Popper's ‘Rationality Principle’ and ‘Epistemic’ Rationality: an Attempt at Reconciliation

ABSTRACT

‘Situational Analysis’ (SA) constitutes Popper's methodological proposal for the social sciences. There is an apparent inconsistency in the fact that, on the one hand, Popper maintains that we learn by ‘trial and error-elimination’ and, on the other hand, he declares that his ‘Rationality Principle’ (RP) has nothing to do with the epistemic account of rationality implicit in his ‘evolutionary theory of knowledge and learning’. We claim that the most likely reason Popper makes such claim is that his formulation of SA is better interpreted as an extension of his view of the ‘method’ of history to the theoretical social sciences. In light of it we argue that the implementation of SA in the latter may require the adoption of an expanded ‘Rationality Principle’ (ERP) whereby scientists assume that individual actors exhibit both practical and epistemic rationality. We identify two variants of SA we denote as SA1 and SA2 according to whether they are grounded on ERP or RP respectively and discuss some approaches in the social sciences that are representative of each of them. We conclude that the adoption of ERP helps mitigating the absence of social interaction in Popper's formulation of SA.

KEYWORDS:

• Isomorphism
• learning
• methodology
• Popper
• rationality
• situational analysis

Acknowledgements

This study is largely the result of a research project I initiated during a six months visit to the Center for the History of Political Economy at Duke University which ended in early 2017. I would like to thank the Faculty and Staff of CHOPE for their help and hospitality during my visit. This manuscript has also benefited from comments and suggestions by two referees and the Editors of the journal to whom I would like to express my gratitude. Of course, any remaining errors are entirely my own.

Notes

1 Early presentations of SA and RP can be found in The Open Society and its Enemies (Popper [1943] 1966; ch. 14, especially page 97), in The Poverty of Historicism (Popper 1957, sec. 31–32), in the English translation (Popper 1985) of a French paper, and in the book titled Objective Knowledge (Popper 1972, 178–179). However, the most detailed discussion of RP appears in the chapter titled `Models, Instruments, and Truth: The Status of the Rationality Principle in the Social Scienceś included in The Myth of the Framework (Popper 1994a).

2 This stands in contrast to neoclassical economics and the narrow version of ‘rational choice theory’ which are based on a narrow notion of rationality whereby actors are assumed to maximise utility or profits.

3 This idea also appears in different parts of Popper's work. For instance, in The Open Society he notes that SA ‘is, in fact, the method of economic analysis’ (Popper [1943] 1966, 97) and in an essay titled The Logic of the Social Sciences (Popper 1992), he remarks that ‘the logical investigation of the methods of economics yields a result which can be applied to all social sciences’.

4 However, several commentators have recently criticised Popper's claim that SA is a generalisation of the methodology of neoclassical economics to the rest of the social sciences (Gorton 2006; Palacio-Vera 2019, 2020).

5 The one commentator who proposes to reformulate RP, albeit for reasons other than ours, is Wettersten (2006, ch. 10) who stresses that Popper views rationality as being problem-solving, critical, and social and proposes to unify RP and his ‘theory of knowledge and learning’ in a encompassing approach that exhibits these features. Unfortunately, space limits preclude a discussion of his proposal.

6 In this respect, Popper's use of examples from economics is illustrative. In section 3 of this last work on SA (Popper 1994a) he expounds the example of a pedestrian (named Richard) who has to circumvent some obstacles to cross a road and catch a train. Then, in section 5 he notes that the example of the social situation Richard faces ‘contains almost all the relevant elements of situational analysis in economics’ (Popper 1994a, 170) and subsequently mentions the theory of perfect competition, the pure theory of monopoly and of duopoly. However, there is a difference between the first one and the last two: while firms in a perfectly competitive industry are atomistic units who, as in Richard's example, do not formulate expectations about other firms’ behaviour because they take market prices as ‘fixed’ (i.e., they are ‘price-takers’), both a monopoly or the firms that conform a duopoly need to formulate expectations about the demand for its output, in the former case, or about future demand and the rival's behaviour in the latter case which implies that these last two theories implicitly take account of social interaction through actors’ expectations about other actors’ future behaviour.

7 In the case of ‘social interaction’ this is all the more curious since Popper underscores the importance of social institutions for the coordination of the actions of individual actors in his early works.

8 Sections 2 and 3 of this manuscript contain several excerpts originally published in Palacio-Vera (2020, sections 2 and 3), copyright © 2020 by SAGE publications, reprinted by Permission of SAGE publications, Inc.

9 As Albert (1979, 11) notes, the birth of neoclassical economics implied a problem shift in the field in the direction of decision logic which left behind the psychological foundations of Classical Political Economy. The resulting ‘logic of choice’ is the method Popper associates to the ‘complete’ rationality and information benchmark he denotes as the ‘zero method’. Yet, as Albert (1979, 13) recognises, the attempt to explain the workings of a market economy by means of this type of model ‘takes place in a cognitive, motivational, and institutional vacuum’. In other parts of his work, however, Popper recognises ‘social institutions’ as an essential ingredient in any ‘situational model’ (Popper 1994a, 167).

10 As we explain below, further requirements may need to be satisfied.

11 If the environment is ergodic then statistics calculated from past time series or cross-sectional data will be reliable estimates of the space statistics that will occur at any future date. As Davidson (1991) puts it, in such environment the ‘future is a statistical reflection of the past’.

12 This implies that the comments we made above in the context of Olson's work about the ‘ergodicity’ assumption carry over to the theory of imperfect competition.

13 This is recognised by Robert Lucas, a Nobel Laureate in Economics, when he notes: ‘Technically, I think of economics as studying decision rules that are steady states of some adaptive process, decision rules that are found to work over a range of situations and hence are no longer revised appreciably as more experience accumulates’ (Lucas 1986, 218, emphasis added).

14 As Albert (1979, 19) recognises, this is the essence of the Keynesian criticism of neoclassical economics: the co-existence at the macroeconomic level of negative feedback mechanisms (i.e., adjustments in relative prices) and powerful positive feedback mechanisms (or cumulative processes) where the latter tend to push the system away from (full-employment) equilibrium.

15 Popper refers to these feedback mechanisms in his Poverty of Historicism (Popper 1957, 89) and denotes them as the ‘Oedipus effect’ after Sophocles’ Oedipus the King tragedy. The former is Popper's main line of defence of indeterminism in the social realm. This argument has also been recently stressed by a former student of Popper: George Soros (2013).

16 Yet if the ‘bubble’ in question ends up bursting then expectations can be said to be initially ‘self-fulfilling’ before they become ‘self-defeating’ when the ‘bubble’ eventually bursts.

17 A key feature of theories that seek to capture this type of social situations is the assumption that rationality is common-knowledge in the sense that every individual actor believes that all other actors are rational, they also believe that every other actor believes that all other actors are rational, and so on and so forth.

18 An ontological implication of ‘shifting’ equilibrium models is that the environment where actors operate may not be ergodic which is the reason why ‘learning’ may not allow actors to attain equilibrium.

19 The classical discussion of ‘self-fulfilling expectations’ is precisely based on the case of bank-runs (Merton 1948).

20 This was the case in the U.S. economy before the creation of the U.S. Fed in 1913 and, especially, after the adoption of Federal Deposit Insurance in 1933 as part of Roosevelt's ‘New Deal’ public policy measures.

21 Mass media may have contributed to ordinary people's knowledge of how bank-runs set off. For instance, the scene of a bank-run in a local savings cooperative in a small town in the US in the 1930s that appears in Frank Capra's classic movie ‘It's a Wonderful Life’ (1946) may arguably be blamed for it, at least partly.

22 A discussion of recent developments in ‘modern behavioural economics’ is in Angner and Loewenstein (2012) and Camerer, Lowenstein, and Rabin (2004). In this approach the normative benchmark is often associated to subjective expected utility theory. By contrast, in Boudon's ‘cognitive’ approach the normative benchmark may not be explicit.