One of the things I probably need to define in the book is “what counts” as a teaching machine. I’d rather do so as broadly as possible without losing any meaning or significance to the phrase. In more ways than one, it’s a minefield, and I already get snide comments from some constructionist corners when I write about Sidney Pressey, for example, insisting that “that machine doesn’t teach!” B. F. Skinner felt the same way about Pressey’s Automatic Teacher, incidentally.
In the introduction to Teaching Machines and Programmed Learning: A Source Book (1960), a collection of articles penned by some of the best known theorists and practitioners in the field (including both Skinner and Pressey), A. A. Lumsdaine lists these as the three key properties of “teaching machines”:
First, continuous active student response is required, providing explicit practice and testing of each step of what is to be learned.
Second, a basis is provided for informing the student with minimal delay whether each response he makes is correct, leading him directly or indirectly to correction of his errors.
Third, the student proceeds on an individual basis at his own rate – faster students romping through an instructional sequence very rapidly, slower students being tutored as slowly as necessary, with indefinite patience to meet their special needs.
The devices thus represent a way of providing a pre-programmed study-practice combination which simulates, in partially or fully automated fashion, the functions of a private tutor in recitation and practice, with immediate correction of errors and feedback to the student.
As this description perhaps implies, teaching machines need not be machines at all. And indeed, many weren’t.
Following his failure to commercialize his Automatic Teacher – in part because of the expense of manufacturing machines – Sidney Pressey continued to work on various paper- and card-based versions of his idea during and after World War II. The important thing was always, as Lumsdaine’s list makes clear, providing the student with immediate feedback.
Although he built a machine that could take students through the same material, Norman Crowder wrote his series of TutorTexts, “scrambled textbooks” that directed students to turn to different pages depending on how they answered exercises and questions at the bottom of explanations.
Skinner co-wrote a “programmed textbook” too – The Analysis of Behavior – with his colleague James Holland. (There are no multiple choice questions in this one, of course.)
And I think I’d like to make the argument, as I have previously, that the Science Research Associates’ famous SRA Reading Cards were also part of this post-WWII trend in individualizing education through self-instruction. (Ideally I find something that explicitly ties this product to the “teaching machine” trend – perhaps the link is IBM, which acquired Science Research Associates in 1964. I don’t know.)
So why call these “teaching machines” if a fair number of them weren’t mechanical at all?
The easy answer: that’s because what their inventors and promoters called them. That’s how they were advertised. The more elaborate answer involves… well, it involves my original quandary: how will I define “teaching machines”? I think I should emphasize, in part, the other phrase that arose side-by-side with the machines – it’s there in the title of Lumdaine’s book: “programmed instruction.” The difference there between the two phrases – what’s the noun and what’s the adjective in each – is illustrative perhaps. Is it less than the device is mechanical and more that the teaching is automated?
You’d think by this stage in the game, I’d have this more sorted out…