Reinforcement
|
|
In operant conditioning, reinforcement is the presentation of a stimulus contingent on a response which results in an increase in response strength (as evidenced by an increase in the frequency of response). This concept has been criticized as circular, since it appears to argue that response strength is increased by reinforcement while defining reinforcement as something which increases response strength. Non-circular definitions have been proposed; for example, reinforcement can be defined as consummatory behaviour contingent on a response. Reinforcement is the key concept and procedure in the experimental analysis of behavior.
Whether the definition is circular or not, the study of reinforcement has produced strong, reproducible results.
| Contents |
Schedules of Reinforcement
The effects of different schedules of reinforcement have been extensively studied. These schedules are:
- Continuous, in which a reinforcer is presented after every desired response,
- Fixed ratio, in which a reinforcer is presented after every nth response,
- Fixed interval, in which a reinforcer is presented after the passage of a specified length of time from the beginning of training or from the presentation of the last reinforcer, provided a response has been made during the period,
- Variable ratio, in which the number of responses between reinforcers varies, but on the average equals a predetermined number, and
- Variable interval, in which the time between reinforcers varies, but on the average equals a predetermined time.
Ratio schedules produce higher rates of responding than interval schedules, with variable ratio scales producing the highest rates of response. Variable ratio schedules produce the greatest resistance to extinction, which is the decline in response strength following the cessation of reinforcement.
Classifying Reinforcers
Positive reinforcement is the contingent presentation of a stimulus following a response, resulting in an increased likelihood of the response occurring in the future. Negative reinforcement is the contingent withdrawal of a stimulus following a response, resulting in an increased likelihood of the response occurring in the future. Unconditioned reinforcement, also called primary reinforcement, is the presentation of primary reinforcers, stimuli which are inherently reinforcing (such as affection, food, sex, or sleep). Conditioned reinforcement, also called secondary reinforcement, is the presentation of a stimulus which has acquired reinforcing power through association with primary reinforcers. Social reinforcement is a form of conditioned reinforcement in which the reinforcer involves some sort of interaction with others. Positive punishment is the contingent presentation of a stimulus following a response, resulting in a decreased likelihood of the response occurring in the future, whereas negative punishment is the contingent withdrawal of a stimulus following a response, resulting in an decreased likelihood of the response occurring in the future.
Shaping
"Shaping" involves the delivery of a positive reinforcer (or removal of a negative reinforcer) contingent upon successive approximations of a desired response. That is, an organism's increasingly accurate productions of the desired response are reinforced. In training rats or pigeons to depress a lever or peck a key; for example, reinforcement will initially be contingent on simply turning toward the lever or key. As training progresses, the response reinforced becomes progressively more like the response desired by the trainer. Behaviours developed through the reinforcement of successive approximations to the eventual desired behaviour are called shaped behaviours and the process is called shaping.
References
External links
- An On-Line Postitive Reinforcement Tutorial [1] (http://psych.athabascau.ca/html/prtut/reinpair.htm)
- Strategies from Parenting Prescriptions [2] (http://www.parentrx.com/strategies/)
- Karen Pryor's Clicker Training [3] (http://www.clickertraining.com/home/)
- Reinforcement Theory & Artificial Intelligence [4] (http://rlai.cs.ualberta.ca/RLAI/rlai.html)