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

Negative Reinforcement

Techniques > Conditioning > Negative Reinforcement

Description | Example | Discussion | See also

Description

Negative reinforcement is guiding the learning about stimuli and actions around avoiding things that are not liked. When an action to avoid something unpleasant seems to work, then it is more likely to be used next time.

Negative reinforcement acts to increase an action using a stimulus that is not liked (an aversive stimulus). We act to reduce the things we do not like, and learn the best ways of doing this.

In brief, negative reinforcement is about removing bad things while positive reinforcement is about adding good things. Negative reinforcement is not about punishment.

Negative reinforcement can have unexpected and problematic results, which means it should be avoided where possible. For it to work in reducing an action, the subject must associate the aversive stimulus with the undesirable action, which usually means catching the subject in the process of the action (preferably at a very early stage).

When using negative reinforcement, make sure you:

• Make the discomfort happen at the same time as the action that is unwanted (or wanted).
• Make the discomfort such that it leads to the desired learning and action, such as cessation of unwanted activity.
• Create the minimum discomfort necessary.
• Avoid creating discomfort that leads the subject into extreme action such as panic, flight or attack.
• When the desired way of behaving is achieved (or even approached), immediately stop the aversive stimulus.

Note this final point: removal of the aversive stimulus. This is the key learning point as the subject learns that cessation of discomfort (which is a good thing for it) is caused by them stopping some action (which is a good thing for you).

With great care, gentle aversive stimuli can be used to create desired actions, such as kicking a horse with your heels or tugging rein to make it turn. Note that when the horse starts moving or turns sufficiently, you should stop kicking or pulling the reins.

Example

Ineffective negative reinforcement: A dog makes a mess when the owner is out. The owner shouts at the dog when the mess is discovered. The dog is confused and does not connect the disapproval with making the mess.

Effective negative reinforcement: A dog jumps on a kitchen table. The owner puts trays on table. Then dog jumps up and is upset by the noise and confusion of the trays falling. It mentally connects this aversive stimulus with the action of jumping up and does not do so again.

Effective negative reinforcement: A child is made to sit and think about things they have done wrong, which makes them feel discomfort. When they start to act out again, a stern gaze may be enough to make them stop.

Discussion

Negative reinforcement was a very important skill for our ancestors as it kept them alive. It helped them to avoid and cope with predators and other dangerous situations. Today, we still try to avoid things we do not like, although our actions may not always be helpful or desired.

In the same way that positive reinforcement leads to an increase in actions that lead to pleasure, negative reinforcement leads to actions that avoid discomfort. Both are forms of reinforcement, which is defined as increasing a certain way of behaving.

Negative reinforcement is often confused with punishment. Reinforcement always changes how a subject behaves. Punishment that does not change how a subject behaves is not negative reinforcement. This typically happens when the punishment does not happen at the same time as the actions that it seeks to reduce. Very often, punishment happens too late for reinforcement to occur. B.F. Skinner defined punishment as what happens when an action results in the loss of something desirable. Taking a toy away from a pet or child can hence be understood as punishment (even if you are just tidying up and do not intend to punish).

For punishment to be an aversive stimulus that acts to create negative reinforcement, it has to be associated with the actions that led to the aversive stimulus. Shouting at a pet can make it frightened such that it runs away. More shouting can make the pet more frightened, to the point it actively avoids the person who shouts at it.

Punishment can cause a range of coping actions, many of which add complexity and can have longer-term implications, for example where the subject develops a psychological disorder of some kind. A particular problem, especially in humans, is that people find aversive treatment unfair and may seek revenge, perhaps in underhand ways that avoid any further punishment.

The response to an aversive stimulus is to take any action that will avoid it, including doing things that are undesirable. Such actions may include:

• Ignoring the stimulus, pretending it is not happening.
• Running away or generally trying to avoid uncomfortable situations.
• Reactively try to reduce the stimulus on the spot.
• Fighting back aggressively.

A way of understanding negative reinforcement is as follows:

Situation/Action --> Perceived discomfort (aversive stimulus) --> Action to reduce discomfort --> Reduced discomfort --> Learning

We are in a situation or may act in a certain way. Something happens that makes us uncomfortable. We do our best to reduce the discomfort. We may learn several things here:

• Some situations where we believe we have done nothing wrong lead to discomfort. If we can learn to recognize the situation beginning to happen, we may be able to avoid the discomfort.

  • Learning: When people start shouting, discomfort will follow. Best to get out early.

• Some things that we do lead to discomfort, so in order to avoid the discomfort we must not do these things.

  • Learning: Making a mess leads to uncomfortable shouting. Best not make a mess (or maybe best not let them find our mess!).

• When we experience discomfort, there may be a way we can reduce this. When we are uncomfortable, we just need to repeat the avoidant action.

  • Learning: When we are uncomfortable, running away quickly seems reduce discomfort, at least for the moment.

• The best way to reduce discomfort may vary with the situation and our abilities. If we can learn what works when, we can be more successful in reducing discomfort.

  • Learning: When the door is open, running away is best. When the door is closed, wailing might work.

If we do not connect the discomfort with our actions, we cannot learn to reduce these actions. A problem with discomfort is that it grabs attention. We feel an immediate pressure to reduce it, rather than learning to prevent it happening in the first place. This is a key reason why using aversion is often ineffective as a means of training, and that positive reinforcement is a far better method than negative reinforcement.

See also

Reinforcement, Positive Reinforcement, Coping