Boudewijn de Bruin - Popper's Conception of the Rationality Principle in the Social Sciences





Published as "Popper's Conception of the Rationality Principle in the Social Sciences." Karl Popper: A Centenary Assessment: Selected Papers from Karl Popper 2002: Volume III: Science. Ed. Ian Jarvie, Karl Milford and David Miller. Aldershot: Ashgate, 2006. 207-215.


Popper's kritizistische Methodenlehre of trial and error applies to natural and social sciences alike. As a solution to some problem P1 researchers develop a tentative theory TT1. This theory is put to severe tests that may result in its refutation by pointing out errors. Such an elimination of errors EE leads to a new problem situation P2 challenging the researchers to come up with a new tentative theory TT2, and so on and so forth. But social and natural sciences not only share the formal aspects of scientific progress captured by this "tetradic schema," they are also similar on a substantial level. A theory, roughly speaking, is a model together with a set of forces, principles, laws that make it running and, to use Popper's term, "animate" it. It is a pack of initial conditions together with universal laws.

The interrelations between models and animating forces in the social sciences are given by Popper's "logic of the situation." Roughly, to explain an action of a (social) agent the social scientist has to provide an analysis of the situation in which the agent acted. This is the model. The rationality principle animates this model by declaring that the agent acted adequately and rationally in the situation.

Boudewijn de Bruin

Boudewijn de Bruin is assistant professor in the Faculty of Philosophy of the University of Groningen. De Bruin did his undergraduate work in musical composition at Enschede, and in mathematics and philosophy at Amsterdam, Berkeley, and Harvard. He obtained his Ph.D. in philosophy from The Institute for Logic, Language, and Computation in Amsterdam. His doctoral dissertation, under supervision of Johan van Benthem and Martin Stokhof, was on the logic of game theoretic explanations. read more

Popper has explained the details of this view in a lecture at the Department of Economics at Harvard in 1963,[1] as well as to some extent in The Poverty of Historicism.[2] Some brief notes on model and animating force.

The model, that is the situation in which an agent is located, Popper characterizes by (i) the physical and social restrictions the possible actions of the agent are subject to, (ii) the goals or aims of the agent, (iii) the knowledge and information the agent possesses. The crucial task for a social scientist in giving a situational analysis is to describe and characterize this situation. For Popper this amounts to a "logical" analysis. The aim is not to give a "psychological" description of the three components of a situation by, for instance, retrieving the goals an agent was consciously aware of, but rather to lay bare by logical analysis a whole scale of (possibly unconscious) goals that are relevant for the situation and determine it objectively.[3]

The animating force, that is the rationality principle, Popper describes as "the assumption that the various persons or agents involved act adequately or appropriately—that is to say, in accordance with the situation" (Models 169). This principle has to be interpreted "logically" as well. It "does not play the role of an empirical explanatory theory" (Models 169) because social theorists should "pack [their] whole explanatory theory into an analysis of the situation—into the model" (Models 169).


An animating force that does not play an empirical role in a theory, could that possibly make sense in the light of Popper's views on falsifiability? "My views on the rationality principle," Popper writes in a section later added to the 1963 Harvard lecture, "have been closely questioned" (Models 177). His interlocutors were not satisfied with his description of the rationality principle as a non-empirical, metaphysical, or methodological principle rather than as an empirical hypothesis. What was his answer? First, modifying his original statement a little, he argues that the rationality principle has, indeed, to be viewed as empirical. He regards it "as an integral part of every, or nearly every, testable social theory" (Models 177). Second, he puts forward and defends the methodological rule that whenever social theories are falsified the rationality principle should not be blamed. It is a "sound methodological policy to decide not to make the rationality principle . . . accountable" (Models 177, emphasis his).

This, it seems to me, is a coherent answer to the question. Its plausibility, however, depends on the plausibility of the methodological rule Popper introduces. I will discuss the arguments Popper gives in its favor, but before doing that I will briefly discuss Popper's notion of methodological rule.

[1] I am very grateful to Professor David Miller for giving me access to copies of typoscripts of various versions of Popper's paper, as well as for providing me with the following historical data. "Models, Instruments, and Truth" is based on a lecture that Popper delivered at the Department of Economics of Harvard University on February 26, 1963. In 1963 and 1964 two sections were added. A small extract circulated in the London School of Economics in 1967 and 1968. This English extract was translated in French as "La rationalite et le statut du principe de rationalite," in E.M. Claassen (Ed.) (1967), Les fondements philosophiques des systemes economiques: Textes de Jacques Rueff et essais rediges en son honneur 23 aout 1966, Paris: Bibliotheque economique et politique, Payot. (The translator is unknown.) A Spanish translation of the French translation appeared a year later. A revised version of the English extract was published in D.M. Miller (Ed.) (1983), A Pocket Popper, London: Fontana. An even further revised version, then, appeared under the title of "Models, Instruments, and Truth," in M.A. Notturno (Ed.) (1994), The Myth of the Framework, London: Routledge. This is the version I will quote from.

[2] See also K.R. Popper (1957), The Poverty of Historicism, London: Routledge.

[3] The different definitions of "situation" Popper introduces are examined in M. Lagueux (2002), "Popper and the Rationality Principle," paper presented at the Karl Popper 2002 Centenary Congress, Vienna, 3 July-7 July 2002. Cf., I.C. Jarvie (1998), "Situational Logic and its Reception," Philosophy of the Social Sciences 28:365-380.

In The Logic of Scientific Discovery[4] Popper introduces a hierarchy of methodological rules. On top of this hierarchy a "supreme rule" stands guiding all other rules. Lower rules are accepted when they conform to higher rules. The hierarchy is not a deductive system in that methodological rules are not deduced from other rules. Furthermore, they do not regulate the transformation of linguistic formulae as the rules of logic do. They are rather conventions comparable with the constitutive rules of chess. This means that accepting or rejecting a methodological rule is a human decision.

The supreme rule that lies on top of Popper's system is prompted by the demarcation principle which identifies empirical with testable, and hence with falsifiable. The supreme rule says that the other methodological rules "must be designed in such a way that they do not protect any statement in science against falsification" (Logic 11). In other words, a methodological rule has to contribute to the testability of theories.

What to think of the rule not to give up the rationality principle? Immunizing the rationality principle against falsification, does that really add to the testability of social theories? Let us take a closer look at the arguments Popper gives for this methodological rule. I will treat them relatively elaborately because I think that they form an important aspect of his views of rationality. They are five, and I will comment upon them in passing.

(i) Blaming the model for falsification of the theory teaches us more than blaming the rationality principle (Models 177).

Popper suggests that learning that the principle of rationality is false provides us with less insight than learning that the situational model is no good. Once we blame the model, we learn that something was wrong with the original sketch of the situation, and that to come up with a new tentative theory we have deepen our analysis of the agents' restrictions, knowledge, and goals.

[4]K.R. Popper (1935/1959), The Logic of Scientific Discovery, Vienna and London: Verlag von Julius Springer and Hutchison & Co.

The argument seems plausible, but only given that one adheres to the logic of situational analysis. For Popper does not compare social theories of his kind with those resulting from competing methodologies where the animating forces (but also the models) may very well be quite different.[5] Comparing an alternative social theory with one of Popper's kind, however, would mean to compare the alternative with the (Popperian) combination of model and animating force. In such a comparison a separation of the model and the rationality principle would not make sense.

(ii) The model is better testable than the principle of rationality (Models 178).

This is a rather unfortunate, and strictly speaking senseless, reformulation of the first point. Again, to test the principle of rationality you need the situational analysis. Ironically enough, not even ten lines below the statement of this argument Popper remarks that "we must not forget that we can test the theory only as a whole" (Models 178).

(iii) The principle of rationality, though false, is sufficiently close to the truth (Models 178).

That is, an empirical refutation of a social theory (model plus principle) will generally be due to the model rather than to the principle: "the main responsibility will normally attach to the model" (Models 178). Again this does not make sense by the book, since the principle of rationality is not refuted on its own but always together with a model. If, moreover, Popper means that social scientists hold the model liable in most of the cases, then that is nothing less and nothing more than the statement that social scientists normally follow the methodological rule he wishes to defend. That is, however, not a good argument since an argument should show us that the methodological rule is compatible with rules higher up in the hierarchy.

(iv) Giving up the principle of rationality leads to arbitrary model-building (Models 178).

It is clear though that giving up the principle of rationality means giving up situational logic. To see, then, whether the models become arbitrary we would need a comparison with other methodologies—with problems as before.

(v) Most social theories share the principle of rationality (Models 178).

That means that theories that survive testing will include the principle of rationality most of the time. The point of this argument seems to be that the methodological rule is not very far-reaching in the sense that it only rules out a small fraction of contemporary social science. Again, though showing that the acceptance of the methodological rule will not lead to a universal reformulation of the social sciences, it does not become clear what is wrong with theories inspired by methodologies that are not based on the principle of rationality or assign a different role to it.


The conclusion is that Popper's answer to his interlocutors' worries about the status of the rationality principle, although coherent, is not backed up by very plausible arguments. In the remainder of this paper, I will examine the reasons Popper may have had to answer in this way. My strategy is to contrast Popper's position with a competing picture. I will start with a short presentation of this alternative. Then, I will lay bare the motivation behind Popper's own answer by showing that particular views—first, on action, second, on the unity of the sciences, and third, on the falsifiability of rationality—prevented him from choosing the alternative. I should emphasize that it is not my intention here to defend the contestant; it only appears as part of an exegetical strategy.

To introduce the alternative view on the principle of rationality, I start with a brief recap of Popper's conception of the principle of causality; that is, the assertion that every event can be causally explained or deductively predicted. Popper mentions two interpretations of this statement. It is tautological as long as you require that an explanation or prediction do not refer to "arbitrary" laws; it is metaphysical (that is, not falsifiable empirically) if it means that all events are ruled by laws. In Popper's philosophy the principle of causality obtains the form of a methodological rule. The rule says that one should always try to explain in causal terms and to search for laws (Logic 78), and that one should try to leave nothing unexplained (Logic 36). This rule is, of course, part of the hierarchy of methodological rules discussed before with the supreme rule of falsifiability on top. A bit lower we find, inspired by the principle of causality, the rule of causality; it imposes on science (natural as well as social!) the formal restriction to work in a deductive-nomological framework (Logic 12). Much lower, inspired by the principle of rationality, we find the rule not to give up the rationality principle; it imposes restrictions on the content of social theories (viz., that they all contain the rationality principle as animating force).

[5]One example from many is R. Boudon (1990), La logique du social, Paris: Hachette.

The alternative outlook on the rationality principle that I wish to introduce as part of my exegetical strategy, now, raises the rule of rationality to the same level in the hierarchy as the rule of causality; it renders it a similarly formal restriction on theories. The alternative methodological rule says that one should always try to explain human behavior in terms of reasons. It is not, or at least not immediately, a restriction on the content of social theories.[6]

To show why Popper did not advocate the alternative formulation of the methodological rule, I will conclude with a discussion of his ideas on action, on the unity of the sciences, and on the falsifiability of the rationality principle.

(i) Action

The tetradic schema of theoretical rationality as a process of trial and error (the conception of science as problem solving) just as well applies to the practical rationality of actions, or so Popper claims. Where in the theoretical case researchers develop a tentative theory as a solution to some (scientific) problem, in the practical case agents choose actions as solutions to (practical) problems. And where tentative theories have to be adequate, actions have to be adequate as well; that is, agents act rationally or adequately in the situation they face.

The analogy between theoretical and practical rationality, however, only works well for the first two steps of Popper's tetradic schema. Having put forward a tentative theory to some problem, the third step of error elimination has the researchers severely test their tentative theory. By pointing out errors such testing may result in falsification of the theory. What counts as a falsification? To avoid a psychological foundation of knowledge ("the doctrine that statements can be justified not only by statements but also by perceptual experience" (Logic 25)), Popper takes a falsification to be a clash between so-called "basic statements" and the logical consequences of the theory; not, that is, a clash between experience and the theory. To avoid subjectivism, the researchers have to agree intersubjectively on truth of the basic statements.

Whether a theory is an adequate answer to a problem depends, then, on whether it clashes with sentences intersubjectively agreed on. Can we say something similar for actions? I believe we cannot. Where a clash between a theory and a basic statement is a simple and objective matter of logic, it is not at all clear where to look for the analogs of clashing in the practical case; and where to look for the analog of intersubjectivity, and the analog of basic statements. A strict and literal application of the tetradic schema to practical rationality in which adequacy of action is defined in terms of intersubjective falsifiability does not make sense.

Falsification in a broader sense is to show that a proposed solution for a problem is no good. An action would then be falsified (and hence inadequate) if it did not solve the problem. But this less strict reading of the analogy does not get us very far either. We cannot take it to be the case, for instance, that an action solves a problem whenever the agent's goals are reached as it is quite possible that the problem situation can impose such restrictions on the agent that he can only make the achievement of his goals more or less probable (but not certain). And conversely, the very reaching of a goal does not make the action performed an adequate solution to the problem because the goal may have been reached by accident, or by the help of fate.

The conclusion is that the analogy between theoretical and practical rationality does not hold water. Whereas for theoretical rationality Popper uses a clear and well-defined notion of adequacy in terms of intersubjective falsifiability, for practical rationality he does not provide us with more than a primitive notion. Popper does not seem to be aware of this problem, and that explains, I believe, why he does not choose the alternative rule of rationality in terms of reasons.

A related sign of the problems of his views can be seen in his illustrations of human actions studied by sociologists, historians, or anthropologists. Popper almost always uses examples of a rather theoretical character. He describes Galileo's allegedly irrational choice of a theory of tides that denies any lunar influence as a choice that, given Galileo's knowledge and scientific aims, was entirely rational. He describes the misconceptions about reality of the "madman" as a rational reply to situations that are in some sense rare and pathological. He describes a general's acts of war as the rational result of his "limited experience, limited or overblown aims, limited or overexcited imagination" (Models 178). Certainly these cases are all examples of decisions that can be explained as a rational and adequate reply to a particular problem situation; situations, however, in which the problem is to choose a theory rather than an action. Certainly the three components of a situational analysis (environmental restrictions, goals, information) can easily be given; Popper's examples, however, are no ordinary instances of agents acting. What these examples lack is a rather essential feature of human action: the wish to change the world in certain respects.

(ii) Unity of Science

The rule of causality is the driving unificatory force in Popper's system of science. It requires that all explanations fit the deductive-nomological pattern; and so it gives, if conscientiously followed, formal unity of scientific explanation. If we now replace Popper's methodological rule about the rationality principle by the alternative rule of rationality to explain human behavior in terms of reasons, then this unity collapses. The rule of causality looses its hold on social theories; the explanation of human actions does not invoke universal laws any longer; the deductive-nomological approach is given up in favor of reasons; that is, the natural sciences bow down to causes, the social sciences to reasons. And this goes against Popper's aspirations of unity in the sciences.

Nevertheless, Popper seems to have been aware of the impossibility in the social sciences to take literally his methodological recommendations to phrase social explanations in terms of universal laws and initial conditions. His method of situational analysis, his usage of "typical" initial conditions (a deviation from the tenets of The Logic of Scientific Discovery we have not emphasized much up to this point), is exactly intended to find an answer to this dilemma. For he writes, "only in this way can we explain a social event," later adding between brackets, "only in this way because we never have sufficient laws and initial conditions at our disposal to explain it with their help" (Models 168). Here Popper comes close to voiding the methodological rule of causality (to explain in a deductive-nomological manner) for the social sciences.

A reason for Popper, then, not to embrace the alternative view on practical rationality may have been that he would have to give up the formal unity of the sciences embodied in the rule of causality. However, the alternative picture does not go against Popper's broader and unifying idea of science as an enterprise directed at "understanding." It is interesting to quote from the fourth Chapter of Objective Knowledge. Expressing a kind of unity it nevertheless comes quite close to the alternative; that is, to positioning the rationality principle and the causality principle at the same level in the methodological hierarchy:

"as we understand men in virtue of some rationality of their thoughts and actions, so we may understand the laws of nature because of some kind of rationality or understandable necessity in them" (Objective Knowledge 184).

(iii) Rationality Falsified?

Exalting the principle of causality to a methodological rule, it might be argued, will be up against much fewer conceptual problems than lifting the rationality principle to a rule. In the former case a tautological or metaphysical principle turns into a rule, whereas in the latter case an empirical (and allegedly falsified) statement is transformed into a rule. And this, one might argue, works mischief; and that is why one should not opt for the alternative view on the rule of rationality.

However, Popper need not argue like this. Careful consideration of the supposed falsifications shows that the alternative rule of rationality is more true to the structure and the mechanism of social explanations, even from Popper's slant. He presents the following (basic?) statements as falsifications of the rationality principle: "flustered drivers trying to get out of a traffic jam, or desperately trying to park their cars when there is hardly any parking space to be found, or not at all" (Models 172). But without knowing the exact goals of the drivers we cannot say anything about their rationality. Without a situational analysis, that is, the rationality principle is not only, as Popper says, "almost empty," (Models 169, emphasis his) but also rather powerless. Returning to the earlier discussion, a theory—"we must not forget" (Models 178)—can only be tested as a whole.

Altogether it does not make sense to say that the rationality principle is falsified on its own; but that does not mean that it is not an empirical statement. However, there are good reasons to doubt whether the kind of principle of rationality that Popper discusses is really empirical at all. Without spelling out any details, one could as well phrase it as a metaphysical principle that all actions have reasons; and the resemblance of such a formulation to the metaphysical principle of causality (viz., that all events are ruled by laws) would not be accidental.

[6] I am indebted to P. Mongin (2002), "Le principe de rationalite et l'unite des sciences sociales," Revue Economique 53:301-323.

Phrasing practical rationality as a metaphysical principle has as an additional advantage that what Popper says in his first (of the five) arguments for his methodological rule makes sense for the alternative rule of rationality as well. Recall that he defended his rule never to give up the rationality principle in the face of falsifying evidence by stating that "we learn more if we blame the situational model" (Models 177) and that "our model is far more interesting and informative" (Models 178) than the principle of rationality. A metaphysical reading of practical rationality makes this understandable as well. What we find interesting in an explanation of human actions, or what we find informative, is the specification of the goals and the knowledge of the agent, the physical and social restrictions he is subject to; that is, the reasons why the agent carried out the action. It is, indeed, much less interesting that the agent acted on reasons, that he was rational, that he acted adequately or rationally in the situation in which he found himself. An explanation of someone's action is, then, the same as reporting his reasons, his goals, his information, his restrictions, his misconceptions, etc. Of course it can be very hard and troublesome to give reasons for an action—and sometimes one may be tempted to say that a person acted without reasons—but that is nothing more than the expression of one's own failure to explain the action.[7]

[7] I would like to thank Maurice Lagueux, Martin Stokhof, and two anonymous referees for their written comments on an earlier version of this paper. I would like to thank Philippe Mongin for inspiring discussions about his paper on the principle of rationality and the unity of the social sciences.