issues which form the metatheoretical basis for theory construction is analyzed. ... "common language" in psychology can reasonably occur only within models.
Reprinted from L1FE·SPAN O~VElOPMENTAL PSYCHOLOGY, RESEARCH AND THEORY ACADEMIC PRESS INC., NEW YORK
e..197cJ
CHAPTER
5
Models of Development and Theories of Developmenf HAYNE W. REESE UNIVERSITY OF KANSAS LAWRENCE, KANSAS
WILLIS F. OVERTON STATE UNIVERSITY OF NEW YORK BUFFALO, NEW YORK
ABSTRACT Models, which originate in metaphor, exist on several levels ranging from all inclusive metaphysical models to narrowly circumscribed models of specific features of theories. Models at the more general levels form the determining logical context for modelsat lower levels. This categorical determinism stretches from metaphysical levels through scientific theories, to the manner in which we analyze, interpret, and make inferences from empirical evidence. Two radically different models which have had a pervasive effect upon the nature of 1 The first listed author wishes to emphasize that this paper represents a collaborative effort, and not a senior-junior relation between the authors.
115
116
Hayne W. Reese and Willis F. Overton
psychology generally and developmental psychology specifically are the organismic and mechanistic world views. The history and nature of those models are discussed and the manner in which they become transformed into corrolary issues which form the metatheoretical basis for theory construction is analyzed. Theories built upon different world views are logically independent and cannot be assimilated to each other. They reflect different ways of looking at the world and, as such, are incompatible in their implications.
.. ."
I. INTRODUCTION
Many psychologists appear to believe that the operationalists and logical positivists had given them a categorical imperative demanding, among other things, that their concepts have a certain form of definition and that the truth or falsity of propositions be determined in certain specified ways. However, as Bridgman himself pointed out, " ... there need be no qualms that the operational point of view will ever place the slightest restriction on the freedom of the theoretical physicist to explore the consequences of any free mental construction that he is ingenious enough to make. It must be remembered that the operational point of view suggested itself from observation of physicists in action [Bridgman, 1954 (in P. G. Frank, 1961, pp. 79-80)]." (Also, see P. G. Frank, 1954, in 1961, pp. 25-26.) The point is that operationalism is not a guide but rather a history; as a history it may serve a guiding function, but no better than any kind of history can serve as a gUide. (It is perhaps worth noting explicitly· that operationalism is not a philosophy of science, but, in Bergmann's 1954 terms, only a footnote in the logical analysis of the formation of scientific concepts.) The operational analysis fits the activities of some psychologists, but not the activities of others. It describes what behaviorists try to do, but not what "cognitive" psychologists such as Piaget and Werner try to do. The purpose of this paper is to examine the differences between these approaches. We will see that the crucial difference is so fundamental and broad in its implications that syncretism is impossible, and the only rapprochement possible is like the parallel play of preschoolers in that the protagonists are separate, but equal and mutually tolerant. We will see that the frequently discussed search for a "common language" in psychology can reasonably occur only within models Across models there can never be a common language because the very nature 0 different models dictates different language. We will first consider models and theories in general, then the concept of development, and finally the mechanistic (reactive organism) and organismic (active organism) models in the study of development.
•
5
Models and Theories of Development
117
II. MODELS AND THEORIES .1
A. Models I. Levels ofModels
".
In science, a "theory" is a set of statements, including (a) general laws and principles that serve as axioms, (b) other laws, or theorems, that are deducible from the axioms, and (c) definitions of concepts. A "model" is structurally separate from a theory, but is functionally part of its axioms (see Lachman, 1960; Toulmin, 1962). According to A. Kaplan (1964), "The proper concepts are needed to formulate a good theory, but we need a good theory to arrive at proper concepts [po 53]." As with other paradoxes, this "paradox of conceptualization" is resolved when it is recognized that there are levels of concepts, models, and theories. Any theory presupposes a more general model according to which the theoretical concepts are formulated. At the more general levels, the concepts are generally less explicitly formulated, but they nonetheless necessarily determine the concepts at lower levels. This categorical determinism stretches from metaphysical and epistemological levels "downward" through scientific theories, to the manner in whieh we analyze, interpret, and make inferences from empirical evidence. The most general models, variously designated as "paradigms" (Kuhn, 1962), "presuppositions" (Pap, 1949), "world views" (Kuhn, 1962; Seeger, 1954), and "world hypotheses" (Pepper, 1942), have a pervasive influence throughout the more and more specific levels, as noted by Kessen (1966) and others before him (Black, 1962; Pepper, 1942; Toulmin, 1962). The different levels of models are characterized by different degrees of generality, openness, and vagueness. At one extreme are implicit and psychologically submerged models of such generality as to be capable of incorporating every phenomenon. These metaphysical systems are the world hypotheses referred to above. They are basic models of the essential characteristics of man and indeed of the nature of reality. Examples, which will be considered in detail in Section IV, are the mechanistic and organismic world hypotheses. In the mechanistic world view, the model for all phenomena is the machine; in the organismic world view, the model is the biological organism and its activities. Less general, more specific, and more explicit are the models that have been called "ideals of nature" or "principles" (Toulmin, 1962), "paradigms of science" (Kuhn, 1962), and "suppositions" (A. Kaplan, 1964). They are not "philosophical" models, that is, they are not part of the logic of science, but rather are part of the psychology or pragmatics of science (see Bergmann, 1954; Griinbaum, 1954; Margenau, 1954; Pap, 1949). These might be called "pragmatic models." We will have little further to say about them in this paper. At the other extreme are the relatively specific and precise models that in the
Hayne W. Reese and Willis F; Overton
118
past have most commonly been connoted by the term "model." These can be called "theoretical" models, provided it is kept in lnind that this designation is ' meant only to distinguish them from the more general metaphysical and "pragmatic" ~models. The "theoretical" models include, on a scale from general and abstract to specific and concrete, (a) conceptual models formulated solely in verbal language, (b) analogue models, and (c) scale models.
.'"
2. Theoretical ModelS
a. The Nature of Theoretical Models. It is generally agreed among 'philosophers of science, and, scientists that general theoretical models are metaphorical (e.g., Black, 1962; Braithwaite, 1953; Chapanis, 1961; Ferre, 1963; Lachman, 1960; Rashevsky, 1955; Schon, ,1963; Seeger, 1954; Toulmin,1962). In Braithwaite's words, for example, "Thifilting of scientific theories by meaqs of models is always as-if thinking [Braithwaite,j95-3,.p. 93]." Model terms are not the same as tll~r~tical terms. ACCO~~ to Margenau (1954), nonempirical criteria establish concepts as an i~terna11y consi.stent set or theory, ,but cannot' validate, the concepts, The validatioh of cOllCepts is an empirical matter. However, he was asserting that ,e~ery'valid concept, which by' definition must, belong to a satisfactory theory, must have two definitions, one formal or ."constitutive" and one instrumental or oper;itional.2 The s~tuation actually; appears to be more ceniplex than that, in, a way, yet in another way more simple. A concept can haveonlr,O:ne ·definition, and tWCl definitions must define two different concepts (e.g., Bergmann, 1957). Therefore, the concept defined formally and the concept defmed operationally are not necessarily c6U~psibte, ;and are certainly not identical on any logical argument. Hence the sUnplif1catl~n. The. complexity is that there are not two, but three kinds of conc~pts that must be considered. There are the formally OJ rationally defined theoretical concepts,' the operationally defined descriptive concepts, and the concepts that describe, characteristics,¢the model. It is not necessary to assert that the three kinds of concept hlWe the same referent, especially if "referent" is Early, training trials Later training trials
Sti~ulus 1,'" .AP,proach Response "" Reward "Stimullls:2 ~ Approach Response ....... Nonreward Stirnill~·i'~ Approach Response (sH R) Sthmtl~~-+ Inhibitory Response (sIR)
Figure 1. A model of inhibition in discriminative learning.
2 Bergmann (1951) has pointed out that strictly speaking a concept cannot be "valid" or "invalid," only arguments can be. However, it seems clear enough that Margenau meant something like "establish the probable truth value of stareirterrtscaaiaUliAg a concept" when he spoke of "validating a concept."
•
5
.'
Models and Theories of Developmen t
119
interpreted as immediate sensation, or directly observed data. Rather, it is meaningful to assert that certain concepts, one of each kind, are correlated. Thus, "inhibition" is a theoretical concept in Hullian theory related to the empirical concept of inhibition, or reduction in response frequency. The concrete model that illustrates the correlation might take the form shown in Figure ,I. How well the model works is not our present concern; essentially, the model says that an approach response is conditioned to the stimulus that is associated 'with reward and an inhibitory response is conditioned to the stimulus that is associated with nonreward. b. Functions of Theoretical Mod{!ls. (1) Representation. Chapanis said, "Recently ... I noted a title about a model of hypothesis behavior in discrimination learning (Levine, 1959). When I read the article, however, I was disappointed to find that the author's -'model' was, primarily ali empirical method for solving a set of simultaneous .equations [Chapanis, 1961, p. 118] ." Actually, Levine did describe a model,which represents behavior as describable by "hypotheses" (in the sense used by l
