Monday, April 30, 2018

Old and New Problems with Behaviorism in EdTech

About 100 years ago, research in Education (like formal, disciplinary "Research") was relatively new, and there was a rift between a philosophical approach to education research on the one hand (e.g., John Dewey), and a scientific approach on the other (e.g., Seymour Pressey and Edward Thorndike) (Lagemann, 2000).

Actually, before that, if you go back another 50 years or so to the mid-1800’s,  public education in the U.S. was spreading like wildfire.  Things like mandatory k-12 education and the Morrill Act which provided land grants helped spread education to the masses (well, white people that is).

The rise in public education led to a teacher shortage that led to more women teaching overtaking men as the predominant gender (this demographic shift persists today with 80% of k-12 teachers being white women- higher ed becomes more male dominated).  Some of these women started doing scholarly research, emphasizing the nurturing and supportive aspects of education.  

In the 1900’s, industrialization was changing every aspect of our society.  New methods of scientific experiments were being applied to all kinds of things including the management of people and organizations. (Think Taylorism, right?)

Folks like Pressey and Thorndike started creating machines for their scientific education research.

These "education scientists" basically told the women: “thanks, we’ll take it from here” and kind of patted them on the head.  Men and their scientific methods began dominating education research (at colleges and universities), while women continued to teach in the lower grades without conducting research (Lagemann, 2000).


The early behaviorists believed that people were like animals or even physical objects and that everything could be explained by natural laws of cause and effect.  They also believed that observable things like memory recall were all that mattered in education.

This is in contrast to Dewey, who talked about education as “democratic” and as a shared social experience.  "Education is a process of living" he once quipped, "not a preparation for future living" emphasizing the experiential or intrinsic value of education rather than the pre-occupation with "results" or "outcomes."

Meanwhile, education policy makers were intrigued by these new methods of punishment and reward and how things like propaganda could be used to indoctrinate youth and instill a sense of national pride through education.

But, as Dewey might suggest, education should respect peoples’ freedom.  Sharing different viewpoints against a backdrop of “traditional” knowledge should be a function of education.

Punishment, reward, and memorization – that’s how we train parrots.  And people aren’t like parrots.  Right? Well, for B.F. Skinner, people were like pidgeons. Actually, he used rats at first, but pidgeons lived longer.  Skinner thought his behaviorist predecessors got some things wrong:  It’s wasn’t just about automatic responses in individuals (like dogs salivating when a bell rings), but there were environmental factors as well—what Skinner called “operant conditioning.”  If we could pinpoint just the relevant environmental factors (and Skinner thought there weren’t that many) we could understand (and control) all human experience. Skinner also believed that freedom was a myth and that thoughtful self-reflection was useless.

The problem with behaviorism is that it works.  At least on some level.  Sure, people respond to stimuli.  We eat when we’re hungry.  We feel sad or happy when things are intended to make us feel sad or happy. The problem is that there are way way too many variables at play in very complicated human experiences. And when data and design is used to manipulate and control these superficial behaviors, it can be objectifying. Now you might say behaviorist tools in math learning or game theory is useful or even fun. 

Sure.  But.

This can be a slippery slope.  Design always contains bias, and to acknowledge that bias, the tools and systems should be totally transparent about what they are doing and how they work.

Predictive systems are only as good as the things they measure- and if they measure things like memorization or if they promote things like superficial behaviors (like “nudging or prodding”) then maybe they aren’t supporting good education.

Behaviorism is present in software design because it assumes the brain is like a computer or a machine.  It isn’t.

Software designers should focus on developing tools that reflect a different set of values like creativity, empathy and collaboration.  And there examples of games and software that do this.  Games and "gamification" might also suggest social, collaborative, and creative benefits of the activities. 

Behaviorism implies that it’s OK for people to be manipulated and controlled.  And behaviorist technology sometimes hides the people, powers, and ideologies behind them. 

We know all too well that when software is used for manipulation, it can cause problems.  In an age of fake news, advertising algorithms, and privacy violations, we need education researchers and educational technologists to not only consider these problems but also to face them head on. 

Lagemann, E. C. (2000). An elusive science: The troubling history of education research. Chicago: University of Chicago Press.

No comments:

Post a Comment