Questions re:clicker training- order of cues, and luring vs capturing vs physical cue

Go Back   Dog Forum > Keeping and Caring for Dogs > Dog Training and Behavior

Questions re:clicker training- order of cues, and luring vs capturing vs physical cue

This is a discussion on Questions re:clicker training- order of cues, and luring vs capturing vs physical cue within the Dog Training and Behavior forums, part of the Keeping and Caring for Dogs category; ...

User Tag List

Like Tree10Likes

 
LinkBack Thread Tools Display Modes
Old 10-23-2017, 01:39 PM
  #1
Junior Member
 
Join Date: Sep 2017
Posts: 7
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Questions re:clicker training- order of cues, and luring vs capturing vs physical cue

So I have recently started basic obedience classes with my 2 year old standard poodle, Ellie. I have signed up for two different courses from different trainers (eight sessions each, they’re both running at the same time), as well as practicing at home but I feel like I am missing something.

I am familiar with the concept of clicker training (the timing/immediately marking behaviors as correct, conditioning the dog to expect a reward every time they hear the clicker, etc) but I am having trouble understanding a few points. Let’s use training a ‘sit’ as an example in the below questions. My actual questions are bolded in each point and I apologize for the length.

I am confused as to when you introduce a verbal cue. Some places say that you should say ‘sit’ right as the dog is sitting down, right before the click so that they start to associate the two (Karen Pryor, for example). But a LOT of other sources say that you do not want to introduce a verbal command and label the behavior until they the behavior is ‘strong enough’ or until they are ‘reliably offering the behavior'. Still other sources recommend teaching a hand-signal before the verbal cue, but they still say to wait to introduce any sort of signal until the dog is ‘reliably doing the behavior’.

1. Then at what point do I introduce a hand or verbal signal? If you wait until the behavior is ‘strong enough’, how do you classify this?

2. What if I am training both sit and down and neither have a cue yet? If she offers a down or a sit, do I reward both? Is the goal here to get her to offer up the behavior on her own because she knows she will get a treat?


To get the behavior in the first place, Karen Pryor differentiates between luring (treat over the head, causing the dog to naturally sit to get at it) and capturing (waiting until the dog sits and rewarding that). I understand how both of these work but it seems like only luring or a physical cue (poking her butt lightly) are conducive to practicing 10x in a row. But there are a lot of sources that discourage luring because it is hard for some dogs to focus on training instead of purely following the food.

3. So what would a capture-only method of teaching sit look like? Pick up treat, get my dog’s attention, and stare at her until she sits (then click and reward) and repeat?

And the final question is something that I have not been able to find any reliable details about and that is how do you train a behavior like going potty on command? There are tons of articles but the details don’t make sense. It seems like for this type of behavior, you *must* say the cue while the dog is doing her business. It doesn’t make any sense otherwise (how would a dog reliably offer up a potty, hoping for a treat, if there is no indication you are waiting for them to do something).

4. So the final question is how would you train something like potty on command, given all the discussion above? Can or should you use two types of training for different behaviors (like no early verbal commands while using capturing for a sit versus pairing the cue “go potty” immediately when the dog does her business from the very beginning)?

I know this is probably much more detail than I need to think about when starting out training but so far Ellie seems confused and I want to make sure I am doing clicker training correctly. Thank you for any and all advice
Silvestris is offline   Reply With Quote
Old 10-23-2017, 01:58 PM
  #2
Senior Member
 
Artdog's Avatar
 
Join Date: Nov 2015
Location: Stouffville, Ontario, Canada, Earth, Milky Way Galaxy
Posts: 883
Mentioned: 84 Post(s)
Tagged: 1 Thread(s)
Ooooh a whole string of dog nerdy questions that I've always wanted to ask but been afraid to. I have never purely 'clicker trained' (as in shaping only, no luring) and in fact use food luring games to motivate my dog, so I am seriously pumped to hear the answers too (preferably with a coffee in hand) and am glad you asked.
@Shandula I think this one is for you. These questions regard the pure style of clicker training, right? Not just using a clicker as a reward marker.
Artdog is offline   Reply With Quote
Old 10-23-2017, 02:11 PM
  #3
Dog Forum ModeraTHOR
 
kmes's Avatar
 
Join Date: Dec 2009
Posts: 11,909
Mentioned: 486 Post(s)
Tagged: 0 Thread(s)
Quote:
Originally Posted by Silvestris View Post
1. Then at what point do I introduce a hand or verbal signal? If you wait until the behavior is ‘strong enough’, how do you classify this?
When depends on the method.

If luring, you'll have a hand signal as soon as you switch from lure to reward. You can make it more subtle over time or just keep it as is.

If shaping or capturing, you need to wait until the behavior is happening predictably and the way you want. Then you can add the cue by giving it just before your dog does the behavior. Click the behavior then treat.

If wanting to change a cue (perhaps switching from a hand signal for sit or down to a verbal) it's:
New cue (Verbal), then old or current cue (hand signal/body cue), then click the behavior, then treat.

Quote:
2. What if I am training both sit and down and neither have a cue yet? If she offers a down or a sit, do I reward both? Is the goal here to get her to offer up the behavior on her own because she knows she will get a treat?
Quote:

To get the behavior in the first place, Karen Pryor differentiates between luring (treat over the head, causing the dog to naturally sit to get at it) and capturing (waiting until the dog sits and rewarding that). I understand how both of these work but it seems like only luring or a physical cue (poking her butt lightly) are conducive to practicing 10x in a row. But there are a lot of sources that discourage luring because it is hard for some dogs to focus on training instead of purely following the food.
[/quote]

Train only one new behavior in a session.

Honestly, luring sits and downs is going to be by far the easiest way for most people to train those behaviors.

There are a lot of ''shaping purists'' out there... while I absolutely prefer shaping, capturing and targeting with my personal dogs, it is often difficult for the average person and especially green dogs to train this way. We focus on lure to reward for most behaviors in my beginner and pet dog classes. Rarely is there ever an issue with a dog learning this way. And if there is, it's typically an issue with reward placement, not fading the lure, or owner getting out a treat to lure when a dog doesn't respond (this is after switching to reward and having at least hand signal).

I just wouldn't worry about it. I don't like to eliminate such a useful method (luring shows complete behavior, how to move body, can get behaviors very quickly, transitions over to hand signals super quickly, etc.) Luring isn't incompatible with clicker trainer. It is a reward based method. Clickers are redundant initially when actually luring, but a marker is very useful as you transition to hand signal/reward (hand signal with empty hand, dog responds and you click the behavior, then reward). If luring works for you, if it's a good option for training the behavior, use it.


Quote:
3. So what would a capture-only method of teaching sit look like? Pick up treat, get my dog’s attention, and stare at her until she sits (then click and reward) and repeat?

This is down, but capturing a sit would be along these lines.

This was a quick video I shot a long time ago for a different purpose, but it might help you out here.

This guy is clicker savvy. He understands his actions cause the click and tend to repeat clicked behaviors right away, or tries to figure what what caused the click. He was just starting his folding down. Normally I would have waited until he was speedier and more confident in response before adding the verbal but again this was to show someone else this down exercise.

A huge part of successfully using capturing in training sessions, is actually rewarding in such a way as to reset for another repetition.

Here I rewarded not only in position (wanted the emphasis on the down to build speed and enthusiasm) but also to get him up onto a stand. There are times when capturing that I give just a single reward in such a way as to reset for another repetition. I could have used a hand target to reset this dog in a stand for another down however the person I originally videoed for hadn't taught a hand target yet.

Quote:
4. So the final question is how would you train something like potty on command, given all the discussion above? Can or should you use two types of training for different behaviors (like no early verbal commands while using capturing for a sit versus pairing the cue “go potty” immediately when the dog does her business from the very beginning)?


With many dogs you can see that they need to potty via their body language. I tend to just say ''go pee'' or whatever as they are finding a place. Click or verbal marker as they finish. Then reward.

Eventually they tend to get the idea. It just tales longer as it's not normally a behavior they can repeat such as a sit, spin, etc.
Artdog likes this.

Last edited by kmes; 10-23-2017 at 06:18 PM.
kmes is offline   Reply With Quote
 
Old 10-23-2017, 02:55 PM
  #4
Senior Member
 
Artdog's Avatar
 
Join Date: Nov 2015
Location: Stouffville, Ontario, Canada, Earth, Milky Way Galaxy
Posts: 883
Mentioned: 84 Post(s)
Tagged: 1 Thread(s)
Quote:
Originally Posted by kmes View Post
Train only one new behavior in a session.

it is often difficult for the average person and especially green dogs to train this way
Thanks for that. I needed the 1st line, I'm over ambitious, now I know one way I confuse my dog.

The 2nd line--very important about 'green' dogs, especially rescue dogs may find pure shaping sessions stressful & confusing. I didn't know what a 'shut down' dog was until I did a lousy shaping session with my guy when he was new to me. It was only later that I found out 101 Things to do with a Box was not always an easy game to play for a dog that had never done shaping before (<--yeah, me, over ambitious tiger mom syndrome again).

Looking forward to watching the video when I get home.

PS, I did the quote thing wrong, sorry, will italic for clarity.
Shandula likes this.

Last edited by Shandula; 10-23-2017 at 03:00 PM. Reason: me mixed up whoopsie
Artdog is offline   Reply With Quote
Old 10-23-2017, 05:59 PM
  #5
Member
 
Join Date: Oct 2017
Location: Lancashire, England
Posts: 63
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Taking the first part of your message - "1. Then at what point do I introduce a hand or verbal signal? If you wait until the behavior is ‘strong enough’, how do you classify this?".

Dogs don't understand English, they can, however, learn that a particular sound has, in the past, been associated with a particular action and, in that way, they can "learn" words.

Try this.

Get a willing human volunteer (for these purposes, let's call him Joe).
Get a load of dollar bills/pound coins (dependant on location). You'll need a lot.
Get a clicker and a piece of paper and a pen.
Write on your paper an instruction and make it fairly precise i.e., "sit on the blue chair". Write the word "ulem" (Joe doesn't see the paper).

Say nothing.

Condition Joe to the clicker. Stand close to him, he doesn't have to do anything, all you have to do is click and hand him some money.

After 4/5 times, Joe should get the general idea.

Bring him into "the room" and get him to stand in the middle.

Without speaking, without touching him, and using hand signals only get Joe to move across the room to a particular place; get him to do the action that you have written down.

Taking "sit on the blue chair" as the example here, you could walk backwards in the direction of the chair, beckoning him towards you. When you are at the chair, you could indicate that you want Joe to move toward the chair, turn around so that he's backing into it and by a hand signal, indicate that you want him to sit THERE.

The moment Joe's backside hits the chair, click and pay him.

Still not speaking, bring Joe back to the middle of the room and do it again, click and pay him. When you do it the third time, most Joes will go straight to the chair and sit on it, so, click and pay him. If not, repeat until he will go with a simple hand signal in the direction of the chair then, click and pay him.

NOW HE KNOWS THE ACTION YOU WANT, you can introduce your verbal cue.

Back to the middle of the room, say "ulem" and indicate with your hand signal what you want. He should go to the chair based on your hand signal. Repeat and gradually minimize the hand signal. Click and pay each time Joe's backside hits the chair.

"Ulem" is Albanian for "sit". As long as Joe's not Albanian, you have reproduced the process that your dog undergoes. Any word will do, repetition and reinforcement will make it stick.

Bear in mind that Joe is probably an adult and will grasp this quicker than your dog (equivalent to a 2/3yr old child) with fewer repetitions.

Later in the day, take Joe into the same room again. As he walks in, say "Ulem", he'll probably sit in the blue chair!

"2. What if I am training both sit and down and neither have a cue yet? If she offers a down or a sit, do I reward both? Is the goal here to get her to offer up the behavior on her own because she knows she will get a treat?"

If she offers what you want, click and pay her. Then get her up and move her around, stop, stand still and see what she offers. No verbal cue from you. She will offer what is easiest FOR HER (I had a collie from a rescue, his default position at start was to lie down).

Work with what she's offering, click and reward. Get her up and move her around, stop, stand still. With repetition, she will offer the same behaviour. and then, as with Joe (above), NOW SHE KNOWS THE ACTION YOU WANT, you can introduce your verbal cue.

"3. So what would a capture-only method of teaching sit look like? Pick up treat, get my dog’s attention, and stare at her until she sits (then click and reward) and repeat?"

Don't stare at her please - very rude in dog behaviour.

Forget Joe for the moment, he's watching Netflix. Take Mary into "the room" (she's conditioned to the clicker) and stand and wait. She'll probably get fed up waiting for you to say something and she'll probably sit, somewhere. It's close to what you eventually want, so you can click and reward, then bring her back to the middle of the room. Stand still and wait. When she understands the action you want, you can introduce a verbal cue.

"4. So the final question is how would you train something like potty on command, given all the discussion above? Can or should you use two types of training for different behaviors (like no early verbal commands while using capturing for a sit versus pairing the cue “go potty” immediately when the dog does her business from the very beginning)?"

Here, you are looking at capturing, i.e., if you can see that she's going to put her backside on the floor, you can say "sit" and as soon as her backside actually hits the floor, click and pay. If you are looking at toilet training, decide on your cue.

For my dogs, I train two ways. "Murphy toilet" for pee, "Toilet Murphy" for poop.

You KNOW that she's going to the toilet and, generally, she will pee first. As she starts to squat, verbal cue. Let her continue. AS SOON AS SHE'S FINISHED. click and pay her WHERE SHE IS (makes it obvious that that's what you are rewarding).

Then she'll wander around and go to poop so, as above, as she starts to squat, verbal cue. Let her continue. AS SOON AS SHE'S FINISHED. click and pay her WHERE SHE IS.

Makes life easy when you stop for a break in the car, I get out, my dogs go to the toilet, they go back in the car, then I go.

Your dog will most likely give you the answer as to the best way to train, i.e., what works for her. Use what produces the best result for you and her. Each dog is individual.

I've probably missed bits here and there and there are lots of others who can answer your questions, so feel free to ask for clarification.
Ptolemy82 is offline   Reply With Quote
Old 10-23-2017, 10:31 PM
  #6
Senior Member
 
piano88's Avatar
 
Join Date: Sep 2014
Location: Alaska
Posts: 171
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quote:
Originally Posted by Ptolemy82 View Post
Don't stare at her please - very rude in dog behaviour.
If a dog has been consistently reinforced for eye contact, then prolonged human-dog eye contact while training shouldn't be an issue, correct?
Posted via Mobile Device
DriveDog, Artdog and Pantalanomin like this.
piano88 is offline   Reply With Quote
Old 10-24-2017, 02:51 AM
  #7
Member
 
Join Date: Oct 2017
Location: Lancashire, England
Posts: 63
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
@Silvestris wrote "...get my dog’s attention, and stare at her until she sits...".

"Direct eye contact often signals challenge and confrontation" (Jolanta Benal, CPDT-KA, CBCC-KA, The Dog Trainer), hence my comment.

If his poodle has been trained for eye contact, I would think that a benevolent gaze would be more appropriate or, at least, a more appropriate phrase.
Ptolemy82 is offline   Reply With Quote
Old 10-24-2017, 04:57 AM
  #8
Banned
 
Join Date: Aug 2017
Location: Boston metro-area, USA
Posts: 1,885
Mentioned: 46 Post(s)
Tagged: 0 Thread(s)
Lightbulb One behavior, or ONE variable of one behavior, per session.



in clicker or any marker-training, one thing at a time always applies:

- only ONE criterion is used per session.
U can't work on *both* faster sits / less lag, AND straighter sits; pick one, work on that - then, in another session, work on the other.
If U are working on a nice Sit to Front recall, get an auto-sit 1st; then one at a time, get a straight sit, at a specific distance from yer knees, & last, work on fluency / latency / lag. // Lag-time goes last b/c the dog needs to have a good understanding of what we want B4 worrying about how fast we want it.

- only ONE goal-behavior per session, except when luring position changes - then choose any 2, & alternate.
e-g, sit / stand, stand / down, down / sit, down / stand, stand / sit, etc.
The luring motion morphs into a hand-cue; a verbal tag is applied anytime as a label, & said only AS the action is performed, not after / not before.
Once the dog has done the behavior 5X with a lure, work with an empty hand & the same movement; get rid of the lure ASAP & go empty.


Any hand signal, body-posture, environmental cue, etc, can serve as a bridge to help the dog connect verbal cue [which starts as a label pasted on the action AS it's done] with the action. Dogs are very aware of body movements, facial expressions, context, head tilts, etc, & we have to try to be aware of our own casual movement / settings / poses so we don't introduce something that the dog thinks is salient.
I-e, tilting my head to look at my dog when s/he is in heel position while teaching SIT can get confused as part of the cue for sit, & in future, when the dog is directly in front of me, not to my side, & i DON'T tilt my head... the dog might not Sit on "cue" 'cuz they think "Sit goes with head-tilt".

similarly, dogs are very slow to generalize -
it takes a minimum of 5 different contexts B4 a dog even *begins* to generalize a cue / action pair.
E-g, for a dog, "sit" on long wet grass in the early morning does not equal sit on warm dry carpet in the evening does not equal "sit" on cold enameled steel 3-ft off the floor atop the washer does not equal "sit" on damp firm beach sand does not equal "sit" on upholstered chair in the afternoon with the TV on does not equal "sit" on cool slick lino in the vet's waiting-room with dogs, cats, & a parrot plus multiple humans in the room, does not equal "sit" on the sidewalk in front of the house as a bicyclist goes by.
For English speakers, sit = place buttocks on surface.
For dogs, it's pure noise - they have to grok meaning via many experiences, in multiple contexts.

there's a terrific 1-week free "intro to clicker" lessons by e-mail; unfortunately, their website is under revision. // However, here's an excellent article explaining history & practical aspects -
https://canisclickertraining.com/hva-er-klikkertrening/

hopefully, the once-a-day e-mail lessons will be back, soon!

- terry

leashedForLife is offline   Reply With Quote
Old 10-24-2017, 01:25 PM
  #9
Junior Member
 
Join Date: Sep 2017
Posts: 7
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
@kmes Thank you very much! That definitely helps and I appreciate the level of detail you put in when answering. I think I understand a few things more clearly now.
@Ptolemy82 Another very heartfelt thank you! The comparison to clicker training people is very helpful. I have heard of and have played the 'shaping game' with other people but I have not seen it used to explain what the dog might be thinking or feeling during training (like Mary eventually getting fed up and sitting down somewhere). Also I am guilty of looking at her directly for extended periods of time- I try to look happy and not like I was looming or leering at her. My reasoning was that it was to let her know my attention was on her and I am expecting something but it seems I am committing a doggy faux-pas. Noted and behavior ceased!
@leashedForLife Thank you! The one-at-a-time mantra makes sense. I guess I got confused because some articles talk about capturing a behavior you want when it happens spontaneously (like lying-on-the-side 'play dead' final position) and it almost seems that you can capture whatever you want as its happening. I assume that's not the case? So should I only practice things that have cues already (in separate sessions) and *one* behavior I am trying to eventually trying to link to a cue with at a time? Or can I capture two or more behaviors 'in the wild' as long as I only work on one in a dedicated session? Actually I am still not sure what a training session for a captured and more complex behavior would look like, but maybe I should just focus on the basics for now. And thank you for the link! I will bookmark it and keep an eye on the site to see if it comes back.

This forum is fantastic Thank you to everyone who responded so far and I hope that my questions are also helping other people who may have been similarly stumped.
Silvestris is offline   Reply With Quote
Old 10-24-2017, 02:08 PM
  #10
Banned
 
Join Date: Aug 2017
Location: Boston metro-area, USA
Posts: 1,885
Mentioned: 46 Post(s)
Tagged: 0 Thread(s)
Quote:
Originally Posted by Silvestris View Post

... some articles talk about capturing a behavior you want when it happens spontaneously (like lying-on-the-side 'play dead' final position) and it almost seems that you can capture whatever you want as its happening.

I assume that's not the case?
So should I only practice things that have cues already (in separate sessions) and *one* behavior I am trying to eventually trying to link to a cue with at a time?

Or can I capture two or more behaviors 'in the wild' as long as I only work on one in a dedicated session?

...I am still not sure what a training session for a captured and more complex behavior would look like, but maybe I should just focus on the basics for now. And thank you for the link! I will bookmark it and keep an eye on the site to see if it comes back.


a given session can be really short - a few minutes or less; take a definitive break of some kind, & start a new one.
Ex, practice luring Sit / Stand, sit / stand, for 30 or 45-secs; stop & offer a tug toy - play for a minute or less.
Start the next session, & now lure Spin [clockwise], starting with a head-turn to the right & a leftward wt-shift / right paw comes up, MARK / reward...


The original web-addy for the free clicker-lessons was
https://canisclickertraining.com/

If they put them back up [& i hope they do!], don't skip the mechanical practice.
That needs to be done away from the dog, & many folks are tempted to skip it, thinking it's not important - but clicking is a mechanical skill, & is a matter of hand / eye co-ordination B4 it's a training technique.

Any on-line games that build hand / eye or hand / EAR co-ordination are good practice - so are such simple games as dropping a ball, & trying to click BEFORE it hits the ground, or choosing a simple sight-line [Ex, a power pole or street sign in the near distance] & CLICK every time someone / something 'meets' the line of that object [Ex, a pedestrian approaches / click when s/he reaches the line of the pole; a woman with a stroller goes by / click when the stroller's front reaches the line of the street-sign post].

Take the clicker to the mall - click every time U see "a man in a green coat", or "a woman with a red purse", or "every blue backpack". Anything is good - just practice.

A wrist-loop to keep the clicker handy is good - U can wear it for hours, & drop it right into Ur hand, with practice.

CAPTURE can be done anytime, but unless it's a rare behavior, and the dog i'm working with is clicker-savvy, I try not to throw capture into a training session.
Green dogs, especially, are easily confused, so i leave capture to moments when we aren't actually training, & catch spontaneous stuff "in between" sessions. // Don't forget to label actions verbally - labels eventually become cues, but only after they've been firmly attached by the dog.

Environmental cues can be terrific for everyday use -
I've taught many dogs to walk out of the kitchen as soon as the cutting board is picked up, it's a simple cue that food prep is starting, & dogs shouldn't be underfoot. Not a word is spoken, they go to their bed or lie at the doorway & watch, or whatever they were taught.

Clicker / marker training is also very, very forgiving - if U click at the wrong instant, & the dog offers the wrong behavior, don't worry about it. Reward anyway; from then on, don't mark that bit of action, & it will quickly fade.
Everything is fixable, & there are no bad associations.

happy training,
- terry


leashedForLife is offline   Reply With Quote
Reply



Thread Tools
Display Modes


Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On

All times are GMT -5. The time now is 11:04 PM.


Powered by vBulletin® Version 3.8.8
Copyright ©2000 - 2019, vBulletin Solutions, Inc.
SEO by vBSEO 3.6.0 PL2
vBulletin Security provided by vBSecurity v2.2.2 (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
User Alert System provided by Advanced User Tagging v3.1.0 (Lite) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd. Runs best on HiVelocity Hosting.