Unblocking in a runway discrimination problem produced by a surprising reduction in S— reward magnitude: The role of generalization decrement

Bulletin of the Psychonomic Society 1984, 22 (5), 469-471 Unblocking in a runway discrimination problem produced by a surprising reduction in S- reward magnitude: The role of generalization decrement STEVEN J. HAGGBLOOM Arkansas State University, State University, Arkansas Blocking and unblocking groups were trained in Phase 1 on an instrumental discrimination task in which internal cues were relevant and brightness cues were irrelevant. The blocking group received large reward (L) in S+ and nonreward (N) in S- in both Phase 1 and Phase 2 and failed to learn about brightness when brightness cues were added as relevant, but redundant, discriminanda in Phase 2. Groups Unblocking:S and Unblocking:S+ N received small reward (S) or a mixture of Sand N trials, respectively, in S- in Phase I, but otherwise were trained in the same manner as the blocking group. Both unblocking groups experienced a shift to all N trials in S- coincident with the addition of brightness as a relevant cue at the beginning of Phase 2. Brightness cues acquired very substantial, and equal, discriminative control in both unblocking groups and a control group trained only in Phase 2 with internal cues and brightness cues simultaneously relevant. Rats trained on an instrumental, go/no-go discrimination problem on which internal, for example, rewardproduced, cues are the relevant positive (S+) and negative (S-) discriminanda in Phase I fail to learn about brightness cues that are added in Phase 2 as relevant, but redundant, discriminanda (Capaldi, Verry, Nawrocki, 1982; Haggbloom, 1981, 1984). That is, prior acquisition of behavioral control by internal cues blocks (cf. Kamin, 1969) the acquisition of control by subsequently introduced brightness cues. This blocking effect is attenuated by a surprising reduction in S- reward magnitude coincident with the introduction of brightness as a relevant cue (Haggbloom, 1984), an effect called unblocking, which appears to be analogous to unblocking produced in classical conditioning by a surprising increase in unconditioned stimulus intensity (e.g., Kamin, 1969). Neely and Wagner (1974) reported that some instances of unblocking due to a shift in reward conditions could be a consequence of generalization decrement due to a change in reward-produced cues accompanying the reward shift. This possibility is of special concern in the instrumental conditioning situations employed by Haggbloom (1984) and Neely and Wagner (1974) , because of the very substantial behavioral control exercised by reward-produced cues in instrumental conditioning (see, e.g., Capaldi, 1966, and Haggbloom & Tillman, 1980). At present, the only evidence for unblocking in instrumental conditioning is the experiment reported by This experiment was supported by an Arkansas Stat e University faculty research grant to the author . The author 's mailing address is: Department of Counselor Education and Psychology. Arkansas Stat e University, P.O. Box 2127 , State University, AR 72467- 2127. Haggbloom (1984), and that experiment did not control for the contribution of generalization decrement to the unblocking effect. The purpose of the present experiment was to investigate whether generalization deerement contributed to the unblocking effect reported by Haggbloom(1984). Three groups-Groups Blocking, Unblocking:S, and Unblocking:S+N-were trained in Phase 1 on a discrimination problem on which reward-produced cues and certain intertrial-interval(ITI)-related cues were relevant discriminanda. Those cues are identified collectively as internal cues to distinguish them from brightness cues. Brightness cues were constant across trials in Phase 1, and thus were irrelevant. All three groups received large reward (L) on S+ trials. On S- trials, Group Blocking received no reward (N), Group Unblocking:S received small reward (S), and Group Unblocking:S+N received S on two-thirds and N on one-third of its S- trials. In Phase 2, internal cues remained relevant, and brightness cues were added as relevant discriminanda for all groups. Group Control began training in Phase 2 and was treated the same as the three experimental groups. All groups received L on S+ trials and N on S- trials, Groups Unblocking:S and Unblocking:S+N having experienced a reduction in S- reward from S (or a mixture of Sand N) to N on all S- trials. The intermixture of Sand N trials in S- during Phase 1 for Group Unblocking:S+N was the treatment of major interest in this experiment. Its purpose was to provide that group with experience responding to internal cues produced by N trials and thereby preclude the occurrence of generalization decrement due to a change in reward-produced cues when, in Phase 2, S- reward magnitude was shifted from S to N in the unblocking groups. 469 Copyright 1984 Psychonomic Society, Inc. 470 HAGGBLOOM The acquisition of stimulus control by brightness cues was assessed, as in the experiments reported by Haggbloom (1981, 1984), by an opposed -cue test involving the reversal of brightness cues. METHOD Subjects The subjects were 40 male rats, approximately 90 days old at the beginning of training, bred in the laboratory from Holtzman stock . Apparatus The apparatus consisted of two parallel straight alleys 105 em long x 9 em high and wide. The walls and floor of one alley were painted black, and the walls and floor of the other alley were painted white . The last 25 em of each alley constituted a goalbox separated from the rest of the alley by a manually operated guillotine door . The doors and goalboxes were painted the same color as the alley in which they were located . Each goalbox contained an unpainted wooden goal cup. A gray startbox, 9 em high and wide x 25 cm long, could be aligned to permit entry into one alley or the other. The startbox had a gray, manually operated guillotine door. The alleys were divided into three sections over which running times were recorded (respectively, start, run , and goal times) . Start times were recorded from the opening of the startbox door , which triggered a .Ol-sec clock, to a point 30 em into the alley. Run and goal times were recorded over the next 40 and 30 em, respectively. The offset of the first clock and the operation of the remaining clocks were controlled by photoelectric circuitry. Procedure The rats were housed in individual cages and had free access to food and water for at least 2 weeks prior to the 1st day of food deprivation. The food-deprivation schedule consisted of 12 g of Wayne Lab Blox per day and was begun 14 days prior to the start of discrimination training. It continued throughout the experiment, with the amount of food consumed in the runway during training being subtracted from the 12-g daily ration. Water was always available. On each of Days 12-14 of deprivation , rats were handled in squads of four for 4 min per squad. After being handled on each day, the rats received 10 45-mg Noyes pellets in their home cages. Discrimination training (...truncated)