What’s It All About, Alfie? (And Fred)

In the right-hand corner, Professor B. F. “Fred” Skinner brilliant and creative proponent of an old idea—organisms (that’s you!) learn through reinforcement, carrot (big) and (small) stick: “Behavior is determined by its consequences” and “[teaching is] arranging contingencies [of reinforcement] which bring changes in behavior.”

In the other corner is reinforcement-denier, multi-book-author Alfie Kohn, who recently wrote “For nearly half a century, research has raised troubling questions...the studies keep telling us … that rewards, like punishments, tend not only to be ineffective—particularly over the long haul—but often to undermine the very thing we’re trying to promote….Extrinsic motivators (rewards) tend to reduce intrinsic motivation (people’s interest in, or commitment to, what they’re doing).” And, don’t forget “Rewards are tools used by people with more power on those with less”!

This flat contradiction—rewards are essential vs. rewards are destructive—is not untypical of social-science research. It is confusing for the public and diminishes people’s faith in science. After all, both Skinner and Kohn cite research to prove their point and Skinner even did some!

The resolution of this contradiction is in fact pretty simple, if you remember three things:

Reinforcement learning—operant conditioning—is a two-part process. It involves selection, yes—that’s the reinforcement bit. But selection from a repertoire. That is what is called variation in the Darwinian selection/variation process that is operant conditioning. Reinforcement is more like a filter than a sculptor’s chisel.

Skinner famously wrote “[reinforcement shapes] an animal's behavior almost as a sculptor shapes a lump of clay.” Well, no it doesn’t: reinforcement selects from a repertoire.

The repertoire, the variety of possible behaviors in a given situation, depends on the situation. The most important thing about a situation is the type and amount of reward (or punishment) associated with it. In other words, reinforcement has two functions not one: it selects (Skinner’s exclusive emphasis) and it induces a repertoire from which to select: a hungry animal in a food-reward situation will show food related behavior, in a fear-inducing situation, fight or flight reactions and so on and all before its behavior has had any effect, any consequence.

Sometimes, the behavior induced by reinforcement conflicts with the behavior being selected: selecting for X, but get Y, as in paradoxical situations like instinctive drift and superstitious behavior. More usually, the behavior to be reinforced will eventually emerge, be “emitted” in Skinner’s words, and then be selected. But not always!

Reward is not reward is not reward: In other words, contra many economists and behavior analysts, all rewards are not the same. That is the bit of truth in Kohn’s argument. Sometimes the wrong kind of reward will induce the wrong kind of repertoire.

Here is a picturesque example. Imagine a young man, Romeo, offered a choice between two gorgeous identical twins. One of them, call her Juliet, is in love with him. The other, Salome, is a prostitute. A generous donor has paid Salome to have sex with Romeo, which Juliet is also happy to do. Juliet’s motivation is what is called intrinsic. Because she loves Romeo, she will spontaneously behave lovingly towards him. Romeo induces a loving repertoire in Juliet.

The case with Salome is rather different. She views Romeo, a pleasant-enough young man, with more or less indifference. But the cash she has been given now induces her to behave as lovingly towards him as Juliet—and possibly even more skillfully, given her experience.

Which of them will Romeo prefer? My guess is that most Romeos would give Juliet the nod. No matter how skillful Salome, the repertoire induced by love will certainly differ from the repertoire excited by cash. Proximity to one’s beloved induces a different state than an infusion of dollars. The type of reward matters.

The resolution of the Kohn-Skinner conflict is just this. Kohn is right in one sense only: rewards like “stickers, food, grades or money” are not the same as the reward offered by solving a problem on your own or writing a satisfying essay. But the reward does matter. A kid who doesn’t enjoy solving a puzzle or finding a new fact will not learn well. Above all, the kid who does not want to solve a puzzle, who does not have “puzzle solving” in his repertoire before he is ever rewarded, such a kid will also not do well.

Motivation Essential Reads

Why Trying Less and Trusting More Is the Way

How to Make Your Accomplishments Count

Skinner is right only in the sense that reward is involved in learning. He is wrong to assume that the single term reinforcement covers all its effects. Above all, Skinner failed to attend to where the repertoire from which reinforcement must select actually comes from. After all education is not learning to do more and more of the same thing—like a pigeon pecking a key in a Skinner box. Education is learning to do new things. It is learning to do something for the very first time: learning a new word or how to solve a new equation. It is getting the kid to make that kind of leap that is the real challenge for education. Skinner treated learning a new thing in the same way as he treated learning to do more of an old thing. It is in this way that Skinner missed the point about teaching.

But Alfie also misses the point by ignoring reward altogether. To say teachers should never reward their students—no more prizes, even for sports, presumably—is just silly. Adding ‘PC-bullying’ to the argument is both annoying and irrelevant. Are rewards really “tools used by people with more power on those with less”. (Reward must have some effect, then: Kohn undercuts his own case here.)

But he is right that the focus should be on ‘intrinsic motivation’ vague though that idea is. Science can in fact give it a perfectly precise meaning: Intrinsic motivation is just another name for the repertoire that students bring to a learning situation. And that is where much-derided behavior analysis can make a major contribution, by asking the basic question, which is not about contingencies of reinforcement but about history. What kinds of history can we give students to make available the kinds of repertoires that will allow them to learn well?

This is not a question with a single answer. Kids differ. There will be many histories, many experiences—many for each child, many more for all children—that can help students to become curious, to enjoy solving problems and learning new things. Behavior analysts should surely pay much more attention to case histories. Let’s look at good teachers and at bad ones. What do good teachers and good schools do that bad teachers and schools do not, and vice versa?

In biology, natural history came before the theory of evolution. So, in trying to understand education, let’s do more natural history: let’s observe the phenomena before tub-thumping our favorite slogan, be it “reinforcement” or “intrinsic motivation”.