An investigation into the persistence-strengthening effects of differential reinforcement of alternative (DRA) behaviour

Abstract

Ten experiments used hens to investigate the effects of different disruptive events and ways of introducing differential reinforcement of alternative behaviour (DRA), on the persistence of responding. Of interest was the persistencestrengthening effects of DRA on the target, or problem, behaviour, and whether this could be reduced by training an alternative behaviour in a separate context to that of the target behaviour. A DRA was compared to the effects of a procedure where the alternative and target behaviours were trained in the same context (traditional DRA). The different behaviours were pecking at different coloured keys, and responding was maintained by food reinforcement. Baseline training involved a three-component multiple schedule; one component with two yellow keys available (DRA analogue using VI 37.5 s and 150 s) and the other two with one key available in each (analogous to target (blue and VI 150 s) and alternative (green and VI 37.5 s) behaviours trained separately). A disruptor test followed, also using three-component multiple schedules. This disruptor test included a component with the two yellow keys (Concurrent/DRA Component) and one which combined the, previously single, green and blue keys (Combined Component). In each experiment, the sequence of baseline followed by a disruptor test was repeated with different components in each test. Persistence was measured as the responses in the test as a proportion of baseline responses for that behaviour. In Experiment 1 the disruptor was extinction and this showed that separate training reduced persistence of the target behaviour and extinction burst, as in previous research. Experiments 2 and 3 used the same procedures but the hens were exposed to a centre key alone (associated with one of VI 150 s (red centre key), 75 s (white centre key) or 37.5 s (pink centre key) schedule) prior to each baseline and this centre key then acted as the disruptor. Reinforcement was available on all keys in the tests. These found that the red key worked as a disruptor, and there were similar effects on persistence as in Experiment 1, regardless of the red key schedule. Experiment 4, using the same procedures but with no formal disruptor, also found the separate training gave reduced persistence of the target but not the alternative behaviour in the tests. Experiment 5 to 7 examined different disruptors (flashing light, sound, and a separate chamber), all with reinforcement continuing in tests. Target response persistence was reduced as a result of the separate training. Experiment 8 compared the effects on target persistence of separate training with the training schedule thinned prior combining and after combining. This found there was reduced target behaviour persistence when the schedules were thinned prior to being combined. Experiment 9 added a baseline with the target behaviour alone, prior to the previous procedure, and showed the same reduced target persistence even when both the target keys were initially presented alone. Experiment 10 showed the results were robust when each component was associated with a physically separate context. All experiments showed that training a target behaviour in a separate context to the alternative behaviour was successful at reducing the persistence-strengthening effects of DRA.