Dr. Lisa D Blalock

This study examined how the degree of automation provided during training and the general intelligence of the learner operate to influence task learning, and transfer to a no automation environment. An experiment was conducted using the Pasteuriser simulation, with participants assigned to one of three training groups: manual control, user-initiated automation, or automatically-initiated automation. The presence of automation facilitated initial training performance, but hindered performance later in training. When automation support was removed, costs were seen for those trained with automatically-initiated automation. An aptitude-automation interaction was observed, such that automation reduced the relationship between trainee intelligence and training performance. These results suggest that the effectiveness of automation in training will vary not only by the type of automation and the task, but also by the aptitude of the operator.

Participants completed a measure of computer trust and generated responses to the question, “What is like a computer?” These responses were coded as one of seven types ranging from being perceptually similar to sharing similar functions to being cognitively similar. Computer trust scores were associated with generating more responses, particularly of a functional nature. However, further analyses showed that participant sex moderated this effect, with computer trust associated with functional analogies for males, but perceptual analogies for females. The results suggest that computer trust and mental models may develop together and should be considered as important variables in the design and training for future technology.