ‘any theory which makes any substantial use of the concept of coherence, and this arguably includes all theories of knowledge which are even minimally adequate’ (BonJour 1985, p. 154). This admission weakens the ability of foundationalists to use the ambiguity objections against the coherence theory but it does not address the question of epistemic regress. A sceptic could easily interpret BonJour’s admission to be an argument for rejecting all theories of knowledge.

THE ECONOMICS OF RATIONALITY 227

33 Justification of coherence criteria is further complicated by the claim that coherence must involve not only a judgement about the current state of the system, it must also reflect a degree of long-run coherence. BonJour claims that:

the force of a coherentist justification depends ultimately on the fact that the system of beliefs in question is not only coherent at a moment (as a result which could be achieved by arbitrary fiat), but remains coherent in the long run. It is only such long-run coherence which provides any compelling reason for thinking that the beliefs of the system are likely to be true.

(BonJour 1985, p. 135)

While BonJour’s point is well-taken it appears to raise difficulties because the meaning of ‘the long run’ must be specified. Is it too much to expect that our cognitive system remain stable for 10 years, 10 months, or 10 days? However these questions are resolved, the standard for judging whether there is an ‘adequate’ degree of temporal stability in the cognitive system would itself have to be justified, and this would lead to another regress of justifications.

34 Even if the coherence standard could be justified by appealing to a ‘supracoherence’ standard that remained within the frame of a cognitive system, this would not solve the regress problem because the supra-coherence standard must also be justified. See Airaksinen (1982) for a more detailed discussion of the regress problem in coherence theories.

35Lemos is equally critical of BonJour’s definition of coherentism because it leads to an infinite regress. Lemos notes that ‘BonJour’s argument, if directed towards the coherence theory, would require a reason, or in his view a justified belief, that any belief cohering with certain other beliefs is likely to be true; and this belief would itself require a justified belief as its source of justification and so on and so on’ (Lemos 1982, p. 308). An example of this sequence is provided by BonJour in his defence of an ‘observational requirement’. After identifying this requirement as an epistemic belief, he states that the justification for it is that it contributes ‘to the overall interconnectedness and thus coherence of the system’ (BonJour 1985, p. 243). The observational requirement is an epistemic belief that is justified by another epistemic belief, namely that of overall interconnectedness. As Lemos notes, this does not solve the regress problem because the new epistemic belief must also be justified. Goldman makes a similar argument against Williams’ use of a ‘second order’ concept (Goldman 1981, p. 208).

36This criticism also applies to Lehrer’s suggestion that inputs to coherent systems be evaluated on the basis of an acceptance system that is ‘an integrated system of probabilities’ (Lehrer 1986, p. 20).

37An example of the circularity objection is found in Alston (1986, pp. 25–6).

38In a departure from his earlier position, Rescher has more recently suggested that ‘if one is prepared to consider coherence in an idealized perspective—as optimal coherence with a perfected data base, rather than as a matter of apparent coherence with the imperfect data we actually have in hand—then an essential link between truth and coherence emerges’ (Rescher 1985, p. 796). However, this is not a

228 ALTERNATIVE VIEWS OF HERMENEUTICS

practical alternative to the correspondence theory of truth. The internal coherence standard cannot provide an idealized perspective for deciding whether optimal coherence has been reached. To determine whether such an optimum has been reached we would have to jettison the cognitive system and then, from a position external to the system, compare it to ‘the world’.

39See Lehe (1983) on the internal nature of the pragmatic standard.

40Numerous philosophers have made similar references to spontaneous inputs that serve as the terminal point for justifications. Bruzina, for example, states that Husserl insisted ‘that for rational statement, only the moment of spontaneous assent in view exclusively of unqualified evidentness, in accord with intentional orientations, can serve, or else it is not an instance of knowing’ (Bruzina 1981, p. 365). The concept of groundless beliefs as developed by Wittgenstein and others, such as Malcolm (1977) and Nielsen (1981), also appear to be similar. More recently, Lehrer has noted that ‘our judgements of reasonableness are implicitly relative to some assumptions and system that we unreflectively take for granted’ (Lehrer 1986, p. 9). A common element in each case is the implicit appeal to beliefs that provide for us the meaning of a coherence-type test.

41Because some of the inputs are epistemic beliefs, it may appear that hermeneutics is a foundationalist theory. There is, however, an important difference between ‘foundationalism’ and ‘foundations’. Krausz observes that foundationalism ‘is a theory or theories presuming privileged access’ while foundations ‘concern the ways in which we settle cognitive claims about the world(s) into which we inquire’. He notes that ‘[o]pposition to foundationalism does not rule out the consideration of foundations. Indeed, no systematic inquiry can be without foundations’ (Krausz 1984, p. 398). Epistemic inputs can be challenged like other beliefs. Based on the distinction made by Krausz, epistemic inputs can provide part of the foundation for justification without requiring that the theory be foundational.

42Larmore has criticized Gadamer’s hermeneutical analysis because it fails to make proper use of a pragmatic account of knowledge similar to Rescher’s (Larmore 1986). However, in a passage from Truth and Method, Gadamer explicitly adopts the pragmatic view Larmore accuses him of failing to appreciate.

The fact that experience is valid, so long as it is not contradicted by new experiences (ubi non reperitur instantia contradictoria), is clearly characteristic of the general nature of experience, no matter whether we are dealing with its scientific form, in the modern experiment, or with the experience of daily life that men have always had.

(Gadamer 1975, p. 314)

43However, he is careful to distinguish objects and subjects from language. Gadamer notes that it would be erroneous to think that ‘everything’ is only language and a language event. See Gadamer (1975, p. xxii) and Hekman (1986, p. 112).

44Rorty notes: ‘We do indeed need to give up the notion of “data and interpretation” with its suggestion that if we could get to the real data, unpolluted by our choice of language, we should be “grounding” rational choice’ (Rorty 1979, p. 325).

45Other philosophers have taken a similar position. For example, Wittgenstein rejected a ‘correspondence’ theory of language in favour of a view that language