Document Type


Date of Degree

Spring 2010

Degree Name

PhD (Doctor of Philosophy)

Degree In

Industrial Engineering

First Advisor

Geb W. Thomas

First Committee Member

John D Lee

Second Committee Member

Linda Ng Boyle

Third Committee Member

David Watson

Fourth Committee Member

Joseph Kearney


Many supervisory control systems require the operator to solve any problems that the system's automation cannot accommodate. Consequently, this class of systems would benefit from designs and methods which improve operator problem solving performance. Currently, human factors researchers develop designs and methods emphasizing the cognitive capacities and abilities of operators. For the most part, these approaches neglect the emotional state of the operator, although emotion has been shown to have an important impact on performance in many other domains.

This dissertation introduces the modified Multidimensional Problem Solving (m-MPS) Model, a theoretical model predicting how affect, one aspect of emotion, will influence problem solving performance. The model was tested in an experiment in which 32 participants attempted to correct a series of 5 bugs in a computer program. During their task, they received compiler messages with keywords specifically chosen to create a positive or negative affective state. The model predicted that the participants with messages designed to increase positive affect would seek solutions with a more divergent thought process, and this would be indicated with a more diverse set of problem-solving approaches, along with higher scores on a divergent thought measuring test administered throughout the experiment. Those with less positive affect would seek solutions in a smaller, less creative space and demonstrate less divergent thought. Unfortunately, the feedback messages did not appear to evoke an emotional response powerful enough to create a measurable change in emotional state. However, the messages did affect various aspects of the participants' performance in ways consistent with the model, including fewer repeated solutions with increasing divergent thought scores (F(1,423) = 12.39, p < 0.01) and the probability of continuing the problem solving process declines with each unsuccessful attempt (Z = -2.98, p = 0.003). The most compelling result was that participants receiving the negative messages were significantly less likely to successfully complete the problem-solving task (Wald Χ2 = 4.06, p = 0.044). These results suggest that in human-computer interactions, messages are an important factor in creative problem solving performance. Further research is necessary to determine the source of these effects in supervisory control interfaces.


Affect, Human factors, Human-computer interaction


x, 121 pages


Includes bibliographical references (pages 108-112).


Copyright 2010 Kristopher M Thornburg