Mind in Nature: the Interface of Science and Philosophy by John B. and David R. Griffin Cobb, Jr.
John B. Cobb, Jr. is Professor of Theology at the School of Theology at Claremont, Avery professor of Religion at Claremont Graduate School, and Director of the Center for Process Studies. David Ray Griffin teaches Philosophy of religion at the School of theology at Claremont and Claremont Graduate School and is Executive Director of the Center for Process Studies. Published by University Press of America, 1977. This book was prepared for Religion Online by Ted and Winnie Brock.
Chapter 2: Whitehead and the Philosophy of Science by Ann Plamondon and Response by Bernard Rensch
Ann Plamondon teaches philosophy at Loyola University in New Orleans.
The alternative tradition has been referred to as historical relativism (Popper 1970; Hesse 1974, Introduction). It is characterized by the rejection of a logic of science. The members of this tradition2 base their discussion of fundamental methodological issues on arguments from historical examples. That is, with respect to those issues for which the formalist tradition describes a logic, these philosophers seem to deny the existence of a logical structure which transcends particular historical examples.
Many philosophers of science see merit in both traditions and cannot be placed within either. However, among these thinkers, few have seen importance in explicating a general logic of science. One philosopher who has insisted on the value of such a logic is Mary Hesse. In a recent work (Hesse 1974, pp. 6-7) she has defended the attempt to explicate a logic of science on the grounds that its function is three-fold: Such a logic (i) provides criteria for ‘good science’ and is thereby normative as well as descriptive; (ii) as normative, it can show the aim of methodology and the adequacy of methodologies in terms of fulfilling that aim; and (iii) it has as its principal aim understanding, not the suggestion of research techniques.
I am in agreement with Hesse in regard to the value and the function of a logic of science. At the same time, I wish to maintain that a logic of science cannot perform these roles unless it has itself received grounding in a more general, viz., metaphysical, theory. I do not intend to attempt a lengthy defense of this thesis. I hope that it will be sufficient to note that an ‘understanding’ which falls short of understanding on the highest level of generality must be interpreted and criticized by this highest level. Such interpretation and criticism are necessary to show the interconnection of concepts on the lower level of ‘understanding’ and to show that these concepts do not involve incompatible presuppositions. In brief, I am suggesting that metaphysics has an essential role in the philosophy of science -- that of the understanding and the grounding of scientific concepts and methodology. That is, the fundamental concepts of a metaphysical system should give an analysis of the foundational concepts of the sciences in such a way that these concepts themselves provide a grounding -- a general logic -- of the methodology of the sciences.
To perform this task a metaphysics must itself be well elaborated and well criticized. Contemporary philosophy of science is in need of a metaphysics which has already met certain logical (consistency and coherence) and empirical (applicability) criteria. This paper takes as a premise that at present the metaphysics which has best met these criteria is a process metaphysics based on the system of A. N. Whitehead. The aim of this paper is to develop a logic of science from certain concepts elaborated in Whitehead’s metaphysics. In what follows I refer to this logic of science as the ‘process view.’ I do not mean to suggest that all process philosophers would or should be in agreement with it. It is based on an interpretation of Whitehead. The term ‘process view’ was chosen because the concept of organism is central in my interpretation; Whitehead called his process philosophy the ‘philosophy of organism.’
I find the organizing principles of this process view in its basic agreement with the formalist understanding of the doctrine of emergence. An elucidation of the agreement here, however, involves concepts which provide a context for an alternative which challenges the formalist position with respect to the logic of other key issues in the philosophy of science, viz., with respect to the nature and status of laws, the role and justification of induction, the model of theoretical explanation, and the intelligibility of conceptual change.
In this paper I shall discuss the aspects of agreement in the formalist and process views of emergence, suggest the fundamental categories for philosophy of science which result, develop the process view based on these categories, and contrast this process view with the formalist view of laws, induction, explanation, and conceptual change.
According to the formalists, emergence, insofar as this doctrine has validity, is a thesis about unpredictability: (i) the unpredictability in hierarchically structured wholes of properties at higher levels of organization from properties of lower levels of organization and (ii) the unpredictability in evolutionary development of later forms of organization from earlier ones (Nagel 1961, pp. 366-380). These two aspects of emergence usually receive independent treatment. For the process view, there is an important relationship between them. This relationship can best be considered after a discussion of the first-mentioned thesis.
Predictability and unpredictability refer to the possibility of deducing one set of statements from another. The statements which constitute the premises of the deduction are those describing the ‘parts’ of a ‘whole.’ The question is whether or not the properties of the parts (which differ from the properties of the whole they organize), in the specific relationship which organizes the whole, suffice for the prediction of the properties of that whole.
The formalist position is that there is unpredictability in at least two important senses. First, the deduction in question is not possible when the premises are constituted merely of statements about the properties of the parts in that relationship. What must be added are premises elaborating a theory which describes the behavior of those parts in forming wholes. That is, from a statement of a theory describing generally the potentiality of parts to organize wholes and the statement of the condition of a particular organizing relation, a statement describing a whole with particular properties can be deduced. Second, the conclusion as to the properties of the whole may follow from one theory about the potential behavior of parts and the organizing relation in question and yet may not follow from another theory (and the same organizing relation).
At first glance the process view of emergence does not appear to be primarily a thesis about predictability. Rather it seems to be an empirical theory about the modification of parts in forming a whole, or, more precisely, about the reciprocal relationships in wholes between the parts and the plan of the whole. Whitehead expressed this reciprocity of part and whole most clearly in Science and the Modern World.
The parts of the body are really portions of the environment of the total bodily event, but so related that their mutual aspects, each in the other, are peculiarly effective in modifying the pattern of either. This arises from the intimate character of the relation of whole to part. The body is a portion of the environment for the part, and the part is a portion of the environment for the body; only they are peculiarly sensitive, each to the modification of the other. This sensitiveness is so arranged that the part adjusts itself to preserve the stability of the pattern of the body (Whitehead 1925, p. 214).
The concrete enduring entities are organisms, so that the plan of the whole influences the very characters of the subordinate organisms which enter into it. In the case of an animal, the mental states enter into the plan of the total organism and thus modify the plans of the successive subordinate organisms until the ultimate smallest organisms, such as electrons, are reached. Thus an electron within a living body is different from an electron outside it by reason of the plan of the body. The electron blindly runs either within or without the body; but it runs within the body in accordance with its character within the body; that is to say, in accordance with the general plan of the body, and this plan includes the mental state. But the principle of modification is perfectly general throughout nature, and represents no property peculiar to living bodies (ibid., pp. 115-116).
It seems possible that there may be physical laws expressing the modification of the ultimate basic organisms when they form part of higher organisms with adequate compactness of pattern. It would, however, be entirely in consonance with the empirically observed actions of environments, if the direct effects of aspects as between the whole body and its parts were negligible. We should expect transmission. In this way the modification of total pattern would transmit itself by means of a series of modifications of a descending series of parts, so that finally the modification of the cell changes its aspect in the molecule, thus effecting a corresponding alteration in the molecule -- or in some subtler entity. Thus the question for physiology is the question of the physics of molecules in cells of different characters (ibid., pp. 215-216).
There is an important sense, however, in which these passages about modification are referring to the possibility of prediction. Since there is modification of parts to form wholes, the very possibility of explaining or predicting the properties of the whole by those of the parts depends upon a theory about the potential behavior of parts forming wholes, as the formalist interpretation suggests. Further, and still in accord with the formalist position, some theories will be inadequate to the explanation or prediction.
On the other hand, the formalist interpretation affirms the empirical situation emphasized by the process interpretation and lays down the logical structure for explanation and prediction required by this empirical situation.
It would seem, then, that there is a wide area of agreement between formalism and process philosophy about the doctrine of emergence. The most important points of agreement can be listed as follows:
1. Hierarchically structured wholes are not aggregates, because theories about the parts of such wholes (taken individually) do not suffice to deduce their relationship in wholes. Wider theories are required for the deduction; these theories affirm the potentiality of parts to behave differently in and out of the various wholes they can organize.
2. Understanding of a hierarchically structured whole can be achieved by an understanding of its parts only when there is a theory explaining the relations of the parts in wholes. Such understanding is possible because the theory involves statements about the potential relationships of parts in the various wholes they organize. Then the behavior of a part in a given whole can be deduced from its full range of possible behaviors.
3. There is reciprocity in the determination of parts and whole; the parts determine the whole in the sense that the whole is an organization of the parts (the parts in interrelationship), but there is modification of parts according to the wholes they organize and thereby determination of the parts by the whole.
What is important in this agreement of formalism and process philosophy on the doctrine of emergence is that certain presuppositions become obvious -- presuppositions not explicit in a formalist philosophy of science, and indeed, not compatible with its anti-metaphysical stance. The chief of these is the internal relatedness of part and whole. The possibility of deduction (and thereby prediction) set out by formalism has taken account of the theses that parts have potential to act differently in different wholes, that parts are modified according to the wholes they organize, and that a complete description of parts involves reference to this modification. These theses entail that the interrelationship of parts is not distinguishable from their nature. Their possibilities for relationship are constitutive of their nature.
The internal relatedness of part and whole is not reflected in the formalist discussion of laws, induction, explanation, and conceptual change. Hence a logic of science structured on this internal relatedness will constitute a genuine alternative to the formalist logic of science.
In discussing the reciprocity of part and whole, Whitehead used the suggestive terms ‘organism’ and ‘environment.’ ("The body [whole] is a portion of the environment for the part, and the part is a portion of the environment for the body"; "The concrete enduring entities are organisms, so that the plan of the whole influences the very characters of the subordinate organisms which enter into it.") His usage here is clearly a stretching of the ordinary usage of these terms. I wish to suggest that an elucidation of these stretched meanings provides a framework for the discussion of the fundamental issues in the philosophy of science mentioned above.
Whitehead used the term ‘organism’ to refer to physical as well as biological entities. "Science is taking on a new aspect which is neither purely physical nor purely biological. It is becoming the study of organisms. Biology is the study of larger organisms; whereas physics is the study of the smaller organisms" (Whitehead 1925, p. 150). 1 think that there is sufficient evidence that the primary reference of this term is to wholes which are not aggregates. Such wholes, as we have seen, are reciprocally related to their parts. On this interpretation, ‘organism’ can refer to atoms, molecules, even ecosystems; for in each case, there is modification of the parts in accordance with the pattern of the more complex hierarchically structured whole.
This understanding of ‘organism’ makes clear the relatedness of emergence of properties of hierarchically structured wholes and emergence in evolution. Leclerc (this volume) has clearly shown that stability of hierarchically structured wholes (of a low or high level of complexity) is incompatible with the understanding of such wholes as aggregates (see also Leclerc 1972, Chs. 23 and 24; Bohm 1969; Bronowski 1970). This very stability, in turn, is a necessary condition for emergence in evolution. Evolution involves the development of "complex organisms from antecedent states of less complex organisms" (Whitehead 1925, p. 157). But there can be no development of more from less complex on an aggregate view of wholes. An aggregate is no more or less complex than the entities by which it is constituted. An organism composed of subordinate organisms, on the other hand, can display increasing complexity because a new order comes about in the organization.
The presuppositions underlying the unpredictability of properties at higher levels of organization from lower levels (internal relatedness of whole and parts) make possible the development of later forms of organization from earlier ones. Process philosophy is, however, in agreement with formalism with respect to the unpredictability of later forms of organization from earlier ones.
‘Organism’ and ‘environment’ are correlatives. Hence the understanding of ‘environment’ cannot be separated from that of ‘organism.’ Whitehead’s uses of ‘environment’ cluster around two senses, both of which make essential reference to ‘order.’ (1) ‘Environment’ refers to the order expressed by organisms and the order pervading organisms. ‘But the character of an environment is the sum of the characters of the various societies of actual entities which jointly constitute that environment" (Whitehead 1929, pp. 168-169). "The environment automatically develops with the species, and the species with the environment (Whitehead 1925, p. 161). (2) ‘Environment’ refers to the order necessary to sustain order. "Any actual occasion belonging to an assigned species requires an environment adapted to that species, so that the presupposition of a species involves a presupposition concerning the environment" (Whitehead 1929, p. 314). "Also survival requires order, and to presuppose survival, apart from the type of order which that type of survival requires, is a contradiction" (Whitehead 1929, p. 311).
Both senses of ‘environment’ seem to be another mode of stating the reciprocity of whole and parts. They are referring to the internal relatedness of organism and environment. The order of the environment is determined by the organisms it sustains, and the organisms could not remain in (that state of) existence without the order provided by that environment. Each is what it is because of its relationship to the other. Further, what constitutes an environment on one level of abstraction may be an organism to a wider environment. That is, it may contribute to the characteristics of wider environments whose order is necessary for its continued existence. The implication is that there are layers of environmental order.
It seems that the senses of ‘environment’ may be expressed in the following meaning: environment’ refers to the order obtaining in a finite spatio-temporal region relevant to the existence of more special structures of order.3 It is clear that Whitehead intended ‘environment’ to be defined in terms of order. It is also clear that an environment is finite; this follows from his conception of order. Order and disorder are correlatives; order is never complete, merely dominant in some region. It is the reference of order to a finite region that constitutes the justification of setting limits to a consideration of environmental influences. With this understanding of ‘organism’ and ‘environment’ in hand, I wish to turn to the development of the process view of laws, induction, explanation, and conceptual change.
The formalist tradition in the philosophy of science has maintained some version of the view that a law-like statement is of universal form, capable of being put into conditional form. A true law-like statement is a law (Nagel 1961, Ch. 4; Hempel 1966, pp. 54-58). It is maintained, however, that such a characterization is inadequate to demarcate laws as different in kind from accidental generalizations -- statements such as "All the rocks in this box contain iron" and "All the screws in Smith’s present car are rusty." Hence a set of criteria is required to mark off laws from such accidental generalizations. It is generally agreed that the list of criteria supply some necessary, but not sufficient, conditions for demarcation.
Many criteria have been proposed for marking off laws from accidental generalizations. Space prohibits consideration of more than a sampling of the most frequently mentioned criteria, viz.:
(i) Laws are unlike accidental generalizations in that they make no reference to particular places, times, and objects, or are derivable from more fundamental laws which do not make such reference (Hempel and Oppenheim 1948, p. 156; Achinstein 1971, p.30).
(ii) Laws are unlike accidental generalizations in possessing unrestricted generality (Hempel and Oppenheim 1948, p. 156; Nagel 1961, p. 59; and Achinstein 1971, pp. 26-27).
(iii) Laws are unlike accidental generalizations in supporting subjunctive conditionals (Hempel 1966, p. 56; Nagel 1961, pp. 68-72).
(iv) Laws are unlike accidental generalizations in that they function differently (in an explanation) and we therefore have a different attitude toward them (Smart 1968, pp. 63-64; Nagel 1961, pp. 64-66; Achinstein 1971, pp. 46-48; Scriven 1961, p. 100; Pap 1962, pp. 301-305; and Braithwaite 1953, p. 11).
A consideration of these criteria will be simplified if another kind of statement is added to this discussion -- that of a biological generalization, such as "Albinotic mice always breed true" or "All living matter contains DNA." Biological generalizations have been considered by some philosophers of science as different in kind from laws because of the failure they seem to share with accidental generalizations to meet the conditions for law-likeness (Toulmin 1953, Ch. 2; Smart 1963, pp. 52-57 and 1968, p. 92).
Criteria (i) and (ii) are logically related. The mention of particular places, times, objects guarantees the restricted relevance of a statement to a particular spatio-temporal region and thereby the restricted generality of that statement. According to the process view, these criteria cannot demarcate laws, biological, and accidental generalizations because all of these statements implicitly refer to some environment. The environment referred to by accidental generalizations is the most special. For example, the order which must obtain for a generalization such as "All the screws in Smith’s present car are rusty" to hold true is very special indeed. The order presupposed for biological generalizations to hold is less special and that for laws less special still; but an environmental order is presupposed for all. Hence there is an implicit reference to environments and thereby a reference to particular spatio-temporal regions of varying generality in all of these statements. It seems, then, that laws as well as biological and accidental generalizations fail to meet criteria (i) and (ii).
The distinction intended by criterion (iii) can perhaps be clarified by considering the purported differences in the relationship between a law, a biological generalization; an accidental generalization and their corresponding subjunctive conditionals. It is claimed, on the one hand, that "All freely falling bodies fall with constant acceleration" supports the subjunctive conditional "If a body were a freely falling body, it would fall with constant acceleration." On the other hand, the biological generalization "All ravens are black" does not support "If an organism were a raven, it would be black," and the accidental generalization "All the screws in Smith’s present car are rusty" does not support "If a screw were a screw in Smith’s present car, it would be rusty" (Nagel 1961, pp. 68-69). But this distinction also loses validity when the concept of environment is taken into account. The failure comes about because the truth of the subjunctive conditional involves an environment wider than that to which the biological generalization and the accidental generalization are known to apply. It may well be that the conditions of this wider environment are capable of supporting counter-instances to the biological and accidental generalizations and, hence, these generalizations do not support conditionals about this wider environment. This is to say that we believe the biological and accidental generalizations fail to support subjunctive conditionals because we believe we are unwarranted in extending these generalizations to environments for which we lack important information or to environments we believe to have importantly different features than the original environment. However, if we were making such an extension in the case of a law, we should have the same reservations. We believe the subjunctive conditional to be supported by the law only when we assume the environment presupposed by the law to be so encompassing as to include the case of the subjunctive conditional. Hence it seems that criterion (iii) does not suffice to show that the laws and accidental generalizations are different in kind.
The difference in our attitude towards laws and accidental generalizations referred to by criterion (iv) is evidenced by the relationship of each to counter-instances. A law is not readily abandoned in the face of a counter-instance; an accidental generalization is readily abandoned. The reasons that laws are not abandoned are that laws belong to a theory or that the nature of their support makes abandonment difficult. These reasons constitute a reference to systematic import (Achinstein 1971, pp. 46-49; Nagel 1961, pp. 64-66). Laws belong to a theory and thereby systematize other basic uniformities. Laws are based on support other than instances falling within the scope of prediction of the law. Hence a law is not independent of other laws and its abandonment would require change in these laws. In brief the tenacity with which we hold on to a law is an indication of its place in the body of scientific knowledge at a particular time. In contrast, accidental generalizations do not systematize more basic uniformities and are not supported by the same Variety of evidence as laws. Hence abandoning accidental generalizations does not require widespread modification because these generalizations lack the systematic import possessed by laws. However, systematic import is related to environment; the systematizing power of a law depends on the environment presumed. We hesitate to abandon laws because, for a particular environment, they are useful in systematizing other uniformities. In another environment they may fail to systematize, and, in such circumstances, they would be easily abandoned. But it is difficult to maintain that biological and accidental generalizations do not systematize for any environment. Even though an accidental generalization is a low-level generalization and does not ordinarily systematize other uniformities, there is a sense in which it can usefully systematize phenomena in the special environment in which it holds. We abandon biological and accidental generalizations more easily because of the special-ness of the environment presumed; our reluctance to abandon laws is based on their usefulness in an implicit wider environment. Then a distinction in kind between laws and biological and accidental generalizations cannot be based on systematic import. Systematic import is relative to a particular environment, and the ease with which a generalization is abandoned depends on its usefulness and this, in turn, depends upon the specialness of the environment.
This sampling of the criteria should make it clear that the distinction between laws, biological generalizations, and accidental generalizations is a difficult one to define. Each of these statements presupposes a particular set of environmental conditions in order to hold; and when the notion of environment is made explicit, the usual criteria do not suffice to distinguish between them.
The process view of laws is able to account for the difference between laws and biological generalizations, on the one hand, and accidental generalizations, on the other, by the notion of dominant, but not complete, order in an environment. Accidental generalizations are either true or false. They are true if every organism mentioned in the generalization can be so characterized and false otherwise. A true accidental generalization is, in effect, descriptive of an environment without disorder. Laws, as well as accidental generalizations, depend upon the characteristics of the organisms in the environments for which they are formulated. Whitehead used the term ‘immanence’ to express this feature of laws. But although laws are immanent, they merely represent the dominant order of an environment. They do so because of the element of disorder in wide environments.
In contrast with accidental generalizations, the fact of disorder renders the terms ‘true’ and ‘false’ irrelevant to the characterization of laws. The dominant order referred to by a law can tolerate disorder up to a certain point and the law still systematize the organisms in the environment. For an accidental generalization, the failure of the organisms to exhibit the characteristic(s) referred to in the generalization renders it inapplicable because it is false. Yet the failure of organisms to exhibit the characteristic(s) of the dominant order of the environment described by the law need not render the law inapplicable.
On the process view, biological generalizations also represent dominant orders of environments. As theoretical biology advances, there is reason to believe that biological generalizations will be formulated which range over increasingly wider environments. However, since biological organisms require more special environments than do physico-chemical organisms, there is no reason for thinking that biological generalizations will come to represent dominant orders of environments as wide as those represented by physical laws.
The process view of laws is, then, that they are to be conceived as statements of the dominant characteristics of wide, or general, environments. The point at which an environment is wide enough for its dominant order to be described as a ‘law’ is more or less arbitrary. Any particular law is an abstraction of a dominant order on a certain level of generality from the total existing order. This is due to the fact that an environment is bound up with wider environments whose characteristic features are not those dominant in it. But, in turn, the characteristic features of the wider environments are abstractions. Hence even the most general laws are not universal but represent a dominant character of a particular level of a hierarchy of order.
Further implications for characterizing laws can be drawn. First, since there is disorder and laws represent merely the dominant environmental order, all laws should be conceived as fundamentally statistical in character. They are the ‘communal customs,’ the ‘large average effects,’ the ‘average regulative conditions’ to which Whitehead refers. Second, since laws are immanent and since neither the environment nor the organisms it supports are unchanging, an environment can become incapable of supporting the organisms it once supported. New dominant orders, capable of supporting new types of organisms can arise. Then it is clear that an order can pass out of dominance and new dominant orders can arise. Laws are, in brief, capable of evolving.
Hence the process view of laws conceives them as statements of dominant orders of environments. Since environments are finite spatio-temporal regions, laws are restricted and not universal.4 Because of the disorder in any environment, laws are essentially statistical in character. Finally, since laws are capable of evolving, they are non-necessary statements (Cf. Bohm 1957, pp. 137-140 and pp. 146-152. See also Bohm 1969).
On the whole formalists have given priority to deductive reasoning in science. They have pointed out important reasons for calling the role of induction into question. Two general areas of issues are relevant to discussing the process view. First, the role of induction in theory formation is very much a matter of dispute. One problem that has been pointed out here is the occurrence of new terms in theories which are not found in the observational data (or lower-level uniformities) on which they are based and which they systematize. It is not clear how the introduction of these terms can be justified (Hempel 1966, pp. 14-15). Another difficulty is that theory seems to be prior to observation in an important sense, viz., a theory is presupposed in ascertaining the relevant data, or uniformities (Hempel 1966, pp. 11-12; Popper 1963, pp. 46-47). These difficulties have led to a variety of positions about induction within the formalist tradition which range from the view that theory formation is not in its essence inference from particular to general (Goodman 1955, p. 68), to the view that there is no inference involved in theory formation, only conjecture (Popper 1963, p. 192; Hempel 1966, p. 15). The second area of issues concerns the problem of the justification of inductive inference. The formalist tradition conceives this problem tobe the justification of a general principle of induction (a principle which, in conjunction with observational data, could provide a deductive inference to the general conclusion). It is argued that all possible ways of justifying such a principle lead to difficulties: An empirical justification involves an infinite regress, while a metaphysical justification (one based on synthetic a priori categories) involves triviality or circularity (Popper 1959, Ch. I and 1963, p. 47). The charge of circularity seems especially appropriate to the conception of metaphysical method held by Whitehead. Whitehead maintained that the categories of speculative philosophy were in an important sense ‘generalizations’ from experience. In face of the apparent insolubility of the justification of a general principle of induction. formalists have limited the required justification to the role of induction in making valid inferences to predictions; this constitutes the "new riddle of induction" (Goodman 1955, p. 68).
The process view I am presenting takes the problems raised by the formalists to be genuine. Theoretical inference is not in its essence a generalization from particulars; it is not a general principle of induction which requires justification, but rather, a justifiable inference pattern. The process view approaches these problems by an application of the notions of organism and environment. The framework of organism-environment provides an explication not only of the role of induction in prediction, but also a role for induction in theory formation. That is, on the process view, inductive inference to theories and to predictions are essentially related.
The process view of induction takes the following passage from Whitehead (1929, p. 314) to set the solution to the "new riddle of induction": "Thus the basis of all probability and induction is the fact of analogy between an environment presupposed and an environment directly experienced." A reconsideration of the internal relationship between organism and environment will allow the development of a justifiable inference pattern based on this passage.
An inductive inference (such as that involved in theory formation) shifts reference from one environment (E) to another (E*). Each environment is necessary for the existence of its organisms. The organisms of E (o) cannot exist without the order provided by E, and those of E* (0*) cannot exist without the environmental order of E*. Hence an analogy between organisms of the different environments will provide a warrant for an inference as to the analogy between the environmental orders. Further, since laws are a statement of dominant environmental order, there is a context for inferring predictions about the behavior of other organisms in the environment to which reference has been shifted -- organisms not explicitly considered in the original analogies. The inference pattern can be schematized as follows (see Hesse 1968 and 1970; Plamondon 1973):
where E and E* represent environments;
AN represents an analogical inference;
/ / / / / / / / AN represents an analogical relation; and
D represents a deduction from E* to predictions.
This inductive inference pattern fully accounts for the notions which suggested a limitation to the role of induction in scientific reasoning. First, new terms can be introduced in inference to theories because the laws constituting the theory are abstracted from an environmental order developed by analogy from another environmental order. Second, the induction to the new set of laws is not a generalization from particulars; the role of prior theory in guiding observation (of data or selection of uniformities) is accounted for. Third, it is an inductive inference pattern and not a general principle of induction that is being explicated. In addition to the explication of the inference pattern, the categories of a process philosophy of science provide for the justification of the inference pattern as well. Because of the internal relatedness of organism and environment, the positive analogy between organisms of the two environments justifies an inference with respect to the analogical relationship of the two environments. Inductive inference to theories is thereby justified. Prediction of the behavior of (as yet) unobserved organisms in the environment inferred to is grounded, ultimately, in the analogy between the environments. In brief, theoretical inference takes place by way of a model whose order is already known; the environment F is a model for the elucidation of E*.
The reference of the problem of induction to the organism-environment relation seems to me to be essential. Without the internal relatedness of organism and environment, no inductive inference can be justified. I take this to be Whitehead’s meaning when he writes: "The question, as to what will happen to an unspecified entity in an unspecified environment, has no answer" (Whitehead 1929, p. 312).
The formalist view of explanation involves the ideal of deductive form (Hempel and Oppenheim 1948; see also Braithwaite 1953, Chs. l-2; Nagel 1961, Ch. 2; Popper 1959, sec. 12). The explanans consists of statements of antecedent conditions and general laws. The explanandum is a description of the empirical phenomenon to be explained. There are three formal conditions and one empirical condition which must be met for an explanation to be given. The formal conditions are: The explanans deductively entails the explanandum; the explanans contains at least one universal law which is actually used in the deduction; and the explanans has empirical import. The empirical condition of adequacy is that the statements of the explanans have not been falsified. In addition, it is claimed that prediction has the same logical structure as explanation.
It is clear that the process philosophy of science as developed thus far cannot accept the deductive model of explanation. The model requires at least one universal law to be included in the explanans, yet there are no universal laws on the process view -- all laws are statistical in character.
The process view of scientific explanation can be developed from Whitehead’s general understanding of explanation. He maintained that scientific explanation is to be conceived as a species of metaphysical explanation. They share a common aim, viz., the understanding of facts (less general principles) in terms of general principles (more general principles). These explanatory principles are attained by a method Whitehead calls (descriptive) ‘generalization.’ This ‘generalization’ consists in the discovery of generic principles from a study of specific facts (or lower-level principles). Generic principles give ‘synoptic vision.’
The method of generalization by which science and philosophy explain is easily given a negative characterization: This method is not in its essence deductive. Rather the role of deduction is in the verification process -- in testing the scope of the principles arrived at by generalization.
Whitehead’s understanding of explanation adds further support to the impossibility of conceiving explanation in the process philosophy of science in terms of the deductive model. The essence of explanation is not deductive; the explanandum is not explained by virtue of being deductively derived from the explanans. The role of deduction is an auxiliary one bound up with confirmation. This was made clear in the explication of justifiable inductive inference on the process view. In this inference pattern the deductive inference is to predictions which test the proposed dominant order of an environment elaborated analogically, not deductively, from the model of another environmental order.
Any attempt at a positive characterization of generalization is hampered by Whitehead’s occasional references to the impossibility of such a characterization. He uses such terms as ‘self evidence’ and ‘direct insight’ in describing this method. On other occasions, however, he seems to attempt a positive characterization: "Words and phrases must be stretched toward a generality foreign to their ordinary usage; and however such elements of language be stabilized as technicalities, they remain metaphors mutely appealing for the imaginative leap" (Whitehead 1929, p. 6). It is the notions of metaphor and the stretching of meanings which I find to be the key in a model of explanation for the process view. A view of scientific explanation as metaphorical has been developed as a supplement to the deductive model of explanation by some contemporary philosophers of science (Black 1962, pp. 25-47 and pp. 219-243; Hesse 1966, pp. 157-177; MacCormac 1971). 1 shall sketch the model they propose and then relate this model to the process framework of organism-environment.
According to the metaphorical view of explanation, there is a literal description of both the explanans and the explanadum. However, language is used metaphorically when words ordinarily used in the literal description of the explanans are transferred to the explanandum. In this transfer, the meanings of the terms of both the explanans and the explanandum undergo change; they come to be seen as analogous. Explanation consists in a metaphorical redescription of the explanandum. This view is not committed to the thesis that all metaphors explain. A necessary condition for a metaphor to explain is the existence of a positive analogy between the two domains (explanans, explanandum). The function of theory in an explanation is to pick out this positive analogy; theoretical concepts are metaphors in this sense. In the picking out of the analogy, there is a meaning shift which consists in an extension of meaning. This extended meaning corresponds to the generic meaning referred to by the process view of explanation by generalization.
The process view is a metaphorical view of explanation. The essence of explanation is not deduction but a ‘synoptic vision’ gained in finding generic categories from restricted fact or laws. The abstraction of such categories is based on analogy, yet involves new meaning. The ordinary usage of terms is stretched in the generalization. The stretching suggests application of the category beyond the facts or laws from which it arose. A coherent set of generic categories constitutes a theory. The theory is essentially an abstraction of analogies between systems. Theories in science abstract analogies between more special systems; metaphysical theories abstract analogies between more general systems, e.g., the sciences themselves.
The explanatory function of metaphor turns on the stretching of language by the metaphor. This stretching is made possible by analogy and makes possible new hypotheses. The necessary condition of analogy for the explanation links the metaphorical view of explanation to the process framework of organism-environment. This is because the inference potential of analogy is grounded in the internal relatedness of organisms and the environment which sustains them. This suggests that the stretching of language is dependent on the environment as well. In brief, this suggests that meaning is environmentally dependent.
The dependence of meaning on a presupposed environment explains how the stretching of meaning in a metaphor is possible. It is possible because the extended meaning depends, not upon the original environment presupposed for the original meaning, but upon a different environment presupposed for the stretched meaning.
The change in the presupposed environment brings with it changes in the presupposed dominant order. This shift in environmental order underlies the suggestiveness of the metaphor; the new dominant order makes possible new associations of terms -- associations not possible prior to the metaphor. Such associations were not possible because they were not included in the systematic associations of terms in the original environment. They are now possible because of such systematic associations in the new environment.
Hence the role of metaphor in explanation depends not only upon analogy, but also upon the more detailed specification of the presupposed environment brought about in the stretching of meaning by the metaphorical use of language.
VII. Conceptual Change
The formalists make a radical distinction between observational and theoretical language (Hempel and Oppenheim 1948; Nagel 1961, Chs. 2 and 5; Popper 1959, sec. 12). Their view is that there is a fundamental observational language the terms of which have a fixed meaning. This thesis of fixed meaning is referred to as meaning invariance. The observational language is thought to be prior to and independent of theoretical language. The theoretical language, however, is dependent on the observational language; theoretical terms are given meaning by observational terms. Successive (or competing) theories do not, then, give a different meaning to observational statements.
This distinction allows conceptual change to be explained by the deductive model. An earlier theory is explained by its successor by deducing it from the later theory and statements about the domain of application. The deduction presupposes that the meanings of the terms of the two theories are fixed; the deduction could not be carried out if the meanings of terms changed in the successor theory.
The process view is incompatible with this understanding of conceptual change. This is because the concept of environment entails meaning variance. Meaning is environmentally dependent and varies with environmental variation. This constitutes a reversal of the formalist position in an Important sense. Observational meaning becomes dependent on theoretical meaning. The meaning of scientific terms depends on a theoretical framework.
It is clear that the thesis of meaning variance has implications for an understanding of conceptual change. What these implications are is, however, a matter of dispute (see Achinstein 1964; Feyerabend 1962, 1963, and 1970; Hanson 1958; Kordig l971(a) and 1971(b); Shapere 1964 and 1966; Shea 1971; Toulmin 1972). The radical interpretation currently under discussion (Kuhn 1970a and 1970b) is that a change of theoretical framework entails a change of meaning for every term in a scientific theory. Conceptual change is therefore ‘revolutionary.’
The difficulty in conceiving conceptual change as revolutionary is that this understanding seems to make conceptual change unintelligible. Serious difficulties arise with respect to how to conceive of the relationship of the ‘same’ term in successive theories. In particular, if the terms are discontinuous in meaning, it is not clear how one could ever come to understand the terms in the later theory.
The process view does not accept this radical interpretation of the implications of meaning variance. There is not an absolute discontinuity between the ‘same’ terms in successive theories. This follows from an understanding of the ‘change’ that takes place in an environmental change. The change can be understood as a displacement in the environmental order. Such displacement is a displacement with respect to the analogies and similarity relationships of the environment. This presupposes a connection with the order displaced. A discontinuity of order, on the other hand, would be a replacement of one order with another. Such replacement would constitute an unintelligible change.
On the process view conceptual change is based on an interpretation of meaning variance as a function of displacement in theoretical framework. In the displacement of a conceptual framework, the meanings of terms can be extended or stretched. This suggests that the model for conceptual change can be taken from the metaphorical model of explanation.
The view that explanatory categories in science are to be conceived as metaphors has built into it a mechanism for conceptual change by way of metaphor. The same features which account for the explanatory function of metaphor account for the possibility of intelligible conceptual change. These features are, first, the positive analogy of the term in successive theories and, second, the hypothesis-producing potential of placing the term from the earlier into the later displaced framework. Without the first, there seems to be no possibility of understanding the meaning of terms which appear in both theories. The analogies between the terms in the two theories make possible the understanding of the meaning of the term in the later theory. Withou.t the second, no sense can be made of a real change in meaning. The hypotheses suggested by this aspect of the metaphor lead to the development of the framework which gives the term its new meaning. The meaning is changed because the term has a different association with the other terms in the respective framework.
In contrast to the revolutionary view of conceptual change, the process view can make sense of the relationship of the ‘same’ term in successive theories. The meanings of the ‘same’ terms are given continuity by the positive analogy of the terms in both theoretical frameworks. It is just because of this analogy that it makes sense to use the same sign in both frameworks. The meaning of the term in the earlier theory is the basis for the extended meaning of the term in its successor.
The stretching of meaning in conceptual change is essentially the grasping of a clearer generic meaning of which the meanings in the earlier theory are species. Successor theories represent ‘synoptic vision’ with respect to meaning.
VIII. Science and Logic
In concluding I wish to emphasize that the process philosophy of science turns away from the deductivism of formalism. It conceives the logic of science as a thoroughgoing inductivism. This inductive logic is grounded in the fundamental process categories of organism and environment. The application of these categories gives an understanding to scientific methodology not only in the sense that these categories show an interconnection between the fundamental issues and a consistent logic of their solutions, but also in the sense that this methodology has itself been set within the framework of a metaphysical system.
1For example, Carl Hempel, Karl Popper, and Ernest Nagel.
2For example, P. K. Feyerabend, N. R. Hanson, T. S. Kuhn, and S. Toulmin.
3In terms of Griffin’s distinction (below, pp. 123f.) of metaphysical and cosmological principles, I am taking ‘environment’ to be a metaphysical category. Any particular environmental order, however, is cosmological.
4Laws of nature on the process view are cosmological, not metaphysical, generalizations. See Griffin (next article.).
Achinstein, P. 1964. "On the Meaning of Scientific Terms." The Journal of Philosophy 61:497-509.
Achinstein, P. 1971. Law and Explanation. Oxford: Clarendon Press.
Black, M. 1962. Models and Metaphors. Ithaca, N. Y.: Cornell University Press.
Bohm, D. 1957. Causality and Chance in Modern Physics. London: Routledge and Kegan Paul, Ltd.
Bohm. D. 1969. "Some Remarks on the Notion of Order." In Waddington, C. H. (ed.), Towards a Theoretical Biology. 2. Sketches, pp. 18-40. Edinburgh University Press.
Braithwaite, R. 1953. Scientific Explanation. Cambridge University Press.
Bronowski, J. 1970. "New Concepts in the Evolution of Complexity." In Synthese 21:228-246.
Feyerabend, P. K. 1962. "Explanation, Reduction, and Empiricism." In Feigl and Maxwell (eds.), Minnesota Studies in the Philosophy of Science. III, pp. 28-97. University of Minnesota Press.
Feyerabend, P. K. 1963. "How to be a Good Empiricist." In Brody, B. (ed.), Readings in the Philosophy of Science, pp. 3 19-342. Prentice-Hall.
Feyerabend, P. K. 1970. "Consolations for the Specialist." In Lakatos and Musgrave (eds.), Criticism and the Growth of Knowledge, pp. 197-230. Cambridge University Press.
Goodman, N. 1955. Fact. Fiction, and Forecast. Harvard University Press.
Hanson, N. R. 1958. Patterns of Discovery. Cambridge University Press.
Hempel, C. and Oppenheim, P. 1948. "The Logic of Explanation." In Philosophy of Science 15:135-175.
Hempel, C. 1966. Philosophy of Natural Science. Englewood Cliffs, N. J.: Prentice-Hall.
Hesse, M. 1966. Models and Analogies in Science. South Bend, Indiana: University of Notre Dame Press.
Hesse, M. 1968. "Consilience of Inductions." In Lakatos, I. (ed.), The Problem of Inductive Logic, pp. 232-257. North-Holland.
Hesse. M. 1970. "An Inductive Logic of Theories." In Radner and Winokur (eds.), Minnesota Studies in the Philosophy of Science, IV, pp. 164-180. University of Minnesota Press.
Hesse, M. 1974. The Structure of Scientific Inference. London: Macmillan.
Kordig, C. 1971a. "The Theory-Ladenness of Observation." In Review of Metaphysics 24:448-484.
Kordig, C. 1971b. "The Comparability of Scientific Theories." In Philosophy of Science 38:467-485.
Kuhn, T. 1 970a. The Structure of Scientific Revolutions. University of Chicago Press.
Kuhn, T. 1970b. "Reflections on My Critics." In Lakatos and Musgrave (eds.), Criticism and the Growth of Knowledge, pp. 23 1-278. Cambridge University Press.
Leclerc, I. 1972. The Nature of Physical Existence. London: George Allen and Unwin, Ltd.
MacCormac, E. 1971. "Meaning Variance and Metaphor." In British Journal for the Philosophy of Science 22:145-159
Nagel, E. 1961. The Structure of Science. New York: Harcourt, Brace and World.
Pap, A. 1962. An Introduction to the Philosophy of Science. Glencoe, New York: Free Press.
Plamondon, A. 1973. "Metaphysics and ‘Valid Inductions’." In Process Studies 3:91-99.
Popper, K. 1959. The Logic of Scientific Discovery. London: Hutchinson and Co., Ltd.
Popper, K. 1963. Conjectures and Refutations. London: Routledge and Kegan Paul.
Popper, K. 1970. "Normal Science and its Dangers." In Lakatos and Musgrave (eds.), Criticism and the Growth of Knowledge, pp. 51-58. Cambridge University Press.
Scriven, M. 1961. "The Key Property of Physical Laws -- Inaccuracy." In Feigl and Maxwell (eds.), Current Issues in the Philosophy of Science, pp. 91-101. Holt, Rinehart and Winston.
Shapere, D. 1964. "The Structure of Scientific Revolutions." In Philosophical Review 73:383-394.
Shapere, D. 1966. "Meaning and Scientific Change." In Colodny, R. (Ed.), Mind and Cosmos:
Essays in Contemporary Science and Philosophy, pp. 41-85. University of Pittsburgh Press.
Shea, W. 1971 "Beyond Logical Empiricism." In Dialogue 10:223-242.
Smart, J. J. C. 1963 .Philosophy and Scientific Realism. London: Humanities Press.
Smart, J. J. C. 1968. Between Science and Philosophy. New York: Random House.
Toulmin, S. 1953. The Philosophy of Science. London: Hutchinson University Library.
Toulmin, S. 1972. Human Understanding, I. Princeton University Press.
Whitehead, A. N. 1925. Science and the Modern World. New York: Macmillan.
Whitehead, A. N. 1929. Process and Reality. New York: Macmillan.
RESPONSE TO PLAMONDON’S PAPER
By Bernhard Rensch
Bernhard Rensch is at the Zoologisches Institut of the Westfaelischen Wilhelms-Unversitaet in Muenster, West Germany.