Book review: Teaching Machines by Audrey Watters

Technology-supported personalised learning is an old concept, and in an eagerly-awaited book Audrey Watters traces the long impact of the behaviorist ideas that underpin it. David Kernohan takes an eye-opening read

David Kernohan is an Associate Editor of Wonkhe

Stop me if you think you’ve heard this one before.

Teaching is a staggeringly inefficient activity. A teacher moving a whole class through a syllabus can only go at the speed of the slowest learner, leaving the others wasting time waiting. We have the technology to have each student work through the material at their own pace in a personalised way – machines learn from the way students interact with them, and can offer extra practice to those that are struggling or skip repeats for those who have already succeeded.

And studies show that students prefer this way of learning, which is based on the latest research into learning. Sure, the equipment comes at a cost, but think how much students will benefit. And, let’s face it, you might not need to pay as many teachers…

Strictly commercial

What you’ve just read is an education technology sales pitch – wearily familiar now to anyone in or around education. Those with longer memories may remember similar stories from the MOOC wars of 2012, or the dot-com boom era education technology startups. It’s an incantation repeated because it is effective – it sells kit by playing into our own anxieties about the mysterious process we label “learning”.

Education technology journalist Audrey Watters, in her new book Teaching Machines (MIT Press, 3 August 2021) traces this set of ideas back nearly a hundred years: to a basis in the work of Sidney Pressay, and – most famously – B.F. Skinner. Both men spent chunks of the thirties working with American typewriter companies and IBM to bring teaching machines (or, variously, the “didak” or the “instructomat”) to market based on their then-fashionable behaviourist ideas.

In essence, Watters’ argument is that – although technologies and buzzwords have changed – these ideas still form the basis of much contemporary thinking about technology in education, and as much as we might like to imagine an alternative history couched in constructivism and social learning it is behaviorism that continues to dominate our thinking about learning (and, it has to be said, about public policy – a “nudge” is just behaviorism in a Silicon Valley hoodie, after all).

Heir conditioning

If you only know Audrey Watters’ journalistic writing, or her many self-published collections of essays (all things well worth seeking out for any serious student of the history of the future) you’ll be familiar with her passion for the odd stories that seem to coalesce around the men of edtech. And – oh, my word – B.F. Skinner.

It’s the labour-saving semi-automatic crib he invented for his daughter, complete with endless bedroll and space-age perspex shield – latter commercialised as (and I kid you not) the “Heir Conditioner”. It’s the inevitable DARPA-funded pigeon guided missiles. And the endless, countless, iterations of a teaching machine that amounted to little more than a crank to feed cards under the windows in a cut-out metal sheet.

Skinner was a perfectionist, as was Pressay. Neither invented a machine that dominated education in the way that either hoped. The craze for mechanical auto-tutors buzzed through the forties and fifties – not so much in schools but in homes, occupying a niche similar to and a mode identical to that of the travelling encyclopedia salesman. If Sputnik was peering into the american living room, it might see “space age” teaching as a part of the US response.

Read in a certain way, we see an early insight into “third stream” activity, the business of commercialising a discovery with pretentions of science into a product with pretentions of technology was as fraught and painful then as it is now. These days of course, we don’t need to assemble anything but code to disrupt learning – but it is still techne rather than research driven insight that drives breakthroughs, whatever the sales pitch says.

It turned out, as it often does, that the real money was in the software. Specifically the cards fed through the devices that contained the actual learning – there are several stories of manufacturers trying to get publishers interested in producing these before it clicked that selling several packs of cards to a customer would make more money for longer than selling one machine. And it quickly became apparent that you might not need the hardware at all – Doubleday’s “TutorTexts” were a “scrambled” textbook with a similar structure to Choose Your Own Adventure books.

Chomsky and his droogs

And the history is genuinely fascinating. We get a vignette of the role of teaching machines in the civil rights movement, and insights into the dubious validity of the Roanoke experiments (the first systematic trial of education technology was not the triumph it is often painted. We get as far as the very early days of computer-aided instruction – following the thread through to the present day would make for a very different book, and there is a lingering power in leaving the parallels an exercise for the reader.

All of this is helped by the fact Audrey Watters can seriously write – her passion for storytelling and the sheer joy of serious archival research shine through, and the drudgery implicit in Skinner’s endless letters to his Rheem corporate handlers is rendered with a dry and subtle wit. For me, the highlight is a deft handling of the wider implications of behaviorism through a close reading of the responses to Skinner’s later, wounded, anti-“free will” absolutism in 1971’s Beyond Freedom and Dignity – including Chomsky’s famous monstering, and the behaviorism skewered by Kubrick’s A Clockwork Orange.

If you’ve read this far, you probably know whether this is a book for you. At a practical, will-it-help-me-do-my-job, level you’d get inured to some of the wilder techno-deterministic memes, and wise to the behavioristic roots of the push to “personalisation” (and actually, what could be less personalised than rote learning with bits you can skip?). But for those with an interest in the space between theory, technology, learning, and commerce, and with a taste for the curious rhyme of history I couldn’t imagine a more rewarding read.

Teaching Machines by Audrey Watters is available now from MIT Press.

One response to “Book review: Teaching Machines by Audrey Watters

Leave a Reply