Metaphysical Perspectives and Their Role in Science

1. Part I: The Conflation of Metaphysics and Pseudo-Science

The term “metaphysics” usually designates prima philosophia (or philosophia prima). It is, of course, hopelessly metaphysical, for it presupposes a very general hierarchy of abstractions and an accompanying hierarchy of fields of inquiry, rendering some theories primary and others secondary. As is well known, traditional epistemology was unrealistically optimistic: it regarded metaphysical speculations as provable. This meant that the hierarchy itself was, thereby, somehow, in need of proof. The situation is circular beyond repair: we seek to lift our prima philosophia, and the metaphysical fulcrum on which we placed its claim for primacy. Alternatively, we are being led by the logic of our situation down an infinite staircase of justifications: it is metaphysical turtles all the way down.

Since Popper was the first serious non-justificationist philosopher of science, it was reasonable to expect that his system would instantly improve the hopeless status of traditional metaphysics. For in his system metaphysical frameworks were suddenly no longer in need of proof: the empirical and the metaphysical could have been portrayed, then, as parts of a continuum of criticisable and improvable attempts to make sense of our environment. Such a worldview could have easily replaced the traditional futile effort to ground scientific theories on a metaphysical uber-turtle, or a self-lifting fulcrum.

Alas, for the early Popper efforts to purify empirical science from the uncleanliness of metaphysical speculations took precedence. Consider for example the following quote. Please note not only its content, which may be familiar to you, but also its astonishingly positivistic vocabulary and style:

“Any form of empiricism must, above all, demand from

the theory of knowledge that it secures empirical science

against the claims of metaphysics. The theory of

knowledge must establish a strict and universally; applicable criterion that allows us to distinguish between

the statements of the empirical sciences and

metaphysical assertions (“criterion of demarcation”)…

…in case of doubt, how can one decide whether one

is dealing with a scientific statement or “merely” with a

metaphysical assertion? (Or more simply, when is

science not science?)

Popper 1934 2008, p 72. [italics are mine]

In fairness to Popper it should be noted that in the context of the rise of a Heidegger-worshiping philosophy, a rule emphasizing testability must have seemed a breath of fresh air. But it implied a sort of purification rite. Securing empirical science against the irrefutable became so central to Popper’s project, that only in his later life did he turn his attention to the importance of the dialogue between metaphysical systems and science. Instead, the early Popper came to define “the empirical” and “the metaphysical” as conceptual antipodes (although clearly the realm of non-empirical theories included other fields, such as logic, pure mathematics, philosophy and the rest of the theories that Popper placed in the curious wastebasket “pseudoscience”). Here is an example:

…we call non-empirical statements about reality

“metaphysical”. According to this definition, all

non-empirical assertions that state something about

existing, real objects are metaphysical; so too are

all non-empirical assertions that lay claim to being

empirical, thus violating the boundary drawn by

the criterion of demarcation.

ibid. p. 681

In this paper I argue that although Popper’s views regarding metaphysical frameworks developed considerably from the 50s onwards, he never openly gave up some positivistic readings of his demarcation criterion. Consequently, his message about the scientific service of metaphysical frameworks remained unclear and even became increasingly confusing.

Let me quickly mention two weaknesses of Popper’s criterion of demarcation, before diagnosing the reason of its weakness, exploring ways by which it can be improved and, finally, questioning, with Agassi, the wisdom and desirability of doing so.

Popper’s criterion of demarcation blurs the distinction between metaphysical systems that are valuable to science and worthless metaphysical systems which hinder scientific progress. The demarcation criterion is therefore simply too narrow, notes Agassi (Agassi 1991, 3), as it purports to exclude all metaphysical frameworks by declaring them all essentially indistinguishable from pseudoscience. Darwin’s theory of natural selection is perhaps the most thoroughly discussed problematic case: Popper was candidly ambivalent towards its status, and openly critical towards his own unsuccessful attempts to solve the predicament. Clearly Popper did not wish to throw natural selection into the pseudoscience wastebasket. But, strictly speaking, he lacked a theoretical ground for his refusal, one that would be consistent with his demarcation criterion. Sometimes Popper referred to Darwin’s theory, correctly I think, as “little short of tautological” (e.g. Popper 1974, 69). For it states, roughly, that “Survivors survive due to being well adapted”, where “well adapted” essentially means “survived long enough to bring fertile offspring”). Of course, in light of Darwin’s principle we can seek to account for biological environments as comprising cases of survival advantages of organisms over their competitors. But this does not improve the logical status of the principle: being a presupposition to all our biological explanations, it guides research without being empirically testable. So long as it is both scientific and untestable, it refutes the demarcation criterion.

Popper’s demarcation criterion is also too wide, notes Agassi (ibid.): it renders quite a lot of refutable superstitions and silly generalizations scientific. “Holy water would make you immune to illness”, is an example. Or consider the fact that astrologers make numerous failed predictions every weekend. Some Astrologers aspiring for acceptance by the scientific community even indicate their occasional failures and adjust their theories in an attempt to improve future predictions, notes Agassi. Does this render Astrology scientific by Popper’s own standards? Popper’s demarcation criterion, then, is both too wide and too narrow. And yet it seems to offer us a valuable insight. What has gone wrong here?

It has been suggested, rightly I think (e.g., Jarvie 2001, 175; Bar-Am 2014, 693–696) that Popper conflated two distinct criteria of demarcation. The first, in my opinion, is the result of over-ambition. The second is one of the greatest contributions to 20^th century philosophy. Understandably, then, their conflation held up the advance of critical rationalism.

Popper’s excessively ambitious criterion of demarcation presents itself as a clear cut distinction between scientific sentences or theories, and non-scientific sentences, or theories. Popper rightly denies repeatedly that he is delineating “a language of science”, a-la Carnap, of course, and in this sense his discussion far surpasses that of his peers, but admittedly he often gives the impression that his aim is to delineate a semantically identifiable sub-language: the realm of empirical sentences (see e.g. Popper 1959, 37: “…my business, as I see it, is…to define the concepts of ‘empirical science’ and 'metaphysics' in such a way that we shall be able to say of a given system of statements whether or not its closer study is the concern of empirical science”).

On the other hand, Popper’s profound criterion of demarcation distinguishes between scientific attitudes, or behaviors, and dogmatic attitudes or defensive behaviors (e.g. Popper 1959, 50: “…empirical science should be characterized by…our manner of dealing with scientific systems: by what we do with them and to them”). It defines science as the result of non-defensive learning from our errors.

That Popper did not properly distinguish between these separate criteria is clear to anyone reading his Conjectures and Refutations (Popper 1963, 33–65). The immense popularity of this paper is, perhaps, partly to be explained in terms of its palatable conflation of the abovementioned criteria of demarcation. It also may explain why Popper’s readers easily divide into avid admirers and sneering detractors. For in Conjectures and Refutations Popper famously dismisses Marxism, the Psychologies of Freud and Adler, and Astrology as cases of pseudoscientific theories (ibid., p.37). He therefore appears to be claiming that there is something inherently defective with these theories qua theories (and without even providing rough formulations of them ¹ ): they appear to be informative, he argues, while in fact being irrefutable in principle. This appearance, I maintain, is both false and misleading. Astrology, as I have already explained above is clearly refutable and false. Many of Marx’s famous predictions (e.g., that the conditions of the working class must steadily deteriorate) were also refuted, and Popper was clearly a world authority on such refutations (even today no criticism of Marxism comes close to his masterful critique of it in the Open Society). Freud also noted and corrected many of his failures (catharsis theory, to take one notable example, is disproved by the fact that abreaction of traumatic experiences often fails to cure trauma, he famously observed). Needless to say he took such refutations seriously and attempted correct his theory in their light. Deciding whether or not he did so defensively is not within the scope of this paper.

Dogmatism, then, is rarely if ever in the theory per se. It is rarely if ever in the strictly logical, or semantical makeup of our sentences. There is no such thing as an essentially empirical sentence and no such thing as an essentially metaphysical one. Consider Thales' assertion “All is water”. It is metaphysical if we express ignorance regarding the nature of water. But if water is H2O, then it is empirically refutable and false. Confusingly Popper explained these points better than everyone else: he insisted, for example, that defending an empirical theory dogmatically renders it less empirical and perhaps even non-empirical (Popper 1959, 37). Of course by doing so we change the meaning of its terms, and so its semantics, he added. But to assume that we can prevent this, by demanding that we perfectly understand every empirical implication of a theory, is a semantic fiction (a positivistic one). We must allow quite a lot of vagueness regarding the meaning of the terms that we use: this is partly due to the fact that the same language is used by speakers with different metaphysical perspectives. Popper was well aware of all this, of course, (see e.g. Popper 1945 v.2, 19: “This is why our terms make so little trouble… …We do not take their meaning too seriously”). But it is undeniable, I think, that he is often in two minds about the implications to be drawn from this fact. This at least is my understanding as an admiring reader. For example, sometimes he says that the logical-semantic structure of sentences is what makes them empirical (or non-empirical), and at other times he argues that historically metaphysical theories have turned empirically testable. This last sentence invites the question: is it the same theory that was once untestable and eventually became empirically testable? Perhaps it is merely a case of two different theories expressed by means of the same symbols? If it is the latter, why refer to it as the same theory? If it is the former, then we must abandon the effort to “define the concepts of ‘empirical science’ and 'metaphysics' in such a way that we shall be able to say of a given system of statements whether or not its closer study is the concern of empirical science” (ibid., p.37).

Popper could have improved things tremendously, I think, by emphasizing that the logical-semantical criterion of demarcation was simply a catchy but inaccurate attempt to clarify a fundamental socio-psychological observation: science rests on the institutional encouragement of openenss to criticism. It may be speculated that Popper avoided this crucial clarification (best articulated in Jarvie 2001) because he regarded a socio-psychological demarcation of science as a lesser philosophical achievement than a logical-semantical one. Indeed, open admission here would amount to clarifying that the demarcation line between metaphysical theories and empirical ones is not always as clear as he wanted it to be.

Note also how Popper’s second criterion, the socio-psychological demarcation of science, easily handles the criticisms directed against the first criterion. For example it does not condemn Astrology as a pseudoscience. Rather it declares it to be a tragically defensive project which conflates science and magic. And it does not forbid metaphysical frameworks from entering science: they are welcome there, so long as they encourage the process of learning from our errors.

Finally, Popper’s second demarcation criterion has another often neglected merit: it is humble and humbling. Instead of dictating to the world a binary division between good science and bad pseudoscience, it presupposes as a matter of course that all humans have their defensive and less defensive moments, indeed that rationality is an ideal, while encouraging us to kindle our openness to criticism whenever we are capable of doing so. This, in my mind, is the true value of Popper’s philosophy. And it is this lesson that I find Agassi distilling, clarifying and elaborating.

As is well known, from the 50’s onwards Popper gradually sensed that there was something both inadequate and lacking in his early dismissal of metaphysical frameworks as unscientific. He therefore gradually but steadily developed the ground-breaking idea that metaphysical frameworks can be rationally discussed and even rationally dismissed or embraced. This new status for metaphysical frameworks became an inseparable part of Popper’s later philosophy. It made it far more advanced than any other philosophy of science at the time. Oddly, however, Popper avoided acknowledging the fact that this must affect his discussions of the demarcation criterion. The result of all this was that even in his later writings when Popper came to the point where he should qualify his criterion of demarcation as mainly psycho-social, he chooses instead to change course and criticize the misconceptions of his critics who wrongfully ascribe to him the positivistic view that the criterion implies the meaninglessness of metaphysical frameworks. In other words, and this is the crucial point: Popper allowed himself some vagueness towards the status of metaphysics under the correct observation that his interlocutors treated it (and him) even worse. He therefore developed the important idea that metaphysical frameworks are historically and psychologically significant for science. Metaphysical frameworks, he explained, are sometimes suggestive of empirical theories. Atomism, for example, was a relevant background to scientific thinking for centuries cradling many testable ideas until becoming testable in 1905, following Einstein’s theory of Brownian motion (Popper 1994, 118). I will return to this observation in the closing paragraph.

Enters Agassi. The first noticeable change that he brings with him to Critical Rationalism is a candid and straightforward attitude regarding the place of metaphysical systems in science. In his writings Metaphysics becomes an inevitable, integral, and openly welcomed layer of every attempt to make sense of reality. It is presupposed (a-la Collingwood) by every question that we can possibly formulate (although contra to Collingwood it does not imply anything like a clear and distinct hierarchy of presuppositions reaching all the way to absolute presuppositions; according to Collingwood, you may recall, absolute presuppositions are not criticisable: Agassi dismisses all this as relativistic nonsense). The demarcation criterion, qua the science club bouncer, then becomes uninteresting simply because we no longer take interest in securing empirical science against anything, let alone against metaphysical frameworks (without which it would be virtually meaningless). Agassi therefore takes it as a matter of course that some perfectly empirical theories are nevertheless expressions of inferior scientific thinking (astrology) whereas some metaphysical systems (e.g. mechanism) are markedly scientific without being empirical. Good science is openly critical, then, but not necessarily empirical.

Agassi’s unapologetic view regarding the place of metaphysics in science invites the question to what extent does it express itself as a direct and clear disagreement with Popper. Before attempting to formulate some such disagreements, I wish to stress how difficult it is to formulate a clear controversy when not all sides express themselves in non-ambiguous terms. As I endeavored to explain above, Popper was unclear concerning the concessions that he made to some metaphysical systems when declaring that they provide important service in science. One striking example is the first section of appendix *X to LSD (Popper 1959, 420–423) where Popper makes a compelling argument in favor of the view that “similarity, and with it, repetition, always presupposes the adoption of a point of view” (ibid., 421). This is the closest that I am aware of to Popper expressing a candidly Agassian view of Metaphysics: science is impossible without a metaphysical cradle. Oddly, however, Popper confuses his readers by avoiding admitting the obvious: the “points of view” which he discusses are clearly metaphysical points of view. Nowhere are they referred as such. At the time Popper wrote this Appendix Agassi was his research assistant. He reported to me (in a conversation) pressing Popper about his curious avoidance from acknowledging the elephant in the room. Popper, he reports, systematically blocked all attempts to discuss his curious avoidance from referring to these “points of view” as “metaphysical”. Then Agassi asked Popper to be excused from further discussing the details of this appendix with him. Popper, he reports, characteristically ignored his request and pressed on, brilliantly elaborating this fantastic appendix. Similarly, Popper carefully and astutely observes elsewhere that “we can never speak of cause and effect in an absolute way, but must say that an event is a cause of another event – its effect—in relation to some universal law.” (Popper 1957, 124) But he does not allow or discuss the question whether our adoption of a metaphysical point of view renders some true universal statements natural laws and some not. Finally, I was trying to determine if theory-ladenness in Popper openly includes metaphysical speculations. It seems to me that it must, for obvious reasons, even though metaphysical speculations are irrefutable in principle ² . Yet I have found no discussion of this crucial question in Popper.

With Agassi, the answer to this last question is a clear affirmative. The most basic service of metaphysical systems, he argues, is nourishing our sense of proportion, which naturally shapes our observations. And metaphysical systems thus tend to direct our attention towards some unrefuted universal statements, suggesting them as interesting candidates for natural laws, whereas it leads us to ignore other such candidates as uninteresting or even silly. (Agassi 1975, 239: “…what looks a natural necessity one generation may look an accidental necessity at the next - I explain this by reference to our changing metaphysical framework”). Popper could have easily adopted this view. Alas, he avoids openly granting metaphysical systems such an active role in determining our scientific data and agenda and so found himself adhering to the bluntly Aristotelian speculation that laws of nature are logically stronger than merely accidental universal statements (Popper 1959, 431–432).

Similar points arise regarding the question of the progress of metaphysics. Popper 1945 still follows Kant in arguing that metaphysical progress is illusory: “…while physics progresses, metaphysics does not… …the obvious objection against speculative metaphysical systems is that the progress they claim seems to be just as imaginary as anything else about them” (Popper 1945 V.2, 247–8). Later on Popper qualified this contention, of course, but his notion of metaphysical progress remained unclear. Agassi observes a two-directional critical dialogue between science and metaphysics (Agassi 1975, 239: “whereas metaphysics guides us in what we may consider a priori a plausible putative natural law, science guides us in criticizing our metaphysics and replacing it by a better one.” This observation renders metaphysics a markedly progressive endeavor: science places ever stricter demands on traditional metaphysical frameworks thus sharpening them, driving their gradual improvement.

As a concluding example let me mention here the case of Atomism. Popper (Popper 1994, 118) argues that it was a strictly metaphysical framework until 1905, when it suddenly received a clear-cut empirical interpretation. This view is too binary, I feel. Since Popper was extremely knowledgeable in the history of science such over-simplicity should perhaps be attributed to his over-occupation with the demarcation criterion and perhaps also, to his lack of interest in elaborating the role of metaphysical perspectives in science. Agassi is right to point out that such a view does little justice to the history of science. Dalton’s atomism, to take just one example, includes the clearly empirical novel notion of an exact weight of atomic compounds. It is a huge improvement upon the original Democritean system (which, incidentally, also includes quasi-empirical conjectures, e.g., the claim that round molecules are sweet and sharp-edged ones hot, although of course testing this contention was not feasible at the time…). Dalton’s improvements can only be attributed to his scientific new problem situations (e.g., his search for an explanation of the law of multiple proportions). Scientific metaphysical theories are therefore in constant progress in virtue of their dialogue with increasingly improving empirical ones.

To sum up: The later Popper clearly and rightly sought to temper his early positivistic views regarding metaphysical frameworks. He did so in a world that was still groping about positivistic notions of science. It was an impressive step forward. He allowed metaphysical systems to contribute to science while remaining sometimes unclear and sometimes frustratingly silent about the scope of their contribution: he never openly acknowledges that without metaphysical cradles, no word has meaning and no observation report can be made. With Agassi metaphysical systems receive a clear and candid critical rationalist role: they become an integral part of our non-justificationist process of learning from our errors. They shape our observations, suggest our problem-situations, and imply their significance. Popper’s demarcation criterion, he therefore argues, not only fails its intended task, but, more importantly, loses its appeal as a worthy philosophical goal.