Training for Rapid Interpretation of Voluminous Multimodal Data
Abstract
Previous research has not specifically addressed rapid decision-making based on large amounts of data, although a large body of research has identified various biases and characteristic errors in human decision making that promote economy of information processing. The purpose of this series of experiments was to determine whether seven known characteristic error types operate in rapid decision-making, and to determine whether training to identify key contexts in which these errors are likely to occur can reduce their occurrence. Student volunteers across four experiments performed a simulated incident detection task using information in a variety of formats. Results supported the primary hypothesis that each of seven types of characteristic errors of interest occur in rapid decision making. A reduced tendency to commit false alarms occurs as a result of general alerting to the presence of error traps, although individuals who received specialized training to reduce the occurrence of characteristic errors performed no better than a control group. Team task performance greatly reduced undesirable false alarm errors, but also reduced desirable incident hits. Teams that received training reported the fewest hits and errors, although it was unclear whether this was a result of training or the effect of alerting observed in individuals.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 2008
- Accession Number
- ADA480514
Entities
People
- Carl T. Blunt
- Dennis J. Folds
- Raymond M. Stanley
Organizations
- Georgia Tech