Important Note

Tufts ended funding for its Open Courseware initiative in 2014. We are now planning to retire this site on June 30, 2018. Content will be available for Tufts contributors after that date. If you have any questions about this please write to

Tufts OpenCourseware
Author: Fulcrum Institute Development Team

Using Molecular Workbench: Some thoughts on models and simulations and their uses

by Judah Schwartz

Scientists are always proposing models of the world around them and then using these models to help them think, analyze, predict and understand.

What do we mean by models? Are models the same sort of thinking tool as simulations? Are there meaningful distinctions between the two? If so, how do they resemble and differ from one another? Both models and simulations, whatever we might mean by those terms, certainly allow us to pose “what if” questions?

Let’s consider some clear examples – say a scale model of a new school building and a cockpit simulation used to train airplane pilots. Each of these is an artificial environment that refers to some external “reality”. Both contain a collection of entities – rooms, doors, windows, etc., in one case, dials, indicators, switches, etc., in the other. Some of the attributes of these entities and some of the relationships among these entities are incorporated into the model or simulation. For example, the way a door swings may be correct in the model, the material of which the model door is made may not be the same as the real door – the real building will likely be both air-conditioned and heated, the model is not likely to be either. In the case of the cockpit simulation the designers go to great lengths to make the simulated dials and indicators as faithful to the real ones as possible – on the other hand the person using the simulator is not up in the air. We see that, at least in these two cases, both models and simulations incorporate some properties and some relationships of what they are modeling or simulating and do not incorporate others.

In science education models and simulations tend to have two different but related purposes. Usually the purpose of making a model and exploring it is to expose the underlying mechanisms that govern the relationships among the entities. Usually the purpose of making a simulation and running it is to provide users with a surrogate experience of the external “reality” that the simulation represents. Roughly speaking then we can say that models seek to “explain” the complex systems they refer to while simulations seek to “describe” complex situations, often by offering rich multi-sensory stimuli that place users in as rich a simulated environment as possible.

Because of this difference in purpose it is not surprising the makers of models often seek to limit the complexity of their models so as to make the underlying causal and/or structural mechanisms clearer and more salient. In contrast, designers of simulations frequently tend to incorporate as much of the richness and complexity of the system they are simulating as possible so as to make the experience of using the simulation as rich a perceptual experience as possible.

In the educational context this difference in purpose points to different roles for models and simulations. In the elementary grades the emphasis is on having the students exposed to a rich repertoire of phenomena. For these students well-designed simulations can complement direct observation of nature, compensating for spatial and temporal remoteness or mismatch of scale to the human perceptual apparatus. Simulations of a rain forest allow urban students to overcome one kind of spatial remoteness. Simulations of an urban transport system allow rural children to overcome another kind of spatial remoteness. Similarly, simulations of the night sky that show the trajectories of the planets over several years during the course of several minutes help to overcome a mismatch of temporal scale. At the secondary level students are expected to understand and be able to explain the underlying causes and mechanisms of the phenomena they are studying. These are best explored in models that make these causes and mechanisms salient without the overlay of complexity that can so readily disguise them.

How does all this apply to our use of Molecular Workbench (MW) in the Fulcrum Institute?

Molecular Workbench is a tool for making models of molecules that interact with one another. It incorporates a deep scientific understanding of how different molecules exert forces on one another. It is best suited for showing how molecules behave when matter is in its gaseous phase. One can learn a great deal from experimenting with MW. We can vary numbers of molecules, masses of molecules, sizes of containers, and temperature and observe the consequences of these variations. From these observations we can make inferences about how real gases behave. Very often the predictions we make on the basis of our observing the model turn out to be correct. This is the case despite the fact that there are many things that are just plain wrong about MW as a model of interacting molecules. For instance:

  • The containers in MW are 2 dimensional and real molecules move in 3 dimensional space
  • Given the size the of the container the size of the molecules is outrageously exaggerated
  • Molecules are not colored green, blue and purple (or any other color for that matter)
  • Molecules are not circular (or spherical, for the most part)
  • There are far too few molecules for the models in MW to exhibit many of the properties of matter that we are familiar with
  • Real molecules at almost all temperatures move very much faster than the molecules in MW
  • We can’t figure out an “honest” way of getting MW to display solids that maintain their shape
  • We can’t figure out an “honest” way to get MW to display the difference between solids and liquids

Nonetheless, scientists find Molecular Workbench and tools like it to be of great use as they attempt to understand the nature of thermal phenomena on a microscopic scale. It is not a tool for giving users a surrogate experience of what they might see, hear, taste or feel if they were as tall as an oxygen atom.

But because it allows us to explore how a “large” number of small particles affect one another as they move about and collide with one another – and because that exploration allows us to make inferences about how real world systems behave – and because those predictions are often correct – we happily use such models with the understanding that the detail of much of what is in them is inaccurate.