Choice and contingency in the development of behavioral autonomy during instrumental conditioning.

Abstract

In two experiments hungry rats received extensive training to lever press for food outcomes before one outcome was devalued by aversion conditioning and responding tested in extinction. If the rats were trained on a concurrent schedule in which two responses yielded different outcomes, performance during the extinction test was reduced by devaluation of the associated outcome. By contrast, if a single response was trained concurrently with the noncontingent presentations of the other outcome, test performance was insensitive to devaluation of the contingent outcome. This finding demonstrates that training on a schedule that offers a choice between responses that yield different outcomes prevents the onset of behavioral autonomy.

Choice and goal-directed behavior in preschool children

“…A comparable difference was observed by Kosaki and Dickinson (2010), who reported consistently lower response rates for rats that were trained on two responses concurrently than for rats that were only presented with a single response option.…”

“…Although exposure to multiple outcomes per se did not appear to maintain sensitivity to outcome value in the study by Kosaki and Dickinson (2010), their findings leave open the possibility that instrumental outcomes are processed in a different way from events that are not contingent on the individual's own actions. Because only the concurrent group was exposed to two different action-produced outcomes, we cannot be certain that the contingent and noncontingent outcomes in the single-action group received equivalent processing, and consequently, whether the training conditions provided animals in the single-action group with the same outcome experience as those in the concurrently trained group.…”

“…If exposure to multiple, response-produced outcomes preserves processing of their sensory features, and if this is the principal reason why choice training maintains sensitivity to outcome value, then both groups should continue to process detailed outcome representations and remain sensitive to outcome value following devaluation of one of the training outcomes. On the other hand, if the critical feature of concurrent training for maintaining sensitivity to outcome value is that it provides a choice between actions that lead to different outcomes, as Kosaki and Dickinson (2010) suggested, then choice rather than single-action training should be more likely to maintain sensitivity to outcome devaluation.…”

“…Recently, Kosaki and Dickinson (2010) reexamined whether the resistance-to-overtraining effect conferred by concurrent schedules could be explained in terms of the differential outcome exposure maintaining detailed outcome representations. They trained different groups of rats on single-action and concurrent schedules.…”

Habits as action sequences: hierarchical action control and changes in outcome value

“…This automaticity has two manifestations: (i) inflexibility of actions to the offline changes in the value of their outcomes [2,7] and (ii) the concatenation of actions executed together to form action sequences that are then treated as a single response unit [8,9]. These two aspects of automaticity share a similar neural structure (e.g.…”