There is a parenting article circulating right now that aims to outline the shortcomings of positive reinforcement. It’s titled something like “Positive Reinforcement Doesn’t Work Long Term” (I am not going to link it here because I really don’t need five thousand angry parents emailing me about how, as a non-parent, I shouldn’t be talking about parenting–which I’m not doing anyway–and it’s frankly not worth reading) and it is so prolific it has showed up in my feed a few times. An article by a dog trainer with the same title wouldn’t have cracked into my feed; that’s not my friends circle. That an article about parenting with this title DID crack in caught my eye. So I looked it over, and my initial response to the article rang true after I read it. That response is this: positive reinforcement is defined by its function. By definition if behavior isn’t increasing positive reinforcement is not at play. So the title itself is a misnomer; and the content didn’t dive into anything of real merit in behavior change.
So why are we talking about it? Because random parenting articles aren’t the only place this misunderstanding is commonplace.
Quadrants and Procedures
We need to get something straight; all quadrants of operant conditioning (positive reinforcement, negative reinforcement, positive punishment, and negative punishment) are defined by their effect on behavior; their function. If I give my dog a slice of hot dog after he goes to a mat when the doorbell rings, then my delivering that hot dog after a certain behavior is my procedure. Positive reinforcement is taking place if a behavior increases. If the mat behavior increases then my procedure is working, and the mat behavior is being positively reinforced. We can assume that the hot dog is acting as a positive reinforcer.
If my procedure fails to increase the mat behavior, did positive reinforcement fail? No, my procedure failed. The behavior I meant to reinforce was not reinforced.
On the flip side if I have a dog who sometimes fails to hit his two on two off behavior in the agility ring and I decide to remove him from the course if he does so, then leaving the ring at that failure is my procedure for suppressing the premature motion off the board. If that behavior decreases then negative punishment (or, potentially positive punishment) was at play. If this fails to affect my dog’s behavior the procedure failed, not punishment itself.
Do Semantics Really Matter?
Semantics may not matter much but understanding matters a lot, and if we use incorrect semantics we demonstrate that our understanding is truly lacking. If we state that “punishment doesn’t work” we show our ignorance just as much as the next person who states that “that clicker stuff didn’t work.” Punishment and reinforcement work, by definition. We can be much smarter in our procedure design if we recognize that it is our procedure at fault–not the quadrant we believe to be utilizing–if behavior change is not taking place in the way we anticipated.
So, if you’re still with me, I hope you’re understanding now that positive reinforcement does work; if it ever appears that it isn’t working, it’s not actually at play. If you’re having a training struggle that you’re considering a procedural change for, know that the quadrant you might be trying to use isn’t the problem. We can put any of the four quadrants to use in changing behavior; what matters is the procedure we choose. A prong collar is just a hunk of metal, a hot dog just a tube of meat, and a tug just bits of fabric sewn together until they effect behavior.