John Annett, L. Paterson
1966 unpublished
The report falls into three sections, a review of the literature on training for auditory tasks, an account of three experiments comparing ouing and knowledge of results as training techniques for a detection task, and the comparison of cuing and knowledge of results in an intensity discrimination task. The review of the literature indicates some disagreement on the kind and amount of improvement in simple auditory tasks. Some improvement is undoubtedly due to familiarisation with the mechanics
more » ... of listening and responding. Some may be due to changes in response criterion and some may be due to a genuine sensitisation to the auditory signals. The variety of techniques and performance measures does not facilitate straight-forward generalisation, although it is reasonably clear that some kindsof training can be effective. Investigating a previous finding that cuing and knowledge of reoults affect response criterion differently, the subjects in a detection task were required to respond with three degrees of confidence. It was found that subjective confidence is not affected by training, but knowledge of results still produced more "risky" behaviour than cuing as defined by the distribution of detections and false positive responses. Investigating the hypothesis that this difference was due to the necessarily higher rate of responding in knowledge of results an experimentally controlled rate of responding reduced the difference but did not eliminate it. A "cuing" procedure with no signal presented atall led to performance similar to that under "genuine" cuing, suggesting that in this task signal distribution is primarily what is learned and not, as had been hypothesized, the nature of the signal. Post-training vigilance performance appeared not to differ for the two techniques. In fact neither group showed significant changes in detection or false positives over a half hour vigilance OOSSirY±. In the third section cuing, knowledge of results and reduced noise were compared in training intensity discrimination. In this case knowledge of results was effective, the other techniques not leading to improvement over five one-hour sessions. As in the detection task, however, cuing and knowledge of results were distinguished by increasing cautinn and increasing confidence respectively following training. Reproduction of this publication in whole or irn pai1: is permitted for any purpose of the United States Government. NAVTRADEVCEN 4119-1 FOREWORD Purpose One of the critical needs of the Navy is to determine the "tr;ainable factors" in sonar and to learn how to optimally conduct such training. This study is the second phase of such an on-going program of oxperimentation. It is concerned primarily with the techniques of cuing and knowledge of results (KR) on a range of auditory detection and discrimination tasks. In the first phase of this program (Annett and Clarkson, 1964), cuing was found to be more effective than KR in an auditory signal detection task. However, in view of the past history of learning theory it seems so unlikely that knowledge of results should turn out to be an inefficient training technique that its use was continued in the present phase of the study to be certain of the results. Specifically, then, the main purposes of this phase were: 1. To investigate the surprising finding that cuing was found to be more effective than knowledge of results. Z. To clarify the finding that cuing and knowledge of results have different effects in auditory vigilance tasks, namely, that KR results in risky behavior (as shown by a large number of false detections), while cuing results in a more cautious approach (reduces false positive score). Results Among the results discussed in detail in the report are the following: 1. A critical review of the literature revealed extensive disagreement about the nature of auditory learning and about the optimal method of training. t NAVTRADEVCEN 4119-1 Z. The superiority of cuing over knowledge of results found in the previous phase for auditory detection was not confirmed in the present study, KR resulting in a greater, though insignificant, increase in detections. 3. The previous finding that cuing and KR result in different response effects was repeated. That is, KR led to a more lax response criterion, resulting in an increase in false positives, whereas cuing resulted in more cautious behavior, shown by a decrease in errors (false positives). Implications The results of this study are not yet applicable to sonar training. It is still necessary to extend this line of research to more complex signals, up to and including real sonar sounds. Also, longer periods than a half-hour are indicated for studying vigilance. However, this program has formed a baseline for the next phase of research using more complex discriminations. The relative effectiveness of cuing versus KR as training methods may be a function of the specific task being learned. KR appears more advantageous when the task requires the detection of larger percentages of signals, regardless of the number of false responses. On the other hand, cuing gives better results when cautious behavior, that is, fewer false detections are desired. There is also evidence that method of training interacts with stage of training. The determination of the optimal combination of training methods for type of task and stage of training requires further research. The next phase will investigate the effect of increased complexity on these interactions.
doi:10.21236/ad0630260 fatcat:hmcwrkt4yba7vn3kdn2pxozuo4