Knowledge, Objectivity, and the Earlier Debates on Scientific Realism
Debates on realism in science have taken place since the Antiquity (Alai 2008), and they concern two different questions: (I) whether the unobservable entities posited by theories can be known (that is, whether we can have beliefs about them, and these beliefs can be both true and justified); and (II) whether any knowledge we have of them is objective or not.
Question (I) arises from the doubt that since observation is the basis of all our factual knowledge, unobservable entities cannot be known; questions (II) arises from the doubt that since scientific representations are inextricably laden with, or distorted by, the subjective, idiosyncratic and a priori features of human cognition and scientific practice, they cannot convey any reliable information on how their objects are “in themselves”. The same two questions arise even concerning different subjects (ordinary material objects, the mind, etc.), and antirealism on knowledge is generally termed ‘skepticism’, while antirealism on objectivity is called ‘idealism’, ‘subjectivism’ or ‘constructivism’.These two questions are logically independent, so they have been mostly discussed separately, and one can be realist on one and antirealist on the other. For instance Putnam held in (1978b) and (1981) that we can know unobservable entities quite as we know ordinary observable objects, but this is not knowing an absolutely mind-independent reality; a similar stance has been taken by some “perspectival realists” (see below). Nevertheless, certain powerful arguments for realism work in both debates, so that one may find it hard to be realist on knowledge but not on objectivity, or vice versa (Alai 2005, 2006: 214-216, 2009, forthcoming). Moreover, as we shall see, the two debates somehow converge, in the sense that some of the most plausible forms of realism on each question (selective realisms on the one hand, and perspectival realisms on the other) basically agree on the possibility of objective but partial knowledge.
Here I shall examine the state of contemporary debates mainly on the knowledge question, and more briefly on the objectivity question.Since the Antiquity and up to the beginning of the last century, the typical antirealist stand about scientific knowledge was that we can at most aim at “saving the phenomena”, i.e. form true and justified beliefs about observable entities. A frequent corollary was instrumentalism: to the extent that theories apparently describe unobservable entities, they should not be interpreted literally as descriptions, because they actually are (or should be) just practically useful instruments, computing devices to predict future phenomena or directions for technological applications, and as such neither true not false.
However, instrumentalists should explain why theories are so useful and successful, if they don’t offer an at least approximately true description of the underlying reality; moreover, what justifies the claim that the literal interpretation of scientific statements as descriptions is wrong, and that scientists which aim at finding the truth are wrong and should pursue a different goal? (Alai 2006: 220). The only possible justification would be the claim that saving the phenomena is the best we can do, but even so instrumentalism seems to be a dubious and idle corollary. In fact, it isn't a live option anymore, except perhaps as concerns quantum mechanics (e.g., Wigner 1967).
Logical positivists tried to bypass the epistemological debate on truth and justification by a “linguistic turn”: their verificationist theory of meaning entailed that, appearances notwithstanding, scientific theories did not actually speak of unobservable entities; the meaning of theoretical terms was entirely reducible to possible observations, so in principle they could be replaced by a purely observational vocabulary (Carnap 1923).
However, eliminativism was abandoned when it became clear that non-observational terms cannot be completely defined by observational terms, as it happens already with simple “dispositional” terms (such as ‘soluble' and ‘fragile') and the irrational values of physical measurements (Carnap 1936: § 7; Hempel 1952: II).
Moreover, non-observational terms play a necessary role in systematizing experience, predicting future observations, and leaving the way open to the discovery of new properties (see Hempel 1958); moreover, their elimination would leave numberless unexplained coincidences (Smart 1963: 39; Psillos 1999: 72-73). Finally, verificationism itself was abandoned in the second half of the XX Century, especially as a consequence of Quine's (1951) criticisms.Therefore van Fraassen (1980: II, 1) explained that linguistic questions are no longer live issues between scientific realists and antirealists: antirealists may grant that theories must be read literally, as purportedly true descriptions of unobservable entities, but deny that we have compelling reasons to believe that they are true. According to his “constructive empiricism” all we need to believe is that a theory is empirically adequate, i.e., it “saves the phenomena”. Thus he brought back the debate to the epistemic question on which it had focused for about 20 centuries, and in so doing he set the agenda for most subsequent discussions on the knowledge question up to this day (Alai 2006: 217-220).
2