In this series of blog posts I am exploring the relationship between scientific research and our approach to education. In the first two posts I looked at the current suitability of the field of education for having the methods and insights of science applied to it, and found it lacking in two general areas: Clarity of definitions, and clarity of purpose.
In the next two posts I am going to come to things from the position of science, to look at how it might be suited for studying an area such as education, and just what the limits might be – not simply currently, but perhaps intrinsically.
In this post I am going to give a bit of a primer to the scientific approach, and in particular the constraints it faces in terms of studying areas of human activity such as education. In the next post I will be making more general claims regarding the actual limits of what it will ever be able to achieve in for us and our educational endeavours.
So firstly… What is science? See if you agree with my bluffer’s guide…
At its core, science is a way of finding out more about things which increases detail and levels of certainty.
Its methods are varied, but essentially require careful observation, as different ingredients in a situation are strategically isolated and manipulated.
In the process it identifies patterns, from which it tries to infer the most probable causal relationships.
In exploring patterns, scientists create theories – plausible explanations of the patterns drawing on existing evidence, and which ideally make predictions which can allow the theory to be tested further.
From a purely logical sense, science can never give us absolute certainty regarding lines of causation, nor exhaust the possible limits of a situation. Consequently, the strength of a theory is assessed according to how well it fits within the existing body of theories, how much explanatory power it possesses, and how parsimonious it is in terms of the number of assumptions it has to make.
A good theory will allow some possibilities to be ruled out (falsified), and in the process of accumulating and replicating results increase our confidence in the probability of a theory being true to the point that – for everyday purposes – it is a useful shorthand to treat it as a true representation of a situation, or a scientific ‘law’.
If the subject under investigation is too difficult to observe directly, scientists employ models – metaphorical situations supported by theory, which function as similarly to the study area as can be conceptualised.
Sometimes different factors being studied can be shown to vary together (correlation), but whether one causes the other, or both are caused by something else, remains hard to pin-point.
What can science give us?
In discovering patterns which point to some genuine cause in nature, science can – at the very least, satisfy a level of human curiosity – expanding the cognitive satisfaction we experience at finding out something new. In the process, it might educe from us a new way of acting towards people and situations in the world.
It might also allow us to make predictions about what is likely to happen in the future, and – in its highest practical form, it might allow us to take actions to control aspects of the world around us – possibly through the creation of new technologies.
‘Hard’ vs ‘Soft’ Science
Scientific fields and endeavours are sometimes characterised as being ‘harder’ or ‘softer’. Generally speaking, a science is ‘harder’ the more that it can precisely isolate and manipulate individual phenomena, with a high degree of predictability and replicability, often in tight consonance with mathematical logic.
In distinguishing sciences this way, ‘natural sciences’ – those that study natural phenomena in and of themselves – tend to be viewed as hard. Whereas ‘social sciences’ – those which study the behavioural phenomena arising through the interactions of processes within the mental lives of people, and between multitudes of people, – tend to be viewed as soft.
Increasingly, the simple catagorisation of a discipline such as Psychology as ‘soft’ is problematic, because the field now encompasses and draws upon such a wide variety of approaches and areas of focus. Neuropsychologists for example work at the interface between brain biology and observable behaviour, and much cognitive psychology functions within highly rarefied laboratory settings, increasingly informed by what does and doesn’t work in the practical applications of artificial intelligence. Indeed, many people working in these areas would these days prefer to be identified under the new interdisciplinary area of Cognitive Science.
The ‘lab’ approach to trying to gain hard evidence
Scientific research into education is similarly going to be a multi-faceted beast, focused on different areas, at different levels and using different techniques. As with psychology, whilst we may gain a lot of information from existing areas of brain science, and we might be able to ‘grow’ some theories through isolated and artificial laboratory experiments, exactly how things work together in real-world educational settings may remain elusive.
At the ‘harder’ end of the research scale, there is work which extrapolates from more general research into brain functioning and cognitive architecture, whereby scientists deliberately try to isolate individual processes which are going-on in the brain and see what this might be able to teach us as educators about dealing directly with the brain through these mechanisms.
This is of course particularly of interest when trying to understand and provide ways to address Special Educational Needs – or atypical functioning of one kind or another, and there is an increasing wealth of research data now in terms of how the brain appears to go about its business, due to highly controlled investigations into our functional constraints in particular situations.
Some of this research has proven exceptionally helpful in terms of identifying what appear to be universal characteristics of human cognitive functioning – the limits of our attentional system and ‘working memory’ for example.
However, we have evolved such that our perceptual and cognitive systems best function in a flexibly dynamic interaction in a rich 3D natural environment. What many lab experiments into humans achieve are demonstrations into what humans can do, but not necessarily what we do do in everyday situations. Similarly, they demonstrate the flaws and limitations of how our brain functions when situations are highly constrained, but these may not prove to be constant limitations in the real world. It could be argued that the human being is the most incredible ‘work-around system’ that has ever existed on this planet.
An obvious example of this would be the situation with the myriad optical illusions which entertain and perplex us in the contemporary world. It is far more rare to see (or at least fall prey to) optical illusions in the natural world around us than it is to see them in 2D images or computer animations. This is because illusions generally arise due to a poverty of information in our perceptual field. In general, we function in a world with binocular vision which can draw on motion information whenever it needs it, and a wealth of environmental light affordances which together enable the brain to piece-together a functionally accurate view of the situation in front of it.
What we find in the lab may not really allow us to make concrete practical conclusions then. A significant problem which cognitive scientists have faced over time is that – in the process of becoming ‘harder’ as a science – in the move towards increasingly tight isolation of variables and processes in a laboratory – they run the risk of loosing what is termed ‘ecological validity’ – the results may not reflect what does happen when these processes are situated in a normal human context.
The ‘situated’ approach to gaining ecological validity
And hence, much research goes into looking at real-world settings, but these situations start to greatly reduce the ‘hardness’ of the science which can be done. It is much harder to properly isolate and exactly replicate individual processes within a classroom.
Social Scientists have gone a long way over the decades in trying to develop techniques which allow them to derive the most objective benefit from studying real-world processes, but it becomes very difficult to go much further than to derive general patterns of correlations.
For example, if we were to use the gold standard medical trial model of research into whether a ‘treatment’ has significant benefits (the double-blind randomised controlled trial) in deciding on whether an educational technique has real benefit, you would need to do the following:
- Ensure that you had a very large sample size
- Ensure that fully informed consent had been given for participation in the trial by either the participants or their legal guardians.
- Ensure that your two (or more) groups of participants were randomly made-up and (hopefully) equivalent in all key areas which could impact on the outcomes of the trial.
- Consistently and equivalently apply the circumstances of the different experimental conditions over as long a period of time as possible.
- Ensure that none of the subjects undergoing the trial knew the significance of which grouping they were in (i.e. whether they were meant to be a control group, a placebo group or the group undergoing the treatment being studied.)
- Ensure that none of the administrators of the trial knew the significance of who was in each group…. In other words – they don’t know what they are doing to each participant (this is the double blind part)…
If trials for drug treatments haven’t gone through such a procedure, you can bet that the claims coming from their findings will be picked-apart from one angle or another.
It is of course very hard to create such circumstances in educational settings (particularly number 6) in a way which doesn’t have frustrating degrees of fudging at one level or another. Even if you could manage to properly satisfy every feasible aspect, the effort and time involved in doing such research makes it a highly slow, laborious and inconvenient process for many schools to be involved in.
Consequently, for much educational research – if being reported honestly – any findings would be introduced by many concessions to the circumstances in which the results were obtained, and therefore in which we could gain most confidence in them being relevant. Effectively they would say “For this particular group, of this kind of children, done in this way, in this setting, for this duration, at this time of the school year, and being taught by teachers who have this kind of knowledge and predisposition…etc. etc. it was found that there was this particular statistical chance that this approach….etc.” And then you have the problem of replicating the same circumstances to find out how robust those particular findings were…
These factors are of course messy, and should constantly be a source of concern for people involved in professional educational research. Consequently, there has been a big move in social science to go for ‘meta-analyses’ – attempts to say “well if you group together all the bits of research into the same phenomenon [assuming that all of those bits of research were indeed investigating the same thing – conceived and operationalised in the same way – see Part 1] then this will help to balance out all of the flaws contained in the individual bits of research, and any remaining statistically significant effect will show that something is definitely going on there. Well, maybe….
If laboratory research can only give us pieces of the ‘capabilities jigsaw’, without helping us see just how these things come-into play (or not) in real world social situations, large-scale field research can only give us forever contentious general patterns – constantly in need of additional clarification and contextualisation to really tell us something we can use with any great confidence. And so we are on a project to constantly improve, refine and extend both kinds of research such that they might one day meet in the middle and then… and then what…?
Well, I have fundamental reservations as to whether both wings of the scientific enterprise as outlined will ever get us to the place – in the social science orientated side to things – which our science-fiction orientated minds assume it will.
Firstly, it is not because I think that human experience and interactions are somehow the wrong kind of ‘stuff’ for scientific study, as in they can’t in theory be reduced to mechanisms within the natural world. In this sense I think that human functioning can be reduced (if useful to do so) to natural science. The key differences between social and sub-atomic processes are degrees of complexity, and levels of organisation.
Furthermore, I don’t think that the problem is simply that we have more processes to distinguish from each other – that the increase in complexity just makes them harder to spot because there are more things going on – like trying to find a product in a big shop that you usually spot easily in a small shop.
Rather, my concern is the explosion of complications you get through the interactions between different levels of elements, the reflexivity of the human subject matter, and the constantly shifting uniqueness of every situation which is looked at.
Unpacking these concerns will be the focus of Part 4.