How we change what others think, feel, believe and do

Variable Reinforcement

Techniques > Conditioning > Variable Reinforcement

Description | Example | Discussion | See also

Description

Variable reinforcement means changing the reward that is given for a given action.

Types of variation that may be used include:

• Gradually reducing the reward so the action is completed only on cue.
• Changing the amount or type of reward (after perhaps starting with a more desirable reward).
• Gradually reducing the frequency of rewards so the subject sustains action on the hope of reward.
• Replacing a basic reward, such as food, with a simpler reward, such as praise.
• Giving a larger reward for completing a chain of actions.
• Increasing the gap between action and reward.
• Using a lure, showing the reward before the action.
• Using a surprise bonus to boost hope and motivation.

When varying reinforcement, always do this intentionally and on a planned schedule, rather than randomly using different methods or as a matter of convenience rather than based on creating designed ways of behaving.

When varying reinforcement, take particular care in watching the change in how the subject behaves, including delays before completing desired actions, performing unwanted actions, etc.

Example

A dog trainer rewards the animal with a bit of food for lying down on command. Over time, the trainer gives food less and less often until the dog lies on a simple signal.

A company gives a bonus for meeting sales targets. Over time these have got less. A new chief executive wants to change the culture to a more 'go getting' style, so reduces base pay and increases possible bonus. She also introduces a 'sales person of the month' award. Over time, the bonus system is varied carefully until the optimum motivations are found.

Discussion

A reinforcement is a reward given after a desired action is completed. The subject will quickly learn that action leads to reward. Reducing the reward may make the subject anxious, leading to such as quicker performance. The adaptability of many species means that the subject will quickly get used to any new schedule of reinforcement, making variation in reward successful as the subject lets go of previous expectations.

When you are training many subjects over time keep a record of what variable reinforcement works for you and what does not. In this way you can know from evidence what works best for you and your subjects.

A variable ratio reinforcement uses a reinforcing reward after an unpredictable number of cued actions. This is in contrast to a fixed-ratio schedule, where a reward is given after a set number of actions. Gambling has a variable ratio reinforcement as the player does not know when they will win or lose. Much of sales works like this as the sales person does not know which deals they will be able to close.

A variable interval reinforcement uses a reinforcing reward after an unpredictable period of time. This is in contrast to a fixed-interval schedule, where a reward is given at a particular time (such as a dog's dinner). When a reward can appear at any time, the subject may perform their own experiments, trying different actions to see if these will hasten the reward. In other words, the unpredictability of the reward

See also

Reinforcement, Fading, Surprise Bonus, Reinforcement Schedules