Carl Westin

5-7 Discussion 109 While raw data are aggregated for the purpose of developing meaningful meta- information, there is a risk for inflation if this process is continued by further extract- ing meaningful information of meta-information to create “meta-meta”-information (as captured by the human-computer interface bottleneck 225 ). In the end, the aggre- gated information presented to the operator may be too distanced from raw data that the embedding transparency is lost. 5-7-2 Personalizing transparency In this study, personalization was restricted to the automation’s resolution advi- sories (i.e., conformal or nonconformal). In addition to conformance, there are sev- eral other interesting personalization aspects that deserve recognition and further research. This includes the automation’s communication style (e.g., automation eti- quette 114 ) and its interface characteristics, such as how information is portrayed by symbols, colors, contrast, and resolution. The questionnaires revealed differences between participants in their percep- tions of the two interfaces, indicating that the degree of transparency appropriate for one participant was not appropriate for another participant. Normally, such dif- ferences are considered negligible and disregarded in interface design. However, the human-machine interaction would likely benefit from personalizing interfaces to the individual’s preferences and needs. For example, by the system automatically adapting to the individual. Alternatively, as with adaptable automation, allowing the individual to influence the interface, and determine how much information that is presented, how it is structured, aggregated, and communicated. A relatively sim- ple method to achieve this is by allowing users to directly control the appearance and functions of the interface. A more advanced method for achieving this can be achieved by the system silently monitoring the user and change the interface ac- cording to the user’s way of working. 5-7-3 Varying SSD transparency The manipulation of SSD transparency in this paper did not vary the internal trans- parency of conformal and nonconformal resolution advisories in relation to Wang and Benbasat 209 three explanation types: why and how a solution was derived, and the tradeoffs considered. Explanations could, however, have been used to reflect information on whether the advisory was based on the participant’s own solution style and patterns in problem-solving, and how this style was determined. As such, a ‘how’ explanation would explain the reasoning underlying how the conformal res- olution advisory was selected: for example, a vector was chosen because this is how you typically solve these types of conflicts. A ‘why’ explanation would justify the