Two types of behavioral contrast in discrimination learning

Two groups of pigeons received daily discrimination training at two values on a line-tilt continuum. S+ (VI 1) and S- (EXT) intervals alternated, and a 30-sec criterion of no responding to S- was required before S+ returned. Rates of responding to S+ showed two separate contrast effects: at an intermediate stage of training a high peak rate appeared which declined, later in training, to a stable level still in excess of the VI baseline rate. The peak rate was correlated with the total number of responses to S-, while the final rate was not; suggesting that the peak rate and final rate may not be functions of the same variable. These results were compared with performance on a red-green discrimination where the two stages were not so clear. A line-tilt discrimination was repeated with fixed length S- intervals terminated by TO, and showed the same contrast magnitude in the final rate without any peak. The peak rate was interpreted as an effect of the ;punishment' contingency where responding to S- prolongs S- for 30 sec, while the final rate was taken to be analogous to previous demonstrations of contrast.     

Stimulus- and pellet-induced drinking during a successive discrimination

In three experiments, interim water drinking was examined in rats exposed to a multiple schedule whose two components were extinction and a variable-time 30-s schedule of food delivery. Two different drinking patterns were observed in Experiment 1. Pellet-induced drinking, characterized by high rates of postpellet drinking in the variable-time component, with little or no drinking in extinction, occurred when the acquisition of stable postpellet drinking preceded discrimination training. Stimulus-induced drinking, characterized by a burst of drinking at the onset of extinction, with no drinking during the variable-time schedule, occurred when discrimination training preceded all other experimental conditions. With extended training, stimulus-induced drinking eventually was accompanied by postpellet drinking. In Experiment 2, the rate of stimulus-induced drinking and the number of sessions during which it occurred without postpellet drinking were found to be inversely related to component duration. In Experiment 3, the rate of schedule-induced drinking was found to vary directly with component duration.     

Some temporal properties of behavioral contrast

Pigeons were rewarded on a variable time interval for pecking a translucent key illuminated with either a 45 degrees or a vertical line. The key illumination changed every 2 min during daily 1-hr sessions. When the rates of pecking were stable, reinforcement was omitted in the presence of the 45 degrees line. Responding in the presence of the vertical line increased. This increase did not disappear when responses to the 45 degrees line were once more reinforced, but when reinforcements for responses in the presence of the 45 degrees line were again omitted, responding to the vertical line increased again. After the second alternation of these two procedures, the increased responding to the vertical line appeared when responses were not reinforced in the presence of the 45 degrees line, and disappeared when reinforcement was available during both stimuli. In a second experiment, the key illumination changed between sessions only, so that 1-hr sessions of reinforcement and non-reinforcement occurred on alternate days. Responding to the vertical line still increased when responding to the 45 degrees line was not reinforced, but the increase tended to disappear during the session.     

Effects consistent with behavioral contrast during a multiple schedule for discrimination training on mands for attention

This study used a multiple schedule to assess the effects of an S+ or S- in the absence of rules on excessive mand rates with three school-aged children with disabilities and assessed possible contrast effects occurring in conjunction with the intervention. Each of the three participants was exposed to the presence or absence of an S- (black baseball cap). When it was worn, mand approach rates decreased for all three participants; conversely, mand rates increased to twice the baseline rates when the experimenter was not wearing the black baseball cap. Generalization probes conducted for one of the three caregivers showed similar changes when the caregiver wore the cap. Results suggest that effects consistent with positive behavioral contrast occurred with all participants.     

Some determinants of behavioral contrast in pigeons and rats

In Expt. 1, pigeons trained on a multiple variable interval, extinction schedule, showed a positive contrast effect by comparison with control groups trained with S+ only, provided that the interval between stimulus presentations was short (10 sec), but not when it was long (60 sec). Positive contrast also occurred more readily with an easier discrimination, and its appearance was highly correlated with the temporary appearance of transient contrast effects. Although a longer interval between trials produced an overall increase in rate of responding in subjects trained only with S+, it was suggested that this at best represented a different type of contrast effect. In Expt. 2, rats showed positive contrast to S+ when S− was correlated with a lower frequency of reinforcement, but not when it was correlated with a reduced magnitude of reinforcement. The results were discussed in terms of frustration theory.     

Some effects of discriminable goal-box conditions on the learning of a successive discrimination

In an attempt to confirm and extend a previous reslt, rats were trained on two tasks where a signal delivered at the start of each trial indicated which of two paths through a maze would be rewarded. In Experiment I both paths led to the same goal-box, and it was found that performance was better when the state of the goal-box was different on trials with each of the two signals. In Experiment II the two paths led to spatially separated goal-boxes. It was found that when the states of the two goal-boxes were discriminably different but the state of each of them remained the same from trial to trial, performance was better than when their states varied irregularly. It is suggested that these results have interesting implications for theories of behaviour.     

Stimulus-specific contrast effects during operant discrimination learning

In two experiments, pigeons' responding was equally reinforced in the presence of four line-orientation stimuli. Responding was then reinforced when only two of the four orientation stimuli were present; the remaining two orientations appeared during extinction. Response rates were often highest in the stimulus adjacent to the orientations presented during extinction and often lowest in that orientation adjacent to the orientations presented with reinforcement. These effects were stronger and more persistent when the stimuli were separated by a smaller angle, rendering the discrimination more difficult. These and other data suggest that discrimination training may not be accurately explained in terms of the simple effects of reinforcement and nonreinforcement associated with isolated stimuli, nor by accounts that depend upon stimulus generalization. Recent accounts of contrast that depend upon “emotionality” produced by nonreinforced responding or upon reinforcement-elicited responses are also difficult to apply to these data.     

Successive discrimination training with equated reinforcement frequencies: failure to obtain behavioral contrast

In two experiments, pigeons were trained on two-component multiple schedules in which responding in one component (S₁) was always maintained by a variable-interval schedule. In Experiment I, low response rates were reinforced in the second (S₂) component for six master subjects. This schedule was adjusted to equate reinforcement frequencies in the two components. These subjects were compared to yoked partners, for which reinforcement in the S₂ component was made available on a variable-interval schedule whose value was determined by the master subjects. A similar procedure was used in Experiment II, where the S₂ schedule for master subjects made reinforcers contingent on the absence of responding. No evidence was found in either experiment for a behavioral contrast effect in the S₁ component attributable to response reduction in the S₂ component. A reliable contrast effect was obtained from a group of pigeons given extinction conditions in the S₂ component, which was compared to a group maintained throughout on a multiple variable-interval schedule. The results suggest that previous indications of behavioral contrast in similar situations were probably caused by uneven reinforcement distributions or reflect uncontrolled fluctuations in response rates.     

Some parameters of behavioral contrast and allocation of interim behavior in rats

Two experiments examined the effects of baseline reinforcement rate and component duration on behavioral contrast and on re-allocation of interim behavior in rats. Positive behavioral contrast occurred during multiple variable-interval 10-second extinction (VI 10 EXT) after a multiple VI 10 VI 10 baseline condition, but not during multiple VI 60 EXT following multiple VI 60 VI 60 baseline. Component duration had no significant effect on contrast. These results differed from those found in studies of pigeons' key pecking. Contrast was accompanied by an increased rate of drinking in the changed component, but drinking in the constant component did not decrease. These results are not consistent with the competition theory of contrast, but are consistent with the predictions based on the matching law. However, no current theory seems to account for all instances of behavioral contrast.     

Positive behavioral contrast when pigeons press treadles during multiple schedules

Pigeons were placed on multiple variable-interval 15-second variable-interval 15-second and on multiple variable-interval 15-second extinction schedules in which treadle presses produced food reinforcers. Positive behavioral contrast occurred. That is, rates of responding were higher during the variable-interval 15-second component when the other component was extinction than when it was another variable-interval 15-second schedule. These results contradict the findings of other studies, which failed to find positive contrast when pigeons pressed treadles for food reinforcers. They may also question the additive theories of behavioral contrast that predict that contrast should not occur in this situation.     

Stimulus comparison strategies and task demands in successive discrimination

This research isolates two distinct strategies used to identify stimuli presented at different rates in successive discrimination tasks. The short-term strategy (used at rapid rates) compares current stimuli to immediately prior stimuli present in sensory or other short-term storage, whereas the long-term strategy (used at slow rates) compares current stimuli to a standard stimulus in long-term memory. In Experiment 1, subjects detected a medium-duration tone amid both long and short distractors. Analysis of false alarms indicated that strategy choice is strongly influenced by event rate, and therefore by the presence or absence of sensory traces of preceding stimuli. Experiment 2 was designed to force subjects to use the short-term strategy (to detect targets, present stimuli had to be compared to their immediate predecessors), and varied the event rate. Subjects were able to maintain a high level of performance throughout the task only at the fastest event rate. This pattern of results suggested that when the task demanded it, subjects could use a particular strategy, but if the event rate (and consequently the availability or otherwise of relevant memorial traces) was not favourable to that strategy, then performance was disadvantaged.