Changing Behavior with Operant Conditioning

[Zarkow]

The best way to get the puppy to learn to roll by itself is to bring the candy over its neck so it turns its head and will push itself out of balance. It works surprisingly well.

[KBW2010]

This was amazing. Well done.

[michelamarie100]

too cute!!

[max1morgan2]

How long did it take to teach him to rollover like this? He looks really small in the beginning of the video and then looks much bigger at the end when he rolls over.

[knappsych]

@max1morgan2 According to the dates on the video a month and 10 days passed from the first day to the last demonstration.

[Miikaika25]

you were messing things up after you rolled him over you should give him the treat regardless that way he can be conditioned. I taught my dog all those tricks except the roll over part when she was a puppy, she never learned roll over because I was not very persistent with it, I basically wanted her to know how to sit and stay and come. We have to practice it every once in a while to keep her on her toes.

[knappsych]

@Miikaika25 I agree, my student should have given Jack a treat after each roll, prompted or not, during training to speed learning. Thanks.

[kamanakai09]

This is really great. I got a smaller dog,  I want him to learn how to roll over but hes a savage and will just stare at the treat in a trance. Maybe later in the future if i get a new dog i'll train it when its younger

[JaimeRBCBA]

The SD is delivered AFTER the behavior occurs for some of the learned behaviors; the dog is emitting a chain of previously reinforced behaviors without the SD delivered and the behavior chain is reinforced. Definitley deliver one SD, (then prompt for the specific behavior without more talking, if needed), and reinforce the behavior (or approximation). Yikes, this is a messay discrete trail. Something to learn from, certainly. Glad it's with a doggie and not a human!

[littlelunatuna]

Your student needs to learn to give the command ONCE and only once. Did you notice how the dog was being rewarded for waiting until the student repeated the command (at least) four times?And how many times the dog refused to roll over until the command was given at least four times? And how long it took the trainer to begin giving better cues?

[Gogoleila]

thanks for the video... I was thrilled when he finally rolled over...

[futuredr2014]

Horrible Horrible Owner, that's what she is. Forcing the dog into a command like that is ridiculous. This isn't shaping at all. The only conditioning here is, that I learn to roll over when she says "roll over" 4-5 times (seriously, count them) and then do it as to avoid being forced to roll over.

[Bonyarms28]

she forgot to tell the dog it was doing good when it obeyed her command which would be even more positive reinforcement. Oh well, we get the jest of it. Such a cute puppy!

[ShaylaDanielleMusic]

what kind of dog is that?

[knappsych]

@ShaylaDanielleMusic Unfortunately, I don't know.

[marco00ve]

HE DID IT!!! great

[Lottiesdogs]

Well I have NEVER seen a roll over taught like that!! I certainly wouldn't advertise that I train it that way either...

Why would you? When you can shape it in about 5 minutes??

[knappsych]

@Lottiesdogs You can shape that in 5 minutes? I'd love to see a video. Please post a video response.

[thistlemiss]

If you look at the video I posted below ("Zoë Free Shaped Roll Over Session 5), you'll see what a truly free-shaped roll over looks like.

[mufflein2201]

without any great psychological background (only senoir years biology) and just knowledge because of having dogs by myself: great job! great job jack!

[jylnbair]

out of context on the video. what breed of dog is Jack?

[knappsych]

@jylnbair Sorry, I don't know.

[samsmoti]

so cute when it comes to the camera!!! *pokes*

[samsmoti]

what she should have done was move the treat close to the puppy's head and move it so that it moves is head then moves the rest of his body to rollover, much more effective... at least with several dogs i know

[knappsych]

@samsmoti Excellent tip, thank you.

[amandaleeXx]

hey professor ploski!

[sueontube]

It would have made a lot more sense for your student to say 'roll over' a lot slower and clearer and also to guide the puppy whilst saying roll over every time without exception.

[ThoughtCriminal912]

cute dog. great vid

[catalin30]

knappsych, It is not necessary to be trapped in a single method of training. It's important to have the possibility to make combinations of methods, to get the result that you want in practicing with your dog.

If you don't harm your dog, all the combinations are great. You puppy made a cute and fast roll over in the end... that's important.

[k9sue]

@catalin30 You have to be very careful when mixing methods with dogs. Consistency is sooo important. When you mix methods, many times they create confusion. Shaping does not work well in combination with Luring" or prompting. A dog that is lured or promoted often will not offer behaviors to shape. A dog trained with negative reinforcement or positive punishment will also be reluctant to give shaping opportunities.

[knappsych]

@catalin30 Sorry if the purpose of this video was misconstrued. This isn't supposed to be the best method of teaching a dog to do a trick, only an application of learning principles for an extra credit project. @samsmoti proposes an alternative that seems likely to produce quicker results without resorting to prompting the dog as my student did.

[JFelts77]

the dog is too cute!

[appsychology]

this is actaullu a great example of chaining behaviors!!!! Great job!!!

[fiqichiban]

can anyone explain the differences between classical and operant conditioning...

please!!!

[Onebluemew]

In brief explanation: Classical Conditioning deals with stimulants and responses. Stimulants naturally produce the want response. Example: foods aroma (stimulant) naturally makes your stomach growl (response). Yet, it classical conditioning a neutral stimulus is added to the stimulant to produce the response. So, if foods aroma is the stimulant, while the growling of the stomach is the response, let us say that a horn is the neutral stimulus. . Note: the neutral

[Onebluemew]

stimulus, (NS), does not naturally produce the wanted response. In other words a horn wont naturally make your stomach growl. But with conditioning (aka learning) it will and can. The experiments always begin with the horn being blow, next the foods aroma being exposed to the

air, and lastly your stomach growling (remember, this is the first time you were introduced to the horn and food aroma being teamed together ). Now (since it is the first time) your stomach is

[Onebluemew]

growling from the food not the horn. The second part of the experiment is pairing the horn with the foods aroma (occurs the second time (of the pairing) to as many time as you want to pair them in the experiment) several time to make your stomach growl. Now stomach is learning (conditioning itself) to understand that whenever the horn is heard food aromas will come and therefore make you stomach growl. Making the last part of the experiment: the horn

[Onebluemew]

(alone/without the aroma of food) making your stomach growl.

Conditioning is another way of saying learning. So whenever you see the word, just think learning. And the responses are all natural, like reflexes (involuntary). Now in brief: Operant Conditioning deals with consequences. Unlike Classical Conditioning, operant is very voluntary

[Onebluemew]

(you control your behavior based on the outcome). Note: consequences are not the same as

punishments. Consequence can be good and bad. Example: you get all As, the consequence of

getting all As is your parents buy you a car (positive plus positive). The operant (behavior influence) is that you will get all As again because you want more positive consequences. Next

[Onebluemew]

example: you get all Fs the consequence is that your parents take away the car (positive plus negative). Bet you wont get all Fs again. Nope, because the consequence is negative, which mean you will try to avoid them. Hope this helps! Go to (wiley. com /college / huffman ) Psychology in Action: In Modules, 8th Edition. Then click on the Student Companion Site. Next go to CyberPsych Animations. Lastly chapter 6 classical conditioning.

[knappsych]

Thanks Onebluemew!

[KatrynMerteuil]

I'm actually studying for a test now. lol

Classical are the reflexive and automatic responses that you get when presented a stimulus (i.e: a dog drools when he is about to be fed)

Operant is mostly the behavior one does to obtain a response (I.e: Dog in video- treat after trick- recompense)

[knappsych]

@KatrynMerteuil Nice summary, thanks.

[JaimeRBCBA]

@fiqichiban

Classical: neutral stimulus (bell) paired with a automatic reinforcer (meat powder) makes bell a conditioned stimulus (dog drools when bell is rung vs when meat powder is presented). Operant: a behavior is followed by a positive (reinforceing) or negative (punishing) stimulus to increase (reinforcement) or decrease (punish) a behavior. Classical is respondent, Operant is learned. Look up Applied Behavior Analysis.

[knappsych]

@JaimeRBCBA I try to avoid using the words positive or negative when referring to stimuli because these words are used by psychologists to describe what happens with the stimuli, regardless of their valence. I like to think about the terms in the correlational sense.

PR: more behavior = more appetitive stim

NR: more behavior = less aversive stim

PP: more behavior = more aversive stim

NP: more behavior = less appetitive stim

[Bek21]

such a cute boston terrior :)!!!

i have one too =)

[TherealAshman]

"1. I disagree with your first claim. Although shaping isn't regularly discussed with manual intervention, the behavior is being shaped.

The puppy does nothing."

perhaps you should check what the definition of shaping is. a dog that does nothing and is being forced into a position is not learning by shaping, he is learning by force.

in a shaping session the handler should not touch or even prompt a behavior from the dog, its all about the dogs choice not yours.

[knappsych]

I agree that I needed to check what the definition of shaping is. Thanks to the controversy this video stirred up, I learned something. Using the physical manipulation is not known as shaping, it's prompting. A common example of prompting is telling a dog to sit and then pushing it's butt down.

So you're right, this wasn't a good demonstration of shaping (I changed the title). It is a good demonstration of prompting to facilitate instrumental conditioning.

[TherealAshman]

the dog learned that to avoid being rolled over by you he has to do it himself, thats not shaping, thats classic positive punishment (roll over or else i will do it for you) the fact that there is no pain involved like there would be in say using a choke chain, doesnt make it positive reinforcement and/or shaping by any means. its still positive punishment, training using corrections.

more so, in shaping you do not introduce the cue before the behavior is learned and you never repeat your cue

[RexalQel]

That's right.

[knappsych]

That's not right. If it was just trying to avoid an aversive stimulus, the dog could have ran away.

In shaping behavior, the cue can be critical. You don't just want your dog to roll over all the time, you want your dog to roll over when you say roll over. Educators don't want kids raising their hands on the playground with their friends, but they do in the classroom.

[Candorsmayhem]

@knappsych Not true. You simply don't introduce a cue until the behavior is at it's peak, so the cue isn't confused with what the behavior was before your bar was raised. If this were true shaping, you might click and reinforce for each increment, 1 might be shifting weight to the side, then laying on his side, laying on his back, etc. You increase the criteria necessary to give the treat as the dog demonstrates understanding of what is being desired (IE he repeats the action)

[Candorsmayhem]

@knappsych (continued) Then when he does the finished behavior, roll over, you say the cue AS he is doing the action, approximately 30 times. Then you move the cue to immediately before he was going to do the behavior anyway, for another 30 reps. Then you move the cue to being an actual cue. What you have here is an example of operant conditioning, as all training is, but this isn't positive reinforcement, which makes use of rewarding a behavior, and ignoring the bad. (IE you wouldn't say 'no')

[knappsych]

@Candorsmayhem: You are correct that this isn't an example of shaping (see my earlier responses and the video description). As long as the "cue" reliably predicts reinforcement in a reasonable time frame, many organisms, including humans, rats, pigeons, and dogs can pick up on the conditional relationship between the cue and reinforcement. I'm confused as to why you state that this isn't positive reinforcement. What is your alternative interpretation?

[Candorsmayhem]

@knappsych This is an example of negative reinforcement. The dog does the action, and when he does, the negative stimuli of being pushed into position is removed. Yes, he is then rewarded, but it's not for doing anything. He's simply being rewarded for being in a particular position. It's like when you scold a dog for peeing on the floor afterwards. They don't see that as "I peed on the floor, I'm bad." They see it as "Pee on the floor is bad." Big difference.

[execratexmexbaby]

@TherealAshman Actually, if you wanted to split hairs, the situation that you're describing is incorrect. From that perspective, this would be an example of negative reinforcement (otherwise, the puppy would STOP behaving, not behave at higher rates).

I'm inclined to agree with knappsych: this procedure closely resembles shaping by successive approximations, or may better be described as a two-step prompting procedure (i.e., 1)"Roll over" 2) physical guide).

[knappsych]

@execratexmexbaby Thanks for the backup!

[catalin30]

Hi there! This pup is lovely! If u don't mind, I will suggest to you to shape behavior not saying words at first. Let him to concentrate to the luring and the move, not t your voice. Your voice distracting him.

[knappsych]

Maybe one of my students will try this next semester without using a verbal command.

[StacyBS]

Wow, it looks like this puppy was taught to rollover through negative reinforcement - he learned to roll onto his side to avoid being pushed over by his handler. When rolling onto his side is not enough to avoid the pushing, he goes further. At the end he recognizes that "handler waving a treat in a circle" indicates that pushing-over is about to occur, so he does a roll-over to avoid it. Very nice example of OC with R-!

[knappsych]

If the push was a negative reinforcement, the dog should have just ran away.

[JaimeRBCBA]

@knappsych Ouch. Check your definitions. Negative reinforcement causes a behavior to increase in attempt to avoid an aversive stimulus. Neg. Reinf. is not punishment!!! If the doggie was attempting to avoid the prompt (which may or may not be aversive- although, doesn't seem so), he would roll over faster. In punishment, you would see a behavior reduced. As in, a nonpreferred or aversive consequence would be applied to follow a certain behavior to reduce the behavior (positive punishment).

[knappsych]

@JaimeRBCBA Perhaps I should have been more clear with my statement about negative reinforcement. You are absolutely correct that negative reinforcement isn't punishment. @StacyBS suggested that Jack learned through negative reinforcement.  Instead of saying "If the push was a negative reinforcement" when I should have said "If escaping the push was negatively reinforcing." Sorry.

[1LBRTY]

At day 2 trial 2, the dog's half successful rollover should have been rewarded. You'd be surprised how quickly the dog would start half rolling. And then, continue to shape the behavior towards a full roll.

If you catch them when they are thinking about it, and just beginning the behavior, they will link cause and effect.

"When I do this, I get that..."

[1LBRTY]

Also, dogs are more responsive to hand signals than voice commands. You can train more quickly by limiting commands to only hand signals. Later, link up verbal commands.

[knappsych]

I agree with your insight to reward the dog when it halfway rolled over. That would have been a better example of using shaping to affect behavior.

[aes005]

Ima steal your dog. so cute

[aes005]

If no food is presented to the dog after awhile the behavior will become extinct.  If the pup was put on a intermittent schedule of reinforcement it would be less prone to extinction.... Thats my guess anyways.

[knappsych]

Yep. It can be useful to use a continuous ratio schedule reinforcement to establish the behavior and then shift to a variable ratio schedule to reduce the likelihood of extinction. Of course when shifting to a variable schedule you don't want to make the schedule too different or the organism may experience ratio strain.

[aes005]

haha

"no animals were harmed in the making of this video."

Classical conditioning involves REFLEXES. Reflexes are involuntary and occur at a 100% probability when paired with 'unconditioned' stimulus. Thus there less adaptable. Rolling over is not a reflex.

Operant behaviors occur at less than 100 probability. They happen more or less frequently based on the consequence that follows shortly after.

It has nothing to do with the sequence of events. But of the mechanisms at work.

[knappsych]

Thanks for mentioning the mechanisms.  I had forgotten to mention the reflexive nature of classical conditioning in my earlier responses.

[lawsofeffect]

Just because food is used as a reward does not make this Operant conditioning. This is more a demonstration of Classical conditioning.

[knappsych]

I currently fail to see how this is a better demonstration of CC than of OC. Is her voice the CS and the treat the US? If that's it, then Jack shouldn't come to make the instrumental response of rolling over. S="Roll Over;" R=Rolling Over; O=Treat.

[lawsofeffect]

Operant conditioning = behavior - reward - cue

Classical conditioning = cue - behavior - reward

My statement 'more' was referring to the fact that even in Classical conditioning the marker (your cue) would be neutral.

[knappsych]

Classical (during conditioning): cue - reward - behavior

Classical (post conditioning): cue - behavior (even without the reward)

Operant: cue - behavior - reward

The nature of the behavior is quite different in CC and OC. With CC the behavior is produced in response to the reward or the expectation of the reward. Furthermore, the behavior isn't instrumental in leading to the reward. With OC the behavior in response to the cue leads to the reward.

[lawsofeffect]

This video does not as you claim:

1)show shaping of behavior

2)discuss reward schedule (or timing of)

3)identify what is serving as reward

4)identify what the cue for the behavior is.

All of these would be necessary to represent this video as shaping using OC.

We could agree to disagreeor you could simply have your handler stand up, turn their back to their pup, request their pup to roll over (in a normal tone). Do you honestly think that smart little pup will roll over now?

[knappsych]

1. I disagree with your first claim. Although shaping isn't regularly discussed with manual intervention, the behavior is being shaped.

The puppy does nothing.

After some shaping, the puppy rolls halfway over.

At the end of shaping, the puppy rolls all the way over.

2. You're right, schedules of reinforcement weren't discussed. She started with a continuous schedule and moved to what looked like a variable ratio schedule.

3. She identified the treats as a reward at the end of the video.

[knappsych]

4. The cue is a compound stimulus involving the statement "roll over" and the circular hand motion.

I'm not sure what the pup would do if my student turned their back on the pup and gave the command to roll over. If that's the desired behavior, it could easily be shaped with some generalization training.

FYI: the student is not my handler. I had nothing to do with this video other than creating the assignment and posting the end product.

P.S. I like your favorites list.

[Nades129]

jack is soooo cute

[FobSquat07]

ur using ur hands to push him to roll over. thats not really naturally conditioning him

[Lokussapiens]

Of course not. Prompting?

[cwagner87]

Nice video! May I recommend trying to use a clicker? this "roll over" is very much prompted by physical interaction, and the behavior could have been acquired within a single day had a marker been used. Which part of the training was operational? I understand you were using positive reinforcement, but it seemed more like classical conditioning since the cue was added before the behavior was taught. Maybe adding a marker and adding the cue after the behavior is predictable? Good Stuff!

[knappsych]

Thanks for your comments. Your suggestion to use the clicker as a marker is right on. A marker can make it much easier for an animal to identify exactly what behavior is being rewarded. With classical conditioning, a stimulus follows another stimulus (e.g. a treat follows a click; like you do in "Training the Retrieve"). With operant conditioning, a stimulus follows a behavior (e.g. treat following a roll). Of course CC and OC can come together (e.g. CC[command-push-roll]/OC[roll-­reward]).