thought to be operating in, the life settings. In this instance, the proximal determinants of cued infant protests were studied employing mothers as experimental agents. Alternatively, a passive-observation study or field experiment (which has many of the same constraints as has the laboratory study) in natural home settings might have been mounted. Nevertheless, a naturalistic description (although useful in the preliminary phases of an investigation) is a deficient research method, insofar as it does not permit the inference of causality ( McCall, 1977). This naturalistic tactic was not used as our first alternative because it is routinely found in nature that the magnitude of effect(s) reflecting the phenomena of interest is small relative to the uncontrolled (error) variation there. Hence, passive observation in natural settings with contextual variables uncontrolled ordinarily would give little return relative to investment. And such observation with a preliminary attempt to control, or stratify for, context also was thought less efficient for our purpose than the laboratory study reported here.
In this frame, the researcher must often consider moving between laboratory settings, in which there is relatively much control of the proximate causal dimensions thought to be operating, little independent-variable variation, and few confounding conditions (i.e., high internal validity), and life settings in which there may be very limited control and many varying and confounding conditions. This is particularly the case where the researcher intends to make claims about life settings from the laboratory research, as ideally we would like to do here. Ultimately, at least some triangulation will be required between laboratory-generated mechanisms such as those presented here and results obtained from passive observation under the massive inefficiencies prevailing in life settings. This would validate applying the laboratory-generated mechanisms to the life setting. For the moment, our abstraction of the mechanism from extensive observation in home settings, of contingent-maternal responding as the main proximal determinant of infant protests at maternal departures and separations, together with the inherent plausibility of the logic used, will have to stand for the triangulation ultimately required until such time as that proximal mechanism could be validated in the life setting.
An experimental analysis of contingent maternal behavior that can train and maintain infant protests cued by maternal departures and separations provides a basis for understanding features of social conditioning in early human life, in particular social discriminated operants that comprise, and can index, the attachment process. The research has illustrated an instance of early infant social learning and some of the maternally-mediated proximal environmental conditions apparently responsible for their acquisition and maintenance. At the same time, the results provide a basis for minimizing or eliminating unconstructive