The Use of Teams in Image Interpretation: Information Exchange,Confidence and Resolving Disagreements 1

Abstract

Using the common procedure of having each team member in two-man teams check the interpretations of his teammate, three experiments centered around the following questions: (1) How much knowledge should the checker have of the initial interpreter's work? (2) How accurately can the initial interpreter rate the accuracy of his interpretations and can the initial interpreter effectively designate which of his interpretations need checking? And, (3) how can a third interpreter best be utilized to resolve conflicts in interpretations made by the original two-man team? Variations centered about the amount of information passed from initial interpreter to checker, discussion between team members versus no discussion, consensus versus one-man decision in determining the team product, confidence ratings made by interpreters and confidence levels below which interpretations were checked, and participation of a third team member under varying conditions to resolve conflicts in interpretation. Results were evaluated in terms of completeness of information extracted, accuracy and efficiency.