Training often requires that a sequence of arbitrary behaviors be structured so that they occur in a specific order. This order of occurrence is based on a predetermined continuity in which one behavior must always precede the next in a set sequence. Orderly sequencing is accomplished by making the advancement of the series contingent on the emission of a predetermined response occurring before the next one in the series is selected. To accomplish this goal, each task in the chain is brought under the stimulus control of a dual-functioning signal serving both to reinforce the correct antecedent behavior conditionally and to simultaneously select the next response in the series. This pattern is repeated until the entire sequence of behaviors is emitted and terminally reinforced.
Within the chain, each discriminative stimulus (Sd) provides conditional reinforcement for the behavior that it follows and simultaneously selects the next response. The SD's dual function is an outcome of the way in which a chain is constructed. The chain is built up by connecting the final response with the terminal reinforcer and then adding on successive behaviors up to the origin of the chain. To obtain reinforcement, dogs must perform each response in the chain at the proper time. Each Sd in the chain not only selects the next behavior in the chain but also reinforces the preceding behavior because it advances the dog one step closer to the terminal reinforcer (Sr+).
perhaps the easiest way to show how chaining works is to use a common example. The recall pattern involves a six-part chain: sit, stay, come, sit-front, finish, and sit. Both the terminal response and the origin are sit responses. Between the origin and the terminal response are four chained responses: stay, come, sit-front, and finish. These various components are linked through shared stimulus control and conditioned reinforcement. The chain is terminated with a sit response and final reinforcement at the trainer's left side. The terminal response is under the stimulus control of the Sd "Sit." "Sit" not only selects the terminal response, it also conditionally reinforces the preceding finish behavior. The next link is the finish behavior, which is brought under the stimulus control of the Sd "Heel." "Heel" not only selects the behavior of moving to the trainer's side but also conditionally reinforces the previous link in the chain—sit-front. Sit-front is under the stimulus control of the Sd "Sit," which not only selects the sit response but also conditionally reinforces coming. Coming is under the stimulus control of the Sd "Come," which also conditionally reinforces the dog for staying. The Sd "Stay" selects staying behavior and conditionally reinforces the sit response. The sit response is the origin of the chain and is under the stimulus control of the SD "Sit"
A far more common form of chaining in everyday dog training is called forward chaining. Although forward chaining lacks the behavioral elegance of backward chaining, it does offer many valuable features and several distinct advantages. Forward chaining requires only that a series of responses be performed on cue for a single reward. The forward chain is built up by placing each response component on an intermittent schedule of reinforcement and randomizing its occurrence in the chain sequence. Such a strategy of intermittent reinforcement and randomization results in the development of a hoped-for expectancy occurring equally with each link in the chain—a result that usually translates into a general strengthening of the chain, as well as immunizing component responses against extinction. It should be noted in the case of backward chaining that behaviors near the end of the chain are the most strongly reinforced, since they are the closest to the terminal reinforcer. In the case of forward chaining, responses near the origin of the chain are the strongest with the lure of reinforcement rapidly declining as the chain of responses is extended. Care must be taken, therefore, not to extend the forward chain too rapidly, perhaps causing a dog to quit. An important advantage of forward chaining is that it provides a means to place acquired behavior on an intermittent schedule. For example, instead of reinforcing a dog on every occasion for sitting, forward chaining might require the dog to perform any combination of the series stand, sit, down, sit from the down, and sit-stay for a single re ward. Individual responses comprising the foregoing chain are conditionally reinforced at each step with praise ("Good"), but the final reward is presented only after the required sequence is performed or the whole series is completed successfully. Forward chaining teaches dogs to expect reinforcement after a certain amount of work, regardless of the actual behaviors performed. Also, the order of responses can be easily altered without significantly affecting the viability of the chain. The response order of a backward chain is locked in without much room for variability or change, other than shortening the chain by assigning the origin to a response closer to the terminal reinforcer. Also, if any of the behaviors in the backward chain fail (a highly likely outcome with responses near the origin of a long chain), the whole chain breaks down. Although carefully constructed backward chains can be immunized against extinction, the sheer complexity of the chain and the repeated occurrence of individual component responses without direct reinforcement make backward chains very sensitive to breakdown under natural training conditions.
Was this article helpful?