In the September 2020 episode of the ‘Meet The Education Researcher’ podcast (episode#65), Neil Selwyn spoke with Audrey Watters about her new book – ‘Teaching Machines’ – a history of the automation of education coming out with MIT Press in August 2021. This excerpt has been lightly edited for clarity.
Q: What’s the 30-second elevator pitch for the book?
A: It’s a book about the history of personalised learning, and personalised learning via technology. And I think it’s important for me to talk about the ways in which this is not something that Sal Khan invented with Khan Academy. This isn’t something that we can just now unlock because of the power of computers. Instead, this has been a century-long project by educational psychologists and by business. And really, I think that’s how we’ve ended up with our education technology today – this long history of teaching machines, of personalising education through automation. It sounds incongruous, but that’s how it works.
Q: So it’s the classic question of ‘How did things come to be this way?’
Q: Let’s go through a couple of the main characters in the book. First off – Sidney Pressey. Who is Sidney Pressey, and why are we still talking about him in 2020?
A: Bless his heart … Sidney Pressey was an Ohio State University education psychologist. Like many of the men – overwhelmingly men – working at the time in psychology, which was itself a fairly new field, he had worked [with standardised testing]. His mentor in college at Harvard was Robert Yerkes, who of course helped develop the Army Alpha Intelligence Test that was administered to US soldiers in World War I. That was really the first massive standardised testing project.
So I think it’s important to think about how standardised testing is constitutive of this idea of personalising and automating education. Sidney Pressey made a lot of standardised tests. He was making a lot of money selling schools standardised testing in the 1920s. It was already a huge business. And then he got this great idea – what if we made a machine that could grade the tests automatically? Or, even better, teach students automatically?
And so he first demoed his teaching machine in 1924 at the American Psychological Association annual meeting, and tried very hard to bring it to market and make it a business. But this little event called The Great Depression happened. So Sidney Pressey was not terribly lucky, although he stayed involved in teaching machines. Of course, more famously, a few decades later BF Skinner invented a teaching machine. And BF Skinner much more well-known, and teaching machines are much more associated with Skinner.
Q: Skinner is going to be much more familiar to most of our listeners, but what was he doing specifically in the mid 20th century that can be traced through to the EdTech of today?
A: Skinner is such a fascinating person … a fascinating/terrifying person. I’ve spent the last couple of years spending way too much time thinking about behaviourism and thinking about Skinner’s influence. But I think that he is truly the most influential person in the development of education technology. I think other people would like it to be someone like Seymour Papert – people would like to say that education technology’s origins are constructionist, but I think that really they are behaviourist.
And of course Sidney Pressey was a behaviourist too. Most psychologists at the time were behaviourists. That is, to study learning, you studied behaviour. And Skinner, famously, worked with pigeons in order to do something called operant conditioning, which was training pigeons by giving them rewards in order to elicit certain behaviours from them. So the pigeons learned through rewards, and Skinner had this brilliant idea of ‘what if we build a machine that offers rewards to students so that we can shape their behaviour just like we shape the behaviour of pigeons?’
So by the 1950s, Skinner tried to commercialise a teaching machine product – he tried to sell it to IBM, and then tried to get different manufacturers to make it. And for some time, in the ‘60s, this was an incredibly popular idea. Door-to-door encyclopaedia salesmen sold teaching machines. All this was really founded on the Skinner notion of operant conditioning – rewards for behaviour, and shaping behaviour through machines.
And I think we see that now in tech. We don’t just see that in EdTech, but we see that in all tech as well? Shoshana Zuboff wrote the book Surveillance Capitalism a year or two ago, and really her argument was that behavioural engineering à la BF Skinner is fundamental to how technologies like Facebook and Google … not just the education technologies that we use.
Q: Yes – and it is now couched in terms of behavioural economics and nudging people. As you say, it’s fundamentally about getting individuals to make the ‘right’ decisions and do the ‘right’ things.
A: Yes, the “right decisions” with air quotes … absolutely.
Q: One of the things you mention about Skinner was him cosying up with IBM. Another theme in the book is when corporations get involved in educational technology. So how did that work out in the mid 20th century? I don’t remember IBM’s teaching machines revolutionising schools.
A: One of the things working on this book that was so interesting to me, is we hear these stories that the reason that teaching machines weren’t successful is that teachers were Luddites. The teachers didn’t want them, the schools refused to change. We hear that today, we hear it all the time that education technology fails because of teachers.
But it’s really much more of an interesting story. A lot of businesses were instrumental in stymying and undermining the progress of this technology, because I think they had a different set of priorities than did two professors. This was a professor from Harvard and a professor from Ohio State University who thought that they had created a science, and that science would have a technology, and that science-informed technology would improve education. In contrast, the corporations were like, “Can I sell a million of these? Yes or no?”
And I think that Pressey and Skinner were both really adamant that their machines were scientific. They were incredibly controlling and were not willing to hand it off to a manufacturer. And so I think they just were the biggest pains-in-the-ass for these companies to work with, because they were just adamant that the machines looked a certain way and functioned a certain way … because otherwise they weren’t scientifically valid.
Q: So, in terms of big tech involvement in public education in 2020, where have Google and Microsoft gone right? What’s changed now to make the corporation the dominant partner in all of this?
A: I think that there are a number of reasons. I think that some of it just has to do with the idea that’s been going on for a long time that education is a huge market, that there’s a huge market opportunity. And there were stories about this in the 1960s in which people crowed about the amazing opportunity that this was to make it just a tonne of money off of the education market. But I think that what we’ve seen in the meantime is a shift in power and of money from public institutions into corporations.
So I do think that the rise of a lot of these companies in education is deeply intertwined with neoliberalism. Rather than schools, university or school districts building things internally themselves, the new motivation is always profit. The pressure for schools to turn outside to for-profit companies and vendors to build a supply of things rather than building capacity internally is just incredibly powerful.
And I don’t think that you can separate the rise of the computer in the classroom from these other trends as well. If you ask me about the main themes in my work, it is that’s what’s so important to always remember, is, it’s not just about the tech. It’s the historical context, it’s the political context in which these things happen.
Q: So it is interesting that you mentioned The Great Depression. But in this period, I’m also thinking about the Cold War, the Space Race, Vietnam, civil rights. How is the rise and fall of teaching machines intertwined with these broader societal, political, cultural shifts?
A: That was one of the reasons why I found myself spending so much time with Skinner in particular. Skinner, I would say, was probably the most well-known public intellectual of his time – certainly the best-known psychologist in the US, and globally had a huge reputation. He wrote a bunch of academic articles, sure, but he was also writing for the popular press. He was a guest on some of the nightly chat shows, he was on television quite a bit. People knew who Skinner was. Also, for a while following World War II, I think that Americans were really committed to gadgetising their world.
So Skinner’s ideas about using technology for child-rearing and using technology in education were somewhat controversial, but they were also really part of the whole gestalt of what Americans thought the future was going to be like. The future was inevitably going to be more technological, more machines doing things in the kitchen, in the classroom, as Skinner would say: “ If you can automate the kitchen, you can automate the classroom”. So again, there have been very gendered expectations for a long time in EdTech.
And Skinner wrote a novel. Like many academics who try their hand at writing novels, it’s not great. It’s called Walden Two – about a whole society that was designed with his behavioural engineering ideas. And was published very early in his career, before he was famous, but he was quite popular in the 1960s. And then he wrote a book that came out in 1971, Beyond Freedom and Dignity, in which he talked about the ways in which freedom and dignity, free will were not real. That was quite a shocking argument, and I think a lot of people then started to maybe ask more questions about Skinner.
But at the same time, you ask about the civil rights. There are other things going on. In the 1963, the Mississippi Freedom Summer, a lot of SNCC activists wanted to organise black folks in Mississippi with literacy campaigns so that they could get the right to vote. Because of course Mississippi had egregiously racist policies blocking black people from voting, using literacy tests at the polls.
So they did a campaign for literacy and one of the leaders at SNCC – Robert Moses – thought maybe we’ll use teaching machines. He had been a student of some of the psychologists at Hamilton College, where Skinner had actually graduated himself. And Moses thought that it might be an efficient way to help adults in Mississippi get up to speed with literacy.
But the ‘programmed instruction’ way in which the teaching machines worked ran headfirst into beliefs about freedom, beliefs about agency and pedagogy, and the other kinds of things that some of these activists were doing – including the freedom schools. And so you could see this conflict in the 1960s. This was before Paulo Freire’s work was translated into English, but people started to see that education absolutely needed to be a practice of freedom versus someone like Skinner, who was like education should be a practice of engineering. Those two ideologies really started to clash in the 1960s.
So you could see the student movements really start to resist not just the teaching machines and the idea of programmed instruction, but all of the other machines that they felt were dominating their lives as well. Take Mario Savio’s famous speech on the steps of Sproul Hall about throwing your bodies against the gears and the machines – resisting this idea that life was becoming more and more mechanised, more and more standardised, more and more automated. And so I think a lot of activists and students in the ‘60s really rejected the vision that Skinner had for education.
Q: So history tells us that EdTech has politics and that EdTech is ideological. Clearly we need a historical approach to EdTech, but how do you get education and technology audiences to engage with this history? Education and technology seems so relentlessly fixated on either the present or the near future. No one really wants to look back and talk about this stuff.
A: They don’t. But, I think that what’s frustrating is that at the same time they also have these pat stories about the past that always get invoked – that education hasn’t changed in hundreds of years, that we have an industrial model of education, that we have a Prussian model of education. So they do cherry-pick from history often incorrectly in order to make arguments that now is the moment for change.
So, that’s why I really wanted to write a book that looked at the people who were making those very same arguments a hundred years ago that education hasn’t changed and this idea that education is stagnant, static, unchanging. You hear tech folks say this now, and you hear the Secretary of Education in the US, she says this too. It’s a pretty common story.
In my book I wanted to be able to fill in the blanks with more context – and I also wanted to tell a story that wasn’t about computers. My book actually doesn’t talk about computers. I mean, maybe I mention them in the conclusion, but it’s not a book about computing. I wanted to make it very clear that current ideas about ‘personalised learning’ and ‘individualised instruction’ through technology are really old. And that’s worth thinking about too.
Q: So despite the idea that everything’s on the internet these days and you just have to go to a computer to find things out, you actually did quite a lot of proper old-school archival research for this book? What research did you do, and was it worth the effort?
A: I did. I went to Sidney Pressey’s papers, which were at Ohio State; BF Skinner’s papers, which are at Harvard. I went to the Educational Testing Service in New Jersey to get Ben Wood’s papers. He was successful working with IBM in developing a test-scoring machine. So you said you didn’t know about IBM’s teaching machine. They never built one. But they did build a test-scoring machine. And again, this close proximity between standardised testing and the rise of EdTech was something I really wanted to talk about.
So I did a lot of archival work. It was great, I loved it. It was fascinating to go through the letters, the correspondence that many of these scholars wrote to one another. So BF Skinner was writing letters to Sidney Pressey, Sidney Pressey was writing letters to Ben Wood. Skinner in particular would get letters from all sorts of people. So people that you wouldn’t expect were all in correspondence.
You know, thinking about our contemporary practices, it is interesting to think what will archival work look like a century from now, and what efforts do we want to make now in retaining stuff? I’ve become a big deleter, and so it’s worth thinking about what do you do with your email. Do you delete them, or do you save them in the hope that a hundred years from now someone’s going to want to write a story about them?
Q: The idea of putting all my correspondence in a box and locking it in a vault is just not going to happen! So what’s being lost now is really an interesting question. And also, to help write the book, you actually also bought an original machine from eBay to inspire you. How did the physical presence of that machine actually influence your writing?
A: You read the correspondence of someone like Pressey, who was so excited and so enthusiastic that teaching machines were going to be amazing … and you hear this rhetoric today, students are going to love this, they’re going to be amazed by it. And I think for maybe 30 seconds, students are amazed by it, even today. But any student who’s done Khan Academy exercises – or even the fanciest, jazziest, snazziest iPad app to learn fractions – knows, after a while, it’s just not that intriguing. It’s still fractions. It’s still a worksheet. It’s just not fun. And so, I wanted to actually use one to see what it was like to go through the process of the programmed instruction.
Again, Skinner was really adamant that you take content down to the smallest possible objects. Of course, EdTech thinks you do this today now too – take it down to the smallest possible content object and then, because you only want to reward positive behaviour, you want to make it so that students don’t make any mistakes. Mistakes are aversive, so you want this positive reinforcement. So going through these exercise on the teaching machine is so laborious. I thought I would never make it through MOOC content. But I would much rather go through an MIT class on Electrical Engineering than the teaching machines. It was incredibly dull.
Q: Academics talk a lot now about the ‘affect’ of engaging with these things, and actually, you’re not going to get a sense of that unless you actually do it. That’s really interesting. Finally, you said you didn’t want the book to be an angry book. You wanted it to be an interesting and honest one. How did that work out?
A: It was funny. I turned in my first draft, and it wasn’t what my editors at MIT Press were expecting. I think they were expecting an angry book. And I think that they were surprised. I think a lot of people actually who know my work are probably going to be a little bit surprised at what the book is. It’s a story. It’s not a book of analysis … although, as one does with this sort of genre of non-fiction, there is an introduction and a conclusion and those do contain analysis – don’t worry.
But I wanted to show that there was a really interesting history here. And I feel like in some ways, EdTech is worse off from not knowing its history and not knowing the theoretical underpinnings. EdTech is the worse for not really thinking about itself as not simply this amazing new, innovative practice but something that people have been doing for a very long time. And so I want to fill in some of those gaps with what I think is really an interesting story. And BF Skinner is, if nothing else, an incredibly interesting person. So this story is not dull!