Wikigroaning and shaping

Maybe you’ve never heard of Wikigroaning. 

A term coined in 2007 by internet user fart, Wikigroaning describes the process of comparing a meaningful Wikipedia article against a…less meaningful article. An example: compare the page for physician to the page for the comic book character Dr. Doom. Or, say, compare radical behaviorism to Radical Optimism (a 2024 Dua Lipa album).

Anyway, let’s take a look at the page for shaping.

***

It starts off strong, with a basic definition (“The method used is differential reinforcement of successive approximations.”). Then a block quote from Skinner, followed by an example bulleted list of hypothetical steps in shaping a lever press.

The final paragraph starts off on the wrong foot. 

Shaping sometimes fails.

This is literally true (in the sense that a person shaping animal behavior can fail), but the implication is that “shaping,” conceptually, is sometimes a failure. But the example fails to illustrate the point, because the example is our good friends, the Brelands.

Instinctual drift – a trained pig who roots a coin rather than depositing it for food – is the example of “failure.” Yet the examples are animals that were successfully trained, but whose behavior was not maintained by reinforcement. The shaping process, that is, worked.

These results show a limitation in the raccoon’s cognitive capacity to even conceive of the possibility that two coins could be exchanged for food, irrespective of existing auto-shaping contingencies.

This is not evidence of a limitation in cognitive capacity; we do not know what raccoons can “conceive of”; and this is not a description of autoshaping, but shaping. It’s also written in a strangely hyperbolic way: eliminate the words “even” and “of the possibility” and the meaning of the sentence doesn’t change.

But then we get to autoshaping.

***

Per Mazur and Odum (2023):

The term autoshaping is now used to refer to any situation in which an animal produces some distinctive behavior in response to a signal that precedes and predicts an upcoming reinforcer. 

By all accounts, autoshaping was first reported by Brown & Jenkins in 1968. They discovered that, when a disc lit up near a food tray and preceded delivery of food, pigeons would soon begin to peck the disc. What was unusual about this response was that it was not required – in fact, it was inefficient. The pigeons should, in theory, immediately move their heads to the food tray when the disc is illuminated. 

Even more strange, when pecking the disc resulted in omission of food, the pigeons pecked in about ⅓ of opportunities. Pigeons pecked themselves out of a meal. What a bunch of peckers!

Admittedly, this is a confusing state of affairs. Doesn’t reinforcement…like…reinforce behaviors? Between the Brelands and autoshaping, there is evidence that animals engage in entirely unreinforced, inefficient behaviors.

There is an explanation. But what does Wikipedia have to say?

***

Autoshaping provides an interesting conundrum for B.F. Skinner's assertion that one must employ shaping as a method for teaching a pigeon to peck a key. After all, if an animal can shape itself, why use the laborious process of shaping?

These might be the two worst sentences in the whole entry. Where does Skinner write that one “must” employ shaping to teach a key peck? Autoshaping was reported within Skinner’s lifetime. Perhaps he even mentioned it.

The second sentence is even more unbelievable. Why use shaping if an animal can shape itself? What does this mean? Animals shape their own behavior? Bears in the circus teach themselves to ride unicycles? Is this a literal misunderstanding of a prefix?

The final paragraph of the entry doesn’t do much better:

But, the clearest evidence that auto-shaping is under Pavlovian and not Skinnerian control was found using the omission procedure. In that procedure,[9] food is normally scheduled for delivery following each presentation of a stimulus (often a flash of light), except in cases in which the animal actually performs a consummatory response to the stimulus, in which case food is withheld. Here, if the behavior were under instrumental control, the animal would stop attempting to consume the stimulus, as that behavior is followed by the withholding of food. But, animals persist in attempting to consume the conditioned stimulus for thousands of trials[10] (a phenomenon known as negative automaintenance), unable to cease their behavioral response to the conditioned stimulus even when it prevents them from obtaining a reward.

“Pavlovian” and “Skinnerian” control, pitted against each other like pro wrestlers. Two men enter, one man leaves!

More importantly, the animals do not attempt to consume the discriminative stimulus. The Breland examples most obviously illustrate that these are not consummatory responses: the pig roots, and the raccoon rubs. 

And as we shall soon see, sometimes the stimulus can’t be consumed.

***

Wikipedia has a preference for secondary sources. This is because, in theory, secondary sources have selected for facts and discarded falsehoods. Although ironically, the autoshaping section only quotes primary sources from the 60s and 70s.

Let’s humor them here. Do ABA texts discuss autoshaping? We looked through the most likely sources by searching “autoshap*” (to cover autoshape, autoshaped, autoshaping, etc.):

Applied Behavior Analysis, Cooper, Heron & Heward (2019): No

Behavior Analysis and Learning, Pierce & Cheney (2013, 2017): Yes

Applied Behavior Analysis, Sarafino (2012): No

Behavior Modification, Martin & Pear (2015): No

The Wiley Blackwell Handbook of Operant and Classical Conditioning, McSweeney & Murphy (2014): Yes

Handbook of Applied Behavior Analysis, Austin & Carr (2000): No

Handbook of Applied Behavior Analysis, Fisher, Piazza & Roane (eds) (2011, 2021): Yes

Principles of Behavior, Malott & Kohler (2021): No

APA Handbook of Behavior Analysis vol. 1, Madden (ed) (2016): Yes

Behavior Analysis for Effective Teaching, Vargas (2009): No

Behavior Analysis for Lasting Change, Mayer & Sulzer-Azaroff (2011): No

Learning, Catania (1992, 2013): Yes

Learning and Behavior, Odum & Mazur (2023): Yes

Applied Behavior Analysis for Teachers, Alberto, Troutman & Axe (2022): No

Mostly, the story in these sources is consistent, with varying levels of detail. 

It’s true that birds peck a disc if it lights up before food delivery. In fact, they emit this response even when the disc is not close to the feeder, and even when it is on an adjacent wall. When food is used as the reinforcer, the pigeons peck the disc with their beak open, but when water is used, the pigeons peck with their beak nearly closed – similar to  how they consume food and water, respectively.

But when the stimulus preceding the food delivery is auditory, the pigeons do not peck. 

When a lighted disc precedes a heat lamp for baby chicks, the chicks nuzzle the disc, which they do to their parents, but spread their wings under the lamp. These are two different responses in a common response class.

And when the signal is a live restrained rat (???) other rats do not “consume” the rat. They nuzzle it and engage in other social behaviors, similar to how pups might treat a parent.

Therefore, it is not superstitious behavior, as proven by the omission experiments. It is not stimulus substitution, as proven by the chick and rat experiments. 

So…what is it?

***

Operant and respondent processes are not entirely separate, and no radical behaviorist has ever published anything to this effect. That’s the short answer.

Skinner himself commented on autoshaping, and “instinctual drift.”

Here is one comment on the Brelands, from Herrnstein and the evolution of behaviorism:

Herrnstein was apparently misguided long before The Bell Curve.

And here is part of Skinner’s brief reply in Science regarding autoshaping:

The Brelands strike again!!

Skinner appears to propose two ideas: 1) adjunctive (or schedule-induced) behavior, and 2) reinforcer-related behavior. He does not propose them as separate, but in combination: the schedule of reinforcement may cause animals to treat non-food stimuli as food-related.

This is generally how behaviorists see the issue today. You’re welcome, Wikipedia.

***

Just for the sickos, we will return to the dreaded primary source materials.

William Baum has a similar theory about this phenomenon that he explains with the concept of induction. Put simply, the discriminative stimulus (light, token) evokes a class of behaviors related to the terminal reinforcer (e.g., if food is a reinforcer, a light could induce salivation). That is, salivation was not reinforced by food delivery, but precedes it.

Skinner, with the final word from The phylogeny and ontogeny of behavior: