Negative Reinforcement ≠ Punishment

Friday, January 12, 2018
By Michele Gregoire Gill

Homer Simpson demonstrating the four fundamental types of reward and punishment

Prof Michele Gregoire Gill is back! In her previous post she mentioned that Beeminder, in large part, motivates her via negative reinforcement. If you think that makes her sound like a masochist, or that she must set scary high monetary penalties on her goals, then you’re probably under a very common misconception about these behavioral psychology terms that Prof Gill is about to disabuse you of.

If my grad students are any indication, the difference between negative reinforcement and punishment is a common point of confusion. So let me set the record straight. Operant conditioning, a key part of behaviorist theory, means shaping behavior through positive and negative stimuli. It’s all about rewarding good behavior and punishing bad.

Punishment is what you’d expect: a response to undesired behavior, to deter it. Reinforcement is the opposite: a response to desired behavior to — that’s right — reinforce it. Which is to say, get more of it. Simple enough.

The confusion comes in when we break those down into the following four fundamental types, based on the type of stimulus and whether we want more of or less of the target behavior:

  1. Positive Reinforcement
  2. Negative Reinforcement
  3. Positive Punishment
  4. Negative Punishment

Putting “negative” in front of those words doesn’t mean they’re bad or that they’re negating the word that comes after. Negative reinforcement isn’t the opposite of reinforcement. Rather, “negative” and “positive” indicate the type of stimulus used for the punishing and reinforcing. A positive stimulus is something you add and a negative stimulus is something you take away.

In everyday language, the word “positive” implies something good. But in the case of operant conditioning, positive is used along mathematical lines — it means to add something. So, positive reinforcement adds a reinforcer. Positive punishment adds something aversive. Similarly, negative means to subtract, so negative reinforcement means to take away something aversive, and negative punishment means to take away something good.

Electric shock is a classic positive punishment. It’s something aversive when you exhibit the wrong behavior. A negative punishment could be taking away a favorite toy from a child. You’re removing a positive stimulus. The obvious one, positive reinforcement, is giving someone a metaphorical (or literal) cookie. And finally, the most misused term, negative reinforcement, means rewarding you with the removal of an aversive stimulus. Like the annoying beeping that stops when you put on your seatbelt.

Here’s a table I use with my students to drive that home: [1]

Table showing the four types of punishment and reinforcement

Bringing this back to Beeminder, paying the $5 pledge is a positive punishment. That’s less motivating for me. But stopping the negative stimuli of the warning emails, that negative reinforcement motivates me a lot.

Punishment is generally considered the least effective for motivating behavior change. Behaviorists, including the father of Behaviorism, B.F. Skinner himself, argue that it comes with a host of side effects and often the behavior change doesn’t persist. It does work, though, in the short term, and the painful consequences of having to pay an increasing fine when one derails is an effective consequence to initiate behavior change. It’s just not a sufficient motivator for me.

Positive reinforcement is the most pleasant extrinsic motivator, yet it’s highly individualized. For one person, a slice of cake is motivating, for another, a cash reward, for another, praise. For me, adding a data point to show a goal being met was a type of positive reinforcement, but since much of my data is automated, the visual effect on my graph was less immediate (and thus less reinforcing, according to behaviorist principles). Seeing my dots above the yellow brick road was positively reinforcing as a sign that I was staying “on track” with my goal, but they didn’t have the “oomph” of a true, immediate reinforcer. (Which is why other habit trackers didn’t work for me. There was no consequence of not adding my data, no compelling reason not to ignore logging my progress for the day.)

“Completing my goals each day removes something negative — being yelled at by Beeminder’s email bot”

Here’s where negative reinforcement works for me. Completing my goals each day removes something negative — being yelled at by Beeminder’s email bot. Again, reinforcement is highly individualistic. [2] For me, a people pleaser, I guess that transfers to pleasing the Beeminder Bot. At least it connects with not disappointing myself. So, each day I can relax once my goals are done, knowing I’m “on track” and not going to be in trouble with myself for failing to meet a goal. Thus, negative reinforcement actually reinforces (increases) behavior.


 

In any case, the hidden brilliance of Beeminder, to me, is that it contains multiple means of motivating habits. For some, like me, it may operate primarily via negative reinforcers. For others, the positive reinforcement of the pretty graph suffices. And for others, or for particularly difficult habits, the threat of punishment — paying money in this case — might be sufficient to initiate behavior change. Because that’s too controlling for me, inciting my inner rebel, I keep the fines low, around $5, but that could change if I needed to create a habit that was particularly difficult but critical to establish. For example, if I developed a medical condition requiring a radical change of diet, and I was having difficulty making the appropriate changes, I’d up the ante on the fine and use that as incentive to stick to the straight and narrow with my diet.

As to why one would pay Beeminder rather than just lie or fake data, well, that’s a whole ‘nother aspect of psychology beyond the space I have here to discuss, but one which Beeminder’s founders have addressed before.


 

Footnotes

[1] The fact that my graduate students struggle with this is not an anomaly. Psychologists often coin terms that don’t quite mesh with how they are used outside of the discipline. (More examples: self-concept is different from self-esteem, being an origin means having autonomy, authoritative parenting is good; authoritarian is not.) Given the confusion, there’s been some recent discussion about possibly doing away with these terms, but these are the classic definitions according to behaviorist principles.

[2] Another example from a classroom environment: For many students, candy is a good motivator, but for some, especially those who cannot eat candy, it has little reinforcement value. Similarly, some kids may perceive attention as a positive thing; others, as something to be avoided.


 

Michele Gregoire Gill is a writer of academic works and nonfiction essays and prose, a mother of two young-ish boys, a teacher, and an educational researcher. She works as a professor of educational psychology and program coordinator for the education doctorate in curriculum and instruction at the University of Central Florida. She also serves as chairman of the board for the K-8 charter school she founded in Sanford, Florida. In her spare time, she knits simple things, learns bass riffs for her favorite rock songs, and listens to birds chirping in her backyard. You can find her online at michelegregoiregill.com.

Tags: , , , , ,

  • FBRacer

    My graph changed today, my data no longer has lines between the points and my rosy line is back. I like it, but wished there was info on your blog.