Bias in Avoidance-Yoked Eyelid Conditioning Comparisons

Abstract

<p>Yoked groups have often been used as controls against which to assess the effects of a contingency between a response and some event. For example, Moore and Gormezano (1961) used a yoked control group in an experiment designed to assess the effects of an avoidance contingency in human eyelid conditioning. For the avoidance subjects in their study, an eyelid closure during the CS-US interval precluded the delivery of the air puff US. Each avoidance subject was given a yoked partner who received the air puff, without regard to his responding, on those trials when his avoidance partner received it. In this way, the avoidance and yoked control groups were equated for number and pattern of USs, and differed only with respect to the instrumental contingency.</p> <p>Moore and Gormezano's (1961) finding was that the avoidance group reached a higher asymptotic level of conditioned responding than the yoked group. However, Church (1964) has argued that the results of such instrumental-yoked comparisons are ambiguous since any effects due to the contingency are confounded with the effects of a bias in the design favoring the instrumental subject. The purpose of the present research was to evaluate Church's argument that bias arising from within-pair mismatching in effectiveness contributes to the avoidance-yoked difference obtained in eyelid conditioning.</p> <p>Three rabbit eyelid conditioning experiments involving a total of 228 subjects are described. Pre-yoke matching (Experiment 1), multiple yoking (Experiment 2), and reciprocal yoking and self-yoking (Experiment 3), procedures were implemented. The results of the between-subject, and within-subject, avoidance-yoked contrasts in these experiments support Church's (1964) conceptualization of a yoked subject as one that receives more, or fewer, shocks, and in a different pattern, that it would if delivery of shocks were contingent upon its own behavior. Hence, the major conclusion that is drawn from these experiments is that differences in acquisition performance between avoidance and yoked subjects in the rabbit eyelid conditioning preparation are the result of bias. The findings of this investigation suggest that the instrumental-yoked differences obtained in other research programs may also reflect bias in the yoked control design, and not the effects of the instrumental contingency per se.</p>