Thomas Kuhn, answered from the texts and cited to the page.
Observation is theory-dependent. There is no neutral observation language standing outside the theoretical frameworks scientists work within, no stock of pure-sense-datum reports that could function as the shared empirical basis for adjudication between paradigms. This much, after Hanson and the post-positivist consensus, is settled. The harder claim — and the one I phrased most carefully — is that in a sense I am unable to explicate further, the proponents of competing paradigms practice their trades in different worlds (Structure X).
I meant the limit of explication seriously, not as a rhetorical hedge. The claim sits between two readings: a strong perceptual reading on which the world-as-experienced literally changes (the gestalt switch from duck to rabbit), and a weaker conceptual reading on which the categorial structure within which the scientist organizes phenomena changes while the phenomena themselves remain stable.1
The texts support both at different moments. The strong reading is more memorable; the weak reading is more defensible; my actual position is that the line between them is part of what world change names.2345
The same point could be made about any of our earlier examples. To see oxygen instead of dephlogisticated air, the condenser instead of the Leyden jar, or the pendulum instead of constrained fall, was only one part of an integrated shift in the scientist's vision of a great many related chemical, electrical, or dynamical phenomena. Paradigms determine large areas of experience at the same time.The Structure of Scientific Revolutions, p. 107
If that supposition were exclusively about objective observation of nature, it would present no insuperable problems, but it involves as well the assumption that the objective observers in question understand 'snow is white' in the same way, a matter which may not be obvious if the sentence reads 'elements combine in constant proportion by weight'.The Road Since Structure, p. 169
As a result, scientists with different paradigms engage in different concrete laboratory manipulations. The measurements to be performed on a pendulum are not the ones relevant to a case of constrained fall. Nor are the operations relevant for the elucidation of oxygen's properties uniformly the same as those required when investigating the characteristics of dephlogisticated air.The Structure of Scientific Revolutions, pp. 105–106
Even brief discussion should permit the isolation of two quite different ways in which the term is deployed in my book and thus eliminate a constellation of confusions which has handicapped me as well as my critics. The resulting clarification will, in addition, permit me to suggest what I take to be the root of my single most fundamental difference from Sir Karl.The Road Since Structure, pp. 134–135
What one of these sees as striking parallelism is virtually invisible to the others. The desire to understand how this can be so motivates the present comparison of my view with Sir Karl's. The comparison must not, however, be a mere point-by-point juxtaposition.The Essential Tension, pp. 288–289
Theory-ladenness is settled doctrine. World change is a position I held in the strong form in 1962 and qualified increasingly through the 1970s and 1980s, eventually preferring the lexical-taxonomic articulation in the late papers. The published acknowledgment that I cannot explicate further is honest, not evasive: there is a real limit to what philosophy can decide about whether perception itself, or only the categorial structure that organizes perception, undergoes the relevant shift. Critics are entitled to push on either side of that line.