It was two nice and inspiring, relaxed days. I was looking forward to working Phoebe in a supportive group environment.
Keep Calm was about teaching high-strung dogs to relax. Phoebe can get quite over-aroused when there are toys, high-value treats and clickers present. She would happily work for cardboard, sometimes gets so excited that she can’t think straight and you have to count your fingers after feeding a treat, and she has a hard time taking breaks once she’s in the training zone.
The ideas Simone presented were not new (look at that, conditioning a relaxation mat and combining it with a certain scent, teaching dogs to wait at barriers, clicking relaxed body cues, Karen Overall’s protocol for relaxation). However, she mentioned some interesting studies I hadn’t heard about that backed up these approaches, and it was helpful to get her feedback as we were working through the exercises. I also appreciated her perspective on “fake” relaxation: if I click Phoebe’s tail for wagging more and more slowly, and her hip for moving to the side, she’s doing a trick rather than actually relaxing. However, the body still responds accordingly – and this will eventually have the effect of calming her down.
It was also reassuring to hear that Simone agreed with what I generally do when Phoebe needs to relax in an exciting environment: I keep my rate of reinforcement high, and only gradually lower it, setting her up for success. I have been criticised for this approach, and this is also the reason we stopped agility shortly after starting it: I clicked her for being calm whenever it wasn’t her turn, and was told to not do that but just tether her to the fence and ignore her. However, teaching Phoebe to jump into the leash and bark at the fence until it’s her turn again is the last thing I want – and this is, unfortunately, only too common in the agility community.
Calm (and tired) doggies in the end of day 1.
Advanced Clicker Training
The advanced clicker training day was particularly inspiring. There are numerous clicker trainers I admire for their trick training skills and their approaches to shaping – and everyone is slightly different, which is the most interesting part of all. I’ve done my last shaping workshops with Sue Ailsby, Donna Hill and Deb Jones via the FDSA. I’m particularly a fan of the Sue Ailsby way, which emphasizes splitting a lot and really teaches a dog to problem-solve independently. Compared to Simone Fasel, Sue is a free shaping purist.
Verena teaches walking figure 8-s around two cones.
Simone’s shaping sessions are even shorter than the ones of most other trainers – she recommends 5 treats per session, or 30 seconds. Also, Simone does not increase criteria within one session. If you took pictures of every iteration within one session, Simone’s pictures would ideally look exactly the same.
Sue and many others (me included) train slightly longer (up to 1 or 2 minutes, or up to 30 treats), and may increase criteria within one session – first click for 1 step, then 2 steps etc. Our pictures, put together, would ideally look like the pictures in a thumb-flip book.
I don’t know if one of these methods is actually superior to the other, or if the best method to choose depends on your and your animal’s particular teaching/learning style. In either case, I enjoyed learning about Simone’s methods and her reasons for preferring it. Her explanations always made sense or were backed up by studies.
Another intriguing difference between Simone’s method and other methods is that Simone recommends not shaping more than one behavior with one single prop until the first behavior is really strong and on cue. For example, she would not shape going around a chair, and in the next session (or even on the next day) shape crawling under the same chair. This, she argues, will lead to confusion and frustration in the animal.
Can we please keep working?
Sue’s approach to shaping, on the other hand, is all about watching out for changing criteria. When working with Sue, the animal learns to answer the question, “What gets reinforced right now?” – The animal even learns that the goal behavior might change within one session. Watching Sue’s dogs work this way, and other dogs (including the easily frustrated Phoebe) follow in her footsteps, I don’t agree with Simone that this is necessarily frustrating. As long as your timing is good and you keep the RoR high, it seems to be okay to change the target behavior even within one single session. However, I also see Simone’s point, and I agree that if you lump during the change of criteria, there is a big chance the animal will end up frustrated. Maybe it really is a question of personal philosophy which approach you prefer? Well, that is, until someone does an experimental study on which method is (A) more efficient in teaching an animal a particular behavior and (B) more effective in teaching an animal general creativity and problem-solving skills.
Another interesting thing Simone introduced was to have a different marker for food and toy rewards. So far, I’ve used the same marker and surprised the animal with the reward that was coming. Since toys are of higher value to Phoebe, I assumed that this would work in my favor – she never knew when there would be the fun toy surprise. I used to hypothesize that by means of intermittently reinforcing with a toy, I’d get the strongest possible marker, just like intermittent schedules of reinforcement (think: slot machine) build the strongest behaviors (think: gambling addiction).
Simone, on the other hand, says the same marker should consistently lead to the same reinforcer: when an animal is expecting reinforcer A upon hearing the click, but receives reinforcer B (which is also coveted), the reinforcer becomes weaker. Disappointing expectations, according to Simone, will always weaken your reinforcer, even if the reinforcer the animal receives is similar in value to the reinforcer she expected. This is interesting, and I’ll have to research it some more to see if I want to start differentiating between a marker announcing a treat and a marker announcing a toy.
Simone also suggested using distinct markers for active and calm behaviors. The excitement of the activity gets built into the reinforcer, and when you work on relaxation, it makes sense to use a different marker than when you work on jumps.
Explaining details about the target cup exercise.
Carina asked another interesting question about different markers. She wanted to know whether it made sense to use different markers for all of her dogs. A little while ago, I had asked the same question on the Clicker Solutions list, and was surprised to find that many people did not tend to use different markers or different clickers for different dogs. Simone definitely thinks that different markers are a good idea, because even if dog A is not paying attention to dog B being clicked – even if dog A knows it’s not her turn! -, the neural connections in dog A’s brain will still get weakened by “her” marker sounding in the background without being followed by a reinforcer.
This is particularly interesting now, since Tom gets his puppy next weekend, and I get to help train him! Yey! So I need a marker for Hadley. Since Phoebe and I usually work with the iClick, Hadley will get a box clicker. Phoebe’s marker word is Yes!, and Hadley will get his own word; maybe Top!, which used to be Pirate’s marker word.
While Simone is a big fan of shaping, she is not a fan of (pure) luring which, in her opinion, mainly teaches dogs to be passive and don’t switch off their brains. It was nice to hear this; I also love shaping best – even if sometimes, luring a simple behavior would be faster than shaping it. Still, I don’t think we can generalize that luring always leads to passive dogs. Emily Larlham is a good example of someone who uses lots of luring and has very creative dogs at the same time.
Another topic that was mentioned was the importance of reducing the latency between marker and reinforcer as much as possible. It is till commonly assumed that the click bridges the time between behavior and primary reinforcer, eliminating the need to feed really fast. However, Simone pointed out that this is not the case – you will still need to reinforce really fast. Just as you should ideally mark at the exact moment the dog performs the behavior, you should ideally deliver the treat no more than 0.5 seconds later – and you definitely shouldn’t take more than 1 second. This makes sense to me, but I’d still like to further look into it – especially since the Alexandra Kurland translation I’m currently working on makes an equally convincing case for something different: according to Alex, you have to promptly initiate the delivery of the reinforcer after the click; however, the way you deliver the treat itself can be slow. That is to say, Alex would take a treat out of her treat pouch within those 0.5 seconds, but then take her time giving it to her horse – according to her, the knowledge that the reinforcer is actually coming (hand into treat pouch) is essential for keeping up the strength of the neural connections, while the time between starting and finishing the treat delivery is not.
We also spent some time working on stimulus control and cue discrimination. It was pretty impressive to see a dog hear the difference between “Pfötli” (Swiss German for raise your paw) and “Bötli” (Swiss German for a small boat) – the two words sound almost the same. However, the dog only performed the behavior upon the correct cue (“Pfötli”). Simone pointed out that stimulus control leads to a dog who works more calmly and is less excited. Dogs who have good stimulus control show very similar working styles, no matter whether they tend to be calm or lively in general. Cue discrimination, on the other hand, leads to a dog who is extremely attentive and a concentrated worker. Phoebe and I will have to work on that some more! It’s good to be reminded of these things sometimes.
Finley pays attention during the cue discrimination exercise.
Phoebe makes new friends during lunch break.
Thank you, Simone, for an inspiring training weekend!