Operant Conditioning and Instructional Design

One of the primary influences on instructional design (and online learning as a consequence) is educational psychology. One of the field’s founding fathers, B. F. Skinner, made an argument for the utility of teaching machines in educational settings. While he believed that learning happens outside of formal institutions, he held firmly to the notion that formal education could be (perhaps should be) systematized.

In his chapter “The Technology of Teaching“, B. F. Skinner asserts that:

The application of operant conditioning to education is simple and direct. Teaching is the arrangement of contingencies of reinforcement under which students learn. They learn without teaching in their natural environments, but teachers arrange special contingencies which expedite learning, hastening the appearance of behaviour which would otherwise be acquired slowly or making sure of the appearance of behaviour which might otherwise never occur (429-430).

Traditional instructional design relies heavily on this idea, and on Skinner’s ideas of the utility of teaching machines (the precursor to the modern LMS). Operant conditioning and the manipulation of response to stimuli (or content) are at the heart of theories that support instructional design, from Bloom’s Taxonomy to competency-based learning.

Matthew Kruger-Ross offers this insight into the history of instructional design:

When we discuss the history of educational technology, we often begin in the 1950s-1960s with the entrance of computer technology. The first experiment with education and computers was called Computer Aided Instruction (CAI) and consisted of a learner seated in front of a dumb terminal. The basic computing program presented piecemeal bits of information to the learner. After, the learner was asked to complete a number of questions written specifically to determine if she had learned the content. Because of the limitations of programming languages and computer capacity, in addition to the engineers’ simplified understanding of teaching and learning, CAI redefined learning as driven by clear and concise objectives that could be easily quantified and measured (see Hamilton & Feenberg, 2012).

How much different from CAI is instructional design, even with the nuance added by newer technology? What relationship do they encourage between learner and learning, or learner and computer? In many cases, the methodology hasn’t evolved from a critical pedagogy, but rather from the same CAI principles of traditional instructional design. Monkey see, monkey do, monkey hit submit. Students in adaptive learning or competency-based learning environments may seem to have more “say” in how their learning happens, but knowledge still equates, and is inevitably assessed, according to recall.

Breaking learning into tidy pieces helps make instruction (and instructional design) efficient, even replicable. Teaching machines—whether simplistic CAI consoles or learning management systems with LTI integrations—require teaching techniques that can arrange learning into component parts. Without that, the teaching machines will most likely fail. In the face of intuitive leaps, for example, or that “sixth sense” that teachers often have in a room with students, teaching machines become just machines.

And so out of the theories of Skinner and other educational psychologists came taxonomical breakdowns of what learning looked like. Bloom’s Taxonomy is among the most notable and influential of these.

Bloom’s Taxonomy

According to Wikipedia,

Bloom’s taxonomy is a set of three hierarchical models used to classify educational learning objectives into levels of complexity and specificity. The three lists cover the learning objectives in cognitive, affective and sensory domains. The cognitive domain list has been the primary focus of most traditional education and is frequently used to structure curriculum learning objectives, assessments and activities.

I recently wrote about my experience working with Bloom’s Taxonomy here:

When I began working in course design, as I have written before, my toolbox included almost nothing but Bloom’s Taxonomy and the preternaturally unremarkable Skillsoft proto-learning management system. The click-through courses I designed as part of my job were carefully structured to take a learner from knowledge to comprehension to application to analysis. Ritually, the first module of any course bridged the lowest rungs of Bloom’s Taxonomy, while the second module pushed the learner to think a little beyond mere recall; but usually, application and analysis were saved for level 2 and level 3 courses. If you wanted to be a really good employee, you could pass these higher level courses, showing your supervisors that you were able to do more than, say, a trained dog.

Bloom’s Taxonomy—and potentially any taxonomical view of learning—lends itself to the idea of scaffolding, or leading the learner along through step A and step B in order to reach step C. The most important idea for now, though, is that learning can be looked at as a linear phenomenon, and one that is both predictable and replicable.

In the best applications, the “efficiency” model of education provides students with a consistent approach to learning, one which they can rely upon and one which, because of its consistency, can free them to think in broader terms. The idea is that there are affordances in constraints.

In the worst applications, this model has led to the literal replication of online courses term after term, year after year, until there is less and less for the instructor to do, fewer relatable moments between instructor and student, and automation has taken over the teaching process. And this is what B. F. Skinner warns of when he states that the kinds of instruction provided by a teaching machine could

imitate, and could presumably replace, the teacher. But holding a student responsible for assigned material is not teaching, even though it is a large part of modern school and university practice. It is simply a way of inducing the student to learn without being taught (427-28).

We might argue that the inverse is true, too: that teaching through the machine can become teaching without learning. In other words, regardless of how we feel about Skinner, educational psychology writ large, Bloom’s Taxonomy, or the LMS, uncritical application of any educational model, especially in the digital age, can lead to a misunderstanding (at best) of what happens within the mechanism of technologically-inflected teaching and learning.