Back to book


Stanford encyclopedia entry on Kuhn
A set of web resources on Kuhn, including summaries of the Structure of Scientific Revolutions


This is an essay I set my students for this topic:

"Discuss to what extent a domain of psychology of your choice has achieved the status of Normal science or whether it is pre-scientific. Integrate into your answer evidence published this year.”

Guidance on the essay:

"To answer the question, first we need to determine a domain of research. It does not seem right to ask ”Is psychoanalysis Normal science?” because that picks out a domain by a school of thought, and one key question in establishing whether an area is Normal is whether there is more than one school of thought. So one needs to identify a phenomenon in Nature (which includes humans) as the topic of research that one or more schools might be addressing; for example; ‘the role of the unconscious in personality', or ‘why do people sometimes systematically avoid the truth about themselves'. Identifying such a research problem to define a domain or research area can be problematic because what the problems are is said by Kuhn to be determined by one's paradigm. Nonetheless, there should be some broad description of the problem that picks out a common area of research that could be addressed by different schools of thought (e.g. the nature of self-deception, how children acquire syntax, the nature of depression, etc).

For science to be Normal there should be a community of scientists that are more or less united in addressing that question. Without a community there is no Normal science. How could one determine that there is a coherent community? There may be a journal they routinely use and cite, indicating some coherence. Maybe people in the area routinely cite each other, indicating some coherence. Maybe they attend conferences with the title of the research area, e.g. “depression”. On the other hand, if one found there were two journals in the domain that rarely if at all cited each other then there would be evidence of a lack of a coherent community – but rather a splintering. Maybe citation patterns indicate a set of disparate groups that rarely cite each other. You could briefly look at papers in the area published this year to get a tentative idea of this. Maybe different camps of people attend different conferences dealing with the same research area (e.g. biologically based and cognitively based people attend separate conferences on depression). Perhaps Googling your research area and ‘conference' will reveal some cues, or asking a tutor will inform you on this issue. Do not spend a lot of time looking at citations and conferences – it is a large research project in itself – but some Googling and asking tutors may reveal a tentative pattern. The existence of a splintering in the community as revealed by such an investigation does not in itself indicate a lack of Normal science. But if one established that the different communities had different values and models of good practice, and if they criticized each others' models, then science is not Normal in that domain. One could skip this step of trying to define a community sociologically and go straight to the next step, especially if you were already aware of potentially different communities (e.g. connectionist vs information processing approaches to language acquisition).

Given a community of people, for science to be Normal there should be in that community universally recognized examples of good practice: canonical experiments and/or methods or analyses of a problem. For example, are there textbooks for the domain, or sections in textbooks, and do the textbooks offer a standard set of classic experiments or methods, hailed as examples of excellent practice (i.e. paradigms in the narrow sense)? If textbooks routinely criticize the experiments or methods presented, then they are not paradigms. The paradigms do not need to be experiments per se (they often are in psychology), but any method of study or dealing with data (interviews, ways of extracting data in case studies, subjective judgments of the grammaticality of sentences, statistical analyses of speech corpora, etc). If textbooks do not refer to the domain in question, maybe you find that papers published this year use certain past examples as models of good practice, examples to be explored and varied, examples assumed to be models of what fine research looks like. That would be evidence for the existence of paradigms (in the narrow sense). It may be that one sub-group of a community focuses more on one set of canonical experiments than others, and vice versa in another subgroup. That may indicate a lack of a single paradigm: Or it may indicate just a concentration by some people on some issues rather than others. If the model experiments (say at biological, cognitive or social levels in explaining depression) seem to be mutually accepted, even if not always focused on to the same extent, and there is some notion that the different experiments ultimately link in a coherent way, then we may talk about a single paradigm with multiple levels. But if people using say a biological model criticize those who use a social model, and vice versa, so there is argument over fundamentals, then there is not a single accepted paradigm, and science is not Normal.

Note disagreement itself is not an indication of a lack of Normal science. Scientists will always disagree. But in Normal science there is mutual acceptance of certain canonical experiments or methods – it is just that people will argue about their proper application in particular cases. For example, researchers may accept that the methods for determining the involvement of neurotransmitters and receptor types is the proper way of understanding depression, yet argue bitterly over exactly which receptor subtypes the evidence points to. So arguing over theories is not in itself an indication that science is not Normal. Arguing over fundamentals, over what counts as the best methods for tackling the problem, what correct solutions look like - because there is disagreement over what counts as examples of good practice - does show science is not Normal.

If you find bickering over fundamentals, so science is not Normal, consider if it has always been this way – so the field is pre-scientific – or whether there was a single dominant paradigm and the field is now in crisis! Maybe asking a tutor in the field some pertinent questions will help you decide this issue.

In writing your essay, do not worry if you come to the ‘right' answer – you are doing new intellectual work and even experts would surely argue over what the right answer is – I just want to see that you understand Kuhn because of how you apply his ideas. As long as your arguments show you have taken on board Kuhn's ideas, it really does not matter whether I agree with your conclusions or not. In structuring your essay, please devote the large bulk of the essay to applying Kuhn's ideas to the domain of interest. So the following structure would not work well: A large section giving a textbook account of Kuhn; another section giving a textbook overview of the domain; a short section relating the two. Instead, after a short preamble, get right down to applying Kuhn to the domain and show me your understanding of Kuhn in how you do this. To help you think like Kuhn, try reading Kuhn so your head is full of his examples and phrases as you think about your domain and write about it.

You could – but need not – include some comment on the presuppositions of the essay question, i.e. on arguments for or against Kuhn's analysis of how science does or should work. But this should be kept brief, as the focus on writing an essay should always be on answering the actual question set. "


For tutors: A lecture on Kuhn I gave to our undergraduates this year. Takes about an hour. Feel free to adapt for your own purposes.




A 20-minute talk by Lakatos:
printed as the introduction to Lakatos. I. (1978). The methodology of scientific research programmes: Philosophical papers, vol 1. Cambridge University Press.

Stanford encylopedia entry on using history to justify philosophies of science, with discussion of both Kuhn and Lakatos:
Matheson, C. (1996). Historicist theories of rationality.

A lecture on Popper and Lakatos I gave late 2014.


This is an essay I set my students for this topic:

"Discuss to what extent your project or an empirical paper published this year contributed in a progressive or degenerating way to a research programme, according to Lakatos' account.”

Guidance on the essay:

"To answer the question, first determine the research programme defining the research. The research programme consists of a hard core, protective belt and positive heuristic, so outline what these are. This will require some thought and judgment on your part. The hard core should consist of ideas and theories that people have historically been committed to and resistant to giving up. That is, the hard core cannot be determined just by listing the set of claims made in a paper (i.e. claims made at a one point in time), though the past research discussed in the paper may give you an idea of the persisting hard core. "Chimps and humans share the same body language in expressing anxiety" it is not a hard core until people have used it in a committed way over a period of time to generate specific hypotheses which are themselves disposable. Thus, the hard core must be rich enough to generate a belt of specific hypotheses around it. A simple statement of a well confirmed empirical result ("people prefer goods they have paid more money for") is not a hard core in itself just because it seems resistant to being given up: It does not necessarily consitiute the motivating ideas powering the research programme. A statement of the mechanisms by which the result was produced is more likely to be a hard core (e.g. the mechanism of dissonance reduction: When people have conflicting cognitions, they change one to make them consistent). Different possible mechanisms (e.g. other than dissonance reduction) could provide the cores of different research programmes, all of which predict the same well established empirical result. The hard core is what gives the research programme its identity. The positive heuristic is related to Lakatos' insight that mature science is not just a collection of random statements - there is some unity or coherence to a mature research programme. Thus, the positive heuristic reflects the hard core, it is not an unrelated set of rules for generating theories. The dissonance reduction research programme may contain a positive heuristic like "Consider alternative possible cognitions subjects might have been having in this experiment, and thus which conflicting ones might have been modified". Having identified the research programme, see if the paper makes novel falsifiable predictions (theoretically progressive) and whether they are confirmed (empirically progressive). The prediction will be testing a hypothesis in the protective belt, inspired by the hard core and ideally guided by the positive heuristic of that research programnme."

See also this assessment of several topics from the book.

A task I set students to discuss in lectures is identifying a hard core and protective belt for a research programme in an area of psychology of their choice. Please contact me if you have devised other simple student activities for teaching Kuhn or Lakatos.