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

Reward

Techniques > Conditioning > Reward

Description | Example | Discussion | See also

Description

A reward is something that is received, and liked, as a result of an action. A subject does something, gets the reward and feels good.

A reward may be a pleasant surprise, which is usually the case the first time it is received. It may be a surprise the second time too, but once the mental link is made between the action and the reward, the subject will come to expect the reward after they complete the action, even to the point of spontaneously and deliberately completing the action in order to get a reward.

Rewards are not always desired, for example a piece of food is a good reward for a hungry subject, but not for one which has just had dinner. Subjects can also get bored of any reward or come to expect it as a right rather than being a deserved payment for their action. Rewards can also be simple, such as attention or praise.

A reward word (or phrase) is one that is substituted for an initial reward such that the word itself becomes a reward, making the subject feel good. For example the initial reward of food is paired with 'good dog' so this phrase will in future make the animal feel good, thus removing the need to feed the dog every time it obeys a command.

Remember that reward is what gives the subject good feelings, which is not necessarily what you give. For example you may be giving food, but the subject is actually valuing your attention or just being in a pleasant place. For the reward to be mentally linked to the action, it must be given immediately after the action.

Example

A rat in an experiment accidentally pushes a lever. A bit of food appears. Later it accidentally knocks it again and more food appears. Before long, it presses the lever whenever it is hungry.

An employee at a company does a good job on some work. Her manager praises her. Next time she does good work she tells the manager and stands by, expecting praise.

Discussion

A reward is not the same as a stimulus. In conditioning, the stimulus happens before the action and the reward happens after the action. Having said this, if the reward is expected, it can become an anticipatory stimulus as the subject performs the action in order to get the reward. In this way, a cue such as 'sit' can lead a dog to think about getting a reward of food, which becomes the inner stimulus for sitting. Rewards can also become stimuli for further action, from delighted jumping around to an intended sequence of actions.

Reward is an important part of the learning process. All creatures are pleasure-seeking, and will learn what leads to pleasure so they can do more of the causing action. Of course they also seek to avoid pain.

The satisfaction of needs is a primary means of gaining a reward. Basic needs, including such as food, warmth, affection, acceptance, can be good rewards for conditioning work. We can also have higher needs such as achievement and transcendence, though these may be more difficult to give as a reward. Food, for example, is hence called a primary reinforcer or primary reward.

In conditioning, the purpose of giving a reward is to link not just the reward to the action, but the cue or command that happens before the action. When the cue is strongly linked to the action, the reward may then be gradually 'faded', removing it from use. The subject now will obey the command.

Because animals have limited mental ability, a reward given too long after an action will not be associated with the action. While humans can make longer-term associations, rewards are still more powerful when they are given immediately.

Rewards can easily become extrinsic motivation, where the subject performs the action in order to get the reward. Intrinsic motivation, on the other hand, leads to performing the action for other, inner reasons, such as for the simple joy of doing it. People in particular will work harder for intrinsic reasons. This can lead to managerial skill problems as extrinsic motivation is easier and many use it even though it is less effective. In using intrinsic motivation, the reward must not replace inner reasons. For this purpose, small and personal rewards, such as thanks or small gifts can work well. The person has to explain to themselves why they worked so hard and cannot attribute this to being for the small reward, pleasant as it is, so they conclude that they worked hard because they wanted to and because they liked the work.

Punishment is a reverse of a reward but is easy to get wrong. Punishing a subject for not doing something (or doing it inadequately) may well not lead to them learning. Punishment is aversive, leading the subject to fight-or-flight, which can take many forms. People in particular can have complex responses to punishment, such as storing up resentment and taking subtle revenge. Animals also can develop all kinds of unwanted responses, which very often makes reward-based training methods the best approach.

See also

Action, Extrinsic Motivation, Intrinsic Motivation, Reinforcement