Margaret A. McDevitt

When good news leads to bad choices

Pigeons and other animals sometimes deviate from optimal choice behavior when given informative signals for delayed outcomes. For example, when pigeons are given a choice between an alternative that always leads to food after a delay and an alternative that leads to food only half of the time after a delay, preference changes dramatically depending on whether the stimuli during the delays are correlated with (signal) the outcomes or not. With signaled outcomes, pigeons show a much greater preference for the suboptimal alternative than with unsignaled outcomes. Key variables and research findings related to this phenomenon are reviewed, including the effects of durations of the choice and delay periods, probability of reinforcement, and gaps in the signal. We interpret the available evidence as reflecting a preference induced by signals for good news in a context of uncertainty. Other explanations are briefly summarized and compared.

Competing sources of stimulus value in anticipatory contrast

In previous studies of anticipatory contrast, identical target components (A and B) preceded either a lower (extinction) or a richer schedule. Higher response rates occurred during the target preceding the lower rate of reinforcement, whereas preference was in favor of the target preceding the richer schedule. In Experiment 1, the response and preference measures were positively related when additional stimuli, with no reinforcement of their own, preceded the target components. The effect of these additional stimuli was presumed to be due to their overshadowing of the Pavlovian association between the target components and their following schedules. Experiment 2 also demonstrated a consistent relation between response rate and preference in a conditioned reinforcement procedure. In the absence of a strong Pavlovian association, anticipatory contrast, like other forms of contrast in free-operant procedures, reflects an increase in the value of the target component with an unchanged reinforcement schedule.

When good pigeons make bad decisions: Choice with probabilistic delays and outcomes

Pigeons chose between an (optimal) alternative that sometimes provided food after a 10-s delay and other times after a 40-s delay and another (suboptimal) alternative that sometimes provided food after 10 s but other times no food after 40 s. When outcomes were not signaled during the delays, pigeons strongly preferred the optimal alternative. When outcomes were signaled, choices of the suboptimal alternative increased and most pigeons preferred the alternative that provided no food after the long delay despite the cost in terms of obtained food. The pattern of results was similar whether the short delays occurred on 25% or 50% of the trials. Shortening the 40-s delay to food sharply reduced suboptimal choices, but shortening the delay to no food had little effect. The results suggest that a signaled delay to no food does not punish responding in probabilistic choice procedures. The findings are discussed in terms of conditioned reinforcement by signals for good news.

Inhibition and Superconditioning

Superconditioning is said to occur when learning an association between a conditioned stimulus (CS) and unconditioned stimulus (US) is facilitated by pairing the CS with the US in the presence of a previously established conditioned inhibitor. Previous demonstrations of superconditioning have been criticized because their control conditions have allowed alternative interpretations. Using a within-subjects autoshaping procedure, the present study unambiguously demonstrated superconditioning. The results support the view that superconditioning is the symmetric opposite of blocking.

Discrete-trial vs. continuous free-operant procedures in assessing whether reinforcement context affects reinforcement value.

Pigeons were trained to respond to two alternating concurrent reinforcement schedules. The reinforcement probabilities were .05 and .10 in one component, and .10 and .20 in the other. In one condition, the pigeons received training on a discrete-trial procedure in which the keylights remained illuminated for 5s or until a response occurred. In another condition, pigeons received training on a procedure in which the reinforcement contingencies were the same as in the discrete-trial procedure, but the stimuli were not turned off after 5s or after a response. Following training in each condition, probe tests were presented. In both conditions, the .20 alternative was, overall, preferred to the .05 alternative during probe tests. Following discrete-trial training, there was no reliable preference between the two .10 alternatives. However, when the stimuli remained illuminated during the intertrial interval periods during training, probe tests results showed preference for the .10 alternative that had been presented in the leaner context during training. The pattern of results is consistent with the notion that probe preference can be influenced both by the absolute reinforcement schedules associated with each alternative, as well as changeover behavior developed during training.

Effects of changeover delay on response allocation during probe tests

The present study assessed whether a pattern of responding that develops when choosing between two alternatives generalizes to novel choice tests when alternatives are presented in new combinations. Pigeons were trained on a two-component multiple schedule. In both components, a concurrent variable-interval (VI) 40-s VI 80-s schedule was used. The COD was 1 s in one component and 10 s in the other. The long COD produced consistently longer dwell times than the short COD did. Following training, subjects were presented with four types of probe-test components in which one alternative was drawn from the component with the short COD and one alternative was drawn from the component with the long COD. When the schedule values of the two alternatives were identical (VI 40 vs. VI 40 and VI 80 vs. VI 80), subjects preferred the alternative trained with the long COD (Ms = .78 and .61, respectively). Additionally, subjects preferred the VI 40-s alternative trained with the long COD to the VI 80-s alternative that was trained with the short COD (M = .85). Systematic preference was not observed when subjects were given a choice between the VI 40-s alternative that was trained with the short COD and the VI 80-s alternative that was trained with the long COD. These results demonstrate that a stimulus associated with a longer COD, and thus longer dwell times in baseline training, may be more preferred during probe tests than expected on the basis of the rate of primary reinforcement associated with that stimulus.

The Influence of Intertrial Interval Food on Extinction and Devaluation in Chain Schedules

The authors exposed pigeons to 2 equal 3-link chains by using variable-interval schedules of reinforcement. An intertrial interval (ITI) bisected by free food separated the chains. After baseline training, the authors presented terminal links in a successive discrimination to devalue 1 terminal link: The authors reinforced responses to 1 terminal link and extinguished responses to the other. The authors then presented full chains in extinction, except that they continued to deliver free food during the midpoint of the ITI. There were 2 principal findings. First, across all extinction conditions, responding decreased but did not extinguish. Second, when extinction testing revealed a terminal link devaluation effect in the 3rd condition, responding to the initial link was affected, but not middle-link responding. Overall, the results suggested that ITI food presentations can exert a substantial effect on responding in 3-link chain schedules, and they appear to influence both the pattern of extinction and devaluation effects.