Many recent experiments have shown that children perform an instrumental response with greater speed if this response is partially, rather than continuously, rewarded (Bruning, 1964;Ryan, 1966;Ryan & Moffitt, 1966). Two hypotheses have been suggested to account for these results. Amsel (1958) and Spence (1960) have assumed that an aversive motivational condition, which they term frustration, is produced in circumstances in which S anticipates but does not receive reward. The resultant increment in motivation is taken to be the factor leading to faster asymptotic performance associated with partial reward. Brown (1961) and Marx (1956) have suggested than an associative mechanism, rather than a motivational one, may account for the above results. They assume that, in the past, Ss may have been reinforced for making more vigorous responses in the face of nonreward. In studies with Children, the partial reward operation differed from continuous reward only in that the mechanism dispensing the reward was disengaged on the nonrewarded trials. This associative interpretation would indicate that the more vigorous responding found for partial reward may have been elicited by the inoperative reward dispenser. Such an effect could be mediated by implicit verbal responses of the nature "marbles (rewards) get stuck sometimes." The faster response speeds may then be attributed to the child's attempts to jar the marbles loose by forceful manipulation of the response lever.The present study was an attempt to evaluate these two alternative explanations. An implication of this associative interpretation is that ifnonreward were manipulated in a manner unlikely to indicate a malfunctioning apparatus, there should be no speed differences between partially and continuously rewarded groups. In the present study, nonreward was indicated by the marble color, rather than by witholding of the reward object.