Improving Performance in an Imperfect Target Detection Simulation through Experience

摘要

Past research has shown an increase in performance over time when humans use imperfect automation. Typically, in these circumstances imperfect automation was delivered via a single modality. This study looks to examine the effects on performance when imperfect automation is delivered via multiple modalities which alternate during any one trial. Twenty-seven participants volunteered to take part in a cross-modal sensory target detection task with three trial blocks. The primary dependent variables were the response time and the associated accuracy rates. Results show that response time was significantly faster in the final trial block compared to the first trial block. Results also showed a trend that response time was faster in each subsequent trial than in the previous trial. A five minute exposure (one trial block) to imperfect automation resulted in a 24% decrease in response time while a ten minute exposure (two trial blocks) to imperfect automation resulted in a 38% decrease in response time. In regards to errors of omission, results indicated significantly lower miss rates in the final trial block compared to the first trial block and showed a tendency that errors of omission were lower in each sequential trial block than in the previous trial block. A five minute exposure (one trial block) to imperfect automation resulted in a 45% decrease in misses while a ten minute exposure (two trial blocks) to imperfect automation resulted in a 65% decrease in misses. These results suggest that alternating multi-modal cues produce stronger learning trends in human-automation interaction than previous uni-modal cue studies.