The British Journal for the Philosophy of Science VOLUME VI MAY, 1955 N o . 21 STRIFE ABOUT COMPLEMENTARITY (I)* MARIO BUNGE UNTIL a few years ago only a few physicists questioned the usual interpretation of quantum theory. Their criticisms were doubtless useful, but remained mainly on the philosophical level; no consistent alternative interpretation of the successful mathematical formalism was offered. Now the situation has altered substantially : several realistic, rational, and deterministic interpretations of the same formalism have been advanced. As was to be expected, they are strongly opposed by the upholders of the official philosophy of quantum theory, which is essentially of a positivistic character. The purpose of the present paper is to examine a recent manifestation of this conservative stand- point, namely, the article in which Professor L. Rosenfeld 1 of Man- chester, who is Bohr's best known pupil, criticises the new realistic, rationalistic, and deterministic trends. 1 What is Complementary to What ? The doctrine of complementarity is an interpretation of Heisen- bcrg s uncertainty relations. In the case of mechanical systems, the latter state that it is impossible to know simultaneously, with an arbitrary accuracy, the values of any two conjugate variables, such as the position and the momentum of an electron ; in the case of a radiation field, the uncertainty relations consist of similar statements regarding the electric and the magnetic field strengths. The doctrine of complementarity, far from interpreting such mathematical relations in terms of errors of measurements of objectively existent attributes * Received 16.ii.54 1 L. Rosenfeld,' Strife about Complementarity' (referred to below as SC), Science Progress, July 1953, No. 163, 393, being a revised version of' L'evidence de la com- ple'mcntarite'', in Andre" George (Ed.), Louis de Broglie, physicien etpenseur, Paris, 1953 MARIO BUNGE (as is commonly believed), claims that it is meaningless to ascribe simultaneously an objective position and an objective momentum to an electron, or all of its components to a radiation field. Conjugate quantities were called by Bohr complementary to each other, in the sense that they are (a) both mutually exclusive, since the sharpening in the value of one of them results in a larger uncertainty regarding the complementary quantity ; and (b) both needed to achieve a complete description of experimental results, which the present form of the quantum theory is assumed to yield, at least in the atomic realm. Owing to the fact that complementary aspects are mutually exclusive, it is impossible—thus Bohr argues—to afford a single well- defined picture of atomic phenomena, being on the other hand indis- pensable to split the image of reality into two complementary models, or pictures, which can be applied in succession, never simultaneously in all rigour, and this simply because the aspects covered by each model are not simultaneously observed. In the particular case of entities endowed with mass (such as electrons), one group of variables (position and time) describes the corpuscular aspect, while the group of quantities complementary to these (momentum and energy re- spectively) describes—as can be seen by recalling de Broglie's relation between momentum and wave-length, and Planck-Einstein's relation between energy and frequency—the wave aspect. In this regard, the contention of the doctrine of complementarity is, that microsystems endowed with mass are neither particles, nor waves, nor wavicles, but that they simply are not in themselves, for nothing is supposed to exist apart from the means of observation. Hence, according to com- plementarity, the words ' particle' and ' wave' designate neither material objects nor properties of material objects ; they have no ontological status, but solely an empirical one, for they arc only entities altering the description of certain experiments. Most people believe that the doctrine of complementarity merely expresses the obvious fact that we alter nature whenever we act in order to know i t ; in other words, that when we perform a measure- ment we establish an interaction between a piece of apparatus and the object under consideration whereby we unavoidably disturb the latter. This is a valid interpretation of Heisenberg's uncertainty relations, which the doctrine of complementarity attempts to interpret ; but that conception is contradictory to the doctrine of complementarity, which is not centred in things that are to be observed and that exist STRIFE ABOUT COMPLEMENTARITY before and after the acts of observation, but which is centred in observa- tions—because, it is argued, it would be ' metaphysical' to assume that there is something beyond observational data. It is not merely that the doctrine of complementarity stresses the doubtless active rdle of the experimenter, the active side of knowledge ; it goes beyond this, asserting that observations are the alpha and the omega of knowledge, that there is nothing which is being observed, nothing beyond observation itself. Bohr has carefully and untiringly explained, for almost a quarter of a century, that we cannot attribute an autonomous physical reality (Le. a reality independent of the experimenter) to objects at the atomic scale.1 Philipp Frank, an authorised spokesman of the same philo- sophical trend, has elucidated this point with his usual clarity, explaining that what we call electron is not a bit of matter but a set of symbols : ' The " electron " is a set of physical quantities which we introduce to state a system of principles from which we can logically derive the pointer readings on the instruments of measurements.' * Of course, the same is deemed to be valid for the qualities of things ; thus for instance the momentum of an electron ' has never existed except in so far as we have a set-up which allows the definition of a " momentum".'* Things and qualities of things are said to exist only in so far as features cf experimental set-ups and acts of observations in themselves. Now that we are clear about the operational meaning of the concept of reality, we are in a position to understand what is complementary to what. According to Bohr * two things, experimental set-ups, and their corresponding descriptions can all be complementary. When we have an experimental set-up for determining (' defining', in the positivistic jargon) one attribute, we destroy the possibility of setting 1 N. Bohr, La thiorie ntomique et la description des phtnomhies (referred to below as TA), transl. by A. Lcgros and L. Rosenfeld, Paris, Gauthier-Villars, 1932, p. 51 ; ' Licht und Lcben' (referred to below as LL), Die Naturwissenscluften, 1933, 21 245-50, p. 247 (see also ' Light and Life ', Nature, 1933, 131, 422, 457) ; ' Kausalitat und Komplementaritat' (referred to below as KK), Erkenntnb, 1936, 6, 293-303, p. 295 ; ' Le problcme causal en physique atomiquc ' (referred to below as PCPA), in the collective volume Les noiivellfs thiories de la physique, Paris, 1939, 11-32, p. 25 ; ' Newton's principles and modem atomic mechanics ' (referred to below as NP), in the collective volume ed. by the Royal Society, Neii'toii Tercemiitary Cele- brations, Cambridge, 1947, 56-61, p. 59. 1 Philipp Frank, Foundations of Physics (referred to below as FP), in International Encyclopedia of Unified Science, I , No. 7, Chicago, 1946, p. 54 3 Frank, FP, p. 55 ' 4 llohr, KK MARIO BUNGE up the ' complementary' arrangement which would allow us to determine its ' conjugate ' attribute. Notice once again that it is not the numerical value of the attribute that is changed by the act of its measurement—since this would entail that it had a value before its measurement. In all this we have neither atomic objects nor their attributes considered as things-in-themselves :. complementarists avowedly do not make statements about the real world, they maintain that quantum mechanics does not speak of real objects that are observed but only of experimental arrange- ments.1 On this purely epistemological ground, complementarists have criticised two very common notions. According to one of these, ' Heisenberg's relations say that it is impossible to measure simul- taneously the position and the velocity of an electron '. This is wrong, explains Bohr, because it implies that position and velocity are well- defined attributes of the object, whereas the point is just that we are forced to give up the notion o f autonomous attributes of the object' (selbstandige Attribute da Objektes).* The second popular notion criticised by the complementarists is that ' The electron has no simul- taneously determined velocity and position, these being actually indeterminate'. This interpretation is wrong, says Frank, because it assumes that there is something (the electron with indeterminate properties) that pertains to the real world.3 What is at stake in all this is not the structure of micro-objects, but the whole theory of knowledge with its old struggle between material- ism and immaterialism : complementarity is not a physical but a philosophical doctrine, because it does not refer to matter in motion but to concepts and their verbalisations. As Frank says so amusingly, ' All the confusion is produced by speaking of an object instead of the way in which some words are used '.* This fact, that the doctrine of complementarity is of a philosophical, not of a scientific nature, is not willingly accepted by most complementarist physicists who, like Rosenfeld, regard it as ' the most direct expression of a fact \ 6 But what positivist physicists fail to see is granted by positivist philosophers. Thus Reichenbach in one of his last books wrote : 1 Philipp Frank, ' Philosophischc Dcutungen und Missdeutungen dcr Quanten- theoric* (referred to below as PDM), Erkenntnis, 1936, 6,303-317, p. 308. Sec also Interpretations and Misinterpretations of Modem Physics, Paris, 1938 1 Bohr, KK, p. 297 » Frank, PDM, p. 308 « Frank, FP, p. 55 5 Rosenfeld, SC, p. 396 STRIFE A B O U T C O M P L E M E N T A R I T Y The duality of interpretations thus assumed its final form : the and of de Broglie's discovery does not have the direct meaning that both waves and corpuscles exist at the same time, but has the indirect meaning that the same physical reality admits of two possible inter- pretations, each of which is as true as the other, although the two cannot be combined into one picture. The logician would say : the and is not in the language of physics, but in the metalanguage, that is, in a language which speaks about the language of physics. Or, in another terminology, the and belongs, not in physics, but in the philosophy of physics ; it does not refer to physical objects, but to possible descriptions of physical objects, and thus falls into the realm of the philosopher.1 T h e philosophical nature of all this debate will become more apparent w h e n going over to its central problem, which is also the central problem of philosophy, viz. the question of the relation of subject and object. 2 Esse est Percipi In order to be classified as an idealist one does not need to speak the whole day long about the spirit, or to maintain that life is a dream ; it is enough to maintain that nothing exists or appears by itself, autonomously, independently from some mind. Berkeley explained it long ago in his straightforward way : The table I write on, I say, exists, that is, I see and feel i t ; and if I were out of my study I should say it existed, meaning thereby that if I was in my study I might perceive it, or that some other spirit actually does perceive i t ; [but] as to what is said of the absolute existence of unthinking things widiout any relation to their being perceived, that seems perfectly unintelligible. Their esse is percipi [their being is to be perceived], nor is it possible they should have any existence, out of the minds or thinking things which perceive them.2 Nowadays it is hard to maintain such a subjective idealism in ordinary life ; it is easier to maintain it for a domain accessible only to the specialist—for instance, atomic physics. Thus, w e often find the amusing spectacle that subjective idealism is asserted with regard to microscopic events, whereas some sort of materialism is retained 1 Hans Reichenbach, The Rise of Scientific Pliilosophy, Berkeley and Los Angeles, 1951, pp. 175-176 * Berkeley, A Treatise concerning t\ie Principles of Human Knowledge, § 3 5 MARIO BUNGE for the macroscopic level. The following is an example of this epistemological dualism : In classical physics it is possible to establish a sharp distinction between the system investigated and the means of observation, and therefore to ignore die latter in framing our conception of die pheno- menon. The existence of die quantum of action makes such a dis- tinction impossible because it imposes a limit upon die analysis of die interaction between die system and the apparatus which fixes die circumstances in which we observe it. It is diercfore die indivisible whole formed by die system and die instruments of observation which now defines the ' phenomenon \ x Bohr has sometimes adopted consistently the idealist point of view, extending it to the macroscopic level. He has argued that, since every observation entails a finite interaction with the instrument, ' one cannot attribute to the phenomena nor to the instrument of observation an autonomous physical reality in the ordinary sense of the word '.* He went so far as to approve Heisenberg's remark t h a t ' ordinary (Le. macroscopic) phenomena are in a way engendered by repeated observations '.* But usually he attributes validity to idealism at the atomic level only, a favourite statement being that in the analysis of quantum effects we are faced with the impossibility ' of drawing any sharp separation between an independent behaviour of atomic objects and their interaction with the measuring instruments which serve to define the conditions under which the phenomena occur '.* The central point is thus the negation of the autonomous existence of atomic objects.' Since atomism, that stronghold of traditional materialism, could no longer be rejected (as it was in Mach's days), it became advisable to denaturalise i t : atoms are at last granted a right to existence, but only on the ideal plane, only as ' auxiliary concepts '.• Once materialism has been disposed of, it is easy to dispense with the notion that everything comes from somedung else, that is, with causality. Bohr explained clearly, for once, that the rejection of causality was only a consequence of the rejection of materialism: 1 Rosenfeld, SC, p. 395 * Bohr, TA, p. 51 * Bohr, TA, p. 64 * Bohr, ' Discussion with Einstein on epistemological problems in atomic physics ' (referred to below as DE), in P. A. Schilpp (Ed.), Albert Einstein : Philo- sopher-Scientist, Evansron, 111., 1949, p. 218. See also NP, p. 59 8 Bohr, LL, p. 247 and KK, pp. 294-296 • cf. Mario Bungc, ' Mach y la teorfa atcmica ', Bcletlu del Qui'tnico Peniauc, 1951, N o . 16, 12-16 STRIFE ABOUT COMPLEMENTARITY We have been forced to forego the ideal of causality in atomic physics solely because, as a consequence of the unavoidable interaction between die object of experiment and the measurement instruments— an interaction which cannot be corrected for if these instruments are to allow die unambiguous application of the concepts that are needed for the description of the experiments—we cannot speak any longer of an autonomous behaviour of the physical object.1 Thus, we see clearly that the celebrated crisis of causality is nothing but a consequence of the adoption of an idealist theory of knowledge : it is not a simple result of modern physics, it is a tenet of modern positivism. 3 Sozein ta Phaiuomena The most important point in this controversy is that most scientists, at least when they are doing research, share the materialistic principle of the objective existence of a gradually knowable thing-in-itself, whereas positivism maintains that there is no such ' hidden' reality behind appearance, since the object is exhausted by its perception (nowadays by its measurement). This posirivistic axiom is very old, but in modern times it was first clearly stated by Berkeley,* who maintained that everything is such as it appears to be, there being no such contrast between appearance and reality, for everything is appearance. Hence the methodological prescription : sozein ta phainomena, sctluare . apparmtias, to give account of phenomena (appearances).' This phenomenalist attitude, typical of positivism since Comte, has been adopted by the upholders of the official philosophy of quan- tum mechanics, one of whose best representatives, Heisenberg, has explicitly stated that the quantum theory does not assume the existence of a Ding an sich behind the phenomena (or appearances).4 In a more technical language this is expressed in the principle of observables, according to which physics, or at any rate atomic physics, is only concerned with observable properties—meaning the actually observed ones, with exclusion of all sorts of ' hidden parameters'. Thus, 1 Bohr, KK, p. 298 * Berkeley, op. cic, §§ 87, 88 3 For die early history of this rule, sec Pierre Duhem, Z&ZEIN TA