Fast Learner: The Interplay Between Reinforcement Rate & Criteria
A high rate of reinforcement is often recommended for optimal learning. Although this is a good recommendation, the bigger picture is more complicated. Increasing the rate of reinforcement does not guarantee that optimal learning will take place.
A high rate of reinforcement describes an outcome and leaves unspecified what the trainer does to raise the rate of reinforcement. High rates of reinforcement are a product of well-designed shaping sessions and depend on the criteria for reinforcement, the probability of the correct response, and the trainer’s mechanics. Different strategies for increasing the rate of reinforcement may or may not lead to faster learning. For example, arbitrarily raising the rate of reinforcement can sometimes produce a “stuck” learner or reinforce extra superstitious behavior.
Other ways of increasing the rate of reinforcement, such as asking for easy, well-known behaviors periodically while shaping a difficult behavior, may not improve the acquisition of behavior but may help the shaping process by keeping the animal engaged. Optimal learning happens when high rates of reinforcement are achieved by balancing the criteria for reinforcement with the existing repertoire of the animal. This Session will explore various ways to achieve high rates of reinforcement and their subsequent effects on the acquisition of behavior.