How we change what others think, feel, believe and do
Cue --> Action --> Reward --> Cue --> Action --> Reward
Gets shortened to:
Cue --> Action --> Reward/Cue --> Action --> Reward
Cue --> Action --> Cue --> Action --> Reward
Cue --> Action --> Action --> Reward
Before creating an action chain, make sure that each individual action is under stimulus control, with reliable action on any given cue.
This provides a way of training subjects to perform multiple actions on a single cue. The progression to a complete action chain can be encouraged by giving a bigger reward when the sequence is completed, such as making a big fuss of the subject.
The best way to train an action chain is backwards. Start by establishing the last action, then the penultimate one plus the last one, etc. This makes sure the subject's reliability gets stronger as they progress through the chain, rather than forgetting or getting distracted when later, less-well-established actions are expected.
A dog trainer teaching 'fetch' first gets the dog to hold and release a ball. Then to bring it back from a few paces away, and eventually to run and find a distant ball that they did not see thrown.
A child initially has to be got up, dressed, fed and readied for school. After a while 'time for school' is the only prompt needed.
Homogeneous chains are sequences of similar actions, such as horse jumping over multiple fences. A heterogeneous chain is a sequence of different actions that is rewarded only after the sequence is completed.
Action chains can also happen accidentally when a subject guesses what is needed next, such as when a dog sits on command and then pre-emptively lies down.
Chains can also break down, which is easy to misinterpret as intentional misbehaving. Subjects, however, can forget or be distracted, especially when they are tired or when there are many other competing stimuli.
And the big