Coming to an understanding of scientific concepts is a challenge to all of us, especially when so many of the ideas in science involve things that we cannot see or touch. Helping students develop such an understanding is even trickier. If we look at how we ourselves develop deeper understandings, we can help our students better. We can observe the effects of phenomena, but that does not explain why something happens. We can see, for example, the light come on when we throw the switch in an electric circuit, but we cannot actually see what is going on inside the wires. Other concepts, such as energy, are even more difficult to conceptualise. How then do we come to make sense of scientific phenomena?

It’s a multifaceted process. We engage in investigations. We make observations, generate hypotheses, carry out experiments and record evidence. In other words, we engage in scientific enquiry, but is this enough? No, because if we are really to understand a concept we need to come up with explanations based on the evidence we have recorded, as well as taking into account the findings of others. How do we do this?

One of the most important ways is to try and create models - physical, mathematical, and mental - which fit the observations, measurements and actions we record. New evidence may modify or discard the model. Although the models can become ever more sophisticated, by their very nature they are not the thing itself - a fact which students especially do not realise.

Nonetheless, a model, whether it is the Watson-Crick model of DNA, Einstein’s equations or the model of a cell students have made themselves, helps to develop understanding of phenomena. Indeed, virtually everything we “know” in science is described in a form of model.

So how do we help our students grasp the idea of models becoming more sophisticated and show them that when we alter the model we are not telling them that what they had learned previously is wrong? Einstein, for instance, did not invalidate Newton.

One approach I use is to ask students if they had a toy train when they were very young. I ask them to describe it. Invariably, the description refers to something made of wood or plastic that was pulled along and did not run on tracks. Yet somehow or other it was clearly a train. Asking students to further describe these toy trains as they got older usually progresses to something which ran on tracks. Then there are the increasingly detailed, precisely constructed versions that make up the most expensive electric train layouts - often the one their father keeps in the attic!

Descriptions need not stop there - we can move on to miniature railways in parks, and so on. At each stage the model of the train gets nearer and nearer to the real thing. Eventually, in this example, we can get to a real train, but at no point is the fundamental idea of a train lost or abandoned. Rather, as each model is considered it adds to the understanding of what a train is and to how it works. Each of the different models is another step in the learning process.

Models are not simply mechanisms for describing or explaining things as they are. Models can also be predictive and allow the exploration of “what might happen if ...”. So, returning to our trains, there is huge scope for using models to test designs for new vehicles or the effects of different gradients, and the like.

I like to think that by sharing with students these ideas about how models become more complex it is possible to help them get to grips with developing their own understanding of scientific phenomena. Moreover, it helps them appreciate the increasing sophistication of the models scientists have developed to aid their thinking.

Derek Bell is chief executive of the Association for Science Education