Copyright © 1999 Massachusetts Institute of Technology
The view that scientific knowledge should be integrated has a long history, with roots in Aristotle and the French Encyclopedists. Perhaps as a consequence of the ever increasing specialization in science, scientists have found it important to work in an interdisciplinary manner where they can draw upon the research skills and knowledge bases of scientists trained in other disciplines. Thus, attempts at integration at the level of individual laboratories, and at higher levels such as that of professional societies, have become relatively commonplace in modern science. Cognitive science itself represents one such integratory effort that began with collaborative conferences (the MIT Symposia on Information Theory in the 1950s) and collaborative research groups (the Center for Cognitive Studies at Harvard in the 1960s), and became institutionalized with funding by the Sloan Foundation in the 1970s and 1980s) and the founding of the Cognitive Science Society in 1978 (Bechtel, Abrahamsen, and Graham 1998).
The modern interest in unity of science stems largely from the work in the 1930s and 1940s of logical positivists of the Vienna Circle, especially the logician-philosopher Rudolf Carnap and the sociologist Otto Neurath. (Today the logical positivists have acquired the image of conservatives, but politically they were liberals and social democrats, and scientifically they advocated a pluralist and Enlightenment conception of science.) One of the driving concerns of Carnap and Neurath was epistemic. The nineteenth century witnessed the birth of many new scientific disciplines, es-pecially in the life sciences and social sciences: cytology, physiological chemistry, psychology, anthropology, and so forth. These various disciplines employed different vocabularies and different methods of investigation, raising the question of whether they were generating real knowledge. One way Carnap and Neurath proposed to evaluate the epistemic status of those inquiries was in terms of their unity with other scientific pursuits, especially physics, which, despite the major controversies occurring within it in the wake of Einstein, was taken as a paragon of a scientific discipline. (One prominent forum for the positivists' proposals was the Encyclopedia of Unified Science, which Carnap and Neurath, in collaboration with the American pragmatist Charles Morris, began to edit in 1938.)
The tool for unifying other disciplines with physics was reduction of two different sorts. The first was reduction of theoretical claims of each science to a base class of sentences whose truth value could be directly determined through observation (Carnap 1928). Thus, all legitimate disciplines would be seen to have a similar base in that their theoretical claims were reducible, through logical analysis, to observations. Carnap initially proposed that the observational basis for all sciences lay in phenomenal reports, but eventually he followed Neurath in requiring only a reduction to observational reports of physical states (Carnap 1934). Such a reduction would show that all science relied on similar empirical foundations and tools of logic for securing their theoretical claims. The proponents of this approach never succeeded in developing an adequate logical analysis to ground theoretical claims in observation. The call for such reductions, however, had a significant influence in furthering the behaviorist movement in psychology, because it claimed to rely only on observations of behavior and laws relating such behaviors to stimuli or reinforcements.
The other sort of reduction Carnap and Neurath advanced was theory reduction, wherein theories of a higher-level science would be reduced to those of more basic sciences. (The best source for this view of reduction is Nagel 1961.) Such a reduction required both translation laws that connected the vocabulary of the higher-level science with that used in the more basic science (higher-level entities might, for example, be characterized in terms of their composition from lower-level entities) and derivations of the laws of the higher-level science from those of the more basic science under specified boundary conditions. It is important to note that success in developing such reductions was viewed not as eliminating the reduced theory but as providing epistemic support for it; thus, a reduction of PSYCHOLOGICAL LAWS to those of neuroscience in this view would provide support for the psychological laws. The theory reduction model has been seriously promoted as a framework for unifying psychology and neuroscience by such theorists as Patricia Churchland (1986); unlike the positivists, though, she focuses on the process of theory development, and proposes a coevolutionary research program in which both neuroscience and psychology will evolve until they are unified by a reduction (McCauley 1996). However, as with the attempt to reduce theoretical claims to observation claims, the attempt to reduce higher-level theories to lower-level ones has encountered serious objections. For further discussion, see REDUCTIONISM and Bechtel (1987).
A variation on the reductionist framework for unifying science that rejects the attempt to ground all theories in physics has emerged several times in the development of the study of cognition. This approach tries to show that a common theoretical framework applies at a number of levels of organization in nature, with the unification provided by this theoretical framework. Cybernetics (a theoretical framework emphasizing the role of feedback in maintaining stable states in complex systems; WIENER 1948) and General Systems Theory (von Bertalanffy 1951) both offered general frameworks that were intended to apply to biological and cognitive systems at a variety of different levels. One reason neither program endured was that at the time it was difficult to develop successful detailed empirical hypotheses about cognition within these frameworks; as a result, in the 1960s and 1970s most inquiries into cognition assumed some form of autonomy of the cognitive system from both the underlying neural structures and its situated context in the world. Today advocates of dynamical systems theory, though, have revitalized unificationist aspirations by offering the mathematical and geometrical models of dynamical systems as unifying frameworks (for applications to cognitive science, see Port and van Gelder 1995).
Another model of unification, developed by Lindley Darden and Nancy Maull (1987), rejects the insistence on a common theoretical framework. Darden and Maull propose instead that interfield theories, which establish connections between phenomena that have been studied in two or more fields of inquiry without making any one of them more basic, constitute the vehicle of unification. Such linkages constitute a discovery heuristic in that what is known about one set of phenomena can then suggest hypotheses about the other set of phenomena (Wimsatt 1976). The fruitfulness of such interfield theorizing can be seen within cognitive science. For example, Chomsky's initial proposals for transformational grammars in linguistics were intended to provide a finite characterization of the grammatically well-formed sentences in a language, not as models of language processing. But George Miller and other early cognitive psychologists attempted to employ them as process models in linguistics and to predict reaction times for processing sentences based on the number of transformations required to generate the sentence in Chomsky's grammar. This effort to forge an interfield connection, though initially promising, died when Chomsky revised his linguistic analysis in ways that no longer could account for such data (Reber 1987). Other attempts to develop interfield connections from lingusitics to psychology have proven more successful (Abrahamsen 1987) and current efforts of cognitive grammarians to ground grammatical structures on general cognitive abilities (Langacker 1987) represent the attempt to develop connections in the opposite direction. Moreover, links between linguistics and neuropsychology prompted fruitful rethinking of the classical analysis of APHASIA as analyses in terms of syntax and semantics replaced those based on comprehension and production (Bradley, Garrett, and Zurif 1980).
So far discussion has focused on unification via theories, a natural consequence of the focus on theories in twentieth century philosophy of science. But increasingly theorists are recognizing other forms of unification in science, ones where research techniques and tools become the vehicle of integration. The development of cognitive neuroscience has been fostered in part by the integration of behavioral measures of cognitive psychology (e.g., reaction time studies and error analysis) with tools for studying brain activity (e.g., SINGLE-NEURON RECORDING and evoked potentials). In this light, the recent development of neuroimaging tools (POSITRON EMISSION TOMOGRAPHY and function MAGNETIC RESONANCE IMAGING) is particularly interesting. The resulting images, which indirectly measure neural activity, are interpreted functionally by applying the subtraction method developed by the nineteenth century Dutch psychologist Frans Cornelis Donders, so that images of the brain active in one task are subtracted from those of the brain active in another task, with the intent of revealing the brain structures responsible for the additional component of processing required in the second task (Posner and Raichle 1994).
Although the dreams of unity of science via reduction advanced by the positivists have generally not panned out, unification and integration, viewed in a more patchwork manner, are a routine part of modern science. Scientists regularly rely on phenomena studied in other disciplines to constrain their own studies or explain what seems inexplicable in their field alone. Tools and techniques are widely shared between related disciplines. The resulting picture is a network of local integration, not one of global unification.
Abrahamsen, A. (1987). Bridging boundaries versus break boundaries. Synthese 72:355-388.
Bechtel, W. (1987). Philosophy of Science: An Overview for Cognitive Science. Hillsdale, NJ: Erlbaum.
Bechtel, W., A. Abrahamsen, and G. Graham. (1998). The life of cognitive science. In W. Bechtel and G. Graham, Eds., A Companion to Cognitive Science. Oxford: Blackwell.
Bradley, D. C., M. F. Garrett, and E. B. Zurif. (1980). Syntactic deficits in Broca's aphasia. In D. Caplan, Ed., Biological Studies of Mental Processes. Cambridge, MA: MIT Press, pp. 269-286.
Carnap, R. (1928). Der Logische Aufbau der Welt. Berlin: Weltkreis.
Carnap, R. (1934). The Unity of Science. Translated by M. Black. London: K. Paul, Trench, Trubner & Co.
Churchland, P. S. (1986). Neurophilosophy. Cambridge, MA: MIT Press.
Darden, L., and N. Maull. (1987). Interfield theories. Philosophy of Science 43:44-64.
Langacker, R. (1987). Foundations of Cognitive Grammar. Stanford: Stanford University Press.
McCauley, R. N. (1996). Explanatory pluralism and the co- evolution of theories of science. In R. N. McCauley, Ed., The Churchlands and Their Critics. Oxford: Blackwell, pp. 17-47.
Nagel, E. (1961). The Structure of Science. New York: Harcourt, Brace.
Port, R., and T. van Gelder. (1995). Mind as Motion: Explorations in the Dynamics of Cognition. Cambridge, MA: MIT Press.
Posner, M. I., and M. Raichle. (1994). Images of Mind. San Francisco, CA: Freeman.
Reber, A. S. (1987). The rise (and surprisingly rapid) fall of psycholinguistics. Synthese 72:325-339.
von Bertalanffy, L. (1951). General Systems Theory: A New Approach to Unity of Science. Baltimore, MD: Johns Hopkins University Press.
Wiener, N. (1948). Cybernetics: Or, Control and Communication in the Animal Machine. New York: Wiley.
Wimsatt, W. C. (1976). Reductionism, levels of organization, and the mind-body problem. In G. Globus, G. Maxwell, and I. Savodnik, Eds., Consciousness and the Brain: A Scientific and Philosophical Inquiry. New York: Plenum, pp. 205-267.
Copyright © 1999 Massachusetts Institute of Technology