* Introduction

* Two Puzzles in the Study of Science

* The Hierarchical Model

* The Reticulated Model

* Analyzing Scientific Realism

* The Failure of Scientific Realism

As the full title of the book suggests, the values Laudan is discussing are cognitive, not ethical, so this is not a book about ethics in science, scientific ethics, or anything like that. Instead, this book is mainly inspired by Thomas Kuhn's work (especially "The Structure of Scientific Revolutions") and other similar claims of incommensurability of different scientific paradigms. Laudan evaluates these claims critically, while constructing a theory of scientific change, which is intended to be both descriptive and normative. He also applies this theory in an evaluation of scientific realism, which he then rejects.

Laudan begins with a short historical introduction to the situation, by describing what he calls the "puzzles about science", or the two "crises in philosophy and sociology of science". The first of these is the "puzzle of agreement" - the somewhat obvious fact that the natural sciences, especially when compared to such fields as philosophy, seem to be characterized by a high level of agreement about proper methods and "assertions of their discipline". The philosophers and sociologists of science of the 1940s and 1950s saw this as the central feature of natural science that qualified it as culturally unique field of thought, "to be sharply demarcated from other intellectual pursuits such as philosophy, theology and aesthetics" (pp. 1-2). Laudan calls this the Consensual View.

The other view Laudan calls the "New Wave". This view, the result of a transformation of both philosophy and sociology in the 1960s and 1970s, is preoccupied with dissensus in the natural sciences - which is the second problem in the study of science. By the mid-70s this view had become predominant. This was mainly due to the influence of Kuhn and Feyerabend.

Both views make the mistake of overlooking an aspect of science: the Consensual View ignored (until it could no longer) the disagreements within science, while the "New Wave" ignored (or ignores) the high level of agreement. Laudan attempts to construct a view that can explain both aspects.

In chapter two, Laudan describes the hierarchical structure of scientific debates by focusing on how consensus is formed about factual or about methodical matters. The model he uses is more commonly called the theory of instrumental rationality and its proponents include at least Popper, Hempel and Reichenbach (p. 23). There are three levels in this hierarchy. On the lowest level are disputes about matters of fact, or what there is in the world. "According to the standard account, scientists resolve factual disagreements and thus forge factual consensus by moving one step up the hierarchy to the level of shared methodological rules". These rules "provide directives for ascertaining, at least in a qualitative sense, how much support [...] the available evidence provides for the theories under evaluation" (p. 24). The highest level is that of "the shared aims or goals of science", or the axiological level. This level is needed to resolve the methodological controversies of the lower level. But because this is the highest level, the standard view leaves no rational means to resolve the disagreements of the axiological level.

Of course axiological disagreements are not the only disagreements that cannot be easily resolved. For example, a given set of methodical rules may still underdetermine the choice among factual claims, because "although the rules plus the available evidence will exclude many factual claims of hypotheses, a plethora of possible hypotheses often remains methodologically admissible." But the common criticism that the possibility of several contrary, yet supported factual claims "makes a mockery of the hierarchical model of consensus formation" misses the point, according to Laudan because, he writes, "scientists should be seen, not as looking for simply the best theory, but rather for the best theory they can find" (pp. 27-28). In other words, science is not the practise of inference to the best (possible) explanation, but rather inference to the best available explanation.

Laudan also admits that "cognitive aims typically underdetermine methodological rules in precisely the same way that methodological rules characteristically underdetermine factual choices". That is, there seems to be no way "to show that that there is any set of rules of empirical investigation which uniquely conduces to" the realization of such "familiar cognitive goals as truth, coherence, simplicity, and predictive fertility" (p. 35). But even if we cannot show that a given rule is the best possible rule for realizing a certain end, we can still show that it is one way to realize it and that it is better than known alternatives.

Laudan notes that the hierarchical model fails to observe one more thing: that "factual beliefs [...] shape methodological attitudes, every bit as much as our goals do" (p. 39). For example, knowledge of the placebo effect forces to include double-blind experiments in our methodology in testing a drug's efficacy.

This suggests, according to Laudan, that "the methodology and epistemology of [natural] science [...] should be conceived, far more than they normally are, as empirical disciplines. Crediting or discrediting a methodological rule requires us to ask ourselves whether the universe we inhabit is one in which our cognitive ends can in fact be furthered by following this rule rather than that. Such questions cannot be answered a priori" (pp. 39-40).

In chapters three and four Laudan first criticises what he calls the Covariance Fallacy, then introduces his Reticulated Model and shows how it be used in "dissecting the holist picture of scientific change".

The Covariance Fallacy "assumes that the presence of absence of consensus with respect to factual claims can be used to infer the existence of agreement or disagreement with respect to cognitive aims" (p. 43). The idea is that scientists advocating different cognitive goals also advocate different methodologies (and ontologies) and vice versa. For example, "Kuhn never imagines that there might be fundamental ontological or theoretical differences between scientists who share the same cognitive goals" (p. 44). And if scientists agree about such matters, then we must supposedly assume that they endorse the same cognitive goals.

But this is clearly not true. According to Laudan, revolutions often happen very fast, because a new theory "is dominant with respect to all the then prevalent methodologies and axiologies" (p. 46). So it is possible for a given methodology to be conducive to several different cognitive aims at the same time. The Fallacy assumes that methodologies and ontologies (factual claims) are covariant with certain cognitive goals, so that scientists subscribing to those goals must subscribe to the whole cluster of goals, claims and methodologies - all or none; and if they discard those goals, they must discard the whole cluster at once.

Most proponents of the hierarchical model like Popper, as well as relativists like Kuhn, have maintained that the choice of goals is not rational. "In short, radical relativism about science seems to be an inevitable corollary of accepting (a) that different scientists have different goals, (b) that there is no rational deliberation possible about the suitability of different goals, and (c) that goals, methods, and factual claims invariably come in covariant clusters" (p. 50).

But Laudan's Reticulated Model, of course, shows how it is possible to evaluate the goals rationally and to change parts of the so-called cluster at a time, leaving the rest of it untouched. A goal could be charged with inconsistency, but Laudan also points out two "general modes of criticizing a proposed cognitive goal or set of goals [...]. [O]ne may argue against a goal on the grounds (i) that it is utopian or unrealizable or (ii) that it fails to accord with the values implicit in the communal practices and judgements we endorse" (p. 50).

Laudan defines three different ways how a goal can be utopian:

(1a) Demonstrable utopianism - The goal can be demonstrated to be impossible for us to achieve. (For example the aim of infallible knowledge of the 19th century.)

(1b) Semantic utopianism - The goal cannot be characterized in a "succinct and cogent way"; attempts to do so may be imprecise, ambiguous, or both. (For example "simplicity" or "elegance".)

(1c) Epistemic utopianism - The goal state, while perhaps properly defined, cannot be given a specified criterion for determining when the value is or is not present or satisfied. (For example, it may be impossible to determine whether a theory actually is true or not.)

The second mode of criticising cognitive goals is more or less an empirical matter: if scientists actually, in their actions and choices, implicitly support different claims than the ones they explicitly claim to support, they can be charged with inconsistency, or even hypocrisy and dishonesty. They are forced to change either their explicit goals or their behaviour.

I have now described Laudan's criticism of the hierarchical model, so it is time to show what he offers to replace it: the Reticulated Model of scientific rationality. In the hierarchical model, the justification of the three levels of scientific commitment (factual, methodological and axiological) flows downward: the goals justify the methods and so on.

The Reticulated Model is a little more complicated, but Laudan shows an excellent illustration of "The Triadic Network of Justification" which clearly depicts the relations between the levels of scientific commitment in the model. Methods, theories and aims are represented by the corners of the triangle, the last two on the bottom. The sides of the triangle represent the different relations between the commitments. The simplest of these relations is that of aims and theories - the two must "harmonize". That is, the theories must be conducive to the aims, but the aims themselves have to be such that - according to our best theories - it is possible for us to achieve them (at least in principle). The aims then justify the methods, which in turn justify the theories, just as they did in the hierarchical model. But in this model, the theories also constrain the methods and the methods must exhibit the realizability of the aims.

For example, our knowledge of the placebo effect constrains our methods by forcing us to require double-blind testing. With that included in our methods, it is possible for us to eliminate a source of error from our theories. It is one of our aims eliminate such sources of error, so our new methodology succeeds in exhibiting the realizability of this aim. Our theory and aim are in harmony (at least they are not contradictory, as they could be, if we demanded absolute infallibility as our aim, while our theories tell us that infallibility is impossible to achieve). (p. 63)

There is a mutual dependency between our scientific commitments: changing one of them may cause a change elsewhere in the network also. But "[T]here is no single 'right' goal for inquiry because it is evidently legitimate to engage in inquiry for a wide variety of reasons and with a wide variety of purposes" (p. 63-64). The argument is "thoroughly Heraclitean: theories change, methods change, and central cognitive values shift" (p. 64). But this certainly does not rule out progress, which is always relative to some set of aims: "Thus we can ask whether Newton's theory of light represented progress over Descartesäs optics, without knowing anything about the cognitive aims of Newton and Descartes. Instead, we can (and typically do) make such determinations of progress relative to our own views about the aims and goals of science" (p. 65).

Kuhn claimed that a scientific "paradigm shift" requires a complete change of world views, each of which is incommensurable with its competitor world views. These changes, he said, are rapid revolutions, where the old world view is discarded completely and at once, and replaced by another complete world view. Laudan believes this to be a misdiagnosis caused by the rapidity of scientific revolutions (which he does accept). What Laudan denies, is that the world views can be treated in this holistic fashion. According to him, the reason why it may seem to (some) historians that a world view was overthrown in the above manner, is precisely the rapidity with which it occured: "[L]et us imagine a historian called Tom, who decides many years later to study this episode [the "revolution"]. [...] Surely, Tom may well surmise, here was a scientific revolution if ever there was one, for there was dramatic change at every level [values, rules and theory]. [...] The point, of course, is that a sequence of belief changes which, described at the microlevel, appears to be a perfectly reasonable and rational sequence of events may appear, when represented in broad brushstrokes that drastically compress the temporary dimension, as a fundamental and unintelligible change of world views" (p. 78).

The "revolution" can occur fast, but probably going through several separate phases. In a fictional example provided by Laudan, it begins with a change in theory. (To use the same example as before, this could be something like the discovery of the placebo effect.) A short time later the scientists have deviced a new method, based on this factual information. (They came up with double-blind testing, for example.) Some time later the scientists realize that either the new knowledge (the placebo effect in our example) is in contradiction with their aims, or that the new methodology is not in fact capable of realizing those aims, or that there is divergence with implicit theory preferences. (For example, finding all substances that alleviate the patient's pain becomes waste of time when it is realized that all substances can do this to some degree; or it is realized that instead of trying to find any substance that has a positive effect on pain, many researchers are in fact trying to find out what pain is). So the aims then change to fit the theory, methodology and actual practice better. If the whole sequence takes only a couple of years, it will be very difficult for later historians to notice the nuances in the "revolution".

Finally, in the last of the five chapters, Laudan challenges realist axiology and methodology. He attacks two of the rules and methods of the realist view: the method of "inference to the best explanation" and the "requirement that acceptable new theories must preserve significant portions of the theoretical content [...] of their succesful predecessors" (p. 105). Convergent realism is the form of epistemic realism that claims that "certain forms of evidence or empirical support are so epistemically probative that any theory that exhibits them can legitimately be presumed to be true, or nearly so" (p. 105-106). According to the realist, the goal of science is to find ever truer theories about the world and has done so in the past, especially in its recent history. The first part is the view's normative component, the latter its descriptive component.

The argument Laudan provides for realism goes basically like this: a true theory will be empirically successful; scientific theories are empirically successful; so scientific theories are probably true. (This is an abductive, not deductive inference - hence the qualifier "probably".) It follows from this that earlier theories that were empirically successful were also probably true. A theory's being true means (here) that it's central terms genuinely refer; that is, the entities the theory postulates are actually "out there" in the "real world".

Laudan claims that the realist assumption that referring theories can be expected to be successful is mistaken. He provides examples of theories that we would now consider to be referring or true, but which were unsuccessful in the past, and examples of theories that we would now consider non-referring, but which were successful in the past.

The realist might claim that the success of a theory warrants the claim that at least some of its central concepts genuinely refer, but then he would have no reason to demand that a later theory should retain the central concepts of previous theories. Another way realism can be defended is to claim that a successful theory is at least approximately true or close to the truth. But so far (by the time Laudan wrote the book at least) the notion of "approximately true" has no been defined, so nobody really seems to know what it means. A Popperian attempt to use the theory's truth and falsity contents in the definition fails, because "it is entirely conceivable that a theory might be approximately true in the indicated sense and yet be such that all its thus far tested consequences are false" (p. 118).

Laudan makes some notes about the claim that realism only applies to so-called "mature" sciences. If the realist does not accept that some of the successful sciences of the present are "mature", then he fails to explain why they are successful. Also, "even if we found a science in which corresponding relations existed between the latest theory and its predecessor, we would have no way of knowing whether that relation will continue to apply to subsequent changes of theory in that science" (p. 122).

Laudan even gives examples of theories that we would consider both successful and referring, but not approximately true. In his words, "the realist needs a riposte to the proma facie plausible claim that there is no necessary connection between increasing the accuracy of our deep-structural characterizations of nature and improvements at the level of phenomenological explanations, predictions and manipulations" (p. 123).

Laudan then looks at the historical record, for if scientists actually do attempt to retain the content of earlier theories in later ones, we should expect to see in historical scientific literature "proofs that later theories do indeed contain earlier theories as limiting cases" or "outright rejections of later theories that fail to contain earlier theories" (p. 126). But we do not find either of these. "The willingness of some realists to attribute realist motives to working scientists - on the strenght of virtually no empirical research into the calues that in fact have guided scientific practice - raises doubts about the seriousness of their commitment to the empirical character of epistemic claims" (p. 127).

Laudan states that "some of the most important theoretical innovations have been due to a willingness of scientists to violate the cumulationist or retentionist constraints which realists enjoin mature scientists to follow" (p. 127). This is only for the better, for if "one took seriously the [convergent epistemic realist's] advice to reject any new theory that did not capture existing (mature) theories as referential and existing laws and mechanisms as approximately authentic, then any prospect for deep-structure, ontological changes in our theories would be foreclosed. Equally outlawed would be any significant repudiation of present-day theoretical models" (pp. 130-131).

Also, he points out that new theories should not be required to explain why their predecessors were succesful: if they are otherwise better than the old, we should accept them anyway and if they are not as good as the old ones, we shouldn't accept them even if they did explain past successes.

Finally, Laudan concludes "in the language of earlier chapters" of his essay, that "the realist offers us a set of aims for science with these features: (1) we do not know how to achieve them (since there is no methodology for warranting the truthlikeness of universal claims); (2) we could not recognize ourselves as having achieved those aims even if, mysteriously, we had managed to achieve them [...]; (3) we cannot even tell whether we are moving closer to achieving them (since we generally cannot tell for any two theories which one is closer to the truth); (4) many of the most successful theories in the history of science [...] have failed to exemplify them. In my view, any one of these failings would be sufficient to raise grave doubts about the realist's proposed axiology and methodology for science. Taken together, they seem to constitute as damning an indictment of a set of cognitive values as we find anywhere in the historical record. Major epistemologies in the past (e.g., classical empirism, inductivism, instrumentalism, pragmatism, infallibilism, positivism) have been abandoned on grounds far flimsier than these" (p. 137).