Behaviors are controlled by their consequences
A behavior is always a tool to achieve certain effects, so the effects (consequences) control the behavior. Consequences are those events which occur right after the behavior and are connected with it (they appear simultaneously with that particular behaviors). If you want to train (change a behavior) you have to modificate its consequences.
Consequences of a behavior, which are important for the animal and influence that behavior are called Reinforcements or Punishments.
Consequences leading to the intensification of a behavior are called reinforecements. There are two types of reinforcements:
Positive Reinforcement (R+)
As a consequence of a behavior something good APPEARS (that’s why R+) leading to reinforcement of a behavior (the behavior intensifies)
If we can do something beneficial to us we will do it more often.
In training we use mainly positive reinforcement – it is the most efficient and easiest to control; besides, an animal will easily understand it and will be strongly motivated,
Negative Reinforcement (R-)
As a consequence of a behavior something bad DISSAPEARS (that’s why R-) leading to the intensification of a behavior.
If we can do something, that will make something unpleasent disappear, we will do it more often.
Negative reinforcement is the base of traditional systems of training – something unpleasant (usually pressure) disappears, when an animal does what we want.
Consequences which decrease the frequency of a particular behavior are called punishments; there are two types of punishments:
Positive Punishment (P+)
As a consequence of a behavior something unpleasant APPEARS (that’s why P+) what leads to the extinction of a behavior (decreases its frequency). It is very often reffered to as a punishment.
If we do something, that makes unpleasent factor appear, we will stop doing it.
Negative Punisment (P-)
As a consequence something good DISAPPEARS (that’s whyP -) and results in extinction of a behavior (decreases its frequency)
If we do something that makes something pleasant disappear, we will stop doing it.
If a behavior that was always reinforced is stopped to be reinforced it may go extinct. If a behavior is reinforced variably, then it is less likely to disappear (if we not always expect a reward, we will try untill we get it – just like in case of one-armed bandits).
Before a behavior disappears, extinction burst occurs – a behavior appears with increased strenght and frequency. If we reinforce it then, it will become very strong.