7.4 Modelling in science
When scientists use models there is an unspoken assumption that the model can be applied within a limited range of contexts. Although enormously powerful, models are rarely directly applicable to everyday circumstances. With the Cumbrian sheep farmers, government scientists took the view that the levels of caesium (emanating from the Chernobyl explosion) in Lakeland grassland would soon diminish, such that levels of contaminating radioactivity in sheep would soon diminish. This was a prediction based on the assumption that the behaviour of caesium in the dominantly acid peat soil of Cumbria would be similar to its behaviour in the better-known circumstance of alkaline clay soils. But eventually (in fact, after two years of confusion) it became clear that whilst caesium compounds become immobilised in alkali clay soils, they do not become fixed in acid soils. Rather, the element becomes taken up by plant roots and as a consequence enters the food chain, passing through the sheep before being re-deposited on the grass and hence becoming available for recycling.
This meant that the unqualified reassurances initially given to farmers by government scientists of an early end to their problems proved groundless because of the frailties of predictive modelling. The result was that the scientists lost credibility in the eyes of the farmers; their ‘experiential and personal knowledge base’, as referred to in the Jenkins reading, ran counter to unqualified scientific pronouncements. Scientists had clearly placed excessive confidence in the predictive value of their models. If the farmers had had a greater awareness of the key role that assumptions play when theoretical models are applied in uncertain contexts, they might have been more inclined to challenge the assertion that scientists were making, perhaps by asking uncomfortable questions. The ability to ask such questions ‘requires an awareness of the nature of modelling in the sciences rather than an understanding of caesium chemistry’. Teaching the value and limitations of models in science inevitably represents a formidable challenge, given the subtleties of model use.
Think of more recent public controversies involving science that have attracted media attention. Do such controversies fall into different categories? For one or two of your examples, list a range of ‘ideas-about-science’ that would be required by a lay observer to get to grips with the episode.
In the UK, recent controversies have included the foot-and-mouth crises, the safety of the combined MMR (measles/mumps/rubella) vaccine, the health risk of mobile phones, the use of genetically-modified (GM) crops, climate change and issues such as cloning and ‘designer babies’. What has attracted attention in my local paper has been the potential health risk posed by building a waste incinerator in the Milton Keynes area. What strikes me is how variable such episodes are – and correspondingly difficult to classify. Some have a political flavour that involves decision-making at a level very distant from the population at large. Others (notably climate change) involve possible long-term effects of global significance, but where immediate effects are uncertain. Others involve personal decision-making (the use of a mobile phone or MMR), with a personal assessment of risk to oneself or others.
To take GM crops as an example, where trials of such crops are currently underway, ‘ideas-about-science’ would include a knowledge of experimental design, assessing the quality of data, assessing conflicting data, the validity of models that predict pollen and seed transfer, sources of uncertainty. I'm sure there are others, some related to the specifics of genetic manipulation.
Jim Ryder's comprehensive analysis of 31 cases of science ‘put to use’ has allowed him to identify different types of science knowledge needed by individuals in order to function effectively in particular settings. This view of ‘functional scientific literacy’ fleshes out some of the ‘ideas-about-science’ described in less concrete terms in Beyond 2000. Ryder's list is long and detailed and groups the main areas of science understanding into six areas, including ‘Interpreting data’, ‘Uncertainty in science’ and ‘Science communication in the public domain’. As an illustration, the bulleted list that follows gives the five distinct points related to modelling that Ryder identifies. Necessary though the generalised statements of the type in the list are, no doubt they disguise the huge variation that exists in the specifics of model use in science. The simplified soil model at the core of Chernobyl example would differ fundamentally from the type of modelling used in weather forecasting, for example, where contingency (i.e. chance effects) exerts a strong but unpredictable effect. And what about models of global warming, where both the models themselves and the variables that are fed into them are subject to massive uncertainty? Thus (to take just one example) the predictive power of models varies enormously – all the more difficult then to square such complex simplifications of reality with the other familiar classroom uses of the term – for example with ‘models’ of the human torso in the biology classroom, valuable precisely because of their closeness to anatomical reality.