How do scientists (or anyone else, for that matter) acquire knowledge? Even Aristotle acknowledged that the foundations of knowledge, since they are based in sensory perceptions, must come from induction, even as he laid the foundations of deductive logic. Unfortunately, philosophers (including Aristotle) have tended to prioritize deductive logic as inherently superior to inductive logic.
Enter the scientific revolution ... which was built primarily on the application of inductive reasoning. Empiricists and skeptics, who by and large supported this revolution as a superior way to gain knowledge, began to realize that the emphasis on inductive reasoning posed a bit of a problem. This problem of induction was best formulated by the Scottish philosopher David Hume. To use the classic example of skepticism, even if you have seen 10,000 sunrises, this does not actually provide the basis of certainty that the sun will rise tomorrow. in the early twentieth century was transformed into the foundation of the modern philosophy of science which, at its worst, denied that science could actually discover truths about the world.
In this 2010 book, The Logical Leap: Induction in Physics, physicist David Harriman tackles the problem of induction in physics head on. His basic argument is that a proper method of inductive reasoning is founded not merely upon taking a series of perceived results and creating a model to account for them, but of specifically outlining the series of causal relationships that result in the phenomena in question. If done correctly, with each step in the inductive process coming out of the foundational understanding of existing concepts.
To support his claim, Harriman painstakingly details the method by which great scientist such as Galileo Galilei and Sir Isaac Newton (as well as fairly detailed accounts of Kepler's laws of planetary motion and the development of atomic theory) actually developed their theories. He shows that they did not merely create models to describe inexplicable phenomena, but thought about what sort of causes might be at work in order to create the models. Though Newton had no idea why gravity worked (that would have to wait for Albert Einstein), he had a clear idea of the concept of gravity and how gravity behaved, and it was that concept of gravity as a real thing that guided his theoretical investigation ... with Newton actively engaged in modifying the theory as needed to conform to empirical evidence.
The book is a powerful and well-articulated approach to the problem and I think should be required reading for any scientist, particularly those interested in the foundations of knowledge, scientific validity, and the philosophy of science. The problem of induction is a thorny one and this approach to resolving it has a lot of merit ... though it is certainly not a slam dunk that will convince all rationalists and skeptics.
There are a number of issues in the book which warrant some warnings and considerations.
First, the book's arguments are firmly rooted in the foundation of Ayn Rand's objectivist epistemology, which involves a theory of concepts. I have read little of Rand's philosophy and most familiar with it only in the political realm, where I've got to say that I'm not personally very impressed by the policies her supporters advocate. As such, I don't know how broadly accepted her "theory of concepts" is and whether there are perhaps attacks against this epistemological approach that might pose a hindrance to Harriman's argument as well.
Second, Harriman doesn't provide much detailed context in the existing philosophy of science. He seems to consider his rationalist and skeptical empiricist adversaries to be so fundamentally wrong-headed in their thinking that presenting an appraisal of their internal logic and arguments isn't needed. A reader of the book would be led to believe, for example, that Karl Popper's theory of falsifiability and Thomas Kuhn's concept of "scientific paradigms" are incorrect, but they wouldn't be able to explain those ideas particularly well unless they've read elsewhere on the subject.
Third, Harriman's arguments from the history of science are themselves particularly limited, and it's not clear how they can really be extended beyond their historical perspective into areas relevant for modern scientific inquiry. He is dismissive of modern research methods, which focus strongly on identifying empirical probabilities instead of causal relationships.
Consider, for example, medical research that seems to require this sort of probabilistic approach. The most it can ever yield, it seems to me, is the notion that if you take drug X then your chance of getting disease Y will decrease by Z percentage on average. Other forms of scientific research - sociology, economics, psychology, and so on - would seem to be in a similar pickle. By Harriman's view, it seems like they would be unable to make a clear claim to ever applying legitimate inductive reasoning, because the causal relationships in these systems are so complex that they cannot be investigated in the same way that a chemical compound can be studied.
One big failure of Harriman's argument is his cursory treatment of quantum physics. Quantum physics represents the greatest challenge to his argument, I think, and he glosses over it, indicating that the thinking behind it is wrong-headed but not actually proposing an alternative approach. But he does acknowledge that the mathematical formalism of quantum physics has matched empirical verification with flying colors, unheard of in any other area of human knowledge. While he might not like that quantum physicists do not understand or investigate the root causes of the quantum phenomena they're working with, the fact that the formalism matches experiment this precisely cannot be viewed as a meaningless accident. Clearly, the quantum physicists are doing something right! It is dismaying that he does not devote as much attention to the failures of quantum mechanics as he does to the successes of atomic theory, because I think there are potentially some insights to be gained there.
Not surprisingly, he has some words to say about the big bang theory and string theory as well, viewing them as areas where science is profoundly overstepping the bounds of what can be known. He strikes me as a little quick to judge these as failures and I think some further analysis of them (more than the couple of pages they are afforded) would definitely go a long way toward fleshing out the theory of inductive knowledge that he advocates.
Despite these flaws, as described above, I feel like the book helps illuminate the power of inductive reasoning within the physical sciences. It is not conclusive, but rather represents a good first step in a revised way of looking at the problem of induction.
Publication Date: July 2010
Publisher: New American Library
Introduction, Preface, 7 chapters, References, and Index
Galileo had fought the Church in order to expel religious faith from the realm of science; Newton fought his fellow scientists in an effort to expel the arbitrary as such, including secular claims. The appeal to faith is the demand that ideas be accepted on the basis of emotion rather than evidence, and it is therefore a species of the arbitrary.
When we have a properly formed concept, one that unites concretes by clearly defined essentials, we are often in a position to know at once when an attribute discovered by study of some instances is applicable to all instances.[...] By doing so, one is claiming: "This is now part of my knowledge of X, i.e., this is true of all X's--including the vast majority of them that I will never encounter." A person who refrained from induction would find that his words did not designate concepts at all; they would be reduced to sounds.
Although a valid conceptual framework does not guarantee the truth of subsequent generalizations, the errors of generalization committed by scientists can usually be traced to some inadequacy in their conceptual framework. When scientists overgeneralize (i.e., extend their conclusion beyond its legitimate range of validity) it is often because they lack the concepts necessary to identify important distinctions.
Mathematics is not a "pure," isolated string of abstractions and deductions--and if it were, it would be nothing but a useless game. Rather, it is the science of relating quantities to one another, quantities that are ultimately related to perceivable objects.
It is important to recognize that the causal knowledge necessary to prove a generalization is not the same as the causal knowledge from which the generalization can be deduced. It is a common error to substitute the latter for the former in an effort to reduce all logical reasoning to deduction.
The rationalists typically appeal to the symmetry, elegance, and beauty of their mathematical theory, claiming that such pleasing aesthetic features imply real insight into the world of Forms (Plato) or the mind of the Creator (any religion). Like the skeptics, they introduce a breach between physical causation and mathematics. The skeptics do so because they regard causes as unknowable and mathematical formalism as arbitrary; the rationalists in physics do so because they hold that the fundamental causes are nonphysical and that the laws of nature can be grasped independent of sensory contact with the physical world.
A science is an integrated body of knowledge, and in the physical sciences this integration is made possible by mathematics.
[...] scientific knowledge is not the floating abstractions of rationalists or the perceptual-level descriptions of empiricists; it is the grasp of causal relationships identified by means of the inductive method.
In physical science, qualitative truths are mere starting points; they can suggest a course of investigation, but that investigation is successful only when it arrives at a causal relationship among quantities. Then the power of mathematics is unleashed; connections can be identified between facts that had previously seemed entirely unrelated[...]. Mathematics is the scientist's means of integrating his knowledge.