Mental Model Assessment for Education

The concept behind mental models was first formed by Kenneth Craik, a Scottish psychologist, who in 1943 wrote “the mind constructs ‘small-scale models’ of reality that it uses to anticipate events, to reason, and to underlie explanation” (Craik, 1943, cited in Johnson-Laird, Girotto, & Legrenzi, 1998, Introduction, para. 1). Johnson-Laird, one of the foremost authorities of early mental model theory, defines mental models as “psychological representations of real, hypothetical, or imaginary situations” (Johnson-Laird et al., 1998, Introduction, para. 1). His seminal text Mental Models (1983) has been the theoretical base cited throughout the literature. Though definitions and ideas about mental models vary widely, the general concept is that mental models “describe a cognitive mechanism for representing and making inferences about a system or problem which the user builds as he or she interacts with and learns about the system.” (Borgman, 1986, p. 48).

 

Mental models are not mental pictures or physical models of a system (Johnson-Laird et al., 1998), but rather the underlying knowledge structure that allows an individual to construct their perception of a system or content domain. Holland, Holyoak, Nisbett, and Thagard (1986) describe models as “assemblages of synchronic and diachronic rules organized into default hierarchies and clustered into categories” (cited in Kearsley, n.d., para. 3). These categories are comprised of three types of knowledge: declarative, structural, and procedural. Declarative knowledge is the “knowing what”. Individuals can know about something, but not necessarily what to do with it nor why. Structural knowledge represents the connections, or networks, between declarative knowledge. This is what allows humans to build schemata and mental models for any particular subject. Lastly, procedural knowledge is “knowing how to do” something, utilizing the connections made of knowledge generated through experience (Jonassen, Beissner, & Yacci, 1993). Thus humans can use their knowledge base and perform meaningful actions. Structural knowledge is the key to mental models and how they assist individuals in how they perceive a system or content domain, providing the underlying rules and connections. 

 

Operationalization of this knowledge structure is what provides individuals with the ability to understand a system through possession of a causal model and a runnable model. Knowledge of the system’s components and rules/operations allows an awareness of how the system works (causal), and an individual can mentally operate the system to predict actions and results (runnable) (Jonassen et al., 1993; Norman, 2002; de Kleer & Brown, 1988). When this internal understanding matches that of the actual system design, correct operation can occur (Norman, 1983; Norman, 2002). 

A good conceptual model allows us to predict the effects of our actions. Without a good model we operate by rote, blindly; we do operations as we were told to do them; we can’t fully appreciate why, what effects to expect, or what to do if things go wrong. As long as things work properly, we can manage. When things go wrong, however, or when we come upon a novel situation, then we need a deeper understanding, a good model. (Norman, 2002, p. 13)

Mental models are necessary to deal with problems and novel situations (Jonassen et al., 1993; Norman, 2002). They facilitate correct operation or functioning within a specific content domain, but more importantly they provide the capability for predicting what might happen based on certain actions. To merely learn a procedural task or memorize a list of information requires no more than rote rehearsal. To go beyond this and successfully apply or use knowledge in a different way necessitates understanding the fundamental principles and relationships among relevant knowledge so as to formalize potential actions and forecast the results. What happens when understanding is incorrect, as it often is to some extent? “If you are actually doing the task and there is a problem, they (models) let you figure out what is happening. If the model is wrong, you will be wrong too” (Norman, 2002, p. 71). Borgman (1986) agrees that appropriate models are “helpful and perhaps necessary” when they are correct, but that performance will suffer when the model is inadequate. Thus for individuals to solve problems and learn to operate complex systems, they must possess accurate structural knowledge of that system or content domain. “Domain-specific problem solving relies on adequate structural knowledge of the ideas in the domain being explored” (Jonassen et al., 1993, p. 10).

Mental models are messy, ill-defined, inaccurate, and incomplete. They are constantly evolving as individuals encounter new experiences, compare them to what they have previously stored in their models, and then alter their conceptual image accordingly. Johnson-Laird states “cognitive scientists have argued that the mind constructs mental models as a result of perception, imagination and knowledge, and the comprehension of discourse” (Johnson-Laird et al., 1998, Introduction, para. 1). Similarly, Donald Norman explains “in interacting with the environment, with others, and with the artifacts of technology, people form internal, mental models of themselves and of the things with which they are interacting. These models provide predictive and explanatory power for understanding the interaction” (Norman, 1983, p. 7). Norman provides a few generalizations about his observations from studying mental models (Norman, 1983):

• Mental models are incomplete.
• People’s abilities to “run” their models are severely limited.
• Mental models are unstable: people forget the details of the system they are using, especially when those details (or the whole system) have not been used for some period.
• Mental models do not have firm boundaries: similar devices and operations get confused with one another.
• Mental models are “unscientific”: people maintain “superstitious” behavior patterns even when they know they are unneeded because they cost little in physical effort and save mental effort.
• Mental models are parsimonious: often people do extra physical operations rather than the mental planning that would allow them to avoid those actions; they were willing to trade-off extra physical action for reduced mental complexity. (p.8)

In other words, people do not carefully organize and file in their minds a complete blueprint for any one system or content domain. Rather they accumulate and assimilate an assortment of concepts, rules, and relationships as they perceive them to make sense at the time. These can and do change over time, but often original perceptions and beliefs persevere even in the face of contradictory evidence. People are inherently rational beings, but not completely logical as we might assume. Mental models “are often constructed…with a kind of naïve psychology that postulates causes, mechanisms, and relationships even where there are none” (Norman, 2002, p. 38). This has crucial implications for many fields of concern including education, professional training, user-interface design, system design, etc.

Advancement of mental model theory has been concurrent with modern approaches in the fields of cognitive psychology and computational research, specifically artificial intelligence and human-computer interaction (Gentner & Stevens, 1983; Jonassen, 1995). The quest to understand what happens in the mind, how individuals process, store and recall information, and how people think about things is at the root of mental model theory development and study.

 

There is no such thing as a physical “mental model”. The concept is a theoretical construct, so there are several notions about how they are formed and operate. The prevalent opinion about the basis for mental models is that they consist of organized knowledge structures—the concepts, rules, and connections that were discussed earlier. This network of knowledge and its complex relationships is at the heart of the assumptions made by Carley & Palmquist (1992):

• Mental models are internal representations.
• Mental models can be represented as networks of concepts.
• The meaning of a concept for an individual is embedded in its relations to other concepts in the individual’s mental model. (p. 2)

Knowledge structure, or structural knowledge, is also referred to as cognitive structure. The general concept is based on the storing of particular chunks of information in a way that is associative and particular to the individual (Jonassen et al., 1993). Individuals inherently try to make sense of the environment around them (Bruner, 1966) and therefore develop their own personal “account” of what it all means. These representations, or mental models, can and do vary considerably amongst individuals. Information processing theory and schema theory support the notion of a mental structure that people build over time. This structure of specific content domains is what allows people to recall enough concepts, rules, and relationships to process a current situation. Due to limited information processing capability, mental models do not store everything, but rather contain (hopefully) enough information necessary to “run the model” and in effect see what might happen and determine what actions are required. “When individuals construct mental models…they make explicit as little as possible, and they focus on that information which is explicit in their models. Concomitantly, they fail to consider possibilities that lie outside their models.” (Johnson-Baird et al., 1998, Focusing in decision making, para. 2). This underscores the importance of fostering well-developed models that encompass a wide array of experiences and relevant connections in a content domain. 

 

Another way of looking at mental model development is that individuals tend to map new experiences or knowledge onto existing structural relationships (Jonassen, 1995; Gentner & Stevens, 1983). When individuals are confronted with a new phenomenon, they first attempt to connect it with a prior experience or schema that is perceived to be similar in some way. For example, when teaching students about electricity flow it is most common to explain this concept in terms of flowing water, pressure being analogous to voltage and flow equating to current. Mapping similar characteristics is effective for knowledge retention and enhancement of existing mental models to understand the new concept (Gentner & Stevens, 1983; Jonassen & Henning, 1999; Staggers & Norcio, 1993). Parush (2004) quotes Donald Norman as agreeing that a “metaphor is a stepping stone to a mental model” (para. 19). From the perspective of teaching people new technologies, Brandt (1999) believes that “analogy in particular has been shown to be an effective tool for teaching a conceptual understanding of technology” (p. 42). However, to underscore the point that mental model research is not an exact science, Borgman (1986) investigated user’s ability to understand and operate an online information retrieval system, a new technology at the time, and whether participants would be able to express their view of the system in any sort of analogy such as a card catalog (which had been the analogical model used during the brief training session). They were not able to do so, and several reasons were speculated including insufficient time to develop a model, methodology issues in attempting to extract participants’ conceptual models, and the fact that the relatively simple tasks perhaps did not need a mental model. Thus, in spite of the fact that most researchers agree about the effectiveness of analogy and metaphor, mental model extraction still depends on the situation and method(s) of measurement.

 

Mental model theory also provides an alternative explanation for human reasoning and inference. As opposed to a step-by-step, train-of-thought process, evidence shows that humans are simply not that logical. Decision-making is based on scattered, incomplete information in a way that is often incorrectly focused. When confronted by a situation requiring action or other response, humans fall upon their particular schema and mental model that seems to apply. Sometimes their own model is insufficient or inaccurate, leading to erroneous decisions as is evidenced through such situations as Three Mile Island and other human-centered disasters. (Seel, 1999; Gentner & Stevens, 1983; Johnson-Laird et al., 1998; Norman, 2002). Williams Holland Stevens, in Gentner and Stevens’ Mental Models (1983), determined that mental models were a critical component for human reasoning. Forbus (Gentner & Stevens, 1983) studied how people qualitatively reason about space and motion physics. He found that “the models we use…seem to be simpler than formal mechanics and appear to be based on our experience in the physical world” (p. 53). Instead of applying algebraic theorems, people tend to form a visualization of the phenomenon which offers relationships among the objects involved as well as the ability to “interpret these relationships.” Thus an individual develops a structural network of objects, concepts, and relationships that seems to describe and predict a physical phenomenon.