Talk:Shaping (psychology)

From Wikipedia, the free encyclopedia

Contents

[edit] Feedback vs. shaping

The article includes mention of feedback processes and makes an attempt to distinguish between feedback and the procedure of shaping behavior. While it might not be as close to the most typical examples of shaping, the argument used in that claim is incorrect. For example:

"Successive approximation should not be confused with feedback processes, which increase the accuracy of a response without changing response strength."

As is already mentioned in the article, shaping constitutes the reinforcement of gradually more accurate forms of a behavior with the desired behavior as the end goal of the procedure. So accuracy is very much what is target of shaping procedures because it involves the gradual changing of the response's form.

"For example, a music student may tune a saxophone by playing successive notes into the microphone of an electronic tuning device, each time adjusting the length of the saxophone until the tuning device says the note is on tune. This procedure, however, does not increase the probability that the student will continue playing the instrument or reduce the time the student requires to tune the instrument."

The problem with this argument is that the overall probability of playing a musical instrument is not the target of the shaping procedure. In the example given, the targeted response is the form of the musical note being played. At the end of such a procedure, the probability of performing the desired response (producing the desired musical note) is in fact increased. What is being targeted is not the playing of the instrument or the time it takes to tune the instrument, but the performance of the note itself.

Instead, one way that it could be argued that feedback isn't quite a form of shaping is that it does not involve successive approximations being reinforced between the performance of the baseline behavior and the target behavior. For example, the tuning device doesn't give the student confirmation of having gotten one step closer to hitting the right note, only once the correct note is performed (or I could be wrong as I'm not a musician). It is probably more an example of extinction-induced variability with reinforcement happening only once the target response occurs. Extinction-induced variability is only part of the shaping procedure, however. Lunar Spectrum | Talk 05:51, 15 December 2006 (UTC)

I think we're just splitting our hairs differently, but I agree the example could be better. I was trying to avoid the example that was already in the article (presented as an example of shaping), but I can't remember why. If you can suggest a better example I'm pretty sure I'll agree it's better. Incidentally, the analog tuning device I used many years ago in high school did confirm that you were getting closer to the note.
As I recall, the original example was a game like 20 Questions, which might be a better example. John FitzGerald 13:43, 15 December 2006 (UTC)


Hello. A simple example of the difference between feedback and shaping can be had by vastly exaggerating the two categories to see how radically different they are.

"Feedback" means some change in the situation that is in response to your action. For example, if your finger is on a button, and pressing it causes a loud noise, we can say "it gave you feedback on your key press". If as you press it harder it makes a successively larger noise, it might be described as "differential feedback".

What is called feedback can fit into numerous categories of consequence, but notably it can include punishment. Feedback is often a euphemism for punishment. A bad parent might defend their criticism and abuse by saying, "I was only giving the child feedback for their behavior." Shaping almost always relies on the use of positive reinforcement, and so it can be distinguished on those grounds quite easily.

Feedback, in common usage, is commonly commingled with notions of reinforcement through successive approximation or "shaping".

Shaping is a technical term that refers to the creation of (often) complex repertoires through a series of intermediate steps.

As noted above it typically relies on positive reinforcement. It uses small steps and intermediate goals towards a more complex goal.

See JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2004, 82, 311–316 NUMBER 3 (NOVEMBER) THE ETYMOLOGY OF BASIC CONCEPTS IN THE EXPERIMENTAL ANALYSIS OF BEHAVIOR by JAMES A. DINSMOOR http://seab.envmed.rochester.edu/jeab/articles/2004/jeab-82-03-0311.pdf

Florkle 07:48, 20 December 2006 (UTC)Florkle

Yeah, I think that makes the most sense. Feedback is definitely an alternative term for "consequences." I remember back when I was adding to the article on Operant Conditioning that I tended to use the term "feedback" and "consequences" interchangeably in order to use more familiar terminology. I'm going to look over what's been said and try to change that paragraph along those lines. If it sounds redundant we can always remove it later. Lunar Spectrum | Talk 03:27, 21 December 2006 (UTC)

[edit] Request for expert

Someone who knows more about learning theory needs to fix this article. I was trying to find a good definition of shaping, and although I don't know what exactly it is, I DO know that this article is written with an editorial slant towards positive reinforcement. Also, the author appears to use negative reinforcement and punishment interchangeably, which is incorrect, as I certainly hope an article on operant conditioning would clear up.

Although lots of people advocate positive reinforcement only training methods, it does not necessarily produce the best results, and positive does not mean good. Punishing a dog for digging into the trash (say, with a can full of pennies poised to fall and make a scary noise) will generally stop the trash-digging behavior much faster than say, providing a treat every time the dog ignores the trash. Beating a dog to make it more likely to fight can be considered positive reinforcement. Positive reinforcement = providing something which increases behavior. Negative reinforcement = taking something away, which increases behavior. Positive punishment = providing something which decreases behavior. Negative punishment = taking something away, which decreases behavior.

If anyone out there can explain shaping better and rewrite this article, it would be much appreciated!

Agreed. Positive and negative reinforcement have been updated to read as reinforcement and punishment as that seems to be the terminology in the Operant conditioning article. Also, I tagged it for an expert.--Serf 16:32, 9 December 2005 (UTC)

I'm certainly no expert, but I've taken a a few college-level courses on Psychology and have a pretty firm grounding in learning behaviour. That said, I think the article is more or less factually accurate, but is definitely biased against the use of punishment--the use of punishment in shaping is still subject to debate, but when administered correctly can speed up the shaping and learning process. Mance 01:34, 19 December 2005 (UTC)

[edit] Request for expert

While I know being a psychology student doesn't necessarily qualify me as an expert, I felt I knew enough about the topic of shaping to give a better view for the Wikipedia Community. I decided to pretty much scrap the whole old article, since it was pretty baised. There's not a whole lot of research into which type of reinforcement or punishment is most effective in shaping. Also, I think the first author means to say "positive punishment" when he says "negative reinforcement". It's a common mistake to do that.

I feel like this is a good summary of shaping, with the exception of one topic, Autoshaping. This is a very small detail in psychology, involving ONLY pigeons, so I left it out for the timebeing. I will go back later and add it when I know a little bit more about the topic, but feel it will be better as a "see also" at the bottom, instead of inclusion in the actual article.

My M. A. in psychology is 35 years old, but what I was taught is that a game like Hot and Cold is an example of feedback rather than shaping. For one thing, the use of the words hot and cold does not increase the response strength of writing your name on the board. That is, the person who has supposedly been reinforced for writing his name on the board is no more likely after the game is over to write his name on a chalkboard. The words hot and cold are information, not reinforcers. The section about applications is far too sketchy. Maybe I'll go look something up and see what can be done. John FitzGerald 15:37, 21 May 2006 (UTC)

[edit] Disputed factual accuracy

I added the disputed factual accuracy tag for the reason given in the previous post – I don't believe the example provided is an example of successive approximation, because it does not describe a process which affects response strength. Some day I'll have the time to work on this article, not only providing a better example but fleshing out the applications section, but until then this tag will have to suffice. John FitzGerald 13:46, 28 May 2006 (UTC)

I put in an actual example of shaping and a discussion of how it differs from feedback. It's been a long time since I trained rats, though, so improvements may be necessary. At any rate, the example i now an actual example of shaping. John FitzGerald 14:01, 29 May 2006 (UTC)

[edit] Shaping is not the only way to change behavior

I am no expert either, but in response to the comment that perhaps punishment will change the behavior of a dog who is digging into the trash, shaping is the process of using successive approximations which is rewarding each behavior that is closer to the target, or desired, behavior. It in no way involves the use of punishment such as shaking a can of pennies. This does not mean, however, that shaping is the only way to change behavior. Nor does it mean that punishment is not an equally effective means of changing behavior in the appropriate circumstances. The definition here is merely that - a definition of what shaping is. It involves successive approximations and rewards to change behavior. That's it.

[edit] Move?

Shouldn't this be called shaping? It's shorter and pretty much as precise. WLU 19:08, 27 February 2007 (UTC)

And more common. John FitzGerald 14:48, 28 February 2007 (UTC)
I also agree that the name of the article should be "Shaping." The phrase "reinforcing successive approximations" is the definition of shaping and has no business being the title of an article. As for moving the article, there is no other article about shaping, so the only other place to put it would be in the Operant Conditioning article ("shaping" in search results in a disambiguation page). I would personally like to see this article have a better chance at being fleshed out before moving it, though. I would like to think that it merits its own article, just like Reinforcement currently does. But in the end it might have to be moved anyway. Lunar Spectrum | Talk 01:25, 14 March 2007 (UTC)

I'd say it's far from a stub, I think it could be moved as is, with a bit of wording changes. I'd move it to Shaping (psychology). WLU 12:15, 14 March 2007 (UTC)

[edit] merge with reinforcement

Could somebody enlighten me on -- Does shaping occur outside the context of reinforcement to have the merit of an article on its own. Looking at the current title, i think that these two articles could be merged. Kpmiyapuram 14:35, 27 April 2007 (UTC)

Reinforcement is a prety long article already. I think the move to Shaping (psychology) would be a better plan, as shaping encompasses RSA along with other things (breaking down tasks into steps, and that's about all I can think of). Shaping could probably be its own article, and we might be able to move Chaining into that article too. WLU 16:10, 27 April 2007 (UTC)
I have renamed the article. As there have been suggestions earlier on this talk page, the article probably needs to be refined. Thanks. Kpmiyapuram 17:13, 30 April 2007 (UTC)