The Right Tool For The Job: desensitisation, counterconditioning or training?

I’m going to apologise straight away that this post is technical, rather than one for the average guardian. It’s mainly based in some very rich and interesting discussions I’ve had with other dog trainers and behaviour consultants in the last couple of weeks.

The question, as always, is rooted around picking the right training method for the dog (and the guardian!). I’m finding more and more that although all individuals learn in the same ways, some ways we can train are singular precision tools rather than a Swiss Army Knife that can do a lot of things, but none of them particularly well.

I’ve been posting a lot recently about dogs who growl, bark, lunge, snarl, snap and even attempt to bite unfamiliar people either when they visit the property or when we’re out in the street on a walk. Today I’m going to keep my focus on that. I seem to go through trends where I get a run of dogs who exhibit the same type of behaviour. The behaviour I’ve mentioned here, I refer to as ‘Stranger Danger’. as a shortcut This post is going to focus on that single, very specific behaviour. Dogs barking and lunging, snapping and snarling at members of the public. This is a perennial problem for the dogs I work with.

The decisions about which tool to use to change this behaviour is a much broader discussion, however, that is as relevant to many other situations where we might be engaged in changing emotional responses to the world. I just wanted a handy example. I can’t think theory without the actual practical bits. It’s only when I can see it in concrete reality that abstraction makes sense to me. So Stranger Danger dogs it is.

Stranger Danger is not just about barking or lunging at unfamiliar people in the street. It can also be a fearful reaction, with the dog freezing or cowering when people approach, and then turn to barking or even biting. I even see it with dogs that people assume are super-social – often gundogs – and what I see are dogs who manically feel the need to assuage and befriend everyone present to diffuse threat. All of these behaviours – even though we might class one dog as aggressive, another as fearful or panicking, another shut down and another as a social butterfly, these can all be rooted in fear of unfamiliar people. All the teenage dogs jumping up on guests can be just as much about Stranger Danger as barking and lunging can… just another way of expressing nervousness around new people.

Does it matter how you modify this behaviour? Is training bespoke or does one-size-fit all? Are there really different methods that might be more fitting?

All my training programmes consist of three elements.

The first is management. Whatever problem the dog has, it’s our obligation as guardians, caregivers or trainers to make sure everyone is safe. It’s not normal for dogs to treat all people outside the home as if they’re dangerous, armed robbers. Nor is it normal for them to be so afraid that they shut down completely. And no, it’s not normal for them to be so needful to make contact that they’re screaming behind a baby gate and then jumping all over guests when they’re finally allowed to make contact. We have to make sure the unfamiliar people are safe, and we need to make sure our dogs are safe. Not to do so is a legal issue on the one hand and a welfare issue on the other.

That means leads, muzzles, distance, security, barriers. Yes, even for the ‘friendly’ dogs. No person – especially not a dog trainer – should be put into a position where they can be harmed. A dog can’t bite strangers if they’re on two leads, muzzled, behind a secure fence. Nor can they knock Nana over in a frantic bid to modulate their anxiety by leaping all over her.

So management is the first strand. That might include distracting the dog. It might include avoiding the situations they struggle in. It might involve securing the dog. Management is fine. It’s not a treatment, but it’s fine. Plenty of dogs, including my own, go through various parts of life being managed. It’s not lazy. It’s just life and priorities. Sometimes, we might do more management than other times.

Behaviour modification is the next strand. Behaviour modification essentially means putting into place a programme to change the dog’s emotions or behaviour. This is what most of this post will be about. Behaviour modification is a fairly limited toolbox in reality, with about a hundred variations on a theme. Good trainers have specialist methods and can adapt the variations to best meet the needs of the animal and their guardian.

Less effective practitioners will have a much less broad range of tools. If you’ve only got a hammer, you’ve got to treat everything you come across as if it’s a nail.

Behaviour modification will include skills in habituating animals to new things in the environment, working with them to take the sting out of stuff they’re already reactive to, changing their emotions, teaching them new behaviours. It can include training, of course.

The third strand won’t be appropriate for every dog. That strand is medication, changes of diet or the use of nutraceuticals. Not all dogs will need this, but if they do it makes behaviour modification destined to fail without it. Or, if not destined to fail, it certainly makes it much harder and much longer to see progress.

So today I want to focus on that second strand, and about how I make my decisions about which behaviour modification tool to use when working with dogs who have a heightened sense of Stranger Danger.

Basically the tools I have available are:

  • Desensitisation: exposing the relaxed dog to the scary thing at levels that don’t elicit a response, and gradually exposing them to more and more of the scary thing.

    The key elements of desensitisation are pairing a relaxed state with low doses of the scary stuff. In human therapy terms, desensitisation is often called exposure therapy. According to Domjan (2015) exposure therapies are ‘extinction procedure[s] in which the participants are exposed to cues that elicit fear in the absence of the unconditioned stimuli’ (p. 246). In the lab, this would mean presenting the conditioned cue without the shock. In reality, this means presenting the conditioned cue – in this case, the scary humans – without the unpleasant emotions and without other unpleasant factors like yanking, jerking or reprimanding the dog. If you want to get technical, Domjan seems to be intimating that desensitisation is actually an extinction protocol. Extinction perhaps being very different than a counterconditioning protocol at a neural level, though we don’t know enough to say this for sure.

    Chance (2014) explores how systematic desensitisation is the process by which a conditioned stimulus is paired up with a state of relaxation (p. 98) While it doesn’t help that there are not agreed definitions of these processes, I take desensitisation to mean the pairing of a conditioned stimulus – the scary person- at levels that do not elicit a behavioural response – with a state of relaxation. This is how it was conceived by its progenitor Joseph Wolpe.

    As such, desensitisation is a subset of counterconditioning. There’s an unresolved tension, of course, with how Domjan defines exposure therapies. You can see why: to Domjan, they are not a pairing procedure; to Wolpe, the scary stuff being paired with a positive emotional state is pivotal.

    To my mind, desensitisation processes are not aided by food or other positive stimuli except relaxation, and this is going to become an important part of discussions that follow. I take the same line as Wolpe does. After all, he named it. To my mind, and she may completely disagree, but Grisha Stewart’s Behaviour Adjustment Training is a desensitisation technique in part. Multiple, careful exposures when the dog is in a relaxed frame of mind.
  • Counterconditioning: this is often lumped in with desensitisation, but desensitisation is a subset of counterconditioning if you subscribe to Wolpe’s definition: it is a counterconditioning process in itself.

    Chance says that counterconditioning is an exposure process to reverse the effects of previous conditioning (p. 394). I think this definition is a bit weak though, not least from a neuroscientific perspective. I know, I know… picking and choosing which definitions I like. I’ll discuss why later, but essentially, I quibble with the idea of reversal and I also think counterconditioning in the animal training world has come to mean something a little different: the pairing of the scary person with food. It’s not unlearning. Counterconditioning is definitely not unlearning.

    I don’t think trainers always understand the difference between desensitisation and counterconditioning. This is not helped by very poor quality examples in very popular dog trainer textbooks. Only this weekend, I read a post from a well-respected animal researcher in which she called ‘DS/CC’ a process. Well, no. It’s two processes for me, and it’s those differences I want to discuss today. For Domjan, they’re two processes: respondent extinction and counterconditioning. For Chance, they are two processes: pairing scary stuff with positive emotional states compared to the reversal of learning.

    Whether you agree with the definitions or not, this is how I define the two in relation to Stranger Danger: where desensitisation is pairing up an emotional state of relaxation or safety with the scary people in progressively more challenging circumstances, counterconditioning is pairing up scary stuff with physical unconditioned stimuli, most often food or toys.

    The aim of counterconditioning, then, is that the scary thing comes to predict the arrival of the good stuff. If we’re talking trainer terms, then Jean Donaldson’s ‘Bar is Open – Bar is Closed’ is the most well-known example of counterconditioning.
  • Operant training: basically operant training requires the dog to behave voluntarily and this is the bit that’s open to consequences. This is Skinner and his pigeons. It’s typical dog trainer territory. Sit – get a biscuit. There are ways we can use operant training to make sure dogs are distractable in the face of the scary stuff, to teach them another behaviour they can do instead, to watch us, to check in with us, to stand between our legs if people go past, to do a u-turn, to ‘Look At That!’ … This is where we’re using food to shape behaviour and we can give cues like ‘Watch!’ or ‘Let’s go!’ to get the dog to move.

    We can even use a thing called operant counterconditioning which not only changes the dog’s emotional response but also gives them a behaviour to do instead. When I do Leslie McDevitt’s Pattern Games with my dogs, or I use Deb Jones’ Focus Games, these can either be used to distract and keep the dog busy in the presence of scary stuff, or they can be even used as a way to help the dog cope with the scary stuff.

    Operant training might not necessarily involve cues, clickers or marker words but we often think of it with a cue, a behaviour, a marker and a reinforcer like food. It needn’t be this four-part structure, however. Simply speaking, behaviour and reinforcer is as much as we need. If I play the engage-disengage game with my dogs, for instance, where I reinforce looking away, that is uncued operant training.

These three tools are basically the ingredients of most programmes to help dogs deal with scary people. In one blend or another, they’re mostly what people are using. In order to think about some of the inherent problems of each tool, it’s going to be really important to separate counterconditioning from desensitisation. Of course, you are full on board with the knowledge that Pavlov and Skinner are sitting on both your shoulders when you’re training and you can’t do one without a bit of the other unless you’re engaging in some very restrictive training, but I do think there are techniques that are MORE Pavlov than Skinner, and vice versa. They’re a sliding scale. Some protocols are more heavily dominated by either Pavlovian processes or by Skinnerian ones.

Desensitisation and counterconditioning are not, however, one single process.

Desensitisation is a subset of counterconditioning using a careful, systematic, gradual stimulus gradient of the scary stimulus paired with a state of relaxation or safety. It’s not the stimulus gradient itself. Some people think it’s about the systematic and gradual bit. It is, but that’s only half of it. It’s the relaxation bit that is being associated with the stimulus that it is the key component for me.

For the purpose of this post, it’s going to be really important to remember that they are distinct processes. Desensitisation wouldn’t involve food, for example. Where there is food, you may also be pairing up with a state of relaxation, but that’d be counterconditioning. Where there is a stimulus gradient, you may be using counterconditioning. If you’re using desensitisation, there will most likely a stimulus gradient. Confusing, I know. For the moment, I think the best way to describe it is to think of the DS bit as being foodless, and the CC bit having food (or other positive stimulus).

Now recently, I’ve had a lot of discussions about the role of counterconditioning or even the role of teaching dogs operant things like ‘Look At That!’. Which are best for the dog and their guardian?

I firstly want to say there is sometimes a bit of confusion over them. I’ve heard of very prominent professionals saying counterconditioning isn’t even a thing, that it’s not even possible. It is. It absolutely is. Thousands of animals have gone through fear conditioning with foot shocks in the lab to prove it is, so let’s not breed deliberate ignorance. In itself, therapies to counteract PTSD are counterconditioning, so let’s not just write off a load of actual science by spouting nonsense.

What is true is we don’t know much about the neural mechanisms by which counterconditioning works exactly. Yet. That said, we are pushing the boundaries of our understanding all the time. There is evidence from applied psychology about respondent extinction that suggests it might be more a case of reconsolidation of fear memories (see Brain & Behavior by Garrett and Hough 2017 for a primer and for cornerstone research) – we’re actually over-writing neural networks during a period of malleability. Maybe. Perhaps when we undertake counterconditioning, we’re just learning exceptions to the initial rule formed by our first learning experiences.

What we also know is that initial learning – the original conditioning – forms a type of rule about the world and also a strong neural connection that, in the case of fear conditioning, is designed to keep us safe, and that forgetting is not what is going on. If you think counterconditioning is about causing ‘forgetting’, it’s time to get your cognitive psychology books out and read the chapters on forgetting. Counterconditioning is not ‘un-learning’. This is why I have problems with Chance’s definition.

But we are probably rewriting that rule about the world with a load of exceptions as we go through counterconditioning procedures (including desensitisation). We may also, at a neural level, be weakening the neural pathways and building new ones. We’re learning exceptions.

Another thing that I’ve heard recently is that counterconditioning doesn’t work because it makes animals more sensitive to the trigger. Say for instance you have a dog who is afraid of scary people and you do Jean Donaldson’s ‘Bar is open – Bar is closed’ game when the scary person comes into sight, where scary thing appears, the bar opens and food rains from the sky, and then the bar closes and food stops the moment the scary thing is out of sight. Some people say that this is sensitising the dog to the trigger. You can see their logic, I guess. In their view, it makes the dog look for scary people.

The trouble is that this naive psychology is not accurate. Sure, if counterconditioning is done badly, then you might make the dog more reactive to the scary stuff. What you’re probably doing is flooding the dog and seeing an extinction burst though. What they’re doing is not counterconditioning. This view about it sensitising the animal is also not evidenced in the lab, however, where counterconditioning with food has been much more effective than simple extinction procedures. Food does help. Their claim that sensitisation happens is not borne out in science.

What trainers who say counterconditioning sensitises the dog probably mean is that the use of food makes the dog actively look for the conditioned cue – in this case, the scary human – because of the food that will inevitably happen when they see the scary person. I think what they are actually referring to is an attention and salience process, not a sensitisation process. They’re different psychology chapters from forgetting or from fear conditioning and counterconditioning. They’re different processes. What I think they mean when they say the dog becomes more sensitive to the trigger is that the scary stuff is becoming more salient. I guess what people who think it can sensitise dogs to a stimulus are really saying is that they think the addition of food can make a stimulus more noticeable. It’s about attention and valence.

To a degree, I think this idea that food might make the scary people more salient might hold some truth. I certainly see it with some dogs. I know dogs who, upon seeing the conditioned cue – the scary human – look back to their guardians, salivating at the thought of the sausage that is about to materialise. If you’re waiting for a payday that’s dependent on a cue, then you’re going to look for the cue. That is evident. From my own anecdotal experience, I think this is sometimes true. I think the food can also hinder the process if it comes to dominate the process. I’ve known dogs who are so fixated on the sausage potential that they barely notice the scary stuff.

Here, counterconditioning fails because the dog is barely aware of the scary stimulus. Sometimes, that’s fine. Distraction is a useful tool. I’ve certainly used it a million times. But distraction involves no learning. If I truly want the dog to feel differently about the scary human, distraction will get in the way. Today, I stopped at a breaker’s yard to drop off some parts and Lidy barked and lunged at the guy who came to carry the box. Later, we stopped at the supermarket and I picked up some sausages because I’m that type of guardian. When I opened the packet, a lady walked past much closer than the guy had been in the breaker’s yard. Lidy didn’t even notice. It was all about the anticipation of the sausage. On the other hand, though, we always have sausages at the supermarket. The sausages simply overshadowed everything else. I think, if you’re doing counterconditioning on every excursion, you might get a dog who’s fixated on the sausages. Make sure you secrete them carefully, I’d say, and you probably won’t have this overshadowing.

This problem is as equally true for dogs who are working operantly – trick training or obedience training in public, if you will – where the patterns and the routines and the food stop them engaging with the scary human. I’ve known dogs who were so lost in a ‘Watch me!’ that they failed to even notice the scary stuff go by. Again… fine, if distraction is your aim. If the dog isn’t noticing the stuff because you’re involved in your own routines, then it’s not really doing anything by way of learning. This can be amazing if your dog builds up such a trust in you that they only have to look at you and work with you to feel safe, but that’s about your relationship, not about the world. However if you only engage in operant training when scary stuff happens, then I think that also may end up making the scary stuff more salient. Counterconditioning, because the very nature of it is to only do it when the scary stuff is present, runs an ever bigger risk of forming links in your dog’s mind that make the situation more salient than not.

So perhaps, then, counterconditioning may make things more salient, especially at first. I don’t know. That’s just my experience. I do find those dogs who are highly fixated on the food lose that sensitivity the more they do. I think it’s sometimes a phase you work through with some dogs: ‘Look! There’s the stuff! Where’s my goodies?’ But I don’t think it is a bad thing. What you have then is a dog who is ready to switch to operant cues. When a dog notices the person and then looks back for a sausage, well, then you can ask them to do something else. I wait for this moment, rather than trying to avoid it.

Does the food increase the level of arousal in general? Maybe. I think this is mostly true in dogs where the scary stuff was already salient, though.

For instance, I work with a lot of car-chasing dogs, and I’ve been the guardian of one too. Sometimes, where I’ve paired the car up with food, I never had a moment where I thought with Flika that she was ‘noticing’ cars to get food. Or, even, that she was more sensitive to the presence of cars. Her level of ‘noticing’ cars neither diminished nor increased. Her reactions decreased, though, for sure. They weren’t more salient, but she shouted at them less and chased them less.

However, for my non-car-reactive boy Heston who walked with her, HE certainly knew that cars meant the bar was open when cars came by and HE came to expect food. In other words, cars didn’t become more salient for Flika, but they sure did for my boy who had no previous interest in cars in eight years of his life. Before, cars were a non-thing to him, as innocuous and meaningless to him as trees, bushes and benches. In fact, if I were to imagine him loose in a road, he’d be surprised to find cars around him. What I was doing was conditioning a neutral stimulus, with him.

The level of initial reaction and salience, as well as the usual frequency of reaction informs my choice of counterconditioning as a tool. If the dog is already having a huge reaction to the scary stuff, well, counterconditioning can’t make it more salient. It was already 100% salient. Cars were 100% salient to Flika. If there was a car, she noticed it. Counterconditioning – adding food when she noticed a car – couldn’t make it more salient. The mechanic was 100% salient to Lidy. Adding food when she saw him couldn’t make him more salient.

But, if the dog is not reacting 100% of the time, or the reactions are fairly mild, then if I add food, I may run the risk that I’m actually conditioning a dog to the stimulus by making the pairing of stimulus > food more strong. For instance, in Pavlov’s famous experiment, when the white coat guys came in the room, pairing them up with food 100% of the time will have made the white-coated assistants more salient. That’s what happened to Heston with cars. The food that inevitably arrived for Flika (and him by default) increased the attention he paid to cars.

So if the dog is not reacting 100% of the time, or even 50% of the time, then I may not choose pairing the stimulus up with food because I may actually be conditioning them, rather than counterconditioning them. I’m creating an association rather than changing their emotional response to the stuff.

Because I’m still wrangling with this in my head, here’s another example. Lidy very occasionally alerts to large birds walking in fields. So if she sees a crow, very occasionally, she’ll start to stalk it. That happens probably once in twenty times. Now if I open the bar on crows, well, I’m making crows more meaningful than they currently are. I can see why, if I start adding food when she alerts on a crow, she may start noticing crows more.

In this case, where the reaction is mild or occasional, I’ll probably choose desensitisation alone. Pairing up a relaxed state with low doses of the stuff. I did this with Heston when he was a pup and barked at cows. We walked at a distance with a few very slow cows in our sight for a short period. Gradually we increased the length of time we were around cows, the friskiness of the cows and we decreased the distance to cows. When a cow recently stuck her head through the hedge, Heston didn’t even notice. They were a non-thing. It might as well have been a part of the bush or a curious flower. Cows don’t trip his neural connections of things that are worth noticing. They are just part of the fabric of our environment. Now.

What I’m getting at, I think, is that the aim of desensitisation is to lead to habituation. It’s to help animals return to a state of neutrality. Repeated exposures at low doses and we get used to things. Does the addition of food in counterconditioning actually make the stimulus more salient? I’d argue it may, if the dog wasn’t always finding it salient in the first place.

So that’s my first decision: is this dog reacting to the stimulus often enough where we need to countercondition them and change their emotional response? Or am I aiming for neutrality and habituation?

For example, Lidy’s reaction to people was so extreme that respondent counterconditioning – bar is open, bar is closed – and operant counterconditioning – Leslie McDevitt’s Look At That game – didn’t have the risk of making the scary people more salient. They were already salient. If she noticed them, she reacted to them. Also, given the severity of her reactions and the length of time she had practised the behaviour, coupled with poor breeding of an already sensitive breed of dog, and an extreme lack of appropriate socialisation, and a highly-traumatic single event in her adolescence and I’d like to put money on the fact scary humans are always going to be salient. If you have a pathological fear of zombies or clowns, I’d say they’re probably going to be salient to you compared to other things. You might say, ‘Hey, there’s a clown!’ and have a milder reaction, but those clowns are never likely to become a non-thing, I’d argue.

So if the scary stuff is never going to be a non-thing, then why not add food? Why not countercondition or train around them? You can’t make them more salient. You might as well make the scary clown a cue that ice cream will shortly arrive.

The irony is that from time to time, I do wonder if scary humans and scary dogs are becoming non-things for Lidy. I was prepared for a lifetime of salience. I was happy with Look At That rather than BAT. I was happy with her cuing me for a u-turn if she needed one. I was happy to scaffold her and give her a support structure of pattern games and focus games if it got too much for her. Yet there are more and more frequent times when she clearly notices the people but they are non-things, not worth paying any attention to, and she’ll go back to sniffing the grass or walking or even showing her best sass moves off. We walked past people in the woods the other week without treats, cues, scaffolds, patterns, worries or reactions. They were a non-thing. The problem with desensitisation if it is an extinction protocol as Domjan suggests, is that extinction is weak and faulty process open to renewal and recovery where the initial behaviour comes back with friends.

I guess what I mean to say by this is that there are times for a more pure, foodless desensitisation process where you’re just ‘being’ around stuff at low doses. There are dogs for whom this is all they need. I worked with a pointer last Tuesday and adding food to the mix was just not necessary. He just needed a carefully scaffolded exposure process.

This has a lot to do with his guardians, who were not clicker people whatsoever, and I mean that kindly and without judgement. Their timing was lousy. It was massively complicated. The food got in the way. She was trying to hold a lead and get food out and she was all thumbs. So I may choose a technique depending on the guardian’s skills.

I may also choose desensitisation over food-based techniques with dogs who have a long history of life without humans and where they have been responsible for making their own judgements. Flika was one of these dogs – she wasn’t used to relying on a human to make decisions. I did choose a food approach for her because her car chasing was reliable and extreme. She’d had 14 years of making patently bad decisions around cars. But if the dog would, under normal circumstances, make a good choice, then desensitisation is great. It goes without saying that severe aggression is not a good choice the dog has been making. The pointer was another of these dogs. Given the lives of pointers in France, it’s likely he’d largely been left to his own devices and been responsible for his own decisions. Counterconditioning and operant conditioning are guardian-controlled processes. Desensitisation is not, other than putting the dog in the right place at the right time at the right distance for the right length of time. Thus, for independent dogs or dogs who have been responsible for all the decisions in their past, desensitisation may be the more suitable procedure.

Besides the guardian’s natural aptitudes and the dog’s learning history, I also think about the severity of the behaviour and the salience of the scary stuff. That might make me lean towards a food-based method one the one hand, or avoid it on the other.

Breed fits in here too, as well as temperament and individual traits. I find malinois need a lot of structure and scaffolding. Gundogs and shepherd dogs seem to benefit more from the bond that food will give to the guardian more than other breeds. Independent breeds may struggle to profit from food-based methods of counterconditioning or operant training, but I find the gundogs and shepherds need more direction. If I didn’t direct Flika or Lidy, I’m pretty sure they’d fill in the gap with some vastly inappropriate shepherd behaviour. But that depends on the dog, their history and on you.

It also depends on the dog’s appetites, behaviour and preferences. I do love a dog who can take food in public, but if I’ve got to teach that as a skill before I even do any work because the dog won’t eat in public, then desensitisation may be the tool I’m looking for. Likewise for food-obsessed dogs. If I had good food on my person, I was the only salient thing in the whole world to my cocker spaniel. Once she heelwalked for 5km. We could have walked over glass and fire and she’d have emerged wondering why her paws were in such a mess. Food overshadowed everything.

What I’d conclude by saying is that it’s helpful to think of desensitisation and counterconditioning as distinct processes – one with food and one without. Other than that, they have the same rules: small doses, breaks between sessions, clean set-ups, positive experiences, finishing on a win. If Heston gets a bit shouty around stuff, I’m desensitisation all the way. His phenobarbital/steroids combo is so potent in terms of his appetite that he wouldn’t notice the scary stuff anymore. With Lidy, I’m counterconditioning and operant training all the way. Leaving her to make her own choices will mean she’ll probably end up lunging, barking and attempting to grab the offender. Many of my clients are that kind of dog: they need the structure or their behaviour is so severed that they’re unlikely to make better choices.

There have been many interesting conversations which, while they might not address this topic overtly, address it in passing somehow. Much of my thinking here was kind of intuitive until I listened to Sarah Stremming’s Lemonade Conference slot. Initially, I had a reaction to it and thought leaving out the counterconditioning bit was certainly not going to help the majority of dogs I work with. But then I thought there was a lot of wisdom in her thoughts about just working around scary stimuli, in just the same way that Leslie McDevitt’s materials work with dogs who benefit from the structure of training. Here, the scary stimuli are nothing more than environmental chatter. I’ve also benefited hugely from reflecting on Grisha Stewart’s BAT procedures, where I’ve known dogs who needed more structure than it offered and had a long history of making dangerous choices. In reality, I do a little Grisha Stewart, a little Leslie McDevitt, a little Jean Donaldson – and in implicit levels dependent on the dog, their age, their history, their personality, their reaction strength and likelihood, the salience of triggers and the capabilities of the guardian. I do think we need to be mindful, however, that using unconditioned stimuli of a positive valence (food!) may be working on a different neural network based on reward learning, and in that case, it’s going to trump desensitisation hands down. Suffice to say – sometimes I use food and sometimes I do not, and it depends on a lot of things.

I’d like to thank Sonia, Ryan and the DoGenius students for making me formulate these thoughts into some kind of conscious and explicit process rather than just unconscious and intuitive practice. We all get better through reflection and discussion.

If you want the cutting edge on respondent extinction, by the way, our DoGenius extinction course should be right up your street. So many of the nuances get missed and it’s vital that we understand them so that we’re keeping those tools sharp. Only then can we make the most difference for our learners. There are three hours of lectures that cover the most up-to-date research. Definitely worth purchasing if you want to know more.