Eliciting and Representing Mental Models
Mental Models

Across a variety of projects, we have developed methods for identifying knowledge structures that capture, in part, a decision maker's situation understanding and plans.

In these projects, we identified mental models based on critical incident interviews in which officers described actual decision making experiences. This occured in several analytical steps, in which we discovered recurring topics, labelled them, and analyzed their tendency to occur together. A mental model is a set of correlated topics, typically linked by causal relationships. More recent work has explored automated knowledge acquisition methods that supplements and confirm these manual methods.

The following is an example of an enemy intent mental model.

Time Stance

Mental models of time stance are of critical importance in decision making. A time stance is the relationship that a decision maker's action has to the decision cycle of another decision maker. The decision maker can try to influence the other party's actions in a favorable way, try to predict the other party's actions and prepare to act on the prediction, or simply react to them after they have begun to occur. The other decision maker may be the enemy or it may be othes in the decision maker's own organizatiion.

It seems plausible to regard initiative as increasing from reactive to predictive to proactive time stances.

Mental Models and Initiative

A further analysis sheds more light on initiative. The following multidimensional space represents an analysis of correlations among mental models and other variables. Initiative increases from left to right, as we go from reactive to predictive to proactive. However, the three time stances from a triangle, with a time stance at one of its corners. The two dimensions of the triangle are one way of breaking initiative down into components: (1) Is knowedge of another decision maker's action obtained early (proactive and predictive) or late (reactive) in the other party's decision cycle? (2) Is knowedge of another decision maker's action obtained merely by assessment (predictive and reactive stances) or is it obtained by active influence (proacative stance)?

Different mental models tend to occur more often in connection with the three different time stances, hence, are associated with different degrees of initiative:

  1. The reactive time stance typically involves surprise and mental models of alternative explanations of what happended.
  2. The predictive time stance involves models of enemy and friendly intent, models of reliability of sources, and models of rate of movement.
  3. The proactive orientation depends on mental models of the higher-level purpose of the operation. It also involves mental models of enemy and friendly intent, but in a different causal relationship: Freindly intent causally affects enemy intent rather than the other way around.

Mental Models and Expertise

More experienced officers were significantly more likely to use mental models of higher-level purpose and the proactive time stance than less experienced officers. There was also a tendency for more experienced officers to use friendly intent mental models, i.e., to be aware of the plan of one's own organization.

The solid line in the diagram below shows the mental models used in at least 30% of incidents by more experienced officers; and the dashed line shows those used 30% of the time by less experienced officers.

Shared and Non-Shared Mental Models in Teams

Specialization is one method for increasing efficiency. Thus, different teams within an organization, or different subteams within a team, may be assigned different task responsibilities in different locations, and may also receive different training and equipment. On the other hand, coordination within and across teams depends on a significant amount of shared knowledge about the situation and plan. A major problem in teamwork is, How best to balance efficiency and coordination?

In our analyes of mental models, we found significant correlations between the use of different kinds of mental models and the type of unit to which an individual belonged, as well as a significant core of overlap. The figure below shows the mental models that occured in 30% or more of the critical incident interviews for a particular unit type.

As the figure shows, only intelligence officers considered the sources of evidence and alternative ways of explaining it. Officers in heavy and light units were concerned with the conclusions of the analysis (e.g., about enemy intent). They seldom reflected on the sources of an inference. On the other hand, intelligence officers were alone in not consistently referencing the components of friendly plans.

Officers in heavy units tended to be proactive more often than other officers, perhaps reflecting a greater abiliity to influence the enemy. These officers were likely to reflect on higher-level purpose, friendly intent, and enemy intent. By contrast, officers in light units were more likely to be predictive. These officers focused more exclusively on enemy intent, largely in order to avoid concentrations of enemy strength. Finally, specialized units focused more on friendly intent, reflecting their role in support of plans constructed by other units.

These results reflect both the advantages and disadvantages of specialization. Lack of consistent considertaion of friendly plans or purpose permits intelligence officers officers to focus on their own job, but it might on occasion cause intelligence officers to overlook creative opportunities for information collection. Similarly, lack of awareness of evidence interpretation issues might lead maneuver officers to put more weight on predictions of enemy intent than they deserve. By focussing too narrowly on friendly plans, specialized units may fail to anticipate changes in those plans to achieve higher-level purposes in the face of enemy action.

The need to balance specialized efficiency and shared knowleddge has important implications for team training. A flexible process of critical thinking is necessary to decide how deeply to probe into mental models associated with other tasks, and to determine the right balance in each particular context.

Critical Thinking about Mental Models

Mental models and critical thinking are inextricably tied together in decision making. Proficiency is a combination of knowledge, i.e., well-structured mental models, and the ability to think critically about knowledge. Even more importanly, critical thinking is a key element in the acquisition of knowledge. Mental models are not static, but evolve thorugh repeated cycles of critical thinking in which problems of uncertainty are identified and addressed. It is the latter process that is described in the Recognition / Metacognition model.

The critical thinking training that CTI has developed is designed to convey both of these key components of proficient decision making.

See also:


Copyright © 2000-2011 Cognitive Technologies, Inc.
Questions? Comments? Contact webmaster@cog-tech.com