More generally, systems that may be described as chaotic may appear not to be deterministic, because they seem to be unpredictable. In fact, chaotic systems may be deterministic, at least in theory, in the sense that the same result will also be obtained from the same initial conditions; but, the specification of those conditions would clearly exceed current and, perhaps, foreseeable human capabilities. Yet, with such systems, a simulation of a representative or completely plausible outcome is achievable, and such analyses may be of considerable value in the understanding of the nature of the phenomena at issue even if falsifiable predictions cannot practically be generated. See, e.g., Roger Penrose, Shadows of the Mind, pp.21–3. Thus, there are useful and accepted theories that do not and cannot generate unique, specific predictions.
An example: Darwinian theory
Thus, one philosophical challenge to the Popperian test of falsifiability is the fact that certain theories that have been deemed to be very respectable in the scientific community do not lend themselves to disproof by experimentation, because they are not readily usable to predict specific future events.28 These theories have achieved their success largely through their apparent power to explain the past. Probably the most widely known example is Darwin’s theory of natural selection, which has been an impressive tool in generating explanations for the emergence and success of various characteristics among living organisms. Related and similar theories are utilized in the social sciences, particularly in micro-economics. (Of course, the parallels between the models of competition in the marketplace and competition in the natural world are obvious. Indeed, the underlying model or explanatory paradigm emerged almost simultaneously in biology, with Darwin and Wallace, and in economics, with Adam Smith, all three certainly influenced by Thomas Malthus.)29
Indeed, useful theories like the theory of evolution by natural selection do not even purport to make predictions in the normal sense of the word—for example, that such a characteristic will appear in a particular type of organism and will do so in a particular period of time. See, e.g., Toulmin, Foresight and Understanding, pp.24–5. That “flaw” is actually embedded in Darwinian theory itself.30 As also discussed below, Darwin’s theory, as generally conceived, assumes random mutations or alternations as the source of variation, which, by definition, are not predictable. What should we make of these deviations from the asserted norm for scientific theories?
Another example: medical science
Let me turn to another example of differences between the theory of evolution and other sciences that may not strictly meet the condition of falsifiability or disconfirmability. Medical science is generally probabilistic—that is, the predictions are set out in terms of statistical likelihood rather than absolute predictions. However, modern medical science tends to incorporate statistical correlations with theoretical structures into which other empirical evidence can be folded for some greater degree of confirmation. As just one illustration, after the identification in 1989 of an abnormality of the prostate labeled “prostatic intraepithelial neoplasa” or PIN and the observation that the condition PIN occurred with great frequency with prostatic cancer, efforts were made to understand the relationship between PIN and cancer.
For example, was PIN a precursor to cancer in the sense that the same cells displaying the abnormality would likely become cancerous (in a direct causal pathway)? Did PIN tend to occur simultaneously with cancer in different cells as a result of some other underlying causal factor that would lead to both conditions? Or, was PIN a likely event on a progression that led to cancer appearing in other cells? The analysis involved not only sampling of a statistical nature but also the examination of the cell structures arising in both conditions, the types of genetic alterations that occurred and the progression of both conditions. The studies also utilized animal models tested through experimentation. See Chapter 9, “Neoplasms of the Prostate,” by David G. Bostwick and Isabelle Meiers, in Bostwick and Liang Cheng, Urologic Surgical Pathology (2008), pp.443–580.
Again, the predictions could not strictly be disconfirmed, since probabilistic outcomes can be consistent with very unrepresentative results (like finding the toss of a coin coming up heads 10 times in a row), but it would seem that this approach to science is clearly appropriate and meaningful. Indeed, the methodological approaches in medicine are largely different from the deductive theories we normally equate with the physical sciences.31
Of course, one might argue that the methodological approaches in medical science reflect the limitations of our current knowledge, rather than a different ultimate conception of what the science should and, hopefully, someday could be. It is conceivable that with a sufficient understanding of all of the relevant initial conditions (the physiology of the patient) and of the pathogens to which the patient has been exposed, we may someday be able to make predictions that are not just probabilistic.32 Even if that turns out to be the case, the methodological approaches that make use of the knowledge that we do have and can hope to achieve in the shorter term are entirely justifiable.
These alternative approaches to knowledge and understanding can be very instructive and productive. Indeed, a persuasive case can be made that our classically-derived Western view of science and scientific method, as reflected in classical mechanics as well as modern physics, threatens actually to be a serious limitation on or hindrance to comprehensive knowledge of the world. It has been asserted that this traditional scientific view is too narrow and restrictive, ignoring (or excluding) major aspects of human experience (including art, music, many emotions and, perhaps, even consciousness). See, e.g., Feyerabend, The Tyranny of Science, pp.6–12, 54–6, 68–9. Arguably, the dominance of a restrictive view of science based upon the model of classical physics may not just hinder the appreciation of other types of knowledge, it may also hinder the continued development of physics itself.
The Process of Science
The discussion so far is largely analytical, abstract and, necessarily, speculative. Significant contributions to our understanding of science were made during the twentieth century through efforts to examine the ways in which science is actually practiced, how creative discoveries have occurred and how, in fact, scientific knowledge has grown. The examination of the history of science in terms of the emergence and acceptance of new ideas provides useful insights into the methodological questions. One of the more important influences on the developments in the history of science was Thomas Kuhn's concept of paradigm shifts. Thomas S. Kuhn, The Structure of Scientific Revolutions (1962).
In simple terms, Kuhn explained how significant changes in science occur with a fundamental shift in the paradigm by which man understands the phenomena at issue. Kuhn viewed those shifts as revolutions. Before and after the shift, scientist will attempt to resolve experimental and observational problems by modifications and amendments to the dominant theory. At some point, the scientific community, in part influenced by general societal developments, becomes able to see the relationships under study in a new light, in and through a different paradigm. With time, the new worldview then becomes widespread in the educated community. Thus, we have the shift from the pre-Copernicus world to the post-Galileo world (whatever debates there may be about what happened in between) or from the Newtonian world to the world of Einstein.