Constructive Empiricism and Science Education Michael Martin

Michael Martin

8 Gerry’s Landing Road

Cambridge, MA 02138




The controversy between social constructivists and scientific realists both in the philosophy of science and in science education shows no sign of abating.  For many participants in this controversy Social Constructivism and Scientific Realism seem like the only two alternatives.  Fortunately, this is not so. I would like to present a compromise position: the theory of Bas van Fraassen, which he calls Constructive Empiricism.[1]  I will suggest that this theory avoids many of the problems that plague Scientific Realism and Social Constructivism in relation to the philosophy of science. Its implications for science education have not to my knowledge been spelled out. Here I will suggest some implications that should be congenial to both social constructivists and its critics.


Constructive Empiricism

Van Fraassen believes that the primary criterion for judg­ing a scientific theory is whether it is empirically adequate; that is, whether it accounts for the observable data.  According to him, although theories are either true or false, their truth or falsehood is irrelevant for science.

For example, suppose physi­cal theory T for a domain D postulates unobservable entities E that behave in a certain way.  Suppose that either such entities do not exist or else that, although they exist, they do not behave as the theory says they do.  In both cases the theory would be false but, according to van Fraassen, this is irrelevant to science. The important question is whether a theory correctly describes what is observable in D and it could do so even if it were false. It might correctly predict what will happen in experiments in D, correctly describe the empirical laws that operate in D, and so on, even if entities E did not exist or did not behave in the way postulated by the theory.

Besides empirical adequacy van Fraassen uses pragmatic criteria in judging theories.[2] Among these are mathematical elegance, simplicity, wideness of scope, and explanatory power. However, he stresses that these are not criteria of truth and have no epistemic import.  Rather they provide reasons for preferring theories independently of the question of truth.   For example, two theories T1 and T2 might both be empirically adequate but T1 might be simpler than T2.  For this reason T1 might be preferable — other things being equal– to T2 . But, says van Fraassen, simplicity is not an epistemological criterion.  That T1 is simpler than T2 does not mean that T1 is more likely to be true than T2 or that T1 is closer to the truth than T2.


Contrast With Other Positions

Van Fraassen’s view of science should be contrasted with other positions.  First, it is different from Scientific Realism, which is the view that science should aim at developing true theories; that is, at theories which correctly describe unobservable entities and their behavior and by so doing explain observable phenomena.[3] For example, realists believe that physics should try to gain knowledge of the un­observable micro-entities, which constitute ordinary physical ob­jects.  Given this knowledge a physicist can then predict and explain the behavior of ordinary physical objects. Thus, knowing the un­observable behavior of gas molecules, he or she can predict and explain the observable properties of gas. According to van Fraassen, on the other hand, the only important thing is whether a theory of gas is true of the observable properties of gas; it is irrelevant if unobservable gas molecules behave in the way specified by the theory.

Second, van Fraassen’s view differs from Instrumentalism.[4]  The latter view holds that scientific theories are neither true nor false but instruments, tools, devices that are useful in predicting and controlling observable phenomena. On one version of Instrumentalism, theories do not take the form of statements but of rules that enable one to make accurate predictions.   However, rules are neither true nor false. In contrast, on van Fraassen’s view scientific theories are either true or false.  The instrumental dimen­sion of theories rest in the fact that pragmatic criteria are relevant to their evaluation.

Third, van Fraassen’s view is different from Social Constructivism, which maintains that observable facts are not discovered by empirical investigation about the world but are constructed.[5] According to van Fraassen’s view, although one might speak of the unobservable entities of physics as being constructed, this is misleading. For van Fraassen such entities may indeed exist independently of the theories of scientists but whether or not they do is irrelevant to science.  One evaluates the postulation of such entities in physical theory by pragmatic criteria, not in terms of whether they are real.


The Advantages

Van Fraassen’s Constructive Empiricism does not have many of problems that plague Social Constructivism and Scientific Realism. Social constructivism has been faulted for assuming both a relativistic and a subjective account of truth.[6] Constructive Empiricism does not presuppose either relativism or subjectivism, however, since, according to it, scientific theories are either true or false. Van Fraassen’s point is that the only thing that is epistemologically relevant is the empirical adequacy of the theories, that it is whether the theories correctly describe the empirical data.

Social Constructivism has also been criticized for not doing justice to the history of science. It is held, for example, that it can give no account of scientific progress.[7] Constructive Empiricism can, however, give such an account. According to it science is indeed progressing since its theories are now more empirically adequate than they were in the past.

At the same time, Constructive Empiricism avoids some of the criticisms made of Scientific Realism.  For example, it has been objected that Scientific Realism makes metaphysical assumptions that go beyond all possible empirical evidence. It is said that as a result it generates a deep skepticism since it postulates entities that transcend all possible empirical data.[8]  But Constructive Empiricism forsakes metaphysical speculation and is simply concerned with whether scientific theories are empirically adequate. Thus, it argues that whether or not the observable entities of physics are real is beside the scientific point.

Another objection to Scientific Realism is that it neglects the nonepistemic factors in scientific theory choice. However, Constructive Empiricism appeals to several pragmatic criteria in the choice between equally empirically adequate theories. Notice however that although pragmatic criteria are allowed, they play a secondary rote:  it is only after theories are judged empirically adequate that pragmatic criteria enter the picture.


Implications for Science Education

Insights from both Social Constructivism and Scientific Realism can be incorporated into science education within a Constructive Empiricist framework even as some of their excesses are avoided.  For example,  Helge Kragh points out that Social Constructivism’s tendency to put N-rays, cold fusion, polywater and other cases of pseudo-science in the same category as successful cases of science would have a disastrous effect on science teaching. He says that from science we know “for sure” that X rays exist and that N-rays do not.  Kragh says that “ought to be reason enough to exclude N-rays from the curriculum; or if including the subject, then treat it as a case- study of scientific error.”[9]  On the other hand, a Social Constructivist might be skeptical of Kragh’s metaphysical claims and argue that both X rays and N rays are in the same metaphysical ballpark: both are fictional constructions of scientists.

Van Fraassen’s approach avoids the metaphysical controversy altogether by sticking with empirical adequacy. Whatever the reality of N-rays and X-rays, the theories in which they are embedded are different.  Theories that postulate X-rays account for the available evidence whereas theories that postulate N-rays do not.  Hence, theories that posit N-rays would be excluded from a science curriculum or else would be treated as case studies of error.

Social Constructionist in science education has stressed a number of related ideas in regard to science education.  For example, they have argued that students learn best by constructing a representation of the world from within rather than from being fed facts from the outside.[10] However, supposing this idea to be correct, it can be easily incorporated into a Constructive Empiricist framework. From this latter perspective students would be taught to construct theories that are empirically adequate; that is, that agree with the evidence.  Although students would be free to create theories that postulate unobservable entities, they would learn that the reality of these entities is scientifically irrelevant. Students would also be encouraged to use pragmatic criteria to decide between equally empirically adequate theories while learning that such criteria have no epistemological import.

Ernst von Glaserfeld has argued that science educators should not tell students that they have produced a “wrong” solution but instead should put stress on whether the conceptual model developed by the students is “adequate” and “satisfying.”[11] When Scientific Realists hear this they throw up their hands and shake their heads in despair, accusing Social Constructivism of irrational subjectivism. However, whatever von Glaserfeld’s intention, his words can be given a different interpretation.  From the perspective of Constructive Empiricism science educators should not be concerned with whether a student’s model is wrong but only with whether it agrees with the empirical evidence; that is, whether it is empirically adequate. Thus, from this perspective it makes no difference if this agreement is satisfying to the student in some subjective sense.  What matters is if it is satisfying from an empirical point of view: if the student model is satisfied when if it correctly describes the empirical data.

Von Glaserfeld has also said that Social Constructivism is a form of pragmatism.  Thus, not surprisingly pragmatic considerations are brought into his views of science education. For example, he says that a student’s conceptual scheme is not a representation of the real world but is adaptive or viable to the subject’s experience and that science educators should help students to develop such viable schemes.[12] Although I am not entirely sure what von Glaserfeld means here, Constructive Empiricism can give a plausible account of the pragmatic element without falling into the trap of subjectivism. Presumably van Fraassen would say that von Glaserfeld is correct that scientific theories do not represent the real world, in the sense that the reality of their postulated unobservable entities is irrelevant.  He would say they should be judged on their viability in that they accurately describe the empirical evidence.

Of course, such empirically accurate theories have pragmatic implications that are not developed by van Fraassen; in particular, they help one to cope with and get around in the world. Pragmatic considerations also enter the picture directly in regard to choosing between different equally viable theories. Different theories will be evaluated differently in terms of criteria such as simplicity, explanatory power, and scope.  The theory one ultimately chooses will depend on the relative strength of these criteria in different contexts.

Let me give you a concrete example of how Constructive Empiricism might be used in science education. Some science educators have developed “black box” apparatus– variously called hypothesis machines or theory boxes– as classroom tools. An apparatus is presented to the class, but its inner mechanism is hidden. Something is put in the apparatus (the input) and something comes out of it (the output). Students are asked to guess the hidden mechanism that produces the output, given the input. For example, in one apparatus there is a series of parallel transparent tubes. However, the middle section of tubes is covered. When marbles are rolled down a tube they disappear behind the covered section; sometimes they appear again in the same tube on the other side of the covered section, sometimes they appear in a different tube, and sometimes they do not appear at all.  Students are asked to guess what hidden mechanism in the covered section has caused the observed behavior of the marbles.

The inventors of the apparatus argue that it teaches students “indirect observation.”[13] If one pretends that the hidden mechanism of the apparatus is unobservable in principle, it can be argued alternately that this apparatus teaches the basic principles of Constructive Empiricism.  On a Constructive Empiricist interpretation of the black box apparatus, the main job of the students would be to achieve empirical adequacy; that is, to describe accurately the input and the output of the machine. For example, they would describe under what conditions a marble does not appear at all or, if it does appear. They would specify if it appears at random times or if there is some discernible pattern. Student hypotheses about the box’s hidden mechanism might be useful in discovering this relation.  But, under the supposition that the hidden mechanism is unobservable in principle, the truth of such a hypothesis is irrelevant. Rather their hypotheses of unobservable mechanisms would be judged in terms of pragmatic criteria. Thus, the students would not only be learning to achieve empirical adequacy but would also be learning to develop fruitful hypotheses about hidden mechanisms.

Naturally the question remains of how much that is learned under a Constructive Empiricist use of the black box apparatus is transferable to actual scientific practice. However, this question pertains to all science education no matter what approach one takes. Whether science educators take a Social Constructivists or Scientific Realist approach, they must be concerned about whether or not their teachings are applied by their students once they leave the classroom and actually start practicing science. This is where research in science education is useful.  Education researchers could explore the question of whether students who learn the Constructive Empiricism approach using a hypotheses machine like the one described above tend to use this approach in scientific laboratories and beyond.



I have suggested some of the advantages of van Fraassen’s Constructive Empiricism over Social Constructivism and Scientific Realism both in regard to the philosophy of science and science education. However, I do not want to be perceived as suggesting that this approach is free from problems or even that I advocate it. In the philosophy of science critics have raised questions about this approach.[14]






[1]See Bas Van Fraassen, The Scientific Image (Oxford: Clarendon Press, 1980).


[2]  The Scientific Image, pp. 87-89.


[3] See for example, Richard Boyd, “On the Current Status of Scientific Realism”, in Richard Boyd, Philip Glasper, and J.D. Trout, The Philosophy of Science, (Cambridge: The MIT Press, 1991), pp. 195-222.


[4] See, for example, Ernest Nagel, The Structure of Science, (New York: Harcourt, Brace, and World, Inc. 1961), pp. 129-140.


[5]See, for example, Steve Woolgar, Science: The Very Idea, London: Routledge, 1988), Bruno Latour and Steve Woolgar, Laboratory Life (Princeton, NJ: Princeton University Press, 1986).


[6]See for example, Robert Nola, “Constructivism in Science and Science Education,” Constructivism in Science Education, ed. Michael Matthews, (Dordrecht, Netherlands; Kluwer Academic Publishers), pp. 31-60.


[7] See Helge Kragh, “Social Constructivism, the Gospel of Science, and The Teaching of Physics,” Constructivism in Science Education, ed. Matthews, p. 130.


[8] See Larry Laudan, “A Confutation of Convergent Realism,” Philosophy of Science, ed. Boyd, pp. 223-246.


[9]Kragh, “Social Constructivism, the Gospel of Science, and The Teaching of Physics,” p. 135.


[10] See Michael Matthews, “Introductory Comments on Philosophy and Constructivism in Science Education,” Constructivism in Science Education, ed. Matthews, pp. 1-10.


[11]Ernst von Glaserfeld, “Cognition, Construction of Knowledge, and Teaching,” Constructivism in Science Education, ed. Matthews, p. 127.


[12] Ibid., pp. 14-16.


[13]See my discussion in Michael Martin, Concepts of Science Education (Chicago: Scott-Foresman, 1972), pp. 16-17.


[14]See for example, Eran McMulin, “A Case for Scientific Realism,” Scientific Realism, ed., Jarrett Leplin (Berkeley: University of California Press, 1984), pp. 19 -21; Philip P. Hansen and Edwin Levy, review of Van Fraassen, The Scientific Image, in Philosophy of Science, 49, 1982, pp. 290-3; Paul M. Churchland, “The Anti-Realist Epistemology of van Fraassen’s The Scientific Image,” Pacific Philosophical Quarterly, 63, 1982, pp. 226-235.