Summary
In most studies on reinforcement of desired behavior, it is noted that the speed of employee training is primarily determined by the schedule (the frequency and intervals) of support. There are five types of reinforcement schedules: one – permanent reinforcement and four – partial. Of particular interest are the plans of partial reinforcement, which will be discussed further. In reality, supporting all the manager’s desired patterns of employee behavior is not possible. Therefore, partial reinforcement is carried out when encouragement occurs only in some instances. There are four main partial reinforcement schedules: a fixed interval, a specified level, a variable break, and a varying level.
Reinforcement at a Fixed Interval and Ratio
In this case, the employee receives remuneration at specific intervals. If they demonstrate proper behavior every day, mounts can be made weekly. Examples of reinforcement with a fixed interval are regular bonuses and other bonuses (Yongliang et al., 2020). Mounting is performed through a certain number of manifestations of the desired behavior, say, every five times. For example, reinforcement is fixed when an employee is paid $ 5.50 for assembling 10 kg of fish. Most piecework payment systems are based on such a reinforcement schedule.
Reinforcement with a Variable Interval and Ratio
In this case, the employee performs support at different unpredictable intervals. An example is the factory director’s rounds of workshops, during which they personally thank the most diligent workers (Yongliang et al., 2020). The variable-ratio is not the period but the number of repetitions of the desired behavior. The employee can be rewarded through 5, 10 or 20 “right actions” (as in the case of slot machines: a person expects that the jackpot will fall after a certain number of games).
Reference
Yongliang, Y., Kyriakos, G. V., Hamidreza, M., Yixin, Y., & Donald, C. W. (2020). Safe intermittent reinforcement learning with static and dynamic event generators. IEEE Transactions on Neural Networks and Learning Systems, 31(12), 5441-5455.