DID YOU KNOW? DTIC has over 3.5 million final reports on DoD funded research, development, test, and evaluation activities available to our registered users. Click HERE
to register or log in.
Procedural Skills: From Learning to Forgetting
Can we help people forget less by knowing how they learn? Can we decrease forgetting by modifying what they learn? These have been long-standing questions in applied cognitive science and engineering. My dissertation study addresses the decay of procedural skills. A study paradigm was created to investigate learning and forgetting of procedural skills in a laboratory setting. Human participants learned and performed a set of novel spreadsheet tasks that are declarative or procedural, and perceptual-motor or cognitive. To examine procedural skills on learning and forgetting, one group of participants used key-based commands and the other group used a novel mouse and menus to complete the task. Participants were able to learn the task well in four learning sessions, confirming the Power Law of learning. Mouse users did not learn or perform better than keyboard users. Retention intervals (6-day, 12-day, or 18-day) showed clear effects on the amount of forgetting. Two modalities (mouse or keyboard), however, did not provide any statistically different rates of forgetting on the first return. When it comes to relearning (2nd and 3rd returns), mouse users showed significantly decreased mean task completion time, indicating relearning occurred in mouse users more effectively than keyboard users. The ACT-R theory, which is used as the main theoretical background, was tested against human data with regard to learning and forgetting. The skill retention model in ACT-R was developed to predict a mouse users learning and forgetting performance in one subtask. The model predicted the learning performance with r^2 = 0.8 and RMSSD = 1.8, when compared with human data. The skill retention model proved that an ACT-R model is able to predict learning performance. Human performance modeling using ACT-R can be used to evaluate efficacy of a training regimen by predicting learning performance, making contributions to workforce engineering both in industry and in military.
Approved For Public Release