Rather than observing and judging science as a sequence of individual theories, Lakatos instead argues that disciplines are best conceptualized as a series of scientific research programs (SRPs). SRPs have four elements: a hard core; a negative heuristic; a positive heuristic; and a protective belt of auxiliary hypotheses (see “A Brief Guide to Imre Lakatos’s Methodology of Scientific Research Programs”
in Chapter 1 on pp. 19–20). The program’s hard core assumptions consist of unchanging content.15 They are protected by a “negative
new meaning to the phrase ‘badly written,’ and no reading of such abominable writing is ever certain or final.” Stephen Van Evera, Guide to Methods for Students of Political Science (Ithaca, N.Y.: Cornell University Press, 1997), p. 44, note 55.
14. See, for example, John Worrall, “The Ways in Which the Methodology of Scientific Research Programmes Improves on Popper’s Methodology,” in Gerard Radnitzky and Gunnar Andersson, eds., Progress and Rationality in Science (Dordrecht: D. Reidel, 1978), pp. 45–70; Douglas W. Hands, “Second Thoughts on Lakatos,” History of Political Economy, Vol. 17, No. 1 (1985), pp. 1–
16; and Leplin, A Novel Defense of Scientific Realism.
15. Although the elements of a program’s hard core may not change, it is generally agreed that they can become better specified over time. See Alan Musgrave, “Method or Madness? Can the Methodology of Research Programmes be Rescued from Epistemological Anarchism?” in Cohen, Feyerabend, and Wartofsky, Essays in Memory of Imre Lakatos, pp. 458–467;
Henry Frankel, “The Career of Continental Drift Theory: An Application of Imre Lakatos’s Analysis of Scientific Growth to the Rise of Drift Theory,”
heuristic,” a set of propositions that say that this content cannot be directly challenged or tested. For example, one of the least controversial elements of the hard core of the neorealist SRP is that states are the dominant actors in international relations. It follows that the negative heuristic of the neorealist research program includes a restriction against developing theories that give a decisive causal role to non-state actors such as Greenpeace or the United Nations (if a scientist does so, she is no longer operating within the neorealist scientific research program).
Scientific research programs also have a “protective belt” of auxiliary hypotheses that “bear the brunt of tests and get adjusted and re-adjusted, or even completely replaced, to defend ... the core.”16 As Spiro J. Latsis notes, the “protective belt may consist of various types of propositions [ranging] from specific auxiliary hypotheses, accounting for predictive failure, to redefinitions of the conceptual apparatus.”17 In the neorealist SRP, for example, it is common practice to distinguish between scholars who assert that states are power-maximizing revisionists, and those who contend that states defend the status quo and minimize relative power losses.18 It is the particular costs and
Studies in History and Philosophy of Science, Vol. 10, No. 1 (1979), pp. 26–28; and Neil de Marchi, “Introduction: Rethinking Lakatos,” in Neil de Marchi and Mark Blaug, eds., Appraising Economic Theories: Studies in the Methodology of Research Programs (Brookfield, Vt.: Edward Elgar, 1991), p. 12. For the critique that hard cores do not exist in science, see William Berkson, “Lakatos One and Lakatos Two: An Appreciation,” in Cohen, Feyerabend and Wartofsky, Essays in Memory of Imre Lakatos, p. 52. For the opposite critique that the elements of scientific research programs, including hard cores, are so pervasive in all intellectual endeavors that Lakatos’s methodology fails to distinguish science from non-science, see Caldwell, “The Methodology of Scientific Research Programmes in Economics,” p. 99.
16. Lakatos, “Falsification and the Methodology of Scientific Research Programmes,” p. 133.
17. Spiro J. Latsis, “A Research Programme in Economics,” in Spiro J. Latsis, ed., Method and Appraisal in Economics (New York: Cambridge University Press, 1976), p. 23.
18. John J. Mearsheimer, 1994/95. “The False Promise of International Institutions,” International Security, Vol. 19, No. 3 (Winter 1994/95), pp. 5–59, at
benefits suggested by these and other elements of the protective belt, in combination with the unchanging hard core, that lead different realists to predict different state behaviors and international political outcomes.19
The SRP’s protective belt is developed in accordance with the program’s positive heuristic, “a partially articulated set of suggestions or hints”20 that “guides the production of specific theories within the programme.”21 The positive heuristic of the program contains:
a set of ideas about how to “fill in,“ make more precise, draw consequences from [statements about the world], and also how to elaborate on them, introduce new assumptions so that they apply to new fields, and how to modify them when difficulties arise.22
The neorealist scientific research program’s positive heuristic would include the suggestion that scholars develop theories that make 11–12, note 27; Benjamin Frankel, “Restating the Realist Case: An Introduction,” Security Studies, Vol. 5, No. 3 (Spring 1996), pp. ix–xx, xv–xviii;
Joao Resende-Santos, “Anarchy and the Emulation of Military Systems:
Military Organization and Technology in South America, 1870–1930,” Security Studies, Vol. 5, No. 3 (Spring 1996), pp. 193–260, note 34; and Eric J. Labs,
“Beyond Victory: Offensive Realism and the Expansion of War Aims,” Security Studies, Vol. 6, No. 4 (Summer 1997), pp. 1–48.
19. Colin Elman, “Appraising Neorealism as a Scientific Research Program,”
unpublished manuscript, Arizona State University, 1997.
20. Lakatos, “Falsification and the Methodology of Scientific Research Programmes,” p. 135.
21. Worrall, “The Ways in Which the Methodology of Scientific Research Programmes Improves on Popper’s Methodology,” p. 59.
22. Ibid. See also Latsis, “A Research Programme in Economics,” p. 16; and John Worrall, “Research Programmes, Empirical Support and the Duhem Problem: Replies to Criticism,” in Radnitzky and Andersson, Progress and Rationality in Science, pp. 321–322, 328–329. For a list of the elements one might find in a positive heuristic, see Worrall, “The Ways in Which the Methodology of Scientific Research Programmes Improves on Popper’s Methodology,” p. 69.
For examples of positive heuristics in research programs from economics, see A.W. Coats, “Economics and Psychology: The Death and Resurrection of a Research Program.” in Latsis, Method and Appraisal in Economics, p. 54; and Latsis, “A Research Programme in Economics,” pp. 22–23.
predictions about international political outcomes (for example, that balances tend to form in the international system, or that multipolar systems will be more war-prone than bipolar systems). Scholars are also enjoined to value theoretical leverage, and to use as few variables as possible when making their predictions.
The methodology of scientific research programs is not only a means for describing competing research programs. It also provides criteria for comparing and judging whether theoretical innovations or emendations represent progress or not; Lakatos labels these “problem-shifts.”23 An intra-program change or problemshift is one that modifies the protective belt of auxiliary hypotheses of a scientific research program. An inter-program problemshift is one that, contrary to the negative heuristic, changes elements of the hard core, thus moving from one program to another.
Lakatos provides explicit rules for determining whether such inter-program and intra-inter-program problemshifts provide added value. Both kinds of problemshifts are degenerative when they are merely ad hoc attempts to deal with apparently disconfirming evidence. Essential to understanding when Lakatosians consider a theoretical adjustment to be ad hoc is the concept of “novel facts.” In degenerating research programs, new theories merely save the program from disconfirming evidence, and do nothing else. By contrast, progressive research programs are those that offer additional content by predicting, and empirically corroborating, some hitherto unknown, unexpected, or unused fact.
In determining whether a problemshift is progressive or degenerating, Lakatosians commonly refer to three distinct notions of
“ad-hocness”: ad hoc1 refers to a theoretical move that generates no novel predictions as compared with its predecessor; ad hoc2 is used when none of the new theory’s novel predictions have been actually verified by empirical evidence; and ad hoc3 refers to a situation in which
23. Vasquez suggests that by problemshift, Lakatos meant “theoryshift” but did not use that word because it “sounds dreadful.” See John Vasquez, The Power of Power Politics: From Classical Realism to Neotraditionalism (Cambridge, UK: Cambridge University Press, 1998), p. 244, note 1.
auxiliary hypotheses are modified in ways that do not accord with the spirit of the positive heuristic of the program.24 Intra-program problemshifts are evaluated on all three ad hoc criteria. Inter-program problemshifts only have to avoid being ad hoc1 and ad hoc2. We now discuss the three kinds of ad-hocness in detail.
Arguably the most important element of Lakatos’s methodology is the notion that good science requires more than simply salvaging theories when disconfirming evidence is encountered. Problemshifts must be theoretically progressive, and produce predictions of novel facts or consequences, because “explaining things which were already known is ‘cheap success’ and counts for nothing.”25 Lakatos requires novel predictions because:
it is easy (and riskless) to make a theory ‘testable’ ... if one already knows how the tests will turn out.... If a theory T is presently accepted and some new evidence e crops up which is not predicted by T, then it is generally trivially easy to use T and e to generate a new theory T’ which does entail e.26
24. Lakatos, “Falsification and the Methodology of Scientific Research Programmes,” p. 175, notes 2 and 3, p. 182; Elie Zahar, “Why Did Einstein’s Programme Supersede Lorentz’s?” British Journal for the Philosophy of Science, Vol. 24 (1973), p. 101; Alan Musgrave, “Logical versus Historical Theories of Confirmation,” British Journal for the Philosophy of Science, Vol. 25 (1974), p. 20;
Richard Nunan, “Novel Facts, Bayesian Rationality, and the History of Continental Drift,” Studies in History and Philosophy of Science, Vol. 15, No. 4 (1984), p. 272; and Siobhain McGovern, “A Lakatosian Approach to Changes in International Trade Theory,” History of Political Economy, Vol. 26, No. 3 (Fall 1994), pp. 352, 354. These are only three of many different definitions of “ad hocness” found in the novel fact literature. For additional definitions, see Martin Carrier, “On Novel Facts: A Discussion of Criteria for Non-ad-hoc-ness in the Methodology of Scientific Research Programmes,” Zeitschrift fur allgemaine Wissenschaftstheorie, Vol. 19, No. 2 (1988), p. 216, note 45.
25. Hands, “Second Thoughts on Lakatos,” pp. 6–7; see also Leplin, A Novel Defense of Scientific Realism, p. 41.
26. Worrall, “The Ways in Which the Methodology of Scientific Research Programmes Improves on Popper’s Methodology,” p. 49.
This is not to say that amendments directed to solving anomalies are automatically degenerating. According to Lakatos, there is nothing wrong with problemshifts designed to account for discrepant facts, so long as they also produce novel predictions.27 Thus, new theories that merely try to salvage the research program without predicting any new facts are judged to be ad hoc1.
But problemshifts must do more than just predict novel facts. Those predictions must be empirically tested, and if they are not corroborated, the problemshift is judged to be ad hoc2.28 To be sure, applying this criterion in IR requires the attribution of dates to novel
27. See Lakatos, “Falsification and the Methodology of Scientific Research Programmes,” pp. 116–118, 124–125, 133, 169–170, 176, 187; Terence Ball,
“From Paradigms to Research Programs: Toward a Post-Kuhnian Political Science,” American Journal of Political Science, Vol. 20, No. 1 (February 1976), pp.
165–166; Mark Blaug, “Kuhn versus Lakatos or Paradigms versus Research Programs in the History of Economics,” in Latsis, Method and Appraisal in Economics, p. 156; Frankel, “The Career of Continental Drift Theory,” p. 23;
Hands, “Second Thoughts on Lakatos,” p. 4; and McGovern, “A Lakatosian Approach to Changes in International Trade Theory,” p. 353. Problemshifts that are not motivated by empirical puzzles can also turn out to have no novel consequences, and hence be ad hoc1. In other words, anomaly solving is only one motive that might produce problemshifts that do not satisfy the need for novelty. As Martin Carrier notes, “the novel fact debate ... is concerned with the legitimacy of any theoretical modification, regardless of the reason for its being introduced.” Carrier, “On Novel Facts,” p. 206.
28. There are at least two ways of structuring inquiry on the ad hoc2 criterion, championed by Elie Zahar and by Lakatos, respectively. Zahar characterizes a theory as ad hoc2 (at time t) if none of its excess content over its rivals has, at time t, been corroborated. Lakatos, on the other hand, characterizes a theory as ad hoc2 if all its excess content has been refuted. See Zahar, “Why Did Einstein’s Programme Supersede Lorentz’s?” p. 101, note 1. To illustrate, suppose a theory T predicts e, and a subsequent problemshift leads to T’ with predictions e, e’, e’’ and e’’’. According to Lakatos, we cannot determine whether the problemshift is ad hoc2 until e’, e’’ and e’’’ have all been tested. If all are falsified, then the problemshift is ad hoc2. By contrast, Zahar says that if we test for any of e’, e’’ or e’’’, and any of them are corroborated, we can conclude that the problemshift is non-ad hoc2. As Zahar argues, the single-confirmation method is more in keeping with the tolerant spirit of Lakatos’s metatheory than is the complete-falsification method.
predictions, and a decision on how long to wait for the empirical corroboration of at least one of the predicted novel facts generated by the problemshift. Dating the creation of a scientific research program and its series of problemshifts is not, however, always a straightforward exercise. For example, if one were to try to date the inception of a realist research program, one could plausibly argue for circa 400 BC (Thucydides), or 1948 (Morgenthau), or for any number of alternatives in between. With respect to the question of how long to allow for novel facts to be empirically corroborated before declaring the problemshift ad hoc2, it is important to remember that Lakatos’s methodology gives theoretically progressive problemshifts some time before deciding whether the novel facts anticipated by the amendment are corroborated. Lakatos notes that we should not be too impatient, nor require instant gratification. Scientists should not quickly char-acterize problemshifts as degenerative simply because their predicted novel facts have not yet been corroborated.29
Lastly, Lakatos’s methodology also considers a scientific research program to be degenerating, “if it anticipates novel facts but does so in a patched-up development rather than by a coherent, pre-planned positive heuristic.”30 For Lakatos, then, “theoretical adjustment [must]
be governed or constrained by a unified set of principles.”31 This ad hoc3
criterion is designed to catch:
unimaginative series of pedestrian “empirical” adjustments which ... may ... make some “novel” predictions and may even conjure up some irrelevant grains of truth in them. But this theorizing has no unifying idea,
29. Lakatos, “Falsification and the Methodology of Scientific Research Programmes,” pp. 116, 133–134, 151, 155. See also Alexander Rosenberg,
“Lakatosian Consolations for Economics,” Economics and Philosophy, Vol. 2 (April 1986), p. 128.
30. Imre Lakatos, “History of Science and its Rational Reconstructions,” in Buck and Cohen, Boston Studies in the Philosophy of Science, p. 125. See also de Marchi, “Introduction: Rethinking Lakatos,” p. 4.
31. David Moon, “Values and Political Theory: A Modest Defense of a Qualified Cognitivism,” Journal of Politics, Vol. 39, No. 4 (November 1977), pp.
899–900.
no heuristic power, no continuity. They do not add up to a genuine research programme and are, on the whole, worthless.32
This requirement places Lakatos’s methodology squarely within the so-called generative tradition, which suggests that new theories are motivated by, and derived from, a body of previous results and guiding principles.33 According to this view, scientific progress is seldom achieved by boldly striking out into unknown territory. Rather, it arises from concentrated attacks on carefully framed questions and theories generated by extant research programs.
In Lakatos’s framework, these three criteria of “ad-hocness”
determine whether various theoretical adjustments are legitimate. As shown in Figure 2-1, intra-program problemshifts must be judged on all three criteria, while inter-program problemshifts can only be measured by whether they are ad hoc1 or ad hoc2. That is, shifts from one program to another are judged on the extent to which they produce novel facts that are then corroborated.34 Shifts within programs have the added burden that they must also be undertaken in accordance with the program’s positive heuristic; if not, they are ad hoc3.35
32. Lakatos, “Falsification and the Methodology of Scientific Research Programmes,” p. 175. For a critique, see Carrier, “On Novel Facts,” especially pp. 228–229.
33. Thomas Nickles, “Lakatosian Heuristics and Epistemic Support,” British Journal for the Philosophy of Science, Vol. 38 (1987), p. 182. According to Nickles, the third ad-hocness rule runs counter to the first two: ad hoc3 discourages novelty by prohibiting radical departures from the program, while ad hoc1 and ad hoc2 demand bold guessing and encourage bold deviations from the current view, no matter how established. See ibid., pp. 191–198.
34. Worrall, “The Ways in Which the Methodology of Scientific Research Programmes Improves on Popper’s Methodology,” p. 56.
35. Lakatos, “Falsification and the Methodology of Scientific Research Programmes,” pp. 175–176.