In a recent piece by Audrey Watters titled Behaviorism Won, the author talks about two diverging philosophies in education, represented by John Dewey and Edward L. Thorndike. Dewey believed that learning is best done by having the learner engage in a thought-provoking journey of inquiry with the teacher. It stems from the theory of constructivism, which essentially states that guiding the learner to re-construct the knowledge on their own is the best way to learn it. Thorndike on the other hand, came from the school of behaviorism as pioneered by B. F. Skinner, and believed that learning is just a form of behavior training, and is best done via conditioning techniques, or whatever is the most effective way to alter behavior as determined by experiments. As the title of the article suggests, over the course of the last century, the latter school seems to have won, and the author goes into some detail on the history and why of the development. But there is a critical reason which the piece does not quite touch on for the victory of behaviorism, and that is its vastly better scalability.

Behaviorism “won” because it is must more easily scalable, both horizontally and vertically. Horizontal scaling is about how easy it is to replicate. By dialling education down to a science, behaviorism simplifies the process of teaching into a set of clear, definite procedures to follow (e.g. standardized curriculum, teaching methods, and tests), and can be replicated widely with relative ease. Vertical scaling in this case means how effective it works across the whole spectrum of different people. By exploiting the biologically guaranteed ways that all our brains react, behaviorism works on all, regardless of individual differences in personality and preferences.

The “Dewey way” of learning, through project-based inquiry, simply doesn’t scale easily in either direction. It requires more intimate teacher involvement in the learning process of each individual student, so scaling wide becomes much more difficult. The education process being more interactive also means that it is hard to codify into a fixed procedure, and demands higher quality teachers which are in lower supply. There is also greater variation in its effectiveness across a population, as it requires truly active participation from students, so it works much more effectively on those who are naturally inquisitive, and much less so on others.

So even though behaviorism may not be the best way to learn from an individual learner’s perspective, it is the only one (out of the two philosophies mentioned here) that can scale easily to the level that our society needs. More broadly, I think almost all problems eventually becomes a problem of scale, whenever you (as an individual, group, or civilization) are trying to do something non-trivial. Scalability matters, and behaviorism, by exploiting the baser, more simplistic and deterministic parts of our brains, is easier to scale in all directions.