Thomas Nickles

I discuss changes of perspective of four kinds in science and about science. Section 2 defends a perspectival nonrealism—something akin to Giere’s perspectival realism but not a realism—against the idea of complete, “Copernican” objectivity. Section 3 contends that there is an inverse relationship between epistemological conservatism and scientific progress. Section 4 casts doubt on strong forms of scientific realism by taking a long-term historical perspective that includes future history. Section 5 defends a partial reversal in the status of so-called context of discovery and context of justification. Section 6 addresses the question of how we can have scientific progress without scientific realism—how progress is possible without the accumulation of representational truth. The overall result (Sect. 7) is a pragmatic instrumentalist perspective on the sciences and how to study them philosophically, one that contains a kernel of realism—instrumental realism.

Imre Lakatos (9 November 1922–2 February 1974) is the most important philosopher of mathematics and one of the most influential philosophers of science since the mid‐twentieth century. A Hungarian, Lakatos changed his name from Lipschitz to Molnar during the Nazi era and then to Lakatos (“locksmith”). After the war he remained politically active, as secretary in the Hungarian Ministry of Education. Later he was imprisoned as a dissident, and escaped to the West during the revolt of 1956. He studied at Budapest, Moscow, and Cambridge (Ph.D., 1958). During the 15 years preceding his death, he taught at the London School of Economics and Political Science, where he became Professor of Logic in 1969. He was a lively teacher, discussant, and social critic. An inspired circle of friends and colleagues gathered around him within the Popperian stronghold at the London School of Economics.

We begin with some questions. What constitutes a scientific discovery? How do we tell when a discovery has been made and whom to credit? Is making a discovery (always) the same as solving a problem? Is it an individual psychological event (an ahal experience), or something more articulated such as a logical argument or a mathematical derivation? May discovery require a long, intricate social process? Could it be an experimental demonstration? How do we tell exactly what has been discovered, given that old discoveries are often recharacterized in very different ways by succeeding generations? What kinds of items can be discovered, and how? Is the discovery of a theory accomplished in much the same way as the discovery of a new comet, or is “discovery” an inhomogeneous domain of items or activities calling for quite diverse accounts? Must a discovery be both new and true? How is discovery related to (other?) forms of innovation, such as invention and social construction? Can there be a logic or method of discovery? Just one? Many? What could such a procedure be? How is it possible that an (a priori?) logic or method available now has so much future knowledge already packed into it? How could a logic of discovery itself be discovered? How general in scope must a method of discovery be? Must it apply to all sciences, independently of the subject matter (as we might expect of a “logic”), or might it apply only to problems of a certain type or depend on substantive scientific claims? How, if at all, is their discovery related to the justification of scientific claims? Is the manner in which scientists make discoveries at all similar to the way in which they test them? Is this justificatory “checkout” procedure really part of the larger discovery process rather than distinct from it? Can discoveries be explained rationally, or do they always contain irrational or nonrational elements, such as inspiration or blind luck? Are historians, sociologists, and psychologists better equipped than philosophers to explain scientific creativity? Can a methodology of discovery help to explain the explosion of scientific and technological progress since 1600? If there is no logic of discovery, and if discovery is irrelevant to justification, then why include the subject of discovery in the domain of philosophy (epistemology or methodology of science) at all? What could philosophers have to say about it? Are there historical patterns of discovery, for example, that tell us something about the rationality, if not the logic (in the strict sense), of the growth of scientific knowledge?

There is a rough, long-term tradeoff between rate of innovation and degree of strong realism in scientific practice, a point reflected in historically changing conceptions of method as they retreat from epistemological foundationism to a highly fallibilistic, modeling perspective. The successively more liberal, innovation-stimulating methods open up to investigation deep theoretical domains at the cost, in many cases, of moving away from strong realism as a likely outcome of research. The crowbar model of method highlights this tension, expressed as the crowbar compromise and the crowbar fallacy. The tools-to-theories heuristic, described and evaluated by Gigerenzer and colleagues, can be regarded as an attempt by some scientific realists to overcome this compromise. Instead, it is an instance of it. Nonetheless, in successful applications the crowbar model implies a modest, instrumental (nonrepresentational) realism.

Affirming-the-consequent is a well-known fallacy that leads naïve people to believe that a correct prediction shows that they are “on the right track,” the track of truth. Here I outline fifteen subtler forms of deception that I term ‘cognitive illusions’, intellectual perceptions that make strong realism seem more plausible than it is. Like affirming the consequent, some of these items are used as positive arguments by proponents of strong realism. I do not claim that exposing these illusions amounts to a decisive refutation of strong realism, let alone weaker forms of realism, some of which I can accept. But appreciating their force makes strong realism less attractive.

The paper locates, appreciates, and extends several dimensions of Simon’s work in the direction of more recent contributions by people such as Gigerenzer and Dennett. The author’s “crowbar model of method” is compared to Simon’s scissors metaphor. Against an evolutionary background, both support a pragmatic rather than strong realist approach to theoretically deep and complex problems. The importance of implicit knowledge is emphasized, for humans, as well as nonhuman animals. Although Simon was a realist in some respects, his work on bounded rationality, satisficing, problem solving, heuristics, models, and scientific discovery mark him as a pragmatist. Indeed, he should be regarded as one of the great American pragmatists, alongside Peirce, James, Dewey, and a few others.

How should we model scientific decision-making at the frontiers of research? This chapter explores the applicability of Gerd Gigerenzer’s “fast and frugal” heuristics to frontier contexts, i.e., to so-called context of discovery. Such heuristics require only one or a very few steps to a decision and only a little information. While the approach is somewhat promising, given the limited resources in frontier contexts, trying to extend it to fairly “wild” frontiers raises challenging questions. This chapter attempts to frame the issues (rather than to provide resolutions to them), and thereby to cast light on frontier contexts, which have been misunderstood by philosophers, the general public, and funding agencies alike.

Gigerenzer and coauthors have described a remarkably fast and direct way of generating new theories that they term the tools-to-theories heuristic. Call it the TTT heuristic or simply TTT. TTT links established methods to new theories in an intimate way that challenges the traditional distinction of context of discovery and context of justification. It makes heavy use of rhetorical tropes such as metaphor. This chapter places the TTT heuristic in additional historical, philosophical, and scientific contexts, especially informational biology and digital physics, and further explores its strengths and weaknesses in relation to human limitations and scientific realism.

Are we entering a major new phase of modern science, one in which our standard, human modes of reasoning and understanding, including heuristics, have decreasing value? The new methods challenge human intelligibility. The digital revolution inspires such claims, but they are not new. During several historical periods, scientific progress has challenged traditional concepts of reasoning and rationality, intelligence and intelligibility, explanation and knowledge. The increasing intelligence of machine learning and networking is a deliberately sought, somewhat alien intelligence. As such, it challenges the traditional, heuristic foresight of expert researchers. Nonetheless, science remains human-centered in important ways—and yet many of our ordinary human epistemic activities are alien to ourselves. This fact has always been the source of “the discovery problem”. It generalizes to the problem of understanding expert scientific practice. Ironically, scientific progress plunges us ever deeper into complexities beyond our grasp. But how is progress possible without traditional realism and the intelligibility realism requires? Pragmatic flexibility offers an answer.