has repeatedly insisted that, though any particular presupposition may be called in question, it is not possible to call all in question at once and still practice science.
A similar pattern of development regularly recurs in the evolution of physics. Thus, sociologists and historians of science have shown that the acceptance and interpretation of many discoveries involves argumentation, transformative negotiations, one-upmanship, power struggles between competing cliques, and sometimes dirty pool. The interpretation finally accepted often differs from that of the original discoverer.30 From the present perspective, this is neither surprising nor undesirable. Any new purported discovery of fact or law is subject to a complex process of adjustment to empirical sources of information, a network of concepts, past theories, intended goals, and mathematical formulations. However, when something is finally accepted as a discovery, then retrospective realism sets in. The discovery is treated as the finding a fact that already had objective existence, much as. the new continents were already there awaiting Columbus's discovery. Such simplification of inference by acceptance is a necessary feature of progress. Thus, experimenters simply rely on the fact, discovered by Ørsted, that a current in a wire produces a magnetic field, while ignoring the historical fact that Ørsted never held such an interpretation (See Caneva, 2001, pp. 2-5).
This may seem to be advocating a social contract theory of truth. Here, however, it is necessary to be aware of a significant difference between the role of presuppositions in discourse and the foundational role often attributed to axioms of a theory. If the axioms are true, then the conclusions derived from them should be true. In discourse, truth is more of a surface phenomenon, predicated of overt claims and reflecting back on presuppositions. Consider a banal example
This shirt is yellow.(S1)
Acceptance of S1 as true seems to entail
Being colored is a property of fabrics.(S2)
The familiar difficulties with primary and secondary qualities might suggest rejecting the truth of S2 and recasting S1 as
This shirt looks yellow(S3)
As Sellars among others has shown, such an analysis does not work. A white shirt seen under yellow cellophane or in yellow light looks yellow. That is, it looks the way yellow things look under normal circumstances. Since 'looks yellow' presupposes 'is yellow', it cannot supply a replacement for it. There is something of a growing consensus that it is reasonable to accepts S2 as true, but not interpret it as an ontological or scientific claim that competes with the wave theory of light. It is a pragmatic, rather than a semantic, presupposition (See Stalnaker, 1972). In effect, (S2) is treated as true in terms of its supporting role in discourse, rather than as an isolated ontological claim.
Normal discourse involves both general and contextual presuppositions as well as nested implicit rules of inference. Here I will indicate only one aspect. Consider four examples.
1) The engine won't start because the carburetor is flooded with gasoline.
2) The cat's fur is up because she sees a dog approaching.
30 Latour and Woolgar (1986) brought the sociological aspects of discovery to the forefront. Canava (2001) shows how four key nineteenth century electromagnetic discoveries were subject to reinterpretation. The sociological process of reifying discoveries is analyzed in Barnes et al. (1996), chap. 4, and in Pickering.