Abstract : This paper describes a pilot study on how human subjects process information during a diagnostic inference task. The objective was a descriptive/predictive model of the inference task and how that task could be affected by implementation of an automated system. The study directly supported research being conducted by AFHRL on quantitative techniques to predict the impacts that automation may have on operator performance, by defining its interaction with the operator's information processing (Modelling Impacts of Automation on Non-Automated Tactical Command and Control Systems). The pilot study involved testing human subjects who had to infer the identity of two fictitious diseases by sampling up to eight symptom dimensions. A set of process and performance variables were selected for measurement. Signal detection theory served as the data collection design. Results were in line with anticipated outcomes (i.e., certainty increased as more cues were sampled); however, certainty rate of increase was highest for trials where subjects sampled four cues and lowest for trials where subjects sampled eight cues (total number of cues was eight). The pilot study helped formulate a list of critical variables expected to affect the operator's information processing and define plausible relationships between those processes and automation assistance. (Author)