Why facta neither speak for themselves nor resolve the psi controversy: The view from the rational epistemology

Theoretically advanced sciences such as physics and biolo~y have developed two complementary and Beld-ernbracmg epis­ temologies or frameworks for the pursuit of scientific under­ standing. One, the tmpirical f'pistemology. aims to test theories and develop a reliable body of facts and ",a)·world appllcanons, whereas the other. the ratumalrJ'istrmoloi:Y' aims to create and develop theories for explainmg available facts, faeilitatin~ prole­ tical applications. and predicting new facts for future testlng, Both epistemologies arc relevant to the recent psi controversy. which represents much more than I personality difference between "theory-trusting" versus "data-trusting" scientists (Child 1987; see also Rao lie Palmer 1987; henceforth R lie Pl. Both sides of the psi debate. however. have implicitly adopted the empirical cpistemol(llZlJ, without spdlinjt out or systemat­ ically applyinjt the rational epistemology. As a result. many dimensions of the pro-psi argument remain unchallenged, and the two sides often fail to make contact. attaching fundamentally different implications to seemingly straightforward terms such as psi, theory. facts. reliability. and practical applications (see below). I will argue that developing the rational epistemology in both psychology and parapsychology represents another pre­ requisite for admiUin~ psi into mainstream science. F.cts under the two .p/stemologles. Under the empirical epistemology observations speak for themselves, whereas un­ der the rational epistemology, observations only become really secure and unimpeachable when explained by a theory, and often do not count as scientific facts until a plausible theoretical mechanism for explaining them isproposed. Anexample is mid­ nineteenth century bioloJtY. which refused to accept the large body of well-known and reliable observations on selective breeding as scientific fact until Darwfn proposed a plausible theory for explaining these observations, Psi likewise falls out­ side the realm of psychological fact under the rational epis­ temology because no plausible theoretical mechanism for explaining psi data has been proposed. and not because these data are unreliable or fraudulent (empirical epistemology arguments). Unreliability und., the raUana/ epistemology. Experimental details that may be relevant to a particular result are unlimited in number. Faced with the task of communicating such details. together with indefinitely many other potentially relevant as­ pects of prior studies. investigators often omit seemingly minor factors (e. g.. personality characteristics ofthe experimenter; see Rosenthal lie Rabin: "Interpersonal Expectancy Effects" BBS 3:377-86) that maycompletely change the outcome ofan experi­ ment and make replication difficult. Under the rational epis­ temology, failure to know, .understand. or communicate the conditions essential to replicating a result oftell reflects the absence of viable theory needed for summarizing available infonnation. Theor/gina of theory unde, the two ep/stemologl••• Acommon assumption in the empirical epistemology is that theories emerge out of data and that theorizing is premature until large and reliable bodies of data have been accumulated. R tic P not only accept this assumption but add a new twist: that discussion of meta theoretical [rameuiorks such as dualism is currently premature because "whether or not the outcome of psi research ever suggests dualism will depend on the nature of the particular omegic theory (if any) that is eventually validated" (original emphasis. sect. 3.1, p. 624). Because metatheoretical frame­ works are needed for specifying the goals of theory develop­ ment. this new twist altogether precludes developing a theory, Moreover, data are unnecessary and insufficient for creating both metatheoriesand theories under the rational epis­ temology. In fact. many successful theoretical constructs (e.g .• sound waves) were invented hundreds of years prior to any data whatsoever. Finally, experimental data often play less of a role in developing rationalist theories than factors such as elegance. consistency. and "making sense" (see ~tacKay 1988). Pal: A ttt.oret/ca/ or an emplr/calterm? Alcock (p. 562) uses the term "psi" as a (poorly developed) theoretical construct that can "transcend temporal boundaries," and can be used with "no effort. no training, and no particular knowledge." However, R &