More on Reinforcement and Punishment

In our Fundamentals of Animal Learning Course, we are tackling some thorny issues. in another post we talked about the terminology problems surrounding terms like punishment, correction, and discipline.  In response to that discussion, our colleague Dr. Frank McMillan, the Director of Well-Being Studies at Best Friends Animal Society in Utah submitted a  thought provoking comment.  We thought it deserved its own post, so Dr. McMillan’s comment is reproduced below, with this permission.

"While much effort has been made (in both human psychological and animal behavior fields) to clarify the definitions of positive and negative reinforcement and punishment, the distinctions remain unclear. The main problem is the distinction between “positive” (meant to indicate the “giving” of something to the animal, be it a pleasant or unpleasant stimulus or experience) and “negative” (meant to indicate the “removal” of something from the animal, again, either a pleasant or unpleasant stimulus/experience).

But in terms of the individual’s internal experience of net pleasure, the “positive” and “negative” are often not distinguishable, for 2 reasons: (1) defining what is “given” or “removed” is often a matter of semantics, and (2) the giving and removal depend on where the animal is positioned on a pleasure—pain continuum, which is not always knowable.

Consider these examples for each. (1) If an animal is outdoors in freezing weather and is allowed indoors where it is warm, this would (in most cases) be a reinforcement for whatever behavior immediately precedes it. However, is the reinforcement a “giving” of warmth (“positive reinforcement”) or a removal of unpleasant cold (“negative reinforcement”)? If a dog performs an undesirable behavior and is immediately confined in a room by him/herself as a punishment, is this a removal of social companionship (“negative punishment”) or a giving of social isolation (“positive punishment”)? (2) If pleasure can be modeled on a continuum running from extremely pleasant to extremely unpleasant, then where the animal “sits” on this continuum defines the “positiveness” and “negativeness” of any reinforcement or punishment. 

Consider the reinforcement of a tasty food treat. If the animal is in a state of discomfort (i.e., hungry), then eating the food treat is the animal’s attempt to lessen the hunger. In this case, the reinforcement is a “removal” of the hunger (the same thing could be accomplished if the dog had a permanent feeding tube surgically installed in his stomach – inserting food through the feeding tube would “remove” the dog’s hunger). However, if the animal were in a state of comfort (i.e., not hungry), then the dog would eat the treat because of the pleasurable taste (like the person who, even though not hungry, pulls the ice cream out of the freezer and chows down). Here the dog is not trying to rid himself of the unpleasant feeling of hunger, but is rather simply trying to increase the pleasant feeling of tasty food. 

In this case, giving the dog the treat is not “removing” hunger, but is instead “giving” pleasure. Therefore, the reinforcement provided by the tasty treat is a “negative reinforcement” if the dog is in a hungry state (on the unpleasant side of the pleasure—displeasure continuum) at the time the treat is offered, where the treat is “removing” the hunger.   Conversely, the treat is a “positive reinforcement” if the dog is not in a hungry state (on the neutral or pleasurable side of the pleasure—displeasure continuum) at the time the treat is offered, where the treat is “giving” pleasure. 

Hence the “positiveness” and “negativeness” of the reinforcement is wholly dependent upon where on the pleasure—pain continuum the dog’s current mental state exists. In all cases the dog will behave in a way that moves him toward pleasure and away from displeasure, but in each case of being offered a reward the dog will be “starting” at a different place.

As a further explanation, consider 2 scenarios: as you continue to provide treats the dog becomes “full,” which ultimately then reaches the point of satiation and the additional consumption of treats loses its reward value (and can even become an aversive if the treat were forced); and the old technique of “increasing the reward value” of a food treat by withholding the dog’s regular food is simply moving the dog further toward the displeasure end of the pleasure-discomfort continuum, such that the dog’s need to “remove” the more intense hunger feeling makes him “more motivated” to behave in a way that gains him the treat.

A final comment is that very often the reinforcement is a combination of both “positive” and “negative,” meaning that the dog eating the tasty treat is attempting to both alleviate (“remove”) the unpleasant state of hunger – and hence make the reinforcement “negative” by definition – AND gain pleasure from the tastiness of the treat – and hence make the reinforcement “positive” by definition."   by Dr. Frank McMillan

We discuss this very issue in our "Fundamentals of Animal Learning Course".  Although the argument has been made by some (Baron and Galizio 2005)  that there is no difference between negative and positive reinforcement, Murray Sidman (Sidman, 2006) and others disagree. 

 We want to thank Dr. McMillan for his thoughtful comment and for his permission to post it here.