Switching from Respondent to Operant Learning

In the last two posts, I’ve been taking you through respondent conditioning and respondent counterconditioning. Big terms indeed. The key question we’re looking at today his is perhaps the biggest one in working with highly emotional dogs: when do you switch?

Honestly, it makes me despair when I see people who are still trying to use respondent counterconditioning alone and they’re 8 months into a programme but they’ve seen so little progress that the dog is practically over threshold the moment they step out of the door. Something is profoundly wrong with their timing and their understanding that’s leaving their dogs stranded as emotional jetsam, tossed about on the tides of life.

If dogs live such short lives – and some bigger dogs have perhaps only 8-10 years with us – then spending a year of their life where you’re making so little progress as to be negligible means that a very long section of your dog’s life is spent facing triggers on a daily basis and still feeling pretty icky about them. I don’t think I can live with that.

We owe it to our dogs to make progress as quickly as possible. Spending portions of their life too uncomfortable to even pee outside because they’re that anxious is no way to live.

If you’re really, really struggling, please consider medication. This is a welfare issue. Medication won’t help your dog overcome their issues without concurrent training, but it will help you work faster once you get the right medication.

Stop thinking supplements, as well, please. I supplement my dogs’ diets. I use herbal remedies. I have no beef with using non-pharmaceutical products. I use them for physical support and emotional support.

What I hate is seeing dogs who are floundering being left without medication because we’re worried it’s not ‘natural’. Stupidly, people would rather pay a small fortune for herbal remedies that are unproven and untested, often where the product is not held to the same scrupulous standards as pharmaceuticals, and we owe it to our dogs who have severe anxiety to consider how we can help them better. Remember also that there are many behavioural medications and you may not hit on the right one the first time. You may need a combination. Don’t stop until your dog has a baseline of relative stability that means they’re not spending vast portions of their day feeling uncomfortable. Sometimes, training and supplements can be like trying to hold back an emotional tide with your hands. We owe it to our dogs not to let them suffer. Please also check that your dog does not have underlying health issues. So many times, ‘reactivity’ is driven by ill health, especially when it’s sporadic.

If your dog needs desensitisation because whatever flips their switch is exciting, you also need to on-board frustration tolerance activities and impulse control activities. These are not the same. The former is about learning how to tolerate life’s frustrations when you can’t get or do what you want. The latter is about learning to control your own body. A month of these in the home and in low-arousal spaces outdoors can really make a huge difference.

Counterconditioning should not take long. We can’t talk in one breath about single event fear learning, where dogs learn in one single episode to associate a trigger with something unpleasant, and not understand that counterconditioning shouldn’t progress at glacially slow speeds. It makes no sense. We know so much about fear conditioning and counterconditioning from laboratory and applied settings that we can say that it shouldn’t take that long.

If it does, then the previous post will definitely help you up your game.

I’m looking for such high value food and such well-controlled, straightforward set-ups that we’re talking a few 5-minute sessions over a week or so for the dog to truly know what you’re up to. That’s the moment to switch to operant conditioning.

There’s another reason that you should also switch to operant, Skinnerian methods. Remember, these are driven by the learner.

Pavlov is about how the world works on us. We are jetsam on life’s tide. The light turns green and we put our foot to the pedal. The light goes red and we stop. We sniff pepper, we sneeze. We get a puff of air in the eye, we blink. We chew sour sweets, we salivate. The cute boy in class talks to us, we blush. We’re weak little response machines bobbing around on life’s ocean.

It’s powerful stuff, but it’s not empowering. All we’re doing is learning new things that make us respond in particular ways.

Pavlov is one-to-one. One trigger. One response.

And it’s an all-or-nothing or a graduated response, but it’s a helpless kind of response. You can’t do anything about it. Trigger, response.

Unless, that is, you bang in some support in the form of desensitisation and/or respondent counter-conditioning.

Skinner is about how we can change the world. We are in control. We behave if we want access to reinforcers that are valuable to us in that moment, including escape from aversive situations. We don’t behave if we want to avoid aversive situations or if we understand that good stuff gets removed from us. It’s about our needs and desires in the moment. We have choice. Choice is hugely empowering.

One reason I switch as soon as I can to operant methods is because of the effect on the learner.

My old girl Tilly was a prime example. When she arrived with me as a five-year-old, having bitten numerous children and seen a legion of specialists for her behaviour, she was an anxious hot mess. When my partner at the time stood up, she’d pee on the floor. She’d growl if anyone touched her.

What made a difference for her?

Showing her how to work the world. You want to go out? Tap the door. You want to be petted? Come and sit near us. You want it to stop? Move away.

I can’t ask her how these things made her feel. The world was predictable and it was safe. She worked us like a finely tuned instrument. Soon, she was scraping the water bowl if it was empty and she was telling me she wanted to go out in the middle of the night instead of peeing on the floor. Five years of incontinence and she finally found a way to ask for what she needed. Five years of biting and she finally felt safe enough to keep her stuff and hand it over when asked.

Skinner turns triggers into questions.

Instead of being forced to respond, because Pavlov gives us no choice whatsoever, Skinner says: ‘Would you like to?… It’s up to you!’

Now I’m making this sound more thoughtful than it is. Learners don’t need to be conscious of that question, but we learn that we have choice.

That choice is the bane of our lives if we’re searching for the perfectly obedient dog. Only Pavlov makes us perfectly obedient to the world. Skinner says, if you want the paycheque, go to work. Skinner says, press the lever if you want a seed right now.

So much so that he had to keep his animals at 80% body weight because when that light came on to say that if the animals pressed the lever at that moment, a seed would pop out, if the animals were not hungry, they’d just opt out.

Not so good for a behaviour scientist if your mice won’t run mazes and they’re all, ‘not today, thanks!’

Operant training is a choice that depends on how much we want the stuff or want to avoid the stuff that happened in these circumstances in the past.

It’s not an obligation to respond.

Where you’re turning former triggers into things that are simply cues for other behaviour, it’s mightily empowering. You’ve got choice. The triggers make predictable things happen.

Operant training is an absolute gift for many anxious dogs. This weekend, I was working with one old dog who’s got a few issues related to being as old as dirt, and we pretty much had NO training on offer simply because he’d lived life in control of his own stuff. My girl Flika was like that. Clearly, nobody had ever met her needs and she would take life into her own paws. Door closed? Open it. Door locked? Chew through it. Or find another door. Hungry? Go in the kitchen and help yourself. Bored at a meeting? Rifle through handbags and find some snacks. Both dogs had lived life on their own terms. It may be a struggle and it may not always be rewarding, but you’re the master of your fate.

For other dogs, especially anxious dogs, the world is unpredictable and the world works on them all the time. These are the dogs like Lidy, the first time I met her. She was a head-butting, bitey, circling, frantic, snatching, grabbing out-of-control piece of work. Operant methods have been the absolute making of her.

The first time she realised, ‘oh! That’s how you make the monkey do stuff!’ when she got food for a behaviour … it was a revelation. She looked at me like, ‘I could take your bag and run off into the woods, lady!’ but she cooperated and she worked out how to make me spit out treats. Operant training builds a partnership between the guardian and the dog.

Years ago, the shelter rescued a number of animals from some incredibly grim circumstances. There were three timid collie siblings in the mix. Two of them went to live in incredibly loving homes with wonderful, lovely people. One went to live with a Swiss dog trainer who does heelwork to music. The difference between the three dogs couldn’t be more marked. Two can’t cope well outside of their small social group. Joey, well, Joey performs heelwork in front of huge crowds, travels across Europe and is worlds away from that nervous little guy found cowering in a room eight years ago. He’s not the same dog. He may look like the same dog, but he’s not the same.

What caused that difference?

Well, I think operant training did.

Don’t get me wrong; I am not a fan of training for training’s sake. I am lazy and I’m somewhat ethically opposed to dogs having to perform for the stuff in their life. I think the ethics come from the laziness to be honest.

Operant training builds confidence. It builds reliability and trust in the world around you.

It builds a bond between dog and human.

Joey, when he’s with Sylvia, he’s in her world. All the world, all the crowds, all the games… it’s all just background noise. When he’s with her, the world is predictable. They do their routines and it’s comfortable. All she’s really been doing these last eight years is proofing his trust in her in a variety of diverse and challenging circumstances.

I do the same with Lidy, just on a smaller scale. Mainly because I’m lazy. I do as much as we need to. But four changes in home this year and she’s shown me what that means. When I am there, I am her safety cue.

There’s fallout, for sure. She’s not so hot at coping when I’m not there. But we’re working on that.

Operant training makes her life predictable. It builds her trust in me. It also does in others. Whoever holds the training pouch is now good people. Whoever holds the paté pot is also good people. Including the vet. In a foyer surrounded by barking dogs and moving people. Operant training helps me trust her too. This is one reason that I so completely fell in love with Leslie McDevitt’s Pattern Games. They just work so well.

It’s for these reasons that I switch to operant as soon as I can. Choice. Predictability. Empowerment. The learner has ultimate control over their environment.

I’m going to let you into a dirty little secret… most of the behaviour I see with dogs is actually not respondent.

Or, it’s not as respondent as all that.

It’s definitely subject to consequences.

Well, sometimes.

Poor Loupi the pointer, aged 10 weeks, saw a sleeping kitten and went into full point.

Not so operant.

Lidy couldn’t have cared less about cats until she was in a new environment, one shot out and she caught it. That woke up every predatory urge in her body.

More operant.

Predatory behaviour is an urge that needs scratching. When it’s being scratched, we can put parameters on it and make it stop when we ask. This is often about the games we play with our dog and giving them opportunities to scratch itches as well as learn to stop. Lidy was not so hot on play at first and there was a lot of grabbing, frustration and lack of impulse control. Now, she’ll sit and wait before I toss the toy. Now, she’ll release and lie down when I ask. She’s in control.

What’s happned to her predatory behaviour?

Well, it didn’t dry up altogether!

But there are glimpses where things become a game again.

The other day, we’d been practising a middle before chasing. It’s pretty simple. She sticks her head between my legs and sits. I throw the toy and we play tug a bit. Then the toy goes dead and we repeat.

What happened when she saw a cat? Well, no word of a lie, she did a middle. ‘As if,’ says I, ‘I am going to let you chase the cat just because you sit pretty, my girl.’

But I saw a glimpse of the operant. A fracture where a bright Skinnerian light glimmered through the fissure.

Not a glimpse, actually. Not a fracture. Not a fissure. A huge great gully. A gully where instead of taking advantage of my weakened state (I was bent over bagging up a turd…) and pulling me over as she made for the cat, she took control of her own body and did a predictable thing.

I did reinforce her with a few games of chasing the food. Predatory behaviour is completed where food is consumed. Ten treats thrown in the grass are not that much different than a cat in a bush. We need to remember that. Predatory behaviour is about the action and about eating. That’s why food and toys are your friends here. They can quite often substitute for predatory behaviours, if that’s what you’re dealing with. Of course, it doesn’t hurt that, in adding more interactive play with rules, you’re helping the dog cope with frustration, learn to control their bodies and building your relationship, as well as scratch a biological itch. A quadruple whammy.

If you’re dealing with dogs who don’t feel safe, much of their behaviour has been put there operantly, I hate to say it. They’ve been placed in situations where they’ve either been forced to respond, and they’ve learned that reactivity and aggression keep stuff away. Or they’ve been placed in situations where they just passively learned to sit it out. Food and toys are not your reinforcers here. The behaviour has been maintained by safety and by distance. You can of course add food into the mix, or even play or toys. However, you’re dealing with something very different.

Much of that behaviour will also become superstitious; the dog feels the need to react. Skinner defined superstitious behaviour as any the dog feels the need to perform in order to get a particular consequence, where the consequence was not contingent on the behaviour. In other words, the consequence might be completely unlinked, but the learner feels they need to do the behaviour to get the consequence to happen. One prime example is barking at postal workers and passers-by. Their leaving is not contingent on the behaviour.

Behaviour is in itself very lazy. I’m a perfect example of that. I do as little as I need to to meet our needs. Behaviour does the same. If we can break the chain for superstitious behaviour so that the dog realises they don’t need to do X in order to make Y happen, then the innate laziness of behaviour will eventually take over. If the postal worker goes away anyway whether you bark or not, then what’s the point of barking? All we need to do is set up enough experiments where the dog can see that happen. It’s one reason I think my alert and alarm barking protocol is so very effective. The dog may well simply be learning that they don’t need to behave to get strangers to go away.

Still, it may also be inherently reinforcing for the dog. Heston, for instance, does love a bark. However, he doesn’t actually like strangers passing. He likes to do fanfare barks and excitement barks and other barks of joy. He doesn’t actually like barking at annoying pedestrians or people who come too close to the house.

How do I know?

Because he doesn’t bark like that at other times.

Thus, I can’t control all his barking (and neither would I want to… his joyful big barks are his celebration of life’s glory) but I can minimise his need to bark at passers-by in operant ways.

You’re still here to find out when you can do that, I know!

When you are doing a clean set-up of respondent trials with great stuff, you should find that, if you’re at the side of the dog or behind the dog, there’s a moment when the dog sees the trigger and then turns to you in expectation. If you delay just a tiny moment, you should see them look to where the food is delivered.

That’s the moment.

At that moment, when they disengage from the trigger, you can shape that turn to you.

That’s to say, you can make it bigger. You can turn that fissure into a gully and then into a canyon. You can stretch it out.

I’ll just say here that you do not want to strain that shaping. You don’t want to decide you’re going to count to ten before you give the dog their food for turning to you or looking at you and disengaging. Many dogs will just go back to what they were going to do instead.

Instead of TRIGGER > BEHAVIOUR within a millisecond, you should now have… TRIGGER > MICROPAUSE > BEHAVIOUR.

That micropause can be shaped into something bigger.

If you get a TRIGGER > PAUSE > BEHAVIOUR, you can then insert a prompt just after the pause:


This way, you are substituting another behaviour. Same trigger. Different behaviour.

In fact, a small but philosophical thing has happened. It’s no longer a trigger, it’s a cue.

Let’s take a typical Lidy example. At first it went like this:


Then, we did a nice, clean set-up where a person came into view for a short period at a controlled distance…


Within three trials, she’d sussed it. I left a micropause:


Within two more trials, she turned to me really briefly:


Now, if you’re a hot marker trainer, you’ll be able to capture and shape this or simply shape it. Capturing would look like this:


Shaping would look like this:


You don’t need the marker, in other words.

When would I use a marker?

When it’s meaningful to dogs. Skinner works without marking – shock, horror! If I’m working with a dog who hasn’t got a history of marker words, I may not use a marker at all. How unconventional of me, I know! Markers can make dogs reliant on us, however, to tell them when they’re doing right. That’s my opinion. No marker means the behaviour itself is working it out. If you’ve done my Hagrid method of loose lead walking, I often don’t use a marker word as the dog is waiting for the marker and focusing on that, not what they were doing.

I’d also use a marker if I’ve got a dog who needs a lot of guidance. In my opinion, malinois need a lot of guidance. They freelance terribly if they don’t get it, and immediately revert to the behaviour they’d wanted to do in the first place. German shepherds are much less reliant on the click. Just my experience! I also use a marker if the dog has a number of triggers. It’s efficient.

I don’t use a marker if I have a dog who doesn’t know markers or if they’ve only got one trigger. They know what you’re up to, believe me, if you set it up right. Markers can also be challenging for clients who aren’t so hot at timing. In short, we don’t need to always mark behaviour. It can, in my opinion, put the human more in the driving seat than the dog.

Whether you capture or shape, you’ll also be able to cue other behaviours, or let the dog lead you with a self-selected behaviour.


You can cue whatever behaviour you like here. I’d go for something your dog likes doing and they’ve a good history of doing. This is why I’ve often been teaching a behaviour discretely. I may start with that weeks before doing the respondent set-up. You can cue a ‘Look at me!’ or a ‘Wait!’ or a ‘Watch!’ or a u-turn or a sit or a down or a hand touch or a middle or a peek-a-boo – whatever floats your dog’s boat. However, I would say that moving behaviours are better than static. I’m not a fan of asking dogs to sit and stay while stuff goes past. I’d prefer to be mobile so we can get the hell out of Dodge if the situation makes a turn for the worse. I’m a huge fan of u-turns, since they offer safety, relief from the cue and distance. Since this is what reactive dogs are seeking out anyway, it provides more reinforcement than simply getting yummy food.

Ironically, it was also Hagrid that taught me the beauty of these strategies. We’d been doing u-turns and practising faithfully when one day, he saw two bouncy labradors about 200m away. He turned to me as if beginning a u-turn. We turned around, got out of the way, did our own thing, let the labradors go past and I had a lesson in how dogs tell us they’d rather not bother being reactive or aggressive, thank you very much.

Eventually, you will fade out the second cue:


This is the ultimate goal. The dog sees their cue, does a behaviour that tells you they’ve seen it and you reinforce. As I said, lazy.

Whatever behaviour you choose can go in there:


In that moment, then, where the dog turns to you for whatever you’d been doing, wait for that millisecond of recognition that you’ve not yet delivered the food that you’d been doing before, and stretch the hell out of it. Insert other behaviours when you’ve done that and experiment.

Should you prompt?

What would that look like?

A prompt is a gesture or word that you’d use to encourage the behaviour if it’s not forthcoming:


I don’t tend to prompt unless the dog gets stuck and the trigger is too overwhelming for them. Skinner is voluntary and choice-based. We can ignore prompts and cues. It’s likely that the prompt will fail. Nevertheless, if I see my dog getting stuck in that moment before they’re going to react, I’ll prompt. I was sitting on the step the other day doing some of this with Lidy and two guys walked down with two shi tzus. I could tell by how long it took her to turn away that she needed a prompt. I told her ‘in’ and she went in the house. Seconds later and she would have exploded.

I don’t prompt if I know the dog isn’t even aware of me at that moment. That is a lesson to me that I’ve failed them and put them in too far. I also don’t prompt routinely. The dog is still reliant on me to tell them what to do.

The other important thing about this is then you can also begin to mix in lesser reinforcers. I’m so cheap these days that Lidy does all this for rubbish dog biscuits and kibble, or even – and how cheap is this? – me telling her she’s a good girl. You can’t put respondent counter-conditioning on cheaper food or mix in other stuff or even phase it out. You can with operant.

One thing that I do like to do with dogs with multiple triggers is cue them to be aware of the possibility of the imminent appearance of that cue. Leslie McDevitt’s Look At That is a good example, but as you can tell from ‘that’, it’s not a specific trigger. Lidy know’s her ‘Where’s Wally?’ from her ‘Where’s the Dog?’



Why I like these pre-cue cues so much is that in my opinion, they prepare the dog for what they are about to see. For reasons I’ll explain in future posts, this can help lower the aversiveness of an unpleasant situation and it can also prepare them, making the world predictable. It turns the world into a predictable game. We know from some work by Adam Miklosi’s team that dogs are surprised when their expectations based on scent are not met. From this work, we can extrapolate that dogs do expect things from scent. If they have expectations of who or what to expect based on one environmental stimulus, there’s no reason that they can’t learn to expect different things. Of course, this adds an unnecessary layer of complication for most dogs, but it’s a gift for those dogs who have multiple triggers and I believe it can work to reduce anxiety about what generalised cues like ‘Look at That!’ might mean if the dog has different emotions or different intensity of emotions related to different triggers.

Other times, I’m really just letting the dog process the world themselves. If stuff is under threshold and well controlled, food has no value. In this way, I’m working more with BAT 2.0, where you’re just letting the dog make their own decisions and make sense of the world. For Lidy, for example, all she’s doing is making sense of stuff, and often, if I give her time to make sense of it, she’ll make good decisions on her own.

In this video, you can see that moment really clearly. There were some dogs barking outside (I think!) and she was listening to them. You can see up to 0.22, she’s focused on the sounds outside the house. At 0.22, she looks to me. That’s the switch. That’s my gold. That’s when I know she’s switched from a respondent brain which is jetsam on the tides of life to an operant brain that is able to do things.

What is ‘Good girl?’

A marker that says treats are available.

Where are the treats kept and delivered?

Well, the dogs tell you.

We move away from the dogs barking and we get food.

Dogs barking, here, are no longer a trigger that causes my dogs to bark.

Dogs barking are simply a cue that says, treats will be made available from the cabinet near the door.

As I said, behaviour is lazy. It takes the least unpleasant course and the least effortful course to get to satisfactory outcomes.

Here’s an annotated video of the above:

It’s that moment at 0.22 that I want.

This comes back to why I want clean set-ups. The above video is not a clean set up. I have no control over how long the dogs will be outside the property for. It’s not clean. I don’t know how many seconds it took me to find my phone, switch it on, switch to video and record, but even if it only took me three seconds, that’s almost half a minute of exposure to a trigger.

That’s too long for good respondent counterconditioning. The longer the exposure, the more likely she will be to bark back. Also, as you can see, I’m not counterconditioning. I’m just letting her process it. We move then into operant. We do a lot of ‘just processing’ – I want her to make good decisions and decide that things are nothing to be bothered about without me telling her so. If I were counterconditioning, I’d be feeding her from 0.01 right until the dogs stop. I’m letting her make sense of stuff.

All we’re really looking for, then, is higher latency. Latency is just the amount of time between a stimulus and a response. Low latency is great for computers and the internet… Instant reactions. Low latency is not great for dogs. Instant reactions are not a good thing. When we’re switching from respondent counterconditioning to operant methods, we’re looking to stretch that latency to a high latency. We’re looking for a lag between noticing and reacting. When we’ve got lag, when we’ve got high latency, when we’ve got SCARY STUFF >>>>>>>>>>>>>> REACTION rather than SCARY STUFF > REACTION, we have something to sculpt…

Time. We have time to stretch and sculpt.

We’re also looking for less intense behaviours. Listening intently and tracking is less intense than barking and jumping up at the window. We can sculpt those too.

I did try and find a video where I switch from respondent to operant, only to realise I don’t have that many. Respondent counterconditioning done well means a handful of careful trials that give way easily, naturally and quickly to operant work. It’s probably 2% of my time with a learner. Once your dogs know what you’re up to, you get that head turn whenever you are working under threshold with savvy dogs, whatever the trigger. I don’t have videos because respondent set ups are delicate things, and I need my eye on the dog, not on the camera.

You should end up with moments like this, where something salient happens and your dog looks to you:

The one that follows is an annotated version where I outline the processes. It’s sloppy. It’s chatty. It wasn’t done for professional showmanship – a bit like my training. It’s impromptu and casual and it’s on the spot – a lot like my training.

You’re looking for that tiny moment of ‘Hey, you’re slow with the stuff!’ where the learner turns to you.

At that point, they are beautifully conscious of what’s happening to them and the processes that are affecting them.

They’re no longer jetsam on life’s tide of offensive stimuli. They are dogs who think that offensive stimuli mean treats.

They’re no longer at the beck and call of predatory behaviours. They are dogs who are learning to disengage and do other stuff instead. In the end, a thousand reliable repetitions of ‘find it!’ or ‘get it!’ will always win out over something chaseable that they never get to catch. With predatory behaviours, what we’re essentially doing is putting it on cue, not unlike gundog trainers might. Only our dogs don’t get to chase cars and bicycles… so we need a substitute to pop in there instead. You’ll find lots, I know, on predation substitute training, that will help your dog scratch itches and cope better with moving stuff.

The key is in waiting for that millisecond where the dog becomes conscious that you’ve not delivered the expected unconditioned stimulus. When you’ve got that, you are ready to switch. It might take a few more repetitions, but a dog who is conscious of the game you’re playing is a dog ready to switch.

It’s a beautiful, beautiful moment because you can extend the time your dogs take thinking and processing rather than reacting. You’re stretching the time between the trigger and the response. That latency means you can insert other things if you like, such as another behaviour. It means you can turn triggers into simple environmental cues that tell the dog to do something else instead of what they used to do. It helps the dog make sense of predictable sequences of events.

Of course, where Skinner goes, Pavlov goes too. They’re shackled to one another. Just because you’ve switched from obligatory responses to triggers to optional and voluntary behaviours subject to consequences does not mean Pavlov has packed up his kit and buggered off. He’s still there, helping your dog feel good about you, about the world and about what’s happening to them. If you opt for an incompatible behaviour instead of whatever the dog had been doing instead (you can’t chase AND recall… you can’t lunge at a target AND touch your guardian’s hand) then you’re not only doing an incompatible operant behaviour, you’re also doing an incompatible respondent behaviour – the very essence of both respondent counterconditioning. Thus, you enter the world of operant counterconditioning – the most powerful tool of all.

If you’re a dog trainer looking for ideas on how to work more effectively with your clients, why not check out my book?