Reference Material:
The Structure of Scientific Revolutions by Thomas Kuhn
The ideas in this blog have been built largely around the philosophical worldview of naturalism. We’ve interrogated to some limited extent what that means, mostly by delving into areas (the mind, morality, meaning in life) where a bare-bones physicalism seems to leave us wanting more in terms of explanation or understanding. What hasn’t been said is anything directly in favor of the actual worldview itself, outside of its implications. The brand of naturalism I’d argue for leans heavily on science – it’s results, methodology, practice, justification, etc. So the blog will now take a multi-post turn into the philosophy of science. We start with one of the most well-known works in the field, and from my perspective, one of the most controversial – Thomas Kuhn’s Structure of Scientific Revolutions.
The discussion will be largely critical of the ideas in the book, but let’s start off with some places where Kuhn and I really agree. Some of Kuhn’s general descriptions of how science operates and proceeds I think are quite good. He describes (normal, paradigmatic) science as consisting of “determination of significant fact, matching of facts with theory, and articulation of theory” (34). Each of these activities helps “add to the scope and precision with which the paradigm can be applied” (36), creating “a highly cumulative enterprise, eminently successful in its aim, the steady extension of the scope and precision of scientific knowledge” (52).
That’s how so-called normal science progresses. What about Kuhn’s revolutions? How do we get new theories that expand the boundaries of what we can explain? When new data clash with the predictions of existing theories, do scientists tend to try to adapt the current theory to account for the anomalous data, or do they opt for a riskier approach, developing a new theoretical framework? According to Kuhn, it’s some of both, and that’s probably a good thing. His description of this balance is worth quoting at length:
I’m quite sure I couldn’t have stated that better.
It’s in his other detailed treatments of these revolutions that I think he missteps. In many places he seems to me to exaggerate the differences between two theories on either side of a scientific revolution. He often characterizes two such theories as “incompatible”, “irreconcilable”, or “incommensurable” (103). The new theory displaces the old theory in a “decisively destructive” (102) transformation, rendering the old theory “wrong” (98). To articulate why I consider these characterizations exaggerated, we’ll consider the paradigmatic example which Kuhn treats at some length, the transition in our understanding of mechanics and gravity from Newtonian mechanics to the special and general theories of relativity. Kuhn considers these theories “fundamentally incompatible” – “Einstein’s theory can be accepted only with the recognition that Newton’s was wrong” (98).
The first point I’d make is to reiterate that “incompatible” and “wrong” are, at best, exaggerations. Newtonian mechanics works perfectly well at relatively low velocities and weak gravitational fields – indeed, entire fields of science and engineering rely on this fact (e.g. fluid mechanics, structural engineering, etc.). This means Newtonian mechanics can be considered a special case of relativity, in some sense. Kuhn admits this, but then goes on to strawman the rest of the objection:
He then points out that “some variant of this argument is quite sufficient to make any theory… immune to attack” (99). True, but a bit beside the point. The objection is that Newtonian mechanics is still true across a certain range of conditions, in that it yields the same correct predictions at these conditions as relativity does. And the range of conditions over which Newtonian mechanics works in this way is actually difficult to overstate. In terms of velocities, almost every macroscopic object humans have ever observed or designed can be treated with Newtonian mechanics because such objects are never accelerated to any appreciable fraction of the speed of light. In terms of gravitational fields, Newtonian mechanics will suffice for most applications on the surface of the Earth, but, for example, predicting satellite orbits with sufficient accuracy requires general relativity. In terms of length scales, engineers use Newtonian mechanics (and its theoretical extensions) for the micromechanics of material failure, the structural and fluid mechanics of a skyscraper on a windy day, and everything in between. Again, calling this theory “wrong” and “incompatible” with relativity is really overstating the truth of the matter.
But even if Kuhn admitted that he built up the strawman above somewhat unfairly, it seems he would still be reluctant to accept the force of the objection. The point I’m trying to make relies on the fact that Newtonian mechanics and relativity yield the same predictions across a wide range of phenomena. Indeed, you can use relativity, in concert with some limiting assumptions that define that range of phenomena, to demonstrate why this is the case. Kuhn is aware of this, but thinks there is a “logical lacuna” that undermines this argument (101). His basic idea is that, even after you take relativity theory, add the limiting assumptions, and derive a set of laws that look like Newtonian mechanics, the variables in this new set of laws are still Einsteinian, not Newtonian. He gives the example of mass – “Newtonian mass is conserved; Einsteinian is convertible with energy. Only at relatively low velocities may the two be measured in the same way, and even then they must not be conceived to be the same” (102). The result is that “unless we change the definitions of the variables…, the statements we have derived are not Newtonian. If we do change them, we cannot properly be said to have derived Newton’s Laws”, and so
I have to say my initial reaction to this is one of frustration. Again, the objection to characterizing Newtonian mechanics and relativity as “fundamentally incompatible”, and hence the former as “wrong”, is that across a wide variety of situations, they both yield the same, correct predictions. At low velocities and relatively small gravitational fields, they are both right. The lesson we should draw from the mass example is not that we have failed to show that relativity reduces to Newtonian mechanics at these conditions, but that at these conditions, contrary Kuhn, Einsteinian mass is equivalent to Newtonian mass.
I think the most frustrating thing about Kuhn’s response here is this. Relativity reproduces the successful predictions of Newtonian mechanics at the appropriate conditions. In addition, at more extreme conditions, it makes novel successful predictions for phenomena that are inconsistent with their Newtonian counterparts (e.g. Mercury’s orbit). In this way, relativity builds on Newtonian mechanics in terms of breadth of successful predictions and explanations. The relationship between the two theories is really very well described by Kuhn himself – science is “a highly cumulative enterprise, eminently successful in its aim, the steady extension of the scope and precision of scientific knowledge” (52). Frustratingly, he of course goes on to argue that the revolution from Newtonian mechanics to relativity is in fact not an example of such a successful enterprise.
That’s probably enough of a diatribe for now, but I can’t help but leave you with one more Kuhnian exaggeration. In a discussion of how science is taught, and in particular the relationship between theory and application in science textbooks, Kuhn makes the following, rather astonishing remark – “But science students accept theories on the authority of teacher and text, not because of evidence. What alternatives have they, or what competence?” Kuhn studied physics at Harvard, and I can’t speak to his experience there. But as a former physics student myself, I must say I’m amazed – I can only characterize his description as an outrageous over-socialization of the process of scientific learning. Just one example – yes, I believed the textbook when it gave the speed of light as ~3*10^8 m/s. I believed it because it was, well, a textbook. But I believed it even more, and now have a better justification for believing it, when I ran an experiment to measure it directly, and produced a result within 1% of that value. Kuhn’s statement is such an oversimplification, it approaches completely misleading. Coming from one of the most well-known (at least to me) works in the philosophy of science, I must say he disappoints.
The Structure of Scientific Revolutions by Thomas Kuhn
The ideas in this blog have been built largely around the philosophical worldview of naturalism. We’ve interrogated to some limited extent what that means, mostly by delving into areas (the mind, morality, meaning in life) where a bare-bones physicalism seems to leave us wanting more in terms of explanation or understanding. What hasn’t been said is anything directly in favor of the actual worldview itself, outside of its implications. The brand of naturalism I’d argue for leans heavily on science – it’s results, methodology, practice, justification, etc. So the blog will now take a multi-post turn into the philosophy of science. We start with one of the most well-known works in the field, and from my perspective, one of the most controversial – Thomas Kuhn’s Structure of Scientific Revolutions.
The discussion will be largely critical of the ideas in the book, but let’s start off with some places where Kuhn and I really agree. Some of Kuhn’s general descriptions of how science operates and proceeds I think are quite good. He describes (normal, paradigmatic) science as consisting of “determination of significant fact, matching of facts with theory, and articulation of theory” (34). Each of these activities helps “add to the scope and precision with which the paradigm can be applied” (36), creating “a highly cumulative enterprise, eminently successful in its aim, the steady extension of the scope and precision of scientific knowledge” (52).
That’s how so-called normal science progresses. What about Kuhn’s revolutions? How do we get new theories that expand the boundaries of what we can explain? When new data clash with the predictions of existing theories, do scientists tend to try to adapt the current theory to account for the anomalous data, or do they opt for a riskier approach, developing a new theoretical framework? According to Kuhn, it’s some of both, and that’s probably a good thing. His description of this balance is worth quoting at length:
Individual variability in the application of shared values may serve functions essential to science. The points at which values must be applied are invariably also those at which risks must be taken. Most anomalies are resolved by normal means; most proposals for new theories do prove to be wrong. If all members of a community responded to each anomaly as a source of crisis or embraced each new theory advanced by a colleague, science would cease. If, on the other hand, no one reacted to anomalies or to brand-new theories in high-risk ways, there would be few or no revolutions. In matters like these the resort to shared values rather than to shared rules governing individual choice may be the community’s way of distributing risk and assuring the long-term success of its enterprise. (186)
I’m quite sure I couldn’t have stated that better.
It’s in his other detailed treatments of these revolutions that I think he missteps. In many places he seems to me to exaggerate the differences between two theories on either side of a scientific revolution. He often characterizes two such theories as “incompatible”, “irreconcilable”, or “incommensurable” (103). The new theory displaces the old theory in a “decisively destructive” (102) transformation, rendering the old theory “wrong” (98). To articulate why I consider these characterizations exaggerated, we’ll consider the paradigmatic example which Kuhn treats at some length, the transition in our understanding of mechanics and gravity from Newtonian mechanics to the special and general theories of relativity. Kuhn considers these theories “fundamentally incompatible” – “Einstein’s theory can be accepted only with the recognition that Newton’s was wrong” (98).
The first point I’d make is to reiterate that “incompatible” and “wrong” are, at best, exaggerations. Newtonian mechanics works perfectly well at relatively low velocities and weak gravitational fields – indeed, entire fields of science and engineering rely on this fact (e.g. fluid mechanics, structural engineering, etc.). This means Newtonian mechanics can be considered a special case of relativity, in some sense. Kuhn admits this, but then goes on to strawman the rest of the objection:
But, the objection continues, no theory can possibly conflict with one of its special cases. If Einsteinian science seems to make Newtonian dynamics wrong, that is only because some Newtonians were so incautious as to claim that Newtonian theory yielded entirely precise results or that it was valid at very high relative velocities. Since they could not have had any evidence for such claims, they betrayed the standards of science when they made them. In so far as Newtonian theory was ever a truly scientific theory support by valid evidence, it still is. Only extravagant claims for the theory – claims that were never properly parts of science – can have been shown by Einstein to be wrong. Purged of these merely human extravagances, Newtonian theory has never been challenged and cannot be. (99)
He then points out that “some variant of this argument is quite sufficient to make any theory… immune to attack” (99). True, but a bit beside the point. The objection is that Newtonian mechanics is still true across a certain range of conditions, in that it yields the same correct predictions at these conditions as relativity does. And the range of conditions over which Newtonian mechanics works in this way is actually difficult to overstate. In terms of velocities, almost every macroscopic object humans have ever observed or designed can be treated with Newtonian mechanics because such objects are never accelerated to any appreciable fraction of the speed of light. In terms of gravitational fields, Newtonian mechanics will suffice for most applications on the surface of the Earth, but, for example, predicting satellite orbits with sufficient accuracy requires general relativity. In terms of length scales, engineers use Newtonian mechanics (and its theoretical extensions) for the micromechanics of material failure, the structural and fluid mechanics of a skyscraper on a windy day, and everything in between. Again, calling this theory “wrong” and “incompatible” with relativity is really overstating the truth of the matter.
But even if Kuhn admitted that he built up the strawman above somewhat unfairly, it seems he would still be reluctant to accept the force of the objection. The point I’m trying to make relies on the fact that Newtonian mechanics and relativity yield the same predictions across a wide range of phenomena. Indeed, you can use relativity, in concert with some limiting assumptions that define that range of phenomena, to demonstrate why this is the case. Kuhn is aware of this, but thinks there is a “logical lacuna” that undermines this argument (101). His basic idea is that, even after you take relativity theory, add the limiting assumptions, and derive a set of laws that look like Newtonian mechanics, the variables in this new set of laws are still Einsteinian, not Newtonian. He gives the example of mass – “Newtonian mass is conserved; Einsteinian is convertible with energy. Only at relatively low velocities may the two be measured in the same way, and even then they must not be conceived to be the same” (102). The result is that “unless we change the definitions of the variables…, the statements we have derived are not Newtonian. If we do change them, we cannot properly be said to have derived Newton’s Laws”, and so
the argument has still not done what it purported to do. It has not, that is, shown Newton’s Laws to be a limiting case of Einstein’s. For in the passage to the limit it is not only the forms of the laws that have changed. Simultaneously we have had to alter the fundamental structural elements of which the universe to which they apply is composed. (102)
I have to say my initial reaction to this is one of frustration. Again, the objection to characterizing Newtonian mechanics and relativity as “fundamentally incompatible”, and hence the former as “wrong”, is that across a wide variety of situations, they both yield the same, correct predictions. At low velocities and relatively small gravitational fields, they are both right. The lesson we should draw from the mass example is not that we have failed to show that relativity reduces to Newtonian mechanics at these conditions, but that at these conditions, contrary Kuhn, Einsteinian mass is equivalent to Newtonian mass.
I think the most frustrating thing about Kuhn’s response here is this. Relativity reproduces the successful predictions of Newtonian mechanics at the appropriate conditions. In addition, at more extreme conditions, it makes novel successful predictions for phenomena that are inconsistent with their Newtonian counterparts (e.g. Mercury’s orbit). In this way, relativity builds on Newtonian mechanics in terms of breadth of successful predictions and explanations. The relationship between the two theories is really very well described by Kuhn himself – science is “a highly cumulative enterprise, eminently successful in its aim, the steady extension of the scope and precision of scientific knowledge” (52). Frustratingly, he of course goes on to argue that the revolution from Newtonian mechanics to relativity is in fact not an example of such a successful enterprise.
That’s probably enough of a diatribe for now, but I can’t help but leave you with one more Kuhnian exaggeration. In a discussion of how science is taught, and in particular the relationship between theory and application in science textbooks, Kuhn makes the following, rather astonishing remark – “But science students accept theories on the authority of teacher and text, not because of evidence. What alternatives have they, or what competence?” Kuhn studied physics at Harvard, and I can’t speak to his experience there. But as a former physics student myself, I must say I’m amazed – I can only characterize his description as an outrageous over-socialization of the process of scientific learning. Just one example – yes, I believed the textbook when it gave the speed of light as ~3*10^8 m/s. I believed it because it was, well, a textbook. But I believed it even more, and now have a better justification for believing it, when I ran an experiment to measure it directly, and produced a result within 1% of that value. Kuhn’s statement is such an oversimplification, it approaches completely misleading. Coming from one of the most well-known (at least to me) works in the philosophy of science, I must say he disappoints.