Information Load, Proportion of Relevance, and Relevance Perception.

The effect of experimentally-induced relevance on subjective relevance perception was investigated. Subjects participated in a complex decision-making task. In Condition A subjects received 1, 2, 3 or 4 relevant messages out of a total of 10 messages for any one of four randomized playing periods. Condition B exposed subjects to 1, 2, 3, or 4 relevant messages out of 20, and Condition C exposed subjects to 4, 8, 12, or 16 relevant messages out of 20. All other messages received by the subjects were irrelevant to their task. Messages had been pre-rated by a parallel population who did not participate in the complex task. It was found that subjective perceptions of relevance in complex tasks tend to be overestimates of actual relevance levels, particularly when the proportion of relevant information to total information is low. Further, subjects tend to underestimate actual relevance levels when information load of relevant items greatly exceeds the optimum established by complexity theory research. Author