A Stochastic Model with Applications to Learning

摘要: A stochastic model designed for analyzing data with changing probabilities is presented. On each of a series trials one two alternatives occurs and the occurrence are changed from time to by events. Corresponding class events an operator which represents linear transformation on alternatives. Cases fixed event considered. Recurrence formulas moments resulting distributions provided. These often tedious apply, but first second several bounds provided; these relatively easy compute. The problem estimating parameters discussed. No general solution obtained simplifying assumptions lead interesting special cases detailed procedures parameter estimation One such case arises when there operators commute, implying that have equal limit points or identity operator. method maximum likelihood applied this case. Another case, slope equal, discussed in Section 8. Applications certain kinds animal human learning described. examples given experiments verbal learning, avoidance training dogs, reward rats simple T-maze, behavior subjects two-choice situation.