Behavioral entropy as a measure of human operator misunderstanding

E.R. Boer

LUEBEC, San Diego, CA, USA

Human operator activity in general can be described as a process of controlling and resolving uncertainty and misunderstanding while trying to achieve or maintain a goal. This characterization applies to many levels and types of activity. As operators encounter more uncertainty, suffer from greater misunderstanding, or have reached the limits of their abilities, their activity patterns are generally more erratic and less predictable; their eye movements, hand movement, brain activity, etc. reflect a state that can be characterized as struggling, searching, or trying. These patterns of greater behavioral randomness are, for example, observed when operators do not know:

1. What they are looking at.
2. What the relevant cues are to control a system.
3. What the dynamics are of the system they are interacting with.
4. How to solve a problem.
5. Do not have the cognitive or sensory-motor abilities necessary for the demands of the task.

Thus the common characteristic across these types of situations lies in the relationship between level of ‘misunderstanding’ and behavioral entropy.

The theory of behavioral entropy is introduced and applied to a number of situations from a number of domains to exemplify what measurable human operator signals are most suitable and what characteristics of these signals lend themselves best to be quantified in terms of information content (i.e. entropy). Applications in the areas of perceptual pattern recognition, motor-control skill development, navigational learning, and interface evaluation are used to demonstrate the general applicability of the theory.

In nearly all disciplines where humans are part of an observed system do researchers search for meaningful ways to characterize human performance. The proposed theory of behavioral entropy is one that targets to quantify how well the operator is able to accomplish their task without imposing performance goals onto the measure. This is especially important in environments where human operators need to perform multiple tasks in parallel and have flexibility in how they distribute safety and performance tolerance margins across the many processes they control. Behavioral entropy quantifies operators’ ability to maintain their own safety and performance tolerance margins.

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.

© 2005 Noldus Information Technology bv