Fourteen Ss were exposed to three 50%-reinforcement sclledllills in human eyelid conditioning. The In the typical 50% reinforcement schedule the conditional probability of a reinforced trial following a reinforced trial, P(R IR), as weil as the conditional probability of a nonreinforced trial following a nonreinforced trial, P(R I ~), is . 5. When both P(R I R) and P(R I R) are increased (maintaining the 50% reinforcement ratio) increments and decrements in response prob ability following, respectively, reinforced and nonreinforced trials are larger, but they are sm aller when P(R I R) and p(R I R) are less than .5 (Prokasy, earlton, & Higgins, 1967; Higgins & Prokasy, 1968). The purpose of this study was to extend these observations by determining (a) if differential sequential effects could be obtained when the same Ss are exposed to different values of P(R I R) and p(R I R) in different sessions and (b) whether the sequential effeets would be maintained with extensive training. In all eases, P(R I R) = P(R IR). METHOD The method of transducing and reeording eyelid responses is reported elsewhere (Prokasy et al, 1967). The es was a 600-msee, lOOO-Hz tone, the interstimulus interval was 600 msec, and the ues was a 50-msee puff of nitrogen of suffieient intensity to support a 150-mm column of mereury. A total of 14 Ss, paid for their services, served in 15 conditioning sessions of 360 trials each. The first 10 trials of eaeh session were composed of five eS-alone and UeS-alone trials. The average intertrial interval was 10 sec, varying from 6 to 14 sec. Three 50%-reinforcement schedules were employed, these differing only in that P(R IR) =.3, .5, or .7. The three sehedules were randomized over sessions so that each S reeeived five sessions at each p(R I R) value in an unsystematic order. Through a scheduling error, two Ss received six sessions of the .7 schedule and only four of the .3 schedule. Also, for two Ss the original data for one of the .7 schedules were misplaced and, consequently, not employed in the analyses. All Ss reeeived standard instructions (i.e., to relax and let their responses take eare of themselves) for each session.RESUL TS AND DISCUSSION The basic analyses were conducted on sequential behavior. Response probability following either zero, one, two, or three sueeessive reinforcements was caJculated for the first and last halves of training. These calculations were performed separately for the three different P(R I R) values, thus permitting a eomparison of the first two-and-a-half sessions with the last two-and-a-half sessions for each of the three kinds of schedule. The summary da ta are provided in Table I.