Content-type: text/html ~ Stephen's Web ~ The Science of Education

Stephen Downes

Knowledge, Learning, Community

Jul 12, 2002

This Magazine Column published as The Science of Education in Learning Place July, online Jul 01, 2002. Education Queensland [Link] [Info] [List all Publications]

I think that before anyone can issue an edict as to whether education - or any other discipline - is or is not a science, they have an obligation to read a basic text in scientific method, such as Ronald Giere's Understanding Scientific Reasoning. What you will find is that even in the physical sciences, where certainty is assumed to reign supreme, there is a great deal of uncertainty and even a certain amount of what critics may describe as fudging or massaging the data.

In education, as in any discipline, the problems being addressed occur at different levels of complexity. In physics, for example, the principles governing the displacement of water are very simple. At a certain level. They explain why boats float. But there is an order of complexity. The principle of displacement, for example, does not apply for air in the same way that it does for solid bodies, because the volume of the mass submerged varies according to the pressure of the water. The same is true, to a lesser extent, for human bodies. This explains many failures in the application of the principle of displacement: for example, why people drown when the principle says they should float, why submarines may experience catastrophic failure, why boats sometimes sink.

In education, the same holds true. At a certain level, our understanding of education nears certainty. Almost every American child, for example, can count to ten, name the letters of the alphabet, recite the Pledge of Allegiance or sing a McDonald's commercial. They were not born with this capacity; they were taught, and our ability to teach them these simple things is near perfect. But in education, as in fluid mechanics, there is a deeper order of complexity, and just as we sometimes experience failures in the physical sciences, we also experience failures in education.

So Steve Eskow writes,

If we did have a science of education, we would be able to teach everyone to read, for example, and we cannot. There are thousands of studies of reading, yet the battle between phonics and whole language continues, and will continue. And I am hard pressed to think of a single educational issue that has been resolved by the "scientific method." Can you, or anyone, think of one?

I can state with certainty one theory about teaching people to read: that we will be more successful if we teach them something than if we teach them nothing. Children left on their own with no teaching almost invariably do not learn to read, while children exposed to teaching almost invariably learn to read. Now one may quibble about literacy rates, where we question how much mastery a child has attained. But that is a quibble. The evidence is overwhelmingly in favour of teaching. One may also quibble that this is not a scientific experiement. But how could you argue that it is not, when the generalization proposed is based on observations and measurements? Sure, it's a pretty simple hypothesis, but if you accept that we can know this, then you must accept that we can know something about education.

And of course we know a great deal more. Probably the most advanced educators in contemporary society are advertisers. The science of advertising - and it is a science - is replete with essential principles, one of the most famous being that a product name should be repeated five times in the space of a 30 second commercial. Advertisers are also masters of associative learning and transferrance, whereby the desirable properties of one thing (being attractive to the opposite sex, for example) are transferred to another thing (owning a Ford convertable). Advertisers are successful, and this success is measured to a startling degree of accuracy, which is why advertising is today a billion dollar industry.

Perhaps a more interesting way to but Steve Eskow's point is to ask why academic researchers have been unable to pin down what it is that parents and advertisers have been able to figure out on their own. Why is it that the academic debate continues when, based on observation and experience, it ought to be relatively easy to adduce any number of principles (much in the manner Kepler inferred that the orbits of the planets must be ellipses).

In the course of preparing my daily newsletter I have read thousands of articles and research papers (I have cited 4000 over the last few years, and I reject a half dozen for every one I read). Though I am not prepared to subject this analysis to a rigorous scientific analysis (who has the time? Though it would make a good research project) I can identify at least a few possible causes for this failure:

- Theorists in the field of education do not have a good grounding in scientific reasoning and logic. If important writers such as Steve Eskow can make basic errors concerning the nature of scientific reasoning, the average researcher in the field can hardly hope to do better. It would be useful were educators more deeply grounded in logic and scientific reasoning. It would also be useful were these basic principles valued more by the practitioners - it is not more than once that my arguments have been rejected simply on the ground that I have presented a logical analysis (ring any bells, Steve?).

I could be more detailed on this point. In particular, it seems to me that the vast majority of writers in the field (much less commentators in the popular press) have little understanding of statistical significance, random sampling, and other elements of inductive methodology. It continues to astonish me that respected journals in the field publish papers in which hypotheses are advanced based on researches involving a couple of dozen American university students. Moreover, there is almost no attempt made to replicate previous results in a new environment: the design of almost every study is restarted from scratch.

Those standardized approaches that are attempted - and here I include standardized testing - are derided as artificial and contrived. And of course they are: the point of any scientific experiment is to attempt to abstract some key variables from a rich background. The question remains open as to whether the variables are significant. The question remains open as to whether there are unobserved intervening variables. But these are questions relevant only in the context of a standardized approach, and not as an objection to one.

- Many theorists in the field of education do not have the advancement of knowledge as their objective. This is a sad truth, but easily observable, and easily explained by the confluence of a number of important factors.

The most readily available observation is the example Steve Eskow cited, the debate between advocates of phonics and whole language. From what I have observed, there is as much a political agenda behind the advocacy of one or the other theory (and, if 'Hooked on Phonics' is any indication, there is a commercial agenda as well). It should be immediately and intuitively obvious to any observer that a blended approach, one emphasizing both word sounds, sentence shapes - and for that matter including a certain amount of structural analysis (such as logic, grammar and linguistics) would be the preferred approach. The current debate makes as much sense as a group of linguists arguing whether the domain should be defined solely by the shape of the script or the sound of the syllables.

The debate is absurd, and the existence of such an absurdity is not in any way explained by the research data. The only conclusion that can be drawn is that there are other intervening variables driving the debate toward its current state of polarity. Political debates and commercial interests tend in general to promote a polarization. Both may be found in the field of education. Both, therefore, offer reasonable explanations for the current absurdity.

Other interests also weigh against the advancement of knowledge as the primary objective of researchers. In the current academic environment researchers are rewarded for the publication of original research papers rather than for their contribution toward knowledge. This creates a tendency on the part of researchers to prefer short and small studies, exactly the opposite of what is needed in order to advance knowledge. Were contributions to a common data bank valued as highly (or more highly, since it's actually harder to do) as journal publications then the state of eductaion would improve.

I have identified two major factors that detract from the science of education as it is currently practiced. I could expand on these at great length. I could include many others: the lack of adequate research facilities, the limits imposed by ethical considerations, the lack of a common taxonomy. The conclusion is inevitable: the reason why the discipline of education does not reseble a science is that it is, for the most part, not practiced as a science. Instead, the discipline serves as a proxy for endeavours in many other fields.

Now the factors I have described above are what might be called overwhelming social phenomena. That is to say, they are so engrained in the discipline of education that it seems almost impossible that they would be exised in the name of pure enquiry. It may be that Steve Eskow's assertion, that we are unlikely to ever have a science of education - may be true after all.

But I don't think to. Consider what Eskow says,

We can make no such assumptions about teaching and learning, which is why we have no science of education, and are unlikely to have one. Those living on the mountaintop want to educate for different lives embodying different values than those in the valley, and those in the valley are themselves subdivided into subcultures, which is one reason they resist a national curriculum.

Now on the face of it, this statement is absurd. Of course we can make some assumptions about teaching and learning. Here are some to get the discussion started:

- All the subjects are human. That is, the domain of education is (my ant video notwithstanding) not concerned with non-human life forms or inanimate objects. (It follows that there is a separate, but possibly related, discipline that may be called animal learning, or in the more common terminology, dog training).

- Some of the subjects exhibit known abnormalities. They may have malfunctioning brains as caused by, say, a tumour, a chemical imbalance, or some other disease or condition. For the purpose of a normalized theory of education studies involving these subjects may be excluded, though that said, there may be useful insights to be obtained by comparing how a person with a malfunctioning brain learns (or fails to learn) as compared to humans with fully functioning brains.

- All subjects have prior experiences. That is to say, they exhibit mental states which (may be held to be) caused by their sensory experiences.

- All subjects have a neural structure. There are certain commonalities in human brains. They are composed of billions of neurons, arranged in layers, connected by axons. They display pulses of electrochemical energy. They function differently given different nutritional and other input. They cease to function when sufficiently disrupted.

This is standard knowledge in the field. ASll of it has been shown over centuries of scientific experimentation and observation. Some of it is the subject of further refinement (for example, do we classify people with what is diagnosed as ADHD as abnormal?) and debate. And from this basic knowldge we have been able to derive some principles which - to me - seems to be pretty basic. Such as:

- Learners (at at least some level of learning) demonstrate learning by imitation.

- Repetition works. The repeated presentation of a stimulus that can be imitated will result in its being imitated.

- Association works. The repeated association of two stimuli, where one may be sensed and the other imitated, will result in the other being imitated when the one is sensed.

And so one. This is basic Pavlov, basic Skinner. We move into more detailed problems when we attempt to explain why these phenomena occur (the behaviourists' explanation - that it's nothing more than behaviour - is unsatisfying as it offers no explanation for these phenomena, merely generalization).

In my opinion, our recent investigations into the nature of interconnected sets of computer data points (networks) is instructive. There is evidence - advance most convincingly by Francisco Varela - that many human processes, including cognitive processes, may be explained as patterns of connectivity in electro-chemical networks. There is a body of computer simulations and mathematics, as proposed by such theorists as Boltzman, Rumelhart and McClelland, which describes in finer detail how this could work. There is the work of people like Amos Tversky that shows how such processes may be described as patterns of similarity.

There is much - so much! - to work with. So much, indeed, that I sometimes think that the reason there is no science of education is that it is in the interests of at least some theorists to ensure that there is not one.

But it seems to me that in the end the reasonings in politically and commercially neutral domains - such as neuroscience and matematics - will prevail. That when the structure of the brain and brain functions, for example, become widely known, theories that deny such knowledge, while they may exist, will be widely disregarded.

Stephen Downes Stephen Downes, Casselman, Canada

Copyright 2024
Last Updated: Jun 14, 2024 05:09 a.m.

Canadian Flag Creative Commons License.