Presentation
CANCELED - Is the efficacy of uncertainty communication dependent on the individual? A preliminary analysis
DescriptionIncreasing the transparency of automation, i.e., its understandability/predictability, is theorized to improve human-automation interaction. For example, uncertainty communication is a type of transparency that tells the human that an automated system may not be able to complete its task. However, performance does not always improve once it’s delivered. Of particular interest is to see if scan patterns, multitasking performance, and/or the human’s cognitive abilities, enlighten this contrast. Four hundred ninety-two Naval Aviation trainees completed a supervisory control simulation where their primary task was to correct errors of the three automated systems, each varying in reliability. They also had to respond to chat messages accurately and quickly. Halfway through the mission, a subset of participants received uncertainty communication about the least reliable system while all other participants received generic information. Multiple linear regression models were selected per a stepwise procedure. When no uncertainty communication was given, individual differences in attention control predicted automation error correction rates. When it was given, automation error correction rates were predicted by chat message performance and how much the highly reliable automated systems, i.e., the ones not referenced by the uncertainty communication, were scanned. Clearly, there is nuance to the efficacy of automation transparency.
Event Type
Lecture
TimeFriday, October 17th8:20am - 8:40am CDT
LocationGrand Hall M/N
Communications
Similar Presentations