Axioms and hypotheses make up a two-layer system, in which the axioms are situated in the core, surrounded by the testable hypotheses (theories) as a protective belt. When comparing our theories to reality, it is the hypotheses that we test, not the axioms. In case we cannot find a considerable concordance between the outcomes we generate and reality, we can only reject the hypotheses, not the axioms. Axioms cannot be refuted within the system. Whilst it is true that axioms cannot be traced back to anything in the system, they are not based on anything, they still have a source outside the theory. Axioms do not turn up by deus ex machina, but they directly rise from the interest of the scientist. Axioms can only be analysed on the basis of the purposes underlying a theory. In other words, the only question that makes sense is whether the axioms are appropriate as for achieving the purposes. All this calls attention to the fact that the ultimate grounds of any theory lie concealed in the scientist himself. It is us who create the theory.
Axioms make up the inner protected core not because of us having chosen them for this role. The reason is that testing axioms directly, i.e. without any theory is unattainable. Through a theory that is directly confronted with reality an axiom that bears indirect relation through the theory itself to reality cannot be tested. Naturally, it does not rule out the possibility of treating an axiom of a theory as a testable hypothesis in another system. It was exactly the case with rational expectations hypothesis suggested by Muth (1961). As the econometric tests had considerably confirmed the hypothesis, Lucas (1973) could apply that as an axiom. However, as soon as a hypothesis is put amongst the axioms, it loses its testability. It is in complete consonance with the results of Carnap (1939: 59). It is the same process as what was described by Lakatos (1968) analysing the evolution of modern natural science at multiple places. On account of the positive heuristics and negative heuristics, the testable protective belt built around the hard core prevents the most characteristic and most stable system of theories and assumptions from the critique and from any modifying efforts. The neutrality of the main stream against the institutionalist critique is not only the result of an instinctively followed scientific strategy. We should not forget about that fact that modern macroeconomics after Keynes have never made efforts to set up general theories. It is exceptionally true of Friedman having worked along an eclectic methodology, who was not even worried about the occasional inconsistency between his particular theories. However, it is also true of the subsequent Lucas–RBC-line that has always been directed at analysing only certain individual mechanisms of the complex causal structure.
 Keynes was far from the level of complexity that is the ideal for institutional economics. De Vroey (2016: 74) quotes a passage where Friedman writes that his esteem to Keynes was raised by the fact that Keynes could focus his attention to a small number of key variables.
Carnap, Rudolf (1939): Foundations of Logic and Mathematics. Chicago: The University of Chicago Press.
De Vroey, Michel (2016): A History of Macroeconomics from Keynes to Lucas and beyond. Cambridge: Cambridge University Press.
Lakatos, Imre (1968): Criticism and the Methodology of Scientific Research Programmes. London: Aristotelian Society.
Lucas, Robert E. (1973): Some International Evidence on Output-Inflation Tradeoffs. The American Economic Review, 63(3): 326–334.
Muth, John F. (1961): Rational Expectations and the Theory of Price Movements. Econometrica, 29(3): 315–335.
Imre Lakatos (1922 – 1974)