2. The original argument and its epistemological foundations

At the core of Albert's original critique (Albert, Reference Albert and Topitsch1971[1963]) is the distinction between relatedness to reality, informational content and truth. Any given statement might be evaluated according to these three criteria. The first asks whether a statement bears any reference to existing objects, the second asks whether a statement provides some information about these objects and the third signifies whether a statement is correct, that is in line with the real state of affairs. For instance, while the statement ‘if the weather does not change it will stay the same’ is clearly related to reality as well as true it does not inform about future events and, therefore, carries no informational content (Albert, Reference Albert, Arnold and Maier-Rigaud2012[1963], 297).

In general, the relationship between these three criteria is rather complex. While relatedness to reality is a precondition for any given statement to bear informational content and/or (non-analytical) truth, the latter two aspects are inversely related: a statement with high informational content precludes many possible events and, thus, has a higher ex-ante probability of being empirically false. Conversely, a statement, which is true by certainty, i.e. tautological, cannot fail in an empirical test, since it does not bear any informational content. That is, of course, a general rationale of critical rationalism (Popper, Reference Popper2010[1959]).

With regard to economics, Albert argues that economists and economic theory systematically confuse the criterions of relatedness to reality and informational content and, therefore, end up with a series of theoretical conceptions, which are (nearly) true by certainty, but do not lead to any explanatory progress. Additionally, he contends that economists tend to protect or defend their theoretical results against empirical criticism by epistemologically invalid arguments or manoeuvres, so-called ‘conventional strategies’ (Albert, Reference Albert, Arnold and Maier-Rigaud2012[1963], 300). These strategies are characterized as leading to an immunization against critique (Kritikimmunisierungsstrategien) and are depicted as further weakening the epistemological position of neoclassical economics. He goes on to illustrate these alleged weaknesses with three prominent examples – the law of demand, the quantity theory of money and the theory of growth.

In more detail, Albert's criticisms against the ‘neoclassical style of economic reasoning’ can be grouped into two categories. First, he makes a series of arguments related to the theoretical structure of neoclassical economics. Specifically, he maintains that the utilization of unrestricted ceteris-paribus clauses, the indeterminate formulation of some theoretical conceptions (like the quantity theory of money or revealed preference theory) as well as the introduction of unrealistic auxiliary hypotheses (‘alibi-assumptions’ according to AlbertFootnote ¹) renders theoretical statements untestable. Additionally, he expresses his dissatisfaction with regard to the specification of the intended domain of application for typical economic models, which he perceived as imprecise and vague. At the same time, Albert criticizes the relative neglect of questions of empirical adequacy compared to questions of logical consistency. In sum, he believed these tendencies to foster immunization against critique on a large scale.

All these arguments are connected to the confusion between relatedness to reality and informational content on the one hand and to the distinction between explanatory and auxiliary hypotheses within economic models on the other hand. While the latter aspect remained underdeveloped in Albert's own exposition, it is very helpful for understanding and evaluating his central arguments.

The second group of arguments Albert advances is less of a theoretical, but more of a discursive nature. Albert argues that theoretical economics is rather closed, neglecting to incorporate empirical results. On a disciplinary level, Albert observes a walling-off with regard to related disciplines such as psychology or sociology as well as a systematic neglect of more deviant traditions within economics. Both aspects are in stark contrast to the critical rationalist ideal of a pluralist and open(-minded) debate (Albert, Reference Albert1985), which has much in common with a pluralist attitude to economic research as suggested by some heterodox economists (e.g. Garnett et al., Reference Garnett, Olsen and Starr2010). While theoretical as well as discursive arguments are of equal importance for Albert's reasoning (Albert, Reference Albert, Arnold and Maier-Rigaud2012[1963]), this article concentrates on the former aspect, i.e. Albert's observations with regard to structural elements of neoclassical theorizing.

In this context, most of Albert's arguments can be traced back to a common root, namely the distinction between explanatory and auxiliary hypotheses within economic models. Take, for example, Albert's critical stance on ceteris-paribus clauses: Even in his original article, his critique is mainly focused on how these clauses are utilized rather than the fact, that these clauses are in use. Specifically, his main reservation was on using ceteris-paribus clauses in an indeterminate form simply to defend theoretical ideas against empirical results (‘something exogenous must have happened, because we did not get the expected result’), while neglecting the possibility to derive new auxiliary hypotheses out of these failed tests. This possibility arises out of a pragmatic approach to ceteris-paribus clauses. Basically, any reference to a ceteris-paribus clause as a means for defending a theoretical proposition has to be accompanied by an explicit specification of the reasons for predictive failure. This specification in turn leads to one or more new auxiliary hypotheses giving a more precise description of a model's intended domain of application (Gadenne, Reference Gadenne, Löffler and Runggaldier1999). As a result, Albert's harsh stance on ceteris-paribus clauses is strongly related to the way explanatory and auxiliary hypotheses are treated within economic modelsFootnote ².

And indeed, the overwhelming majority of contributions to economic theory does not even introduce this differentiation at all, but rather treats all model assumptions as epistemologically equivalent premises (‘axioms’). While this makes no difference on the level of mathematical deduction, it is highly consequential for the epistemological interpretation and, thus, the testability of the theorems derived from economic models. Auxiliary hypotheses are preconditions, which have to be fulfilled in case of an empirical test, and therefore specify the intended domain of application for the explanatory statements inherent in such theorems. Let's illustrate this point with a simple example from the physical realm. Take, for instance, Galileo's law of falling bodies, which is unbounded as such. Empirical tests on earth would corroborate it, while empirical tests on Mars would lead to its falsification. But do we know anything about the applicability of his law of falling bodies on Mars? The classic formulation of the law of falling bodies is not specific enough to give us a precise answer on this question. However, from the perspective of Newtonian theory, all this is very simple. Combining a subset of the Newtonian laws with an appropriate set of auxiliary hypotheses allows for deducing the law of falling bodies. In such a formulation, it becomes clear that the Mars is as outside of its intended domain of application (since the auxiliary hypotheses specify Earth to be one of the two relevant mass points) as is an empirical test based on the fall of a parachutist (since the auxiliary hypotheses usually specify the absence of exogenous forces such as air resistance). By specifying the intended domain of application, auxiliary hypotheses also restrict the informational content of the respective theorems. In this case, the original formulation of the law of falling bodies as a single hypothesis has a greater domain of application, more informational content, but is also at greater risk to fail in an empirical test than its Newtonian reconception as part of a more general theory.

In analogy to the natural sciences, we may understand central theorems of economic models as those statements, whose informational content is relevant for judging the testability of the associated models. The typical variant of an economic model thereby postulates a series of basic assumptions or axioms from which a certain theorem is derived

\begin{linenomath}\begin{equation*} \frac{{{\rm A}_1,\,\,{\rm A}_2 \ldots {\rm A}_{r + s} }}{{\rm T}}.\end{equation*}\end{linenomath}

In this raw form, one might judge whether a model is related to reality, in the sense that the axioms contain descriptive elements referring to real objects, but the informational content of the model's theorem remains unclear, since we have no information on how exactly these axioms relate to reality. Is, for example, the axiom of cross-substitutability of inputs implicit in most production functions to be understood as an explanatory assumption, which may be ‘disproved’, or is it to be interpreted as an auxiliary hypothesis restricting the applicability of a model's theorems? So in order to estimate whether there is informational content in T, we need to differentiate these assumptions into a group of r explanatory hypotheses (H) and another group of s auxiliary hypotheses (AH)

\begin{linenomath}\begin{equation*} \frac{{{\rm H}_1 \ldots {\rm H}_{\rm r} \, \wedge \,{\rm AH}_{\rm 1} \, \ldots \,{\rm AH}_{\rm s} }}{{\rm T}}.\end{equation*}\end{linenomath}

What follows from this distinction for our theorem? Well, basically, the theorem T should apply to situations as specified by AH₁ to AH_s and, thus, these auxiliary hypotheses are transferred into the if-clause of the resulting theorem

\begin{linenomath}\begin{equation*} \frac{{{\rm H}_1 \, \ldots {\rm H}_{\rm r} }}{{({\rm AH}_{\rm 1} \, \ldots \,{\rm AH}_{\rm s} ) \to {\rm T}}}.\end{equation*}\end{linenomath}

From the perspective of logical deduction, this transformation is completely equivalent, to the one considered earlier where the auxiliary hypotheses are part of the explanation (Suppes, Reference Suppes1957). However, from the perspective of empirical testing, the decisive difference between these two transformations is that the second variant captures the implications of classifying assumptions into explanatory and auxiliary hypotheses, where the latter serve as specific restrictions for the application of the derived theorems. Since no binding conventions exist in economics on how to precisely apply this differentiation with respect to specific axioms (Albert, Reference Albert1996), a certain amount of arbitrariness is introduced with respect to the informational content of the resulting theorems. An increase in the number of axioms interpreted as explanatory hypotheses leads to a corresponding increase in informational content of the relevant theorem, which is now less restricted (i.e. more general). Conversely, an increase in the number of axioms interpreted as auxiliary hypotheses decreases the informational content.

The question which epistemological role to ascribe to various axioms is particularly vexing in the context of very unrealistic assumptions: If they are interpreted as explanatory hypotheses, the resulting theorem is highly likely to fail and, conversely, if they are treated as auxiliary hypotheses, they render the theorem inapplicable by minimizing its domain of application. So, the basic dilemma is that ‘either the theory is irrefutable or it is far too easily refuted because it is unbounded’ (Archibald, Reference Archibald1963, 227). But even if unrealistic assumptions understood as explanatory statements lead to correct conclusions there is no reason for contentment, since what we got is a classical case of ‘wrong antecedent, correct consequent’. Regrettably, there is no logical justification for judging the validity of the assumptions by the correctness of the conclusion (as suggested by Friedman, Reference Friedman1966[1953]) – indeed any such agreement between theory and reality is to be interpreted as coincidental.

In this context, Albert's critique is based on the assertion that economists do not face but rather systematically evade the dilemma associated with unrealistic assumptions. He specifically argues against ‘alibi-assumptions’, which provide a comprehensive alibi against conflicting empirical evidence, and explicitly refers to unrealistic assumptions that are interpreted as a condition for a successful application (i.e. as auxiliary hypotheses). His accusation of a Platonic style of thinking partially rests on the observation that an interpretation of economic axioms simply as logical conditions leads to a thought-experimental style of theorizing devoid of any substantial hypotheses and, thus, a lack of informational content. Technically, such an interpretation implies to treat all assumptions as auxiliary hypotheses (that is the special case where r = 0).

Albert's critique on ceteris-paribus clauses can, in turn, be interpreted as plea for providing specific explanations in case of a failed empirical test and, consequently, for introducing novel auxiliary hypotheses as a result of this process. Since all of Albert's major criticisms are strongly related to the differentiation between explanatory and auxiliary hypotheses, this distinction allows us to systematize the different shortcomings diagnosed in his classic article. While the mechanics of the alleged immunization strategies differ with respect to their treatment of the basic axiomatic structure of economic models, their result is always a decrease in, or even an elimination of, informational content. The following section tries to differentiate between different variants of immunization against critique as addressed by Albert in a hypothetico-deductive framework (Hempel, Reference Hempel1970[1965]) and, thereby, aims to provide a more systematic outlook on his most important arguments.

3. Five variants of interpreting economic models

In this section, five different possibilities of interpreting economic models are illustrated with respect to their epistemological structure in general and the testability of the resulting theorems in particular. In this spirit, the standard interpretation of economic models (1), which is often complemented by the assumption of a ceteris-paribus clause (2), is compared with three more precise, but also more critical accounts. These are Albert's interpretations of economic models as pervaded with alibi-assumptions (3), a thought-experimental style of theorizing (4) and indeterminate explanatory hypotheses (5).

Usually, economic models are given in the following form, where a theorem T is deduced on the basis of a certain set of unspecified assumptions A

(1)\begin{linenomath}\begin{equation} \frac{{{\rm A}_1 ,\,{\rm A}_2 , \ldots {\rm A}_{{\rm r + s}} }}{{\rm T}}.\end{equation}\begin{linenomath}

In hypothetico-deductive terms, Albert's critique suggests that this conception of economic models is insufficient and, thus, misleading. It has to be put in more precise terms in order to illustrate the epistemological subtleties associated with economic models. More specifically, the various strategies of immunization against critique Albert identifies can be illustrated by more precise variations of the previous formulation. The simplest case of such a variation, one of which economists are well aware, is to complement the previous formulation by adding a ceteris-paribus clause

(2)\begin{linenomath}\begin{equation} \frac{{{\rm A}_1 ,\,{\rm A}_2 , \ldots {\rm A}_{{\rm r + s}} \wedge {\rm CP}}}{{\rm T}}.\end{equation}\begin{linenomath}

If, in such a case, a researcher resorts to the ceteris paribus clause to defend her theorem against conflicting empirical evidence, she has to precisely specify which exogenous forces caused the theory's prediction to fail and suggest one or more appropriate new auxiliary hypotheses to accommodate for the observed interference. Otherwise, it would be a case of immunization against critique via resorting to a ceteris-paribus clause, since the only usage of the ceteris-paribus clause has been to protect a theorem from being refuted.

While unspecified ceteris-paribus clauses provide theories with an alibi against conflicting empirical evidence in a very general way, it is also possible to introduce more specific alibi-assumptions. These assumptions are conceived as unrealistic auxiliary hypotheses, which are never expected to be fulfilled (e.g. the assumption of fully informed individuals) and, thus, give an opportunity to deny any relevance to empirical results, since the specific condition can never be realized. Formally, the introduction of such an alibi-assumption (AH_i) among the set of auxiliary hypotheses (AH₁ to AH_s) can be illustrated in the following way, where the expression non-AH_i represents the act of drawing on such an alibi by referring to the unrealistic character of the specific assumption AH_i

(3)\begin{linenomath}\begin{equation} \frac{{{\rm H}_1 \ldots {\rm H}_{\rm r} \wedge {\rm AH}_{\rm 1} \ldots {\rm AH}_{\rm s} }}{{{\rm (AH}_{\rm 1} \ldots {\rm AH}_{\rm i} \ldots {\rm AH}_{\rm s} \wedge \hbox{non-AH}_{\rm i} {\rm )} \to {\rm T}}}.\end{equation}\begin{linenomath}

The resulting expression postulates the inapplicability of T for any given situation, by imposing a contradiction in its antecedent, which renders its consequent logically true, but empirically worthless. Thus, the resulting argument ((AH₁ . . . AH_i . . . AH_s ∧ non-AH_i) → T) lacks any informational content, since possible settings for a solid empirical test of the resulting theorem are logically inconceivable. Already in his original article, Albert (Reference Albert and Topitsch1971[1963], 417) emphasized that in an extreme form (i.e. in a totally unbound interpretation) ceteris-paribus clauses render a theory tautological. The same also holds for models, which include alibi-assumptions, since their theorems have logically no domain of application as soon as they draw on their ‘alibis’.

So, ceteris-paribus clauses and alibi-assumptions work in a very similar way: They allow theories to evade empirical results, by declaring them as invalid for refuting the theorem at hand – either with reference to unexpected exogenous forces or by interpreting conflicting evidence as outside of the theorem's domain. But Albert also addressed other aspects of economic models, which deemed him to be epistemologically dubious. More specifically, he diagnosed a tendency to substitute questions of empirical adequacy with questions of logical necessity and attacked those proponents of neoclassical economics, who deemed their central assumptions to be true a priori (Albert, Reference Albert, Arnold and Maier-Rigaud2012[1963]). Reaching beyond the critique of pure apriorism in economics, his diagnosis also refers to a kind of self-sufficiency with respect to the relative importance of mathematical expositions versus empirical results.Footnote ³

Even though the prevalence of apriorism in economics has declined and empirical research has become much more prominent – two developments also recognized by Albert himself – his intuition with regard to the relative weight of empirical considerations in terms of theory construction, i.e. model building, still seems valid. The rationale for constructing economic models out of simple and empirically dubious basic assumptions has changed, but not disappeared. In earlier times, this rationale has been justified by stating that ‘only [. . .] the ignorant and the perverse’ (Robbins, Reference Robbins1945[1932], 16) would doubt the main assumptions on economic behaviour. Nowadays, the very same assumptions are justified as the ingredients of a specific, conceptual ‘story’ (Gibbard and Varian, Reference Gibbard and Varian1978; McCloskey, Reference McCloskey1998; Morgan, Reference Morgan2001), a ‘fable’ (Rubinstein, Reference Rubinstein2006) situated in a ‘counterfactual world’ (Sugden, Reference Sugden2000, 18), which is told by the author of an economic model. This ‘story’-view on economic models, which is highly popular in current economic methodology, justifies these assumptions as being part of a potentially interesting thought-experiment, that is a speculative scenario ‘without empirical commitment’ (Hausman, Reference Hausman1992, 6). In hypothetico-deductive terms such a thought-experiment takes none of its assumptions ‘seriously’, i.e. as explanatory statements, but conceives them all as auxiliary hypotheses (the special case of r = 0). In such a thought-experiment, there exist only auxiliary conditions, since no statement is effectively put to the test

(4)\begin{linenomath}\begin{equation} \frac{{{\rm AH}_1 \ldots {\rm AH}_{\rm s} }}{{\rm T}}.\end{equation}\begin{linenomath}

The resulting theorem is true by definition, since it a priori applies only to situations where it is correct with absolute certainty. While the justification for the increased emphasis on mathematical exposition in relation to empirical adequacy has changed, the routine as such, which has been in the focus of Albert's critique, remained.

Especially with regard to the quantity theory of money (in his original article) and the role of rational choice theory in economics (in later publications, see Albert, Reference Albert1998[1967]), Albert suspected that the theorems of economic models are devoid of informational content, since central theoretical propositions in neoclassical theory are already conceptualized in a tautological way. This assertion is rather well-known (Hausman, Reference Hausman1992; Hutchison, Reference Hutchison1965[1938]), Rosenberg, Reference Rosenberg1976) and is based on the observation that, if the utilized hypotheses are indeterminate (H*), because they are already formulated tautologically, they are easily compatible with any state of the world

(5)\begin{linenomath}\begin{equation} \frac{{{\rm H}_{\rm 1}^{\rm *} {\rm \ldots H}_{\rm r}^{\rm *} \wedge {\rm AH}_{\rm 1} {\rm \ldots AH}_{{\rm s1}} }}{{\rm T}}.\end{equation}\begin{linenomath}

A simple example for an indeterminate hypothesis is the assumption of utility maximization with completely flexible preferences, where any behaviour can be explained by envisaging a compatible preference structure. This agnosticism with regard to preference structures makes the resulting hypothesis so flexible that it is devoid of informational content and, thus, may accommodate any potential evidence. Lazear (Reference Lazear2000) provides a list of examples following this approach and Blinder (Reference Blinder1974) offers an entertaining account of its consequences.

This section summarized Albert's main arguments from a more technical perspective. It exposed five varying interpretations of economic models, which are subtly present in Albert's original account as well as in the practice of current economic research. The first of these accounts – the standard interpretation of economic models – is too simplistic to allow a concise analysis with regard to a model's testability. The following two interpretations – where economic models are complemented by a ceteris-paribus clause or an alibi-assumption – allow for declaring conflicting empirical evidence as unsuitable for refuting the model's predictions. In the last two cases – where models are either conceived as pure thought-experiments or contain indeterminate hypotheses – economic models are tautological by nature. The classic interpretation of economic models as hypothetico-deductive systems, introduced in the previous section, serves as another alternative way to interpret economic models.

All these accounts of economic models are explicitly and implicitly present in economic thinking as well as in the history of economic methodology. Milton Friedman's famous defence of unrealistic assumptions in economic theory (Friedman, Reference Friedman1966[1953]) is thereby strongly related to Albert's assault against alibi-assumptions, albeit both authors argue from a quite different perspective. Conceiving economic models as thought-experiments is compatible with an interpretation of economic models as representing stories situated in ‘counterfactual worlds’ (Sugden, Reference Sugden2000, 28). And ceteris-paribus clauses of course represent an age-old topic in economic textbooks and economic methodology (Hausman, Reference Hausman1990).

4. What is Model-Platonism?

What, then, does the idea of Model-Platonism contribute? First, as we have seen in the foregoing sections, Albert's critique allows for summarizing and differentiating various forms of immunization against critique on the level of economic theory. Second, Albert also emphasizes the importance of the discursive level, where he proceeds in a standard critical rationalist fashion and urges economists to be open for criticism of all sorts and to specifically design their scientific institutions in a way assuring this openness. However, along with his critique, also comes a specific view of economic theory, which is not exhaustively represented by analysing different immunization strategies. This section is dedicated to this specific view on economic theory and exposes two vital routines of the ‘neoclassical style of economic reasoning’, which mainly operate in the intellectual background of what has been discussed so far, that is the axiomatic structure of economic models.

Let's start with the term ‘Platonic’, which is to be interpreted quite literally here and has some general epistemological implications.

‘[A] view generally associated with Plato, locates the superiority of a priori knowledge in the objects known. What we know by reason alone, a Platonic form, say, is superior in an important metaphysical way, e.g. unchanging, eternal, perfect, a higher degree of being, to what we are aware of through sense experience’. (Markie, Reference Markie2008[2004])

One obvious interpretation is that Model-Platonism implies that economic models are Platonic, if they take the form of thought-experiments, which use idealized conceptions of certain objects or entities (Platonic archetypes). This clearly sounds like a pretty familiar procedure, although the rationale for the thought-experimental character of economic models (if they are conceived as such) has been transformed over the years from apriorism to story-telling ‘without empirical commitment’ (Hausman, Reference Hausman1992, 6).

This interpretation is already present in the title of his most important book on economics – Marktsoziologie und Entscheidungslogik (‘Market-Sociology and the Logic of Choice’). When interpreting economic models as thought-experiments, that is, as mainly speculative endeavours, these models operate without any obligation to accommodate empirical results. However, if the idealized assumptions employed in these models are interpreted as true forms in a Platonic sense they gain strong empirical relevance, since they are assumed to provide us with a form of knowledge far superior to observational data. The combination of these routines – a thought-experimental style of theorizing and a strong reliance on the validity of a series of idealized basic assumptions – allows for reducing the question of market interaction, which, according to Albert, is an institutional question relating to the sociology of markets, to a question of mechanic calculus.

Following this line of argument, Model-Platonism as an epistemological concept can be understood as the combination of the following two routines: the reliance on thought-experimental style of theorizing as well as the introduction of idealized, metaphysical and, hence, ‘Platonic’ arguments in the form of basic assumptions. The latter element has often been identified with apriorism, which is no longer explicitly used to justify neoclassical assumptions on human behaviour or the workings of markets, and has been replaced by the notion of economic models as devices for story-telling to provide a much less committing justification for using these assumptions. However, we still find resemblances of the Platonic idea of ‘superior insights’ through ‘true forms’ in economic models. Thereby, these insights are sometimes even believed to be generally immune to conflicting empirical evidence – an attitude, which seems truly amazing to observers of economic discourse stemming from the natural sciences, like the physicist Jean-Paul Bouchaud.

‘An economist once told me to my bewilderment: “These concepts [like rationality or equilibrium] are so strong that they supersede any empirical observation”’. (Bouchaud, Reference Bouchaud2008, 1181)

Taking assumptions like utility maximization or market equilibrium as a matter of course leads to the ‘standing presumption in economics that, if an empirical statement is deduced from standard assumptions [. . .] then that statement is reliable’. (Sugden, Reference Sugden2000, 17)

It is worth noting that this idealization in explanatory terms is accompanied by a parallel idealization in normative terms, implying that behaviour according to neoclassical postulates will deliver beneficial results. Table 1 provides a short overview on some of these residuals of apriorism in economic thinking, the superior insights they are supposed to carry and the normative interpretations accompanying these insights.

Table 1.

Platonic forms in neoclassical economics

The ongoing importance of these assumptions is especially evident in those areas of economic research, where empirical results are challenging standard views on economic behaviour like experimental economics or behavioural finance. Some authors even hold that the mere existence of these fields proves the increasing irrelevance of early 20th century concepts such as utility maximization (Colander, Reference Colander2000; Colander et al., Reference Colander, Holt and Rosser2004). However, from the perspective of Model-Platonism, these research-areas are still framed by the ‘superior insights’ associated with early 20th century concepts, essentially because almost all of their results are framed in terms of rational individuals, who engage in optimizing behaviour and, thereby, attain equilibrium. For instance, the attitude to explain cooperation or fair behaviour in experiments by assuming an ‘inequality aversion’ integrated in (a fraction of) the subjects’ preferences is strictly in accordance with the assumption of rational individuals, a feature which the authors are keen to report.

‘No other deviations from the standard economic approach are necessary to account for the evidence. In particular we do not relax the rationality assumption’. (Fehr and Schmidt, Reference Fehr and Schmidt1999, 819)

Fehr and Schmidt have gone to great lengths to evade a sociological or institutional explanation of the observed behaviour and to synthesize the existing evidence with the neoclassical standard model. However, their strategy to vary the preference structure of individuals renders, if applied successively, the neoclassical theory of economic behaviour completely tautological. The hypothesis of rational action itself becomes indeterminate (H*) if simply anything suitable to explain the evidence at hand can be integrated into the relevant preference-structures. In its most extreme form of economic imperialism, this strategy leads to the extraordinary claim that all forms of human behaviour can be explained by a most flexible treatment of preference structures (Lazear, Reference Lazear2000).

Aside this observation, the tendency to explain and interpret empirical evidence with strict reference to the standard model of economic behaviour by assuming appropriate preference structures pervades fields like experimental economics or behavioural finance and allows for flexible treatment of any encountered anomalies. In this context, behavioural finance is interpreted as a collection of exceptions to the standard theory, which thereby preserves its legitimization as a representation of typical and optimal behaviour on financial markets. Neoclassical theory may, thus, incorporate results from behavioural finance as special cases ‘without experiencing any fundamental changes in its own methodology’ (Frankfurter and McGoun, Reference Frankfurter and McGoun2002, 376). Experimental economics, on the other hand, goes even one step further and often, as in the case of Fehr and Schmidt (Reference Fehr and Schmidt1999), follows the basic routine of economic imperialism by consciously invoking specific variants of preference structures accommodating for the observed behaviour.

‘Experimental papers in economics usually describe the data of an experiment and then propose a theory that fits the data if various parameters are suitably chosen’. (Binmore and Shaked, Reference Binmore and Shaked2010, 89)

In the latter quote, ‘theory’ can practically be equated with the neoclassical theory of behaviour, while ‘parameters’ stands for ‘preferences’. Another indicator for the continuing relevance of traditional neoclassical assumptions in this newly arrived fields, is the tendency to design experimental situations with much courtesy to neoclassical theory. One way to do so is to restrict the subjects’ decision space strictly according to theoretical conditions.

‘The well designed laboratory experiment gives a theory its “best-shot” at making accurate predictions. The assumptions of the theory are designed into the lab experiment’. (Croson and Gächter, Reference Croson and Gächter2010, 126)

The development of dictator-games (where player A unilaterally divides an amount of money between himself and another player B) out of ultimatum-games (where player B has veto-power) in experimental economics clearly followed this ‘best-shot’-logic: while the results in ultimatum-games strictly contradicted neoclassical behavioural assumptions, the more unequal dictator-setting proved to be less severe for economic theory. From a critical rationalist viewpoint, such a strategy seems dubious, since theories should be tested as severely as possible instead of being treated kindly by minimizing the risk exposure for a given theoretical statement. However, in case this strategy fails (even a ‘best-shot’ might miss the target), the approach behind Vernon Smith's experiments on market behaviour serves as an alternative to the ‘best-shot’-logic. Smith was well aware that ensuring all relevant theoretical conditions in market experiments did often not bring forth the expected results. So, he resorted to a very specific form of market interaction – the double oral auction – to gain results comparable to the predictions of standard economic theory, even if the institutional setting of double oral auctions differs rather strongly from the standard assumptions on perfect competition. In this case, experimental conditions were not designed strictly according to theoretical presumptions, but rather to induce appropriate behaviour among the relevant test subjects.

‘[T]he methodology of experimental economics was developed mainly in the context of market experiments and was aimed at the active inducement, rather than passive observation of preferences’. (Guala, Reference Guala2006, 262)

Given these observations, it is rather unsurprising that particularly German-speaking researchers, who were still aware of Albert's original critique, first identified the Model-Platonic tendencies prevalent in experimental economics.

‘Who only strives for a neoclassical explanation of laboratory results is in danger to construct a parallel word – equivalent to a Platonic realm – to reality’. (Güth and Kliemt, Reference Güth, Kliemt, Held, Kubon-Gilke and Sturn2003, 315, Translation by the author)

So, while the mere emergence of research areas like experimental economics is sometimes deemed a clear sign for the advent of a new era (Colander et al., Reference Colander, Holt and Rosser2004), a closer look at these fields allows us to illustrate the enduring relevance of the Model-Platonism-topos and, thereby, shows the pervasion of these fields with a traditional neoclassical style of thought.

5. Extending Model-Platonism: two suggestions

This section introduces two novel views on immunization-strategies in neoclassical economics. They are insofar connected to Albert's original critique as they refer to the absence of a thorough differentiation between explanatory (H) and auxiliary hypotheses (AH) among the assumptions of economic models. Additionally, resting on slightly more general observations and arguments, they contain some of Albert's more specific insights as special cases

Before going into the details, let's first illustrate how these two arguments connect to the missing distinction addressed here. Given that there exist no definite rules to transform

\begin{linenomath}\begin{equation*} \frac{{{\rm A}_{\rm 1} {\rm ,}\,{\rm A}_{\rm 2} {\rm \ldots A}_{{\rm r} + {\rm s}} }}{{\rm T}}\end{equation*}\end{linenomath}

\begin{linenomath}\begin{equation*} \frac{{{\rm H}_1 \ldots {\rm H}_{\rm r} }}{{({\rm AH}_{\rm 1} \ldots {\rm AH}_{\rm s} ) \to {\rm T}}}\end{equation*}\end{linenomath}

Another feature of theories coming into play here is that auxiliary hypotheses are normally varied with respect to the domain of application. Applying Newton's laws on Mars requires using different auxiliary hypotheses compared to applying the very same laws on earth. In neoclassical economic thought, however, even this distinction does not fully apply, since any assumption, be it of explanatory or auxiliary character, may be subject to some modification. This observation serves as a foundation for the second argument advanced in this section called axiomatic variation.

Acknowledgement

The author is indebted to Leonhard Dobusch, Volker Gadenne and three anonymous reviewers for providing valuable comments and suggestions.