I consider the title of this section of the conference, “Theoretical Reasons for a Choice” (between the laboratory and everyday life), to be a bit misleading because, really, there is no necessity for choice. The issue should be considered from the perspective of adopting sampling strategies that will support the scientific generalization to be drawn: Do the sampling procedures support the inference from the particular samples chosen to the universes of interest?

Brunswikian representative design

To begin, I shall briefly describe some Brunswikian views regarding research strategies and develop these ideas within a sampling framework. Brunswik (1952, 1956) argued that we should build a science adequate to understand the behavior of organisms in their environment and should adopt research strategies appropriate to that undertaking. Because behavior takes place in a semichaotic medium that contains cues of limited trustworthiness, expressed vicariously, a research strategy different from those usually advocated will be necessary to realize this understanding. This strategy is called probabilistic functionalism; it utilizes representative design and is adequate to the task of conceptualizing and understanding complex behavior (Petrinovich, 1979). Unfortunately, it is not a simple conceptualization, nor is it easy to implement.

The Brunswikian argument can be conceived as an exercise in sampling theory. First, there is agreement between proponents of representative design and systematic design that it is essential to obtain a representative sample of subjects on which to base theoretical conclusions of general applicability.