A quantitative approach to stimulus synthesis

Abstract

The research presented in this dissertation investigated the stimulus synthesis process. By stimulus synthesis is meant the operations which occur when an organism combines and evaluates the information it receives and effects a response on the basis of those multiple sources of information. The stimuli for the experiment were four hundred 35 mm slides, Each slide had four cues which subjects could use to solve the problem. Each cue had twenty different values. The task required the subjects to assign numerical values to each slide as it was presented, After each response. one of three kinds of knowledge of results (KR) was given. The KR was constructed from the cue values in such a way as to describe three functions which previous research indicated might best describe the stimulus synthesis function. These were called the additive. square, and log KR after the arithmetic operations performed to obtain the values. It was assumed a kind of KR which describe a function that most closely approximates the synthesis process would enhance performance more than a kind not as congruent with this process. The experiment tested the synthesis process as a function of age and experience with the task. This was done to determine whether the synthesis process changes as a result of maturation and learning, and to determine if different functions would describe the behavior at various stages of learning. Sixty high school students and sixty grade school students served as subjects. The results tended to support the previous research. The additive KR produced the best performance, followed by square and log KR, in that order. All the differences were statistically significant. The pre-adolescent children performed with more error than the adolescent children, but there was no interaction of age with KR. The Stage X KR interaction was significant. This was attributed to the early plateau achieved by the log KR groups as compared with the additive and square KR groups. Although the results show that among the hypotheses tested a linear model best describes the synthesis process, it was concluded that the actual relationship falls between the square and additive models. It is suggested that a closer description of the function could be obtained if other factors were introduced to remove some of the error variance.