Journal article in School Science Review, Volume 98, Issue 365.

Access (free): Hardman (2017) via

Scope of Paper

The author argues that models are integral to science education, but that they can also be a source of confusion for students.

In constructing the argument, Hardman touches on how scientists use models, how science relates models to the 'truth' of the world, and how models help students to learn science in the classroom.

Key Findings

Students, like scientists, use myriad models to consider and explain the world. We should also teach students to develop their own models –  in this way they can learn both the nature of science and concepts relating to the phenomena that their models describe.

As teachers, we can frame our assessment to be model-based, which gives students the opportunities to express their own models.

More Detail

Models and forms of truth

After 10 years of working with new science teachers, I am still struck by how many of them portray their scientific careers as a series of stages in which they found out that everything they had been told previously was wrong.

Science education builds a student's knowledge of the world through a series of increasingly sophisticated models. This may have the unintended side effect of putting students off science, as they perceive teachers to be keeping them from the 'real' science.

This view of science education suggests that teachers slowly tell students the 'truth' about science – a commonly held view in our culture. However, this view is a caricature of the reality of science education, as well as of science communication in general.

In practice, scientists use models on a scale of realism: at one end, a strong view of realism suggests that models are a very close approximation of the world 'as it is'. At the other end, a weaker view of realism suggests that models are developed for a predetermined purpose, and represent a useful approximation of reality.

Does science lead models, or do models lead the science? We accept scientific explanations based on how well they portray the phenomenon in question, but this doesn't mean that they are 'right' or 'true', or even that they are the final version of an explanation. The nature of empirical science leaves room for the continual evolution of models.

The development of models via the observation-explanation route is not the only option. There are examples through the history of science where models have been generative, i.e., they have predicted or explained things that we are yet to see evidence for. This opposite route, the explanation–observation pathway, has been the basis of large scale funding approval, which in turn has led to major scientific discoveries, for example, in detecting the Higgs boson.

Models can therefore be used to both explain and predict phenomena in science. When models are predictive, there is a danger that a seemingly small detail can have a large effect on a system (the butterfly effect), and result in the model being rather different to the phenomenon. But this doesn't mean that more accurate, more detailed models are closer to reality. Ultimately, all models are reductions, so while it's important for the model to resemble the phenomenon, the level of detail does not always determine how useful a model can be.

Refining models through their intended use

We might say that a photograph gives us a more accurate image of a landscape than a drawing, but a schematic map is likely to be of more use for orienteering.

Scientists frequently use models to focus on one aspect of a phenomenon. As such, we judge the suitability and value of models on how well the model explains a particular aspect of the world. We shouldn't think of models as ways of understanding the universe as it is, rather, as a way of understanding a particular feature of a phenomenon.

We evaluate models based on many criteria: how well they fit existing understanding; how useful they are for their purpose; how far they can be justified to others; how elegant they are; who has proposed the model. The author argues that, given the wide range of criteria for evaluating models, science employs a weak form of realism in developing models. They aren't often developed to describe the world as it is, rather, to help explain the focus of a scientist's attention.

Models as explanations in the classroom

It may not be wise to fully expose young scientists to what Feyerabend (1975) calls the ‘anarchy’ of scientific method: scientists use whatever methods they need to advance science.

In the classroom, we can't let students believe that science is a series of increasingly accurate models, ultimately approaching the truth.

The teacher must decide how to teach the nature of science to their students, which includes understanding the role and nature of models.

The author suggests that models are framed as explanations; simplified representations of phenomena, used by scientists to describe (and predict, and reason) the world as it appears to us. While this might be an idealised view of models, it provides teachers a useful discussion point on how science is, versus how it should be.

Within this framing of models as explanations, Gilbert, Boulter and Rutherford (2000) further delineate different forms of explanation:

  1. Descriptive explanation. Where measurements of a phenomenon are presented; for example, in considering variation in height between members of a class.
  2. Causal explanation. A description of why a phenomenon behaves as it does; for example, why there is variation in the heights of class members.
  3. Predictive explanation. Considering how the phenomenon will behave under specified conditions.
  4. Intentional explanation. For example, in identifying the mode of operation of the AIDS virus with the intention of enabling prevention and cure.
  5. Interpretive explanation. Here we consider what a phenomenon is composed of; much of chemistry consists of these abstractions.

We also need to teach our students how models evolve over time. Models are proposed by individuals or small groups, and may be agreed upon by a wider community. At this point they become consensus. Once tested experimentally and peer reviewed, the model is accepted as a scientific model. Later, a model might be contested, disproven, and ultimately replaced, at which point it becomes a historic model. (The atomic structure model is an example of this progression — Ed.)

It is also useful to distinguish between a model and a theory: that a model describes a particular phenomenon, and multiple models combine to support and develop a theory.

In the classroom we use different models to describe different phenomena that may relate to the same object (e.g. particles as they exist in different states versus how they react with one another). In this way, models are explanations that serve specific purposes, in the context of how they are proposed, developed, and become consensus.

While this oversimplifies the messy and complex nature of the development of models by scientists, it provides students with a coherent and consistent view of models. This in turn allows students to understand how and why we use models in teaching and learning.

Modelling-based teaching

Teachers engage with a huge variety of models every day and are concerned not just with how models relate to the phenomena being discussed but also how students will learn from those models.

While models are useful ways to link science to the classroom, teachers can have a more complex job than scientists in employing them in practice. Something simple, like introducing the heart, can involve many different types of models: we might show a plastic heart (concrete representation), draw a diagram (visual), show a heart monitor display (mathematical), clench our fist and place it near our chest (gestural), or describe the heart as two pumps (verbal).

Students understanding how these models a) are developed, and b) relate to the phenomenon, is just as important as them learning from the models, but is often overlooked in the classroom. We should strive to allow students to both use and develop models in the classroom; in this way they will learn first-hand the process of evolving a model. This is an important part of understanding the nature of science.

See Hardman's reference to Gilbert and Justi (2016), page 95, for a developed example of 'modelling-based teaching'.

Teaching science in this way greatly enhances both student understanding of scientific content and their skills, and also provides a vehicle for developing an understanding of nature of science.

Learning through models – beyond concepts

Modelling-based teaching allows students to learn from the process of engaging with models in a way that approximates to the work of scientists, such that students learn both content and the processes of modelling.

For students to develop models they must learn and use multiple concepts. This relates to the longstanding constructivist theory that students learn science by developing concepts in their long-term memory.

The focus on concepts implies that students acquire understanding of phenomena via the activities they engage with in the classroom. However, there are many examples of students erroneously learning a model as a concept, and subsequently being unable to apply that knowledge to other questions about the phenomenon. They have learnt the model itself, not what the model intends to portray.

To avoid this, helping students understand what a model is and what a model is used for is paramount.

Teachers can subtly change the framing of assessment to identify what students have learned. Instead of asking a student a recall question (how many ventricles in the heart?) we might ask them to explain something (how does the heart pump blood around the body). In answering the first, the student relies on recall alone. To answer the second, the student must use their own 'mental model' to express other models while explaining the phenomenon.

This gives the teacher a more accurate view of what has been learned, and how it has been learned. The author argues that students learn through the models they see, express and evaluate. This relies on them understanding what models are used for, but more importantly, being given opportunities to use them.

In summary: Models, truth and teaching science

A great deal of what both professional scientists and students do is to engage with models, and this is what links learning and doing science.

Models cannot be perfect representations of a phenomenon. Scientists use models for a wide variety of uses, and develop them under a wide variety of motivations. Yet students sometimes think of models as the 'truth', when instead they should be taught to understand them as explanations of the world around us.

We should also teach students to develop their own models – in this way they can learn both the nature of science and the concepts relating to the phenomena that their models describe. As teachers we can frame our assessment to be model-based, which gives students the opportunities to express them.

The author concludes by drawing these two arguments together:

Students learn through the models that they express and with which they engage, and this is not just about the concepts listed in curricula; it is also about how the process of doing science is modelled. Secondly, if we are to present science as a coherent and authentic subject, we must recognise that science is the process of explaining features of the world through the development and evaluation of models. This process is messy, as is the way that students learn in classrooms.