Cristina Bubb-Lewis Lucent Technologies
Mark W. Scerbo Old Dominion University
Adaptive automation refers to dynamic systems which adjust their methods of operation in response to changes in situational demands ( Rouse, 1988). In an adaptive automation system, the human and the machine must work together as partners in order to maintain optimal operation of the system ( Scerbo, 1994). Because of the close relationship between the human and the system, it seems reasonable that human- machine communication would be critical to adaptive automation. Thus, one goal of the present study was to examine the effects of different communication patterns on performance with an adaptive task.
Desirability for Control (DC) refers to one's need to control the environment ( Burger and Cooper, 1979). High DC people are described as decisive, assertive, and active while low DC people are described as nonassertive, passive, and indecisive. High-DC participants have been shown to display higher levels of aspiration, have higher expectancies for their performance, and set more realistic expectations than low-DC participants ( Burger, 1985). In addition, high-DC participants respond to a challenging task with more effort, persist longer, and perform better than low-DC participants.
Although DC has not been studied with regard to team dynamics, it seems to be relevant. A high-DC person might be less willing to act as a team member in solving problems because of their need to control situations. On the other hand, a low-DC person might rely too heavily on their partner. Either of these effects within a team could have a detrimental impact on the efficiency of the interaction, and consequently could affect human-computer interaction in adaptive automation in a similar manner.
The current study used a Wizard-of-Oz simulation ( Gould, Conti, & Hovanyecz, 1983) to study the effects of communication mode, task complexity, and desire for control on performance with an adaptive task. The focus of this paper will be on the DC results.
The simulation involved a"talking" adaptive computer that helped participants complete computer tasks. Four modes of communication were used that differed in the level of restriction placed on communication between the participant and computer. Two levels of task complexity were used with all participants completing both simple and complex tasks. DC was measured and participants were split into high-DC and low-DC groups for analysis. Dependent measures included task score as well as responses on a participant questionnaire
It was hypothesized that as more restrictions were placed on communication, performance would decrease and computer control would increase. Also, restricting communication was expected to make the interaction less efficient and make it more difficult for participants to complete the tasks, thereby leading to lower scores and increased computer intervention. Task scores were also expected to be higher for simple