Kuhn on Popper ,'normal science' and  'paradigms'

[These remarks are mostly drawn from Kuhn's famous work The Structure of Scientific Revolutions]

Popper's view is that science proceeds by learning from mistakes, but what is a mistake? It is easy to see when individuals make a mistake but what about the "mistakes" made by whole systems? The problem is that testing is not independent of theory. Indeed, tests are often suggested by theory and only become possible if one accepts the theory. The results of testing also vary -- disagreements between observations and theory need not lead to the falsification of the theory. Such disagreements can be left as an unresolved puzzle. Moreover, empirical tests alone are not usually sufficient grounds for rejecting a theory, as the dispute between Copernicus and Ptolemy indicates (see below).

Is falsification a strictly logical procedure? It is clear that one statement can be seen as a logical contradiction of the others, but falsification is supposed to include evidence from the real world too. This can present a problem because there is no one-to-one correspondence between statements and the real world -- to say there is a grasshopper on my desk is to apply a set of agreed, learned discriminations and classifications. This can be seen if you invite other people to say whether or not there is a grasshopper on my desk. Problems arise immediately such as people asking:  'Which is your desk?';  'It is that what you mean by a desk [not a table ]?';  'Is that what you mean by a grasshopper  [not a locust, insect, model etc]?'.

We can only use these statements objectively if we agree on these discriminations. This is recognised by Popper when he says that  'basic statements' are used in falsification, not just immediate sense data. He recognises that these 'basic statements' must involve some agreement on using discriminations, but fails to show how these 'basic statements' arise.  [For Kuhn, socialisation, learning, and collective intentions held in the scientific community are the source].

This leads to a further confusion. A theory only gets refuted by a series of 'basic statements' if the scientific community itself agrees first on what a 'basic statement' means, and second whether this particular 'basic statement' is one which raises problems for theory. This is never a purely logical process, but much more akin to a change in seeing the world.

As an example, Aristotle saw a swinging pendulum as a body trying to fall to its  'natural' position. Galileo's breakthrough was to see a pendulum as a very different phenomenon, as the equivalent to a ball rolling down an inclined plane and up the other side. The  'same' observation had very different implications for these two explanations of motion. In this way, observations cannot be seen as some sort of neutral data which can be used to test theories, since observations are closely linked to theories. Theories suggest observations, and conversely, observations involve some  'theoretical' set of perceptions and discriminations, although these are not always explicit. Only when these underlying perceptions and discriminations change can views of the world change, and familiar observations get interpreted differently.

The history of scientific growth itself reveals considerable differences from Popper's account. Revolutionary changes in scientific theories occur for many reasons, not only because the existing theory has become falsified. For example:

(1) Copernicus and Ptolemy. 
(a) the Ptolemaic system was already seen as largely unsatisfactory before Copernicus's theory; 
(b) Ptolemy and Copernicus were both able to explain the existing data; 
(c) the persuasive evidence for the Copernican theory was not forthcoming until long after Copernicus's theory had been accepted by the scientific community -- new mathematical techniques developed on the basis of an acceptance helped to show Ptolemy's errors were greater than Copernicus's.

(2) Priestley and Lavoisier.
(a) the phlogiston theory and oxygen theory [of combustion] co-existed for quite a long time, and the existing evidence was unable to conclusively accept or reject one of them;
(b) the theories appealed to different sorts of evidence (phlogiston theory was able to explain qualitative evidence like why metals are similar while ores are not, while oxygen theory was able to explain weight losses and gains during combustion). In this case, Lavoisier triumphed only after the scientific community had agreed that quantitative data was more important than qualitative data in chemistry. There had been a long process of conversion to this view, rather than a decisive logical compulsion.

In these and in most cases, conclusive empirical falsification of theories was not the decisive factor in explaining changes in science. An appeal to evidence was important, but only after the scientific community had decided what counts as evidence, what sort of evidence they should find, what level of accuracy is acceptable, and what kind of problems they want to explain.

This last point needs to another of Kuhn's key arguments. Unlike Popper, Kuhn says that most scientific activity is not actually about attempts to overthrow theoretical frameworks by falsification and bold conjecture. Most science instead consists of attempts to solve problems ('puzzles') within existing frameworks. This is  'normal science', and it implies the acceptance of existing theoretical frameworks. Without such acceptance, there would be no commitment to the lengthy work of developing applications, measurements and experimental technique. The existence of a puzzle is a case where theory is problematic, but there is an assumption that existing theory can cope, that it will solve the puzzle in the end.

Most scientists spend their lives solving puzzles, but even a failure to solve a puzzle need not cast doubt on a theoretical framework. Failure can result from the individual faults of scientists; it can be left as something that will be solved at some future date; or, it can be seen as a serious anomaly casting doubt upon the whole theoretical apparatus, as Popper thinks. There are no hard and fast logical rules for deciding how to interpret failures.

By contrast, making progress within normal science is undoubtedly important in leading to theory change. It helps scientists pin down, specify and eliminate puzzles which may gradually uncover an anomaly that cannot be solved. This process of steady work inside an accepted theory is the hallmark of science for Kuhn. It is this practice that demarcates science from other approaches, not falsifiability as such.

Science therefore is a characteristic activity that produces the collective pursuit of puzzle solving, in a continuous effort to apply theory, tease out the implications, attempt to be more and more precise, and establish more and more work within an existing framework. Without this background activity of established work, falsifiability in Popper's sense is unlikely.

Kuhn discusses astrology as does Popper. Briefly, astrology is not a science for Popper because it is not falsifiable. It is not falsifiable because it does not take enough risks -- its predictions can always be rescued. For Kuhn, astrology is not a science because it did not yield enough puzzles to solve, and thus did not progress and develop into normal science. Astrology have a number of rules, laws and calculations, but lots of its variables were uncontrollable -- it needed exact dates of birth, data on the exact position of stars and so on. These were never pinned down, and so astrology could never develop precise scientific calculations. It made no progress as a result. Once the basic ideas were established, no further research was possible, and astrology never developed into normal science. As a result, it never managed to properly isolate falsifying examples. This is why it has never been falsified -- it simply disappeared as a possible science, and the scientific community lost interest in it because there were no puzzles left to solve [leaves room for the importance of being able to develop a research programme as in Lakatos).

So, nor sciences are very important part of scientific progress. Normal science is taken for granted in Popper's account. What is it that gives scientists do commitment to do normal science? How do they require and come to believe in theoretical frameworks? Kuhn says that they are educated or socialised into seeing normal science as worthwhile and fruitful. How?

Scientists acquire their knowledge and belief through paradigms. There is some looseness in the use of this term, and some attempt to clarify it in Kuhn's later work. The basic sense of the term refers to ways in which scientists learn to do particular kinds of science. As we have argued above, scientists have to learn to perceive the world in a particular way, and they have to be committed to particular theoretical frameworks. This commitment does not depend on explicit logical rules and arguments, if only for the reason that these are often not available at the time  [and scientists are not logicians or philosophers]. Instead , scientists learn by example. The textbooks they use when they are learning science contain expositions of theories, classical experiments, and examples for them to try for themselves.

Learning science is never a matter of automatically applying techniques, since interpretation is also involved. Only when they have worked through the key examples can they come to see the world as established scientists see it. Trainee scientists need to reflect on what they have read and understood and try to see the world that way. One process is learning to see new cases as the same things as the classic examples in the books -- to recognise an actual pump as the source of 'increased pressure' in theory. [My own favourite personal example involves the great day when I was able to see a particular piece of the landscape in front of me as a  'valley' as described in geography textbooks. I am also constantly reminded of the point when walking in the countryside with my wife: where I see a range of plants varying from little ones to bigger ones, she sees hundreds of species ranging from 'shepherd's purse' to 'hollyoak' ]. The whole process is developed and learned through paradigmatic experiences and examples -- classic examples which scientists use to educate newcomers to see the world as they do.

Overall, then, science is not a completely logical process. There are learned non-logical elements in it too. The scientific community and its activities is vital to understanding the development, growth and change in scientific theory. 

[Here, we can point out what Kuhn has taken for granted -- he needs a good sociological analysis of the scientific community instead of the rather naively functionalist one that he seems to operate with. What actually is the scientific community? Is it socially united or divided? What keeps scientists in it? What do scientists actually do when they go round deciding what to accept as science, and do they do this right from the beginning, or only after they have actually come to write up the results?]