Science, Philosophy of
SCIENCE, PHILOSOPHY OF•••
Philosophy of science as an autonomous subject is a product of the twentieth century. Its development stemmed from the great intellectual challenges of the quantum and relativity theories, but philosophical issues surrounding such theories as psychoanalysis, evolutionary theory, Marxist and capitalist economics, the ethics of human experimentation, and the enormously increased importance of science as an intellectual endeavor led to a great expansion of the field.
Work within philosophy of science tends to fall into two approaches. The first sees science as a testing ground for traditional philosophical problems. Chief among these traditional problems is this: Can we have any knowledge that is certain and in terms of which all other knowledge in the area can be justified (foundationalism), or are all claims to knowledge uncertain (fallibilism)? Within the realm of things that can be known by empirical investigation, it would seem that science has the best claim to secure knowledge. Philosophers of science have thus devoted a considerable amount of time to what kinds of scientific methods are effective in producing such reliable knowledge. On the other hand, many philosophers, especially in recent times, have denied that science does actually produce a privileged body of knowledge, and have argued that all scientific knowledge is a product of its historical and social context.
The second approach to philosophy of science focuses on issues that are peculiar to individual sciences. Of particular interest here is the possibility of reducing biology to chemistry or physics, and of reducing some of the social sciences, especially psychology, to biology. If these reductionist projects were to be successful, then issues that currently appear to be peculiarly biological, such as the question of what makes something a living organism, would turn out to be merely a question of degrees of complexity, and not specifically biological at all. In addition, the moral issues that pertain to humans and animals because of their psychological characteristics would be approached very differently if psychological properties were considered to be unreal or merely disguised biological properties. These differences between the sciences are crucial. For example, a great deal of medical research cannot enjoy the unlimited freedom of laboratory experimentation that is characteristic of physics simply because of the ethical constraints its subjects require. Moreover, the variability of its subjects makes universal laws hard to formulate in biology, in distinction to, for example, astronomy.
Predecessors to Contemporary Viewpoints
It was the logical positivists and logical empiricists of the Vienna Circle (1923–1936) and the Berlin school (1928–1933) who succeeded in placing scientific issues near the heart of the philosophical enterprise. (A classic, albeit sententious, presentation of the logical positivists' views can be found in A. J. Ayer's Language, Truth and Logic, 1946.) For philosophers such as Moritz Schlick, Rudolf Carnap, Hans Reichenbach, and Carl Hempel, all of whom had a scientific education, the task was to provide a foundation for genuine knowledge, and this foundation was to be as secure as the best science of the time. The logical positivists were squarely within the empiricist tradition, which holds that all genuine knowledge must be reducible in principle to knowledge obtainable by empirical methods, and ultimately to that obtainable through the human sensory apparatus. To this empiricist view they added a deep concern with language resulting from developments in logic in the late nineteenth and early twentieth centuries. Although the most famous manifestation of their approach was the attempt to eliminate metaphysical claims through the verificationist criterion of meaning (which asserts that a sentence is factually significant to a given individual if and only if he knows what observations would lead him to accept that proposition as true or to reject it as false), their true legacy has been the view that it is by means of logical analyses of philosophical concepts that genuine understanding is achieved. It is no exaggeration to say that philosophy of science since 1950 has been primarily engaged in a struggle to decide which elements of the positivist monolith to retain, and what should be the replacement approaches for those parts that have been rejected.
An important alternative to the positivist program has been the falsificationist approach of Karl Popper. Although his Logik der Forschung was published in 1934, its impact was muted until the expanded English translation appeared in 1959 as The Logic of Scientific Discovery. Popper set himself the task of providing a criterion that would distinguish between genuine scientific hypotheses and pseudoscientific statements. A key belief driving Popper's work was his view that the traditional problem of induction could not be solved. Most generally, inductive inference involves reasoning from what has been observed to what has not been observed, a characterization that covers inferences from the past to the future, from observed data to the existence of directly unobservable microentities such as prions, and from finite data sets to the universal hypotheses that represent scientific laws and general theories. Justifying inductive inferences was a serious problem for logical positivism, because the verificationist criterion ruled out all universal scientific theories and laws as meaningless, simply because no amount of finite data could conclusively verify these general claims. Popper instead proposed the demarcation criterion that a statement or theory was scientific only if it was falsifiable; that is, it must be possible to state in advance a set of possible observations which, if observed, would result in the statement or theory being rejected. Theories such as astrology and psychoanalysis were, according to Popper, branded as pseudoscientific on the basis of this criterion because they traditionally accommodated themselves to fit any observations whatsoever. To refuse to relinquish a theory in the face of recalcitrant data is a characteristic feature of scientific irrationality. Popper's brand of falsificationism is comprehensive, for it requires that even reports of observations be falsifiable. Thus, in contrast to the positivists' foundationalism, which is grounded in an empirical base that is certain, falsificationism is a deeply fallibilist position, within which claims to certainty are relinquished at all levels of generality.
Popper was well aware of a point often made by the French philosopher Pierre Duhem: In order to draw out testable predictions from scientific hypotheses, one ordinarily needs to assume the truth of various background assumptions and theories (Duhem). Thus, if the prediction turns out to be false, the force of the falsification could be deflected away from the principal hypothesis onto the background assumptions. Hence the need in the above specification of falsificationism to state in advance what would result in the hypothesis being rejected.
Although this strategy removes the force of Duhem's criticism that there are no crucial experiments that can conclusively decide between competing theories, it moves the emphasis away from a method of testing that is based only on logic and empirical data to one where a (human) decision plays a central role, and this introduces a characteristically conventional element into the picture. Falsificationism is primarily a normative methodology, for it prescribes and proscribes courses of action with respect to scientific hypotheses. As historical and sociological studies of science have become increasingly influential, there has been a concomitant emphasis on the need for methodological theories to be descriptively accurate of what scientists do and have done. It is easy to find cases where historically important episodes of science do not fit the falsificationist model, ones where scientists refused to abandon theories in the face of clear counter evidence. The difficult task is to articulate when this furthers broad scientific ends, rather than just narrow personal motives. But to reject falsificationism merely because it is not descriptively accurate of everything done in the name of science would be as misguided as the attempt to turn ethics into a purely descriptive enterprise.
Thomas Kuhn's Work
One of the best known alternatives to the positivist approach is Thomas Kuhn's. Ironically, Kuhn's seminal work The Structure of Scientific Revolutions (1996) was originally published in the positivists' International Encyclopedia of Unified Science (Kuhn, 1955). Kuhn's strategy was to use the history of science as a proving ground for methodological positions in the philosophy of science. This history, Kuhn claimed, could be divided into two distinct types of periods. There were long stretches of normal science punctuated by brief periods of revolutionary science. To illuminate both kinds of science, Kuhn introduced the concept of a scientific paradigm. This concept, in its mature characterization, consists of four components. First, there are the symbolic generalizations, those fundamental laws and principles of a science that underpin all theoretical work in the field, such as the laws of genetic replication or the principle of natural selection of species. Second is the metaphysical component of the paradigm, within which the fundamental kinds of things constituting the subject matter of the science are specified, such as atomistic or field-theoretic assumptions in physics, or a commitment to specifically mental properties, as opposed to material properties, in psychology. Third, there are the value commitments. These not only concern what constitutes an acceptable piece of evidence in the science, but what the appropriate goals are for a science, and what the ethical standards are to which one should adhere. Thus, double-blind studies will be considered the standard methodology for drug trials. Fourth, there are the exemplars, those quintessential successes that a scientific field can point to as evidence for the fruitfulness of the first three elements, as, for instance, Newtonian mechanics could point to its success in predicting the existence of the planet Neptune.
Normal science, then, is science conducted entirely within the framework of a single paradigm, whereas revolutionary science consists in the development of a competing paradigm and the process of a scientific community's transfer of allegiance to the new paradigm. A seemingly inescapable consequence of paradigm change in periods of revolutionary science, and one that is deeply disturbing to many, is that the process of change is determined by neither rational argument nor empirical evidence. Because a change in paradigm necessarily involves a change in at least one of the four components already described, there will inevitably be fundamental differences of opinion about whether the old or the new component is preferable, and the remaining three components will frequently not provide a large enough common ground to resolve the dispute in an impartial way. In this way, paradigms are, to use Kuhn's term, incommensurable. There is then a deep difference between Kuhn on the one hand and both Popper and the positivists on the other.
Equally important is the distinction between internal and external descriptions of science. Within both the positivists' and Popper's approaches, the way in which science proceeds ought to be appraised only in terms of influences that are purely internal to the science at hand, including the construction of theories, the invention of new experimental apparatus, and the verification or falsification of hypotheses by empirical data. Any interference by nonscientific factors, such as economic considerations, political pressure, and religious prohibitions, are to be condemned as illegitimate influences to be resisted in practice, and ignored in writing the history of the science. In contrast, Kuhn holds that not only are such influences usually present and causally effective in propelling or impeding the elaboration of a paradigm, but they are frequently important in fixing the values component of a paradigm. Thus, the religious opposition to research on fetal tissue derived from deliberate abortions, the political pressure to direct funds in molecular biology toward acquired immunodeficiency syndrome (AIDS) research, and the decision to allocate significant financial resources to the Human Genome Project are all part of an externalist appraisal of the scientific research concerned. Inseparable from this externalist approach is the shift in emphasis from scientific theories as logical entities whose existence and appraisal are objective matters, and the truth or falsity of which is something to be discovered, to a position where the opinions of a community of scientists are primary, and acceptance of a paradigm is determined by a consensus in that community rather than by the paradigm's truth or falsity. Coupled with the inclusion of externalist factors, this leads naturally toward a focus on the sociology of science, rather than its philosophy as traditionally conceived.
Some further consequences of the Kuhnian approach are worth mentioning. Because of the incommensurability of paradigms, revolutions lead to schisms in the path of science, with a resulting loss of the notion of scientific progress. Comparative judgments of the kind "Paradigm A is superior to Paradigm B" can no longer be made on a uniform scale of comparison, and what remains is technological progress without any necessary concomitant progress toward the truth. Consequently, what has come to be known as the Whig view of the history of science, which sees the development of science as an uninterrupted triumphal march to the peak of contemporary success, has to be abandoned in favor of a contextually sympathetic interpretation of previous theoretical traditions. Finally, if Kuhn is correct, there is no longer anything peculiarly privileged in the scientific enterprise. The development of art, architecture, music, and so forth can all be characterized in terms of paradigms, normal practice, and revolutionary changes, a feature that has not escaped Kuhn's critics.
Contemporary Work in the Field
Perhaps the most important consequence of the collapse of the positivists' domination in the philosophy of science has been the splintering of the field into a number of subsets. One principal division is between those who continue to hold that there are general principles underlying various scientific methods, and those for whom only local, context-specific approaches are feasible. Certain areas of science still seem to be amenable to the first approach. The nature of scientific explanation is a topic of perennial interest, with various causal and unification approaches (Salmon) serving as the chief contenders to replace Carl Hempel's logical model. How scientific hypotheses and theories are confirmed is the subject of another area of research (Achinstein), with computer-assisted diagnostic procedures in medicine forming a small but important proving ground for inference procedures. There is considerable current interest in causal inference, particularly of the kind used in epidemiology (Pearl). Despite these successes, issues related to the autonomy of particular sciences have increasingly come to the fore. The positivists' orientation towards reducing all sciences to physics, at least in principle, has been replaced by a recognition that at least in practice, and perhaps even in principle, this reduction cannot be carried out. There is now a "philosophy of X" for almost every science, from economics to geology. In particular, the philosophy of biology and the philosophy of medicine are well established subfields with their own problems and methods. Accompanying this trend has been a reduced emphasis on grand unifying theories in favor of local models that capture, albeit imperfectly, the structure of specific systems (Humphreys). This latter approach works well for biological models, within which the sheer number and complexity of the influences on a system and the importance of its historical evolution render simple general theories inadequate.
A second primary division is between those for whom normative, objective, and a priori characterizations of science are desirable and attainable, and those who maintain that such characterizations are inevitably descriptively inaccurate and unrevealing of the true nature of science. Within this latter orientation lie contemporary naturalistic and cognitive approaches to philosophical issues. Philosophers using these methods hold that scientific knowledge from areas such as psychology and evolutionary biology shed more light on why certain methods are successful than can more traditional a priori approaches. For example, instead of specifying a priori the inferences that an ideal reasoner should make in deciding which course of action is appropriate in some clinical setting, a naturalist will investigate the heuristics that underlie reasoning used in clinical practice (Gigerenzer, Todd, and ABC Research Group).
Another dispute is between those who hold that many objects of scientific investigation, such as various psychiatric disorders, are social constructions, and those who hold that there is an objective reality that science investigates (Hacking). Much of this work is interesting and legitimate, but the rejection of traditional norms of rationality has led in certain quarters to a denial that science has any claim to superior methods of investigating the world. The so-called "science wars" between those who seek to maintain the epistemological superiority of science and those who wish to undermine it are an extreme, albeit avoidable (Koertge) consequence of this division.
All of the threads described have made formulating a satisfactory account of scientific progress less easy than it was in earlier periods, especially within the philosophy of biology. The piecemeal framework of models, the attacks on both the rationality of scientific appraisal and the objectivity of reality, the autonomy of multiple sciences—all have made a defense of progress towards a unified scientific account of the world more difficult than one might wish. Nevertheless, mere complexity and locality does not preclude science from accurately describing an objective reality in a systematic and rational fashion.
Philosophy of science and bioethics share a common concern. Each must draw a line between the prescriptive and the descriptive, between what is rational and justified on the one hand, and what is merely popular opinion and prejudice on the other. Both Galileo and Ignaz Semmelweiss were victims of such antiscientific attacks, the first for advocating the correct theory of the solar system, the second for discovering the mode of transmission of childbed fever. It is thus essential to have some clear distinction between fact and opinion, between the rational evaluation of a hypothesis or ethical view and its mere acceptance, between what is ethically justified and the way individuals happen to act. To use a specific example, it is essential to distinguish between what science can do to allow premature babies to survive and how one can evaluate the quality of life they might expect. This, if nothing else, is why the apparently dry and abstract issues of the foundations of knowledge, of internal and external influences on science, and of fact versus convention bear directly upon matters of more immediate concern.
paul w. humphreys (1995)
revised by author
Ayer, A. J. 1946. Language, Truth, and Logic, 2nd edition. London: Gollancz.
Curd, Martin, and Cover, J.A., eds. 1998. Philosophy of Science: The Central Issues. New York: Norton.
Duhem, Pierre. 1906 (reprint 1962). The Aim and Structure of Physical Theory. New York: Atheneum.
Gigerenzer, Gerd; Todd, Peter M.; and the ABC Research Group. 1999. Simple Heuristics That Make Us Smart. Oxford: Oxford University Press.
Hacking, Ian. 1999. The Social Construction of What? Cambridge, MA: Harvard University Press.
Hempel, Carl G. 1966. Philosophy of Natural Science. Englewood Cliffs, NJ: Prentice Hall.
Humphreys, Paul. 2003. Extending Ourselves. New York: Oxford University Press.
Klee, Robert. 1997. Introduction to the Philosophy of Science. New York: Oxford University Press.
Klee, Robert. 1999. Scientific Inquiry: Readings in the Philosophy of Science. New York: Oxford University Press.
Koertge, Noretta, ed. 1998. A House Built on Sand. Oxford: Oxford University Press.
Kuhn, Thomas S. 1996. The Structure of Scientific Revolutions, 3rd edition. Chicago: University of Chicago Press.
Pearl, Judea. 2000. Causation. Cambridge, Eng.: Cambridge University Press.
Popper, Karl R. 1935. Logik der Forschung. Vienna: Julius Springer, tr. Karl Popper, Julius Freed, and Lan Freed (1959) under the title The Logic of Scientific Discovery. London: Hutchinson.
Salmon, Wesley C. 1990. Four Decades of Scientific Explanation. Minneapolis: University of Minnesota Press.