The Operant Conditioning Model Fallacy

Misunderstanding canine Operant Conditioning

Overview

Modern canine training methodologies are based on two fundamental models of behaviour. These are Classical Conditioning (Pavlovian Conditioning[1]) and Operant Conditioning (Skinnerian Conditioning[2]). Skinner[3] is regarded as the founder of operant conditioning based on the Behaviourist psychological model, following the works of Watson[4] and Thorndike[5] and demonstrated with various animals by the Brelands[6]. Simplistically, operant conditioning describes a learning process where behaviours are influenced by consequences. That is, the learning comes from a stimulus or event, if responded to by a particular behaviour/reaction, leading to a consequence, either a pleasant (reinforcing) or unpleasant (punishing) outcome – the fundamental bases of the Behaviourist psychological model.

When translated to the canine training world post the 1950’s era (until then dominated by such dog trainers as Most and Koehler[7]) the Skinner model has been captured as the “four quadrants of operant conditioning”, shown in Figure 1, below. As labelled in this diagram, the terminology bears some explanation, as the quadrant’s scientific/technical descriptor labels do not necessarily mean exactly what one might at first assume. In fact “negative” does not mean bad, it means “take away or remove” (i.e., “minus”); “positive” does not mean good, it means “add-in” (i.e., “plus”); “reinforcement” means to encourage or consolidate a certain behaviour or reaction (desirable or undesirable, although usually the former – an “undesirable reinforcement” example is the dog having self-reinforcing experiences, such as the dog barking at the postman, the postman goes away – success – job done – do the same next opportunity) to occur again in the future as a reaction to the stimulus; “punishment” means to deter a certain behaviour or reaction (usually undesirable) not to occur again in the future as a reaction to the stimulus[8].

The misunderstanding inherent in this model

The common belief in the “four quadrants of operant conditioning” is that they should be applied by a balanced dog trainer[9] with equal regard, aligned with the assumption that to be “balanced” is a good thing (which it may well be in life in general, but not necessarily in this technical model construction) and that every living being encounters life learning opportunities in each quadrant[10]. This latter argument invariably references human examples[11] and really has little to do with understanding optimal dog training methods and, no – sorry – the direct linkage is not obvious. Dogs do learn about “life” in all quadrants through the multi-faceted concept of “socialisation” or “experiential” learning, but that does not translate to a necessity to use the punishment based quadrants of the operant conditioning model in dog training. The catch in this argument is that Skinner’s and, indeed, Watson’s and Thorndike’s, conclusions regarding reinforcing behaviours through consequences, although often quoted, are rarely understood or explained – this “balance” is misunderstood and is an incorrect conclusion (illustration below)[12].

But wait – hold the phone – before the “Balanced Trainers” get back on their horse, I am not about to tell you that your dog must be shielded from all forms of stress (the argument debunked in the companion Articles on this website AVSAB vs Balabanov analysis and Reinforcement vs punishment: A dog trainer's perspective). I am also not about to insist that positive reinforcement is totally, absolutely control or pain free and does not limit a dog’s “agency” (see companion article on this website “Do we unfairly manipulate dogs?”)  – but we need to define “punishment and the linkage to pain” in the dog training context (further discussed below). Nor am I going to argue that you can never say “No” to your dog (that mild aversive commonly referred to as a “non-reward marker”) because I do not regard that as punishment, but part of the “redirection protocol”, arguably appropriate in some circumstances but unnecessary in others (avoid the “Aha – so you admit to using aversives” and read on). Furthermore, the discussion and views expressed are aimed at companionship dog management and training (commonly referred to as “pet dog training”) as well as training for the dog sports of Obedience, Agility, Hoops, Frisbee, Dock Diving, Gun Dog Retrieving, etc., in other words all-inclusive general dog training. Does it apply to the very special military and police dog training of high drive dogs, such as the Belgian Malinois? Well, yes – and no. Yes, absolutely, in the general terms of reference as to the ethics and benefits of positive reinforcement training and teaching dogs to think and respond[13]. No – in terms of the details of what these special purpose dogs are required to learn to overcome/combat and the actions they need to perform in their special jobs.

 We will come to all that – bear with me.

There are three points to consider:

Firstly: The Behaviourist Model was not intended nor stated (possibly despite some Behaviourist follower’s desires) to mean that one had to apply all of the four constructed reinforcement/punishment option classes to teach a sentient being anything. It would be amusing – if it hadn’t caused so much angst in the dog training world – to think that this was the only and superior psychological model, when in fact, at its simplest, it was postulated to be an aid in understanding a theory or concept (as all models are), in this case what might simply be referred to as “cause and effect in reactive or behavioural outcomes”. The model was (and still is) certainly believed to be profound and a valid means of understanding behaviour (complimenting or contrasting with such models as Cognitivism, Constructivism, Socialism and Experientialism) but it was not intended as any kind of rule, instruction, stipulation or strict and singular guide to teaching. Rather, the beauty of the Behaviourist Model is its fundamental simplicity in providing a framework of understanding the relationships between actions and consequences on a learning path – do this and that happens – Skinner did not compose a “little red book” based on Behaviourism that some dog trainers claim must be adhered to. There remain many other options which might come into play separately or synchronously (illustration below)[14].

Secondly: The 4-quadrants are not really equally balanced (as has been taken up by labelling “balanced dog trainers”). This confusion presumably arises because the Skinner “teaching options” are most commonly represented just as shown here, as four equal quadrants of a circle or equal quarters of a square graph. The misdirection is that the actions (positive/negative reinforcement/punishment) were never intended to be utilised in some kind of equal proportion as the image and the label “balanced training” infer. Rather, they illustrate an academic model to categorise certain processes in an effort to clarify what might be considered in the “Behaviourist” psychological model of Watson’s, Thorndike’s and Skinner’s groundwork.  Remember the adage that “all models are wrong; some are useful”[15].

Thirdly: Skinner et.al. really concluded that positive reinforcement (that single quadrant/quarter in the model) is the most effective, lasting teaching methodology, which actually encourages further learning. Remember that much of this groundbreaking “Behaviourist” logic was expounded following generations of thought that forcing an outcome in behaviour was the only way to get anything done – by humans, or any animals.

Now, does this all mean that the Skinnerian Behaviourist Model is so misunderstood and misrepresented that it is therefore irrelevant and should not be considered by dog trainers? Certainly not – the terminology is entrenched in the dog training vocabulary and it remains a most helpful and enlightening model. But then, does it necessarily mean that, if one is to discount the “four equal quadrants” concept, there is only the one option - that of purely positive reinforcement - a dog trainer should consider in a teaching tool kit? Strictly, no, not necessarily, there are certain choices to be made based on appreciating the ramifications of applying each “quadrant action” whilst avoiding the “equal application/balanced” concept – thereby acknowledging here that we would be better to drop the descriptor “quadrant” altogether – it is used because it is part of the common parlance. However, the important point is that the pros and cons of applying all options in the operant conditioning model do need to be carefully considered.

Pros and cons of punishment versus reinforcement

Always bear in mind that we are thinking in two parts. One is the Skinnerian Model of Behaviourism (or some other psychological model which dictates or develops animal learning and responses to stimuli, as in the boxed diagram above) which might be applied or situationally relevant and, the other, is all about how to actually (and practically) train dogs in various circumstances – and what end does the trainer have in mind? Some real-life examples of training logic and outcomes are described below.

Staying with the Behaviourist Model, how do we decide what “training technique” (or part of the Skinner Model) could or should be applied? Let’s start by providing some simple examples of dog training actions which might be applied in each portion of the Skinner Model, as shown in Figure 2 (and later see Figure 3 below). We can then analyse what each quadrant of the model can mean, how it works and what might be the ramifications of application – always bearing in mind that we are not agreeing that the common interpretations of how to apply the total model are correct.

Punishment – positive and negative

Positive punishment in this model (Figure 2) can be understood as directly applying an unpleasant stimulus during/after an undesirable/maladaptive action/behaviour to deter future repetitions of that behaviour.

Negative punishment can be understood as directly removing a potentially rewarding stimulus during/after an undesirable/maladaptive action/behaviour occurs to deter future repetitions of that behaviour.  

Before proceeding, just as we defined “aversives” in the companion Article on this website Aligning "positive reinforcement" with "use of aversives" dog trainers, we need to be clear about the definition of “punishment”. At the extreme, punishment might be the application of a highly aversive tool (prong collar/e-collar) deliberately causing pain (although often tongue-in-cheek passed off by the pseudonym “stimulation”) or as innocuous as a verbal “non-reward marker” (“No”/ “Wrong”/ “Oops”) or a conditioned stimulus/cue, such as “Wait”/ “Stop”/ “Steady”/ “Leave it”. The distinction is that these latter are trained cues the dog will normally respond to, but in a high drive or emotionally charged situation these may be accompanied by, say, leash restraint. Importantly, the guardian needs to be cognisant of the dog's breed, nature, temperament and age in the situational context and then how the dog “experiences” the punishment. For example, the protocol to divert a Malinois from active prey drive will be different to cueing a slightly recalcitrant Spaniel to “Sit”.

Further, there is a somewhat semantic refinement that argues it is not the dog which is “punished”, it is the unwanted/undesirable behavior/action. The logic for this is focused on what “punishment” is in the model – an influence to reduce the likelihood of the behavior being repeated/occurring again. Yes – true – but the impact is the same – the dog receives the punishment; it is not some abstract derivative or embodiment of the behavior or action that is punished, in the end, it is the dog.

As always, remember that we are discussing canine training and behavioural management/ realignment here, not “clear and present danger” safety dramas which may justify an immediate and more severe response (or the mild and arguably justifiable/permissible conditioned “No – Leave It”).

A simple example of negative punishment is not reacting to a dog’s jump-greeting, which will, over time, reduce the likelihood of repeating that action towards the person who reacts in that controlled manner, especially if, for this example, the dog is rewarded for “four on the floor” greetings (adding positive reinforcement to the training regime). To explain – the “negative” here is taking away the human greeting/patting/ acknowledgement of the jumping dog, the “punishment” here is reducing the likelihood of repeated behaviour because there is no benefit in performing the unwanted action - no encouragement to repeat the behavior. This is definitely different to “punishing” the dog with a push, a knee, a kick or a yell, which, perversely, dogs can actually find encouraging, because it is an interaction, even a game.

That being (hopefully) clear and returning to the combined positive and negative punishment quadrants, the issues with both of these actions are:

·       The undesirable behaviour is suppressed at the time, but not necessarily forgotten or permanently transformed[16].

·       There is a risk of the punishment actually being perceived as reinforcing (such as attracting the attention the dog desired) or of creating a response engendering fear or aggression – depending on the dog’s temperament, the situation and how abrupt and severe the punishment is[17].

·       There is a lack of guidance – what is the desirable action/behaviour to replace the undesirable action/behaviour? This is “punishment without learning” and the underlying issue that causes/stimulates the action/behaviour is not addressed.

Reinforcement – positive and negative

Positive reinforcement relies on establishing a clear and beneficial connection of “cause and effect”- that is, if this action/behaviour is demonstrated (mostly on cue) then this reinforcement (reward) will happen. The reward is most effective if marked/bridged with a “Yes” or “click” and properly timed (very soon after the desired response/action/behaviour – usually instantly or at least within a few seconds), thoughtfully proportional and (over time) attention is paid to the intervals of reinforcement (i.e., continuous, fixed, or variable - an extension of the topic, but in brief, essentially delivering improved learning and desire to repeat the action/behaviour).

Arguing against positive reinforcement, “Balanced” dog trainers are fond of quoting Kohn (1993) [18] as evidence which challenges the Skinnerian Behaviourist view, in particular that reinforcement alone is not the answer it is purported to be (inference: by the force-free/positive-only cohort) – in fact it lessens the core interest in learning itself (which, in human terms, might be referred to as “self-actualising”) and undermines the animals “wellbeing”[19]. Two challenges to this evidence to bear in mind in the canine framework are:  

a)     The argument is wrapped in the extremes of constant rewards (in the dog world that would be feeding treats non-stop for every iteration of desired behavior – see comment above on intervals of reinforcement) and;

b)     Kohn’s studies and thesis all relate to humans (ranging from children to workers) and proffers advice to parents, teachers and people managers – not dog trainers. Does that make it all irrelevant? Perhaps not – but caution should be exercised in that automatic assumption that canine brains, temperament and emotions are all identical to those of humans, which (most of us like to think) have a different (more complex/higher) plane of cerebral dynamic operating.

Negative reinforcement requires that an aversive (such as a leash/collar-pop/pressure or e-collar application) is first applied so that it can then be removed (“negatived”) so as to reinforce the action or behaviour the dog then exhibits (such as the “sit” or “recall” following the leash tug or e-shock). This terminology is a source of major confusion in the model, because this quadrant really calls for a “punishment/painful” action to be negatived (removed) – so it feels like it should be relabeled. But remember the meaning of the term in the special model context.

The three aversive quadrants

Admittedly, the interpretation and especially the application of the four elements (quadrants) is rather more complex than they might appear initially, nonetheless, the quadrant model is simplified if each quadrant is seen for what it is – what it represents in terms of applying grades of reinforcement/rewards or aversive/punishing actions in dog training (see Figure 3)[20].

We should note a concept regarding the aversive nature of the three quadrants so labelled in this diagram, viz., the application and aversive actions and results are clearly not the same for each of the three quadrants. So, yes, this is a sweeping, all-inclusive label.

Hopefully, pure positive punishment is readily understood for what it is, albeit nuanced in some interpretations, according to the definition, administration and level of punishment. In any case, positive punishment can be understood to always be aversive – even though there may be substantial variation in severity of application. So, let’s examine the remaining two quadrants and decipher the aversive nature of these and how they are applied.

Negative punishment has been discussed above. We can offer the thesis that it is, in most applications, the “aversive quadrant” with the least “aversiveness” – because it is simply removing something (normally removing a benefit or canine self-rewarding opportunity) without applying a punishment (audible or physical) per se[21].

The “negative reinforcement is really positive reinforcement” argument debunked

Negative reinforcement requires an aversive to be first applied so that it can then be removed so as to reinforce the action or behaviour the dog then exhibits. A simple example is pulling the leash up and back to direct a “Sit”, the leash tug is then released (negatived) once the Sit occurs and that Sit action is therefore reinforced. This is referred to as “escape learning” (because the dog escapes the leash pressure). The construct that believers in balanced training would make is that this escape learning transitions to “avoidance learning”, whereby the “threat” or “anticipation” of the aversive prompts the required response – and therefore this negative reinforcement is really no different to the positive reinforcement process of “cue - bridge/mark - reward”.

Now, this might well seem like a real stretch (even for the convoluted logic the psychology of dog training is used to) but wait – there exists “scientific evidence(!?)”, in the form of some measurements of rapid dopamine[22] release in rats being shocked, then avoiding that shock by pressing a button and measuring a dopamine spike at that point, suggesting that the whole experience is actually favorable/exciting/pleasant[23]. What this “scientific extract” appears to be used to demonstrate is that, in reality, negative reinforcement is just another form of positive reinforcement because (maybe) it’s not really aversive and (astoundingly) “negative reinforcement creates resilience and confidence” and that (quote) “the key to effective dog training lies in understanding the neurobiology of pain and pleasure. By achieving a balanced approach, we can guide our dogs towards greater mental resilience and well-being”[24]. Hhmmmm (?) – readers will have to make their own assessment as to the “rat analogy” from the first quoted research and whether this proves “scientifically” what it is purported to[25] and whether the leap to “prove” balanced dog training as the way to maintain dog brain homeostasis and build resilience is really relevant to your chosen training methodology. Also note that interpreting dopamine concentrations (high or low) is not simple and not purely an indicator of all things good – there are conflicting research views, specifically that dopamine release (linked to cortisol release) can be associated with anxiety-based stress.[26]

This author’s view is that the “negative reinforcement generates dopamine spikes” argument presents a custom-made excuse to use snap-tugs on the leash, a prong collar or an electric shock to train a dog to simply sit or heel and overlooks both the full biochemistry of triggering spontaneous hormone release (in this case dopamine) and the Behavioural Psychology findings that positive reinforcement engenders a desire to learn quickly and learn more – (and, I suggest, think for themselves to make good choices) without any hint of fear of an aversive being involved – and builds the canine-human relationship, trust and bond.

Returning to a parallel between positive and negative reinforcement - yes, the trainer could add a positive reinforcement reward when the desired action is executed following negative reinforcement – but that doesn’t change the procedural pathway – why impose the punishing element in the first place rather than, say, simple luring (see Figure 4 and discussion below)?

In addition to the above “logic” regarding negative reinforcement, there is a belief that animals do not know or experience “fear of pain”, although admitting that fear based on trauma (which could be regular punishment applied in training regimes) is locked into the canine hippocampus[27] and that using aversives in training without incremental and thoughtful implementation is likely to cause fear conditioning and even chronic stress (distress)[28]. Confusing? Somewhat. This is scientific logic that catapults the discussion into the realms of epigenetics and battles between the endocrinology and psychology arenas[29]. Perhaps a better description than “fear of pain” (sought by those who don’t believe in fear of pain) then is “apprehension of a bad outcome” (an outcome to be feared?). The alternative concept (purported to be “scientifically demonstrated”) is that avoidance behavior/reactions are fundamental to survival and are NOT associated with fear, as avoidance reactions can be demonstrated without fear being induced. Of course they can – no real surprise there. The avoidance (with repeated experience) becomes “habitual” (a conditioned reflex not dissimilar to a spontaneous reaction?) just as the Pavlovian conditioned stimulus triggers an automatic response – postulated as a dopamine reaction (an upward spike) rather than fear – and, it is argued that the idea of avoidance based on fear has been “debunked”. Wow! Really? Watching many bad dog trainers with cringing or “frozen” dogs, this seems like a rash generalisation and more than a bit hard to digest, even though it sounds pretty plausible when “proven by scientific evidence” (albeit taken at face value, out of context and ignoring alternate empirical, canine observation-based evidence: see further below) – but make up your own mind of this proof and its relevance to canine behaviour and training[30]. This author can’t quite get there, having witnessed the interplay of canine discomfort, apprehension and fear (call it what you will) leading to cringing body language and shutdown (learned helplessness[31]). Maybe it is all a matter of extremes of application, contextual predicament and dog experiences and learning, but either way there is more to this than what a suite of “dopamine spike rat experiments” in the laboratory tells us – beware the pitfalls of “scientific absolutes”![32]

This in turn opens a Pandora’s Box of research on “the emotional brain” on the general subject of fear as one of many emotions - with an enormous literature (hundreds of research publications) from Charles Darwin (1872)[33] to Le Doux (2012)[34] and beyond, where the debate rotates around the putative “primary (hard-wired) emotions” of (especially) mammals, which (it is argued) essentially link to survival[35] - and then merges into what are innate versus learned emotional responses (nature versus nurture in “emotional survival circuits”). Fascinating stuff, but moving beyond the intended scope of this article.

To put a stake in the ground, it is a very long bow to draw to suggest that applying aversives in dog training builds canine resilience and confidence – rather than instilling apprehension of punishment or aversive treatment and, again – No – it is not suggested that dogs should be sheltered from all forms of stress, just from bad stress imposed by impatient, intolerant trainers. The preferred view is that resilience and confidence are built through broad socialisation – including play - delivering learning and training which provides the canine with challenges it can rise to – calling on the psychological models of no less than Behaviourism, Experientialism, Constructivism and Socialism[36]. That is, positive reinforcement would work every bit as quickly and certainly more humanely, without the risk of overdoing the pressure or pain[37]. So, in summary, can we put this piece of “scientific proof” down to being a bit of neurological science interesting fun-fact, but of questionable help to the dog training world – at least without further thought, especially regarding the complete role of neuro-transmitters, the degrees of application, plus the psychological, contextual and behavioural comparisons[38]?

Quadrant conclusions

The over-riding take-away from all of this dissection of the Skinnerian quadrants is that the positive reinforcement approach is typically more effective for long-term behaviour change and bypasses any negative emotional associations, such as fear of the guardian/handler/ trainer, or at least a lack of positive bonding. As stated in the companion Article on this website “Puppy socialisation and confidence building” – “Animal training should never have followed that type of laboratory science down the path that promoted pain and especially long-term pain {and we might now add “repeated punishment” – both potentially leading to} (distress or fear of pain) to change behaviour. In fact, Skinner himself determined that “positive reinforcement (that single quadrant/quarter in the model) is the most effective, lasting teaching methodology, which actually encourages further learning”[39].There is never any reason to use pain to teach basic manners or compliance to a puppy. Language and gestures can be used far more effectively than any aversive procedure.” 

A note to scientists. In publishing and quoting scientific research findings, scientists have a responsibility to stick to the facts and be cautious about drawing or inferring conclusions beyond what the data actually relates to without accounting for correlating or contradictory evidence. There is a saying about wanting to believe what they like to hear is true – a form of confirmation bias. “Science” is like a point source of light – it needs to be seen in context, repeatedly, from many sides and understood in its sphere of influence and relationship to other data and disciplines, then built upon to find its place in understanding the world – don’t be blinded by the light. This does not mean that no conjectural questions can be raised, of course, but there is a parallel to “be careful what you wish for”, because people are listening and holding your authority in some regard. A similar note might be issued to globally well-known dog trainers who carry the combined weight of credibility and responsibility.

Applying the positive reinforcement training method

Positive reinforcement, at its simplest, is providing the dog with a cue (a “direction”, an “ask”, a “requirement” – avoiding the old term “command”, simply because it relates to outmoded training techniques and just sounds unnecessarily “military”, “demanding” or even “bullying”). In a little more detail, the positive reinforcement training process can be thought of in a stepwise manner, as shown in Figure 4. The training method starts with luring using hand gestures – usually with a food treat in hand – followed in time by a verbal cue[40], moving to either luring the dog into an action/position without food then bridging/marking (“yes” or clicker sound) then rewarding (food/petting/play) then, eventually, at the desired response, the dog will follow either the verbal or body language cue (alone or together) and intermittent rewards are delivered on some set or variable schedule. Mission accomplished – over time, with persistence and patience (not in an edited, 10-minute, clickbait YouTube video).

What flexibility in training might be required?

“Positive only” training - and even genuinely force-free training without a collar and leash[41] - can be shown to be enormously successful with willing puppies and many very willing or compliant breeds. Examples are often shown on social media with Border Collie/Australian Shepherd breed dogs (who are notoriously willing, compliant learners) and delightfully enthusiastic Malinois pups and certain Terrier puppies - and dogs not having any entrenched undesirable behaviours.

However, when faced with a less compliant dog (for example, one which often appears to test or challenge the handler – i.e., simply reluctant to comply[42]) or an older dog which has developed some “less than desirable/maladaptive” behaviours, what options are there? This is where the trainer is tempted to turn to “balanced” training, in which the right or requested behavior is rewarded and the wrong/undesirable behavior is punished. The good news is that this can be done without punishment, dominance or severe correction in the training – and definitely without the excessive force of prong collars or electric shock collars - but rather that an effective training regime must be found. This will worry the cohort of “positive-only/force-free” trainers – but let me explain.

Firstly, keep in mind that:

1.     Puppy owners need to know that rewarding good behaviours as a preference consolidates good dog behaviour.

2.     New puppy owners are deterred from obedience training when they hear about “punishment” – no matter how that word is explained, they just didn’t get a pup to punish it.

3.     There is a concern that “positive-only/force-free” training (a) never tells a dog what not to do and (b) leaves dogs with established bad habits out on a limb – unable to be retrained to correct behaviours – because they need to relearn and differentiate incorrect/correct behaviour outcomes.

The solution

The word “punishment” is replaced with “redirection”.

What does redirection mean – isn’t it simply a “correction - which is another word for punishment, or at least a correction which grades into a punishment”? No – “redirection” adds the concept of “restructure of the training”. Just like when explaining a concept to a person and they don’t understand – we commonly try to explain it by looking at the issue from a different angle (for humans, this might be a variation on the verbal argument, using a different illustrative example, drawing a physical picture, or even acting out the solution).

For example – repositioning a dog following a broken “stay”, then leaving a shorter distance and a shorter time – aimed at establishing success and “positive reinforcement” (reward) outcome totally overcomes the negative connotations of “punishment” for the breach and shows the dog what went wrong and what to do instead, in an appropriate build-up to the final desired outcome – realizing that either the environmental distractions or the duration (in this case example) was beyond what the dog was ready for/able to cope with. That is, when a dog does not follow a cue which the trainer thought the dog knew well, this does not require punishment nor even a non-reward marker. Rather, rethink the training method or dog’s status in the program, the distractions, the impact of the dog’s nature (breed characteristics, behaviours and personality), even other training the dog has been through - and try again – with patience. Yes, there is an element of “control” in this protocol, but it is a long way from what most would regard as “punishment” (remembering our definitions and semantics above).

Always think about what might be the consequences of delivering a punishment (or aversive action) – do you really want to accept the ramifications – the collateral damage – of the punishment administered? If in doubt, revert to the thought that you want to be the dog’s best friend, just as it is yours. Whatever the action you are contemplating, would you do that to your best friend - and will you still be their best friend afterwards?

Real life examples

As a concluding summary with some personal examples illustrating this latter point, of a dog seemingly not obeying the trainer’s cues, or the need to intervene in a behaviour sequence, hopefully readers will understand why punishment – in any variation or at any level - for not following a cue would be inappropriate under certain conditions and when scenario management or a mild intervention might be appropriate.

Example 1: This involves a 2-year-old Dobermann bitch named Andrasta[43] and is in two parts – neither of which elicited reprimand or punishment for the dog acting differently to the handler’s instructions.

a.     When trialing this dog in a formal obedience test, with 10 other dogs of various breeds, during the final test of the long-duration out-of-sight sit-stay, suddenly a group of children playing chase, ran around the front corner of a nearby building at speed. The young boys were chasing the girls – all in harmless good fun – and the girls were screaming loudly (in fun, excitement and, maybe, to enliven the chase). All of the dogs on the nearby sit-stay remained locked in position – except our Dobermann. She jumped to a standing position and barked a single severe warning – eyeballing the boys but without moving from the spot – clearly signaling the threat she perceived and warning that, if required, she would step in to “save” the girls. The children (all part of families familiar with dog trialing) realised what had happened when they heard the loud alarm bark and stopped the chase – exactly what the dog requested – the dog sat again, as she knew that no further action was required, apart from keeping a wary eye on those boys.

Now, I can imagine certain readers thinking “Oh boy, this guy is just making excuses – he didn’t “proof” the dog enough so that it would only activate on command – I could make that dog obey – see - it should have been punished if it ever broke from a Stay – maybe give it an electric collar shock!”. To that I would answer that the aim in this Dobermann training, recognizing the strong instinctive protection nature, physical prowess and remarkable cerebral dynamics of the breed, was to build a confident dog which would demonstrate good manners and respond willingly to cues (the fundamentals of recall, sit/down, stay, fetch, seek-find-retrieve, etc.) - and make good choices, especially in regard to behaviour involving children. We were not trying to build an automaton or robot dog, nor suppress its instincts (if you want that, my suggestion would be to think carefully about the end-game before you have a Dobermann or German/Belgian Shepherd/Malinois as a companion).

Obviously, this dog received zero points for the exercise, because she “broke the Stay” – but, despite some disappointment of not qualifying for the test, we did not reprimand her in any way, because we were so proud of her instinctual judgement, appropriate level of response and self-control (she could have covered the distance to those boys in under 2.5 seconds had she decided direct action was warranted – she made the choice that the warning stopped the chase and that was enough). Interestingly, very few people with contestant dogs understood the subtleties of what had happened – we heard a few “tut tuts”.

b.     On another occasion, this same Dobermann was placed on an off-leash sit-stay outside a small shop in an alley which was a walking thoroughfare for shoppers. After a time, checking on the dog from inside through the shop window, we realised that a group of a dozen or so young school children had descended on this dog and were crowded around, cooing over her and patting her. The dog sat calmly and actually looked a bit embarrassed and gormless – like a black and tan marshmallow - as she enjoyed the attention and adoration, whilst holding the sit-stay position. Suddenly, an unfortunate, very intoxicated man in a disheveled state, came lurching down the alley towards the children, muttering, yelling and gesticulating. The Dobermann stepped straight through the crowd of children – without disturbing them – took up a classic “Defcon 2 Dobermann-ready-crouch” position and emitted one loud, guttural and threatening growl with a flash of teeth towards the man, who came to his senses and left the scene rather rapidly. Message delivered – “These children are in my care – I am their protector - you will not harm them; you will not come any closer!”. The dog returned to the sitting position amongst the children and the petting and cooing continued. When the children finally moved on, the dog remained in place. The “sit stay”, including this interruption, lasted a total of 30-40 minutes.

Although we were on alert and ready to act when the dog growled at the drunk, at no time did we feel compelled to leave the shop to intervene. When we eventually finished in the shop and left the scene, we did not reprimand the dog for breaching the requirements of the “Stay”, but congratulated her on a job well done – and making new friends.

Example 2: This involves my Landseer Newfoundland bitch, Nemesis[44], who is now approaching 7 years old. This dog, typical of the breed, has a natural instinct to “save” people from the water (probably related to a combination of the water-dog, retriever and guardian breed instincts filtering through their genetics as this breed developed). She is a powerful swimmer and she is particularly fond of children – just thinks they are the best thing on earth and wants to meet and greet every one of them she sees – even babies in prams. Playing to this natural talent/inclination, this Newfoundland is trained for water rescue. She will swim out to a “drowning swimmer” (play-acting for training) and, with the aid of a special water rescue dog harness, she will swim to a position to enable the “drowner” to grab the harness, then she will pull them back to shore with her powerful style of “breast-stroke” swimming, aided by her large paws, webbed toes and tail rudder[45] – no action by the swimmer at all, they just get towed to safety. Now – she just loves this activity and it is amazing to watch – she will even take a rope in her mouth and pull an adrift row-boat back to shore. This training is evidenced on a number of YouTube videos linked in the Footnote[46].

However, when walking her on a dog friendly beach, if there is a toddler even standing near the water’s edge, or children in the water, splashing and squealing with delight, I cannot have her off-lead, even on a strict “heel” (for which she is well trained and practiced) because she becomes frantic that the children really need to be kept away from the dangers the water presents to them, or rescued from the water. She stares and whines, trembling and straining with every fibre of her body to take action. This requires management and calming. Imagine the parent’s reaction on seeing this 50kg, hairy great black and white dog hurtling along the shore or plunging through the water headed straight for their kids. How stupid would it be to punish this dog for breaking a “heel cue” because of wanting to rescue these children? So, it is managed and the potential catastrophe of frightened parents (and possibly unwilling rescued children) avoided. Once again, this is not making excuses about a dog which hasn’t learned the heel properly, or learned proper impulse control, it is about being sensitive to the strong instinctual drive of the breed and their wonderful, devoted nature. In the event a genuine water rescue is required, I do not want the dog to think “I was punished last time I tried to rescue someone, maybe I better not do that again”.

Example 3: When my 11-month-old Scottish Terrier bitch (Bonnie Brae – 8 kg) plays with my 8-year-old, large and robust, Russian Blue cat (Dr. Pepper – 6 kg), both give chase in turns and have fun, the cat commonly initiating the game by stalking Bonnie around the sofa, leaping on his “victim” with a swipe then racing off. But, the Scottie (quite typical of Terrier breeds) has no “off-button”, so very occasionally a “Bonnie – No – Leave the Cat” is appropriate to hit “pause” on the chase when Bonnie is getting over-excited - and give Dr. Pepper a chance to leap onto a chair or back of the sofa, to call a halt to the fun and take a breather. Yes, this is applying a mild aversive (with a non-reward marker added to a learned cue) – and it is only marginally classifiable as a safety/emergency situation (remembering that this cat can leap tall furniture with a single bound – and has the advantage of claws, which are never actually deployed in these games). But the action is warranted in the circumstances. Application of a mild aversive, but no fear, lasting trauma or imbedded aversion/avoidance.

What these examples illustrate is that whatever dog training methodology one chooses to apply, it should always be done with patience, perseverance, understanding both the dog breed and the broader issues and, of course, compassion. Think holistically about what you are really trying to achieve.

Summary and conclusion

The “four quadrants of operant conditioning” should not be thought of, nor be applied, by a dog trainer, with equal regard, aligned with the assumption that to be “balanced” is generally a good thing – it is not necessarily so in this technical model construction. The Skinner et.al. conclusions do not support this literal interpretation of the model, because the 4-quadrants were never locked-in as equal and were not set down as an instruction manual - they are merely an illustration of a model. The conclusion of the “Behaviourist” psychological studies and modelling was actually that positive reinforcement (that single quadrant/quarter in the model) is the most effective, lasting teaching methodology, which encourages further learning and, we postulate, enables choices and decision making in unrehearsed predicaments.

There may be options a dog trainer could consider beyond positive reinforcement only – the questions are, “Why” and “How”?  Training decisions should be carefully thought through in terms of what is driving the trainer to consider these measures - is it impatience and frustration – or just the visibility of punishment-based training techniques on social media promulgated by otherwise credible “professional” dog trainers? Then, what is the impact on the human-canine bond and, importantly, the collateral damage of applying the (highly not recommended) punishment-based training regime on any dog, especially when that subject is following strong instinctual motivations in its actions and responses.

To revert to comment in a previous Article on this website (Reinforcement vs punishment: A dog trainer's perspective) - the discussion of training methods for the specially trained military and police dog working dogs (commonly very high drive German Shepherd and Malinois breeds and crosses) does not mean that because such canines are regarded as “tough dogs”, which can handle the rigours and discipline of the jobs they perform, that the weighting of operant conditioning is necessarily to more aversive training techniques. The type of canine suitable for these disciplines means that there are special requirements in these fields of training of dogs of temperament and a level of drive which fall outside the band of what most dog owners encounter and scenarios of danger and confrontation which the average dog trainer just does not get exposed to. So – an open mind is required. There may be more to this than what we can readily see, because the public are not privy to the details and the published research in this field is limited, especially real-life boots-on-the-ground studies (for obvious reasons) [47]. There is, however, strong evidence that a modern-day military/police dog trained with positive reinforcement techniques is more likely to respond to its partner/handler and “think for itself” when confronted with an unrehearsed situation[48], a field rich in further research opportunities[49].

In any event, to return to the theme of this article, the Operant Conditioning Model has been taken far too literally and out of context in the general dog training world – I am simply trying to redress that imbalance.

The mantra is - “if you want to be the dog’s best friend, just as it is yours, then whatever the action you are contemplating, ask yourself would you do that to your best friend and will you still be their best friend afterwards?”

This discussion will be extended in a future article questioning why certain dog trainers so favour the electric shock collar (don’t get me started – yet – that storm is coming!).

References

[1] Pavlov, I. P. (1897/1902). The work of the digestive glands. London: Griffin: (1928). Lectures on conditioned reflexes. (Translated by W.H. Gantt) London: Allen and Unwin

[2] Classical conditioning (also known as “Pavlovian conditioning” or “Respondent Conditioning”) occurs when two stimuli are repeatedly paired: a response which is at first elicited by the second stimulus is eventually elicited by the first stimulus alone. Operant conditioning (“instrumental conditioning” or “Skinnerian conditioning”) is a method of learning that uses rewards (likely to be repeated) and punishment (prone to happen less) to modify behavior. Note that the punishment part of the equation is not advocated in this discussion – rather the techniques involve “redirection” of the dog’s focus and emotional response. See companion articles on this website: Reinforcement vs punishment: A dog trainer's perspective. and AVSAB vs Balabanov analysis.

[3] Skinner, B. F. (1951) How to teach animals; Freeman: originally published with the title “How to Teach Animals” in Scientific American Magazine Vol. 185 No. 6 (December 1951), p. 26: Skinner, B. F. (1938). The behaviour of organisms: an experimental analysis; New York, Appleton-Century: (1951) How to Train Animals; Scientific American, 185(6), 26–29.  (1976) About Behaviourism; Vintage books (Random House) NY; 291pp. Operant conditioning can be defined as modification of behaviour by the reinforcing or inhibiting effect of its own consequences.

[4] Watson, J. B. (1924: 4^th Reprint 2004)) Behaviourism; Transaction publishers; Intro. By G. A. Kimble; 251pp Further reprints 1997/2017; Routledge; 276pp: Watson argued against punishment as a behaviour changing technique on a broad base, ranging from a position against child beating and abuse through to enlightened techniques of factory management.

[5] Thorndike, E. L. (1898) Animal intelligence: An experimental study of the associative processes in animals; Psychological Monographs; General and Applied, 2(4), i-109; Thorndike, E. L. (1911). Animal intelligence. New York: Macmillan.

[6] Breland, K. and Breland, M. (Introduction by Bailey, M.) (1996) Animal Behavior. Republ (2018); Storymakers Inc.; 352pp: Breland, K., & Breland, M. (1951). A field of applied animal psychology. American Psychologist, 6(6), 202–204. https://doi.org/10.1037/h0063451: Breland, K. & Breland, M. (1961). The misbehavior of organisms. American Psychologist, 16, 681-684.

[7] Most, K (Colonel) (reprint 2001) Training Dogs: A Manual; original 1910, translated to English from the German 1954; Dogwise Publ.; 214pp: Koehler, W. R. (1962) The Koehler Method of Dog Training: Howell; republished 1996 Hall & Co. USA: 378pp (Koehler trained a number of dogs for Disney movie roles). Note that we are not stating that everything these famous authors wrote about dog training is wrong – for example, Colonel Most referenced “primary inducements”, which might be seen as the precursor to the luring and reward methods in modern dog training – it is simply that the total concept of dog training has moved on and away from coercive training techniques, without disposing of some past “good ideas” which can still find a place in the positive reinforcement training regime.

[8] To add to the confusion – and seemingly to make the model appear to be more “scientific”, without really adding value – an abbreviation nomenclature is commonly used as R+ (positive reinforcement); R- (negative reinforcement); P+ (positive punishment); P- (negative punishment).

[9] We even hear the description “truly balanced” – as if there is an advanced knowledge and sophisticated version of the methodology. Perhaps this is no more than searching for a new terminology to describe a canine training model which is positive reinforcement based, but allows redirection of undesirable behaviours or responses to cues, without aversive punishment, or a descriptive concept of a “more enlightened” (?) view of finding the right way forward in training a particular dog – the inference being that the trainer is cognisant of the dog breed, age, past experiences, temperament, personality, senses and drives, etc., and what exactly is the targeted outcome of the training – then applying “the right balance of training methods”? But then, sadly, perhaps it actually reiterates the belief in using reinforcement and punishment variants in equal proportions is something a trainer should do, because “the model says so”?

[10] A common misconception is that a “good dog trainer” must apply all four quadrants of operant condition (YouTube example - Operant Conditioning in Dog Training: Explained With EXAMPLES! Suburban K9 Dog Training;  https://youtu.be/oXWf8bA_kWc?si=Lzas2V9746Yyblj8 at 5.33 minutes).

[11] E.g., see YouTube video, Dr. M. Uhde: Positive Reinforcement is NEVER Pain-Free - and 2 Other Dog Training Myths Debunked: https://youtu.be/heyX-5avUHs?si=6-o0iQaPzHHeX4qU

[12] On first learning the “Skinner 4 Quadrant Model”, I concluded that Skinner had a lot of bad stuff to answer for. I later learned that it is the ill-informed interpreters of that model who have a lot to answer for! There are certainly issues that we might find with the Skinner experiments – the application of electric shocks (which sometimes could not be shut off) and a significant level of food deprivation to ensure that the rats and pigeons used were food motivated to respond – not things to be recommended in dog training.

[13] Palman D (2020)  Not Using Force in Police Dog Training; Maine Warden Service; United States Police Canine Association; URL https://uspcak9.memberclicks.net › assets › Resource (PDF); Quote “On the street with a police dog, the dog’s ability to think is an invaluable asset. All of the dog’s natural abilities are put to work for the handler. All the situations that are encountered in real work can never be duplicated during training. I would much prefer to have a thinking dog at my side instead of a preprogrammed machine that has learned to react because he fears punishment”. 

[14] This diagram is for illustrative purposes - not intended as an exhaustive picture of psychological models nor teaching-learning pathways – and the “older” (mostly pre-1912) psychological models referred to as “introspective psychology” by Watson (op. cit.) which focussed on consciousness rather than behaviour, are not shown. Note that although there exists a separation of Pavlov’s Classical Conditioning (focussed on conditioned reflex) and Skinner’s Operant Conditioning (focussed on conditioned response), the two models and explanations work hand-in-hand to offer explanations of many behaviours and reactions of sentient beings to stimuli – taking the two models/theories together delivers a richer explanation than always separating them. Experienced and observant dog trainers will often wonder what role the other models listed might play, if and when they become relevant, even (going out on a shaky limb) on occasions, overshadowing Behaviorism and Pavlovianism in modelling a dog’s reactions to learning.

[15] George Box (1919-2013): see URL: https://asq.org/about-asq/honorary-members/box

[16] Note that a dog may learn not to jump-greet the guardian/handler, but still perform this behaviour when greeting other humans – so multiple training examples will need to be undertaken.

[17] Skinner, B. F. (1976) About Behaviourism; Vintage books (Random House) NY; 291pp: Aversive stimuli which generate a host of bodily conditions are the stimuli which function as reinforcers when they are reduced or terminated. Negatively reinforced behaviour may be strengthened to the point that the subject acts compulsively or aggressively, or may move to escape. When behaviour is punished, the punished behaviour is then displaced by incompatible behaviour conditioned as escape or avoidance. A punished person remains inclined to behave in a punishable way, but avoids punishment by doing something else instead, possibly just doing nothing (at p. 68).

[18] Kohn, A. (1993/2018 – 25^th Anniversary Ed.) Punished by Rewards: The Trouble with Gold Stars, Incentive Plans, A's, Praise, and Other Bribes; Harper-Collins; 480pp.

[19] e.g., YouTube: Dr. M. Uhde: UNTOLD - Science, Ethics, and Dog Training with Ivan Balabanov; https://youtu.be/-MoGINOVV2g?si=R0kARh4X5sMZmrlR] and for a “human” application explanation see YouTube “Optimal work: Punished by Rewards”: @ https://youtu.be/OCtU_glyhGo?si=u299EKVp6ttGVLlX.

[20] An easy-to-follow exposé of these matters around reinforcement/punishment – more in relation to human behaviour example – can be found from Simply Psychology: McLeod, S. (Updated March, 2025) Operant Conditioning: What It Is, How It Works, and Examples; at URL https://www.simplypsychology.org/operant-conditioning.html 

[21] An argument is sometimes presented that withholding a reward a dog was anticipating is itself terribly stressful and therefore severe and cruel. This really is a stretch – we are not talking about locking a dog in a small enclosure without food for days at a time as a “negative punishment”. Keep it real.

[22] Dopamine is a neuro-transmitter and hormone – and precursor to adrenaline. Dopamine is commonly referenced as an indicator of a sensation of wellbeing (pleasure/satisfaction) but it also plays a role in mood, motivation and movement. Dopamine imbalance (high or low) is linked to various mental states, such as depression, psychosis, schizophrenia, as well as aggression, lack of impulse control and even ADHD.

[23]Dr. M. Uhde E.B. (2025?) Why negative reinforcement creates resilience and confidence; YouTube https://youtu.be/BfjVVbYXR8g?si=IA-msqyiN-VqRBNw quotes “Oleson et.al, (2012) J Neuroscience 17;32(42):14804-8; Sub-second dopamine release in the nucleus accumbens predicts conditioned punishment and its successful avoidance {Note: the nucleus accumbens (NAc) is a subcortical brain structure known primarily for its roles in pleasure, reward, and addiction}: Extract to show the relevance (or otherwise) of this piece of “science”: “……rats could initiate an avoidance response by pressing a lever within a warning period, preventing foot-shock. Alternatively, once foot-shocks commenced, animals could initiate an escape response by pressing the lever, terminating foot-shock. This design allowed us to assess sub-second dopamine release events during the presentation of a warning signal, safety periods, and two distinct behavioral responses. We found that release consistently increased upon presentation of the warning signal in a manner that reliably predicted successful punishment avoidance. We also observed sub-second dopamine release during the safety period, as occurs following the receipt of reward. Conversely, we observed a decrease in release at the warning signal during escape responses. Because of this finding, we next assessed dopamine release in a conditioned fear model. As seen during escape responses, we observed a time-locked decrease in dopamine release upon presentation of a cue conditioned to inescapable foot-shock. Together, these data show that sub-second fluctuations in mesolimbic dopamine release predict when rats will successfully avoid punishment and differentially encode cues related to aversive outcomes”.

[24] Quote from Dr. M. Uhde (2023/4) The Pleasure Paradox: When Rewards Turn into Punishment in Dog Training: YouTube https://youtu.be/WbgAe8SLsGY?si=eVn0rVKrgMhguiFr. Similar interesting and attenuated logic is dissected in the Article on this website AVSAB vs Balabanov analysis.

[25] We have an Australian saying that “Bull$#!t baffles brains” – this might be a case of “science baffles behavioural logic”, based not on canine research, but going back to the Skinner Box rat experiment regime and making slippery stepping-stone jumps to canine training based on a single “scientific” quote (compare with further research referred to below).

[26] Anxiety induced stress has been evidenced to be associated with significant dopamine release in animals and humans, closely linked with cortisol release – in the human studies all wrapped in the cloak of past life experiences such as maternal care experiences [Pruessner, J.C., et al (2004) Dopamine release in response to psychological stress in humans …. etc. …. J. Neuroscience; 17:24(11); pp.2825-2831] – so this “dopamine spike indicates everything is all good” is not a simple conclusion to draw in isolation (and see Footnote 21 above).

[27] Dr. M. Uhde: My top 3 brain facts to level up your dog training skills; YouTube https://youtu.be/Bg0fIRdTk3g?si=Tp1Jolfoji93zFRa 

[28] Dr. M. Uhde (2023/24) The Pleasure Paradox: When Rewards Turn into Punishment in Dog Training; YouTube https://youtu.be/WbgAe8SLsGY?si=bvylpcRlNITJJc7J

[29] An interesting contemplation of parallels might be made with the infamous and unethical “conditioned fear stimulus” experiments of the 1920’s (the Little Albert experiments and others) described by Watson (op. cit.). Note that these studies led to the introduction of the concept of “unconditioning” (or “reconditioning”) – now in the dog behaviour world taken into “desensitization and counterconditioning” protocols.

[30] Refer Dr. M. Uhde summary and her analysis in “Avoidance, Mis-diagnosed: How We Got It Wrong—and What Science Reveals” on YouTube at https://youtu.be/gDUnCLTGQ3s?si=3sqMjRhjIQHVKdjQ referred to above.

[31] Yes – learned helplessness is real – see discussion in companion article on this website Reinforcement vs punishment: A dog trainer's perspective.

[32] There are numerous of these experiments reported by a cohort of researchers (mostly involving the key player, Oleson, E. B.); e.g., Oleson, E.B. (2012) et al; Sub-second dopamine release in the nucleus accumbens predicts conditioned punishment and its successful avoidance; Jour. Neuroscience; 17;32(42):14804-8: Oleson, E. B. and Cheer, J. F. (2013) On the role of sub-second dopamine release in conditioned avoidance; Front Neuroscience; 7;7:96: BUT - compare studies such as Overmier, J. B. and Seligman, M. E. (1967) Effects of inescapable shock upon subsequent escape and avoidance responding; Journal of Comparative and Physiological Psychology; 63(1):28-33 (a complex subject, but essentially this research found that inescapable shock interfered with escape-avoidance response to new situations).

[33] Darwin, C. (1872) The expression of the emotions in man and animals: London Fontana Press; Paperback (2009) Penguin; 432pp.

[34] LeDoux, J. (2012) Rethinking the Emotional Brain; Neuron, 73/5, P. 1052 https://www.sciencedirect.com/science/article/pii/S0896627312001298

[35] There are postulated to be at least six of these basic primary emotions, including fear, anger, happiness, sadness, disgust and surprise – but then these are paralleled by “survival circuits” such as defense, maintenance of energy and nutrition, fluid balance, thermo-regulation and reproduction. A complex interplay of primary emotions and survival requirements.

[36] Refer discussions in companion Articles on this website “How Dogs Learn” and “AVSAB vs Balabanov analysis”.

[37] Comments by Michael Ellis in YouTube podcast with Dr. Uhde; “UNTOLD - Merging Science and Dog Training with Michael Ellis [Part 1]”: https://youtu.be/5T3llrHcnC8?si=QNTo2uq6UQuW5VLh especially following at around 42 minutes into discussion relating to dogs attacking their police/military handlers when punishment is applied in this negative reinforcement regime. And consider Lorenz, K (1974) On Aggression; Harper Paperbacks; 324pp: “aggressive energy accumulates …...and…… if blocked, this energy may be redirected to substitute objects, or it may explode on those nearby”.

[38] Dr Uhde is clearly well-educated, knowledgeable and a serious contributor to the dog training debates – appearing in numerous videos and podcasts as a passionate, articulate and entirely engaging researcher and presenter - but that does not seem to help avoid the pitfall of finding a piece of “science” to prove what a person already believes and then apply it with a slight shading but without a full analysis of the ramifications of holding that position. The mishap is compounded by confirming that labelling something as “scientific evidence” is the be-all and end-all – without taking into account the leap from rats to dogs nor psychological evidence or impacts and interpretations – and then accepting plaudits from others for “telling the truth” – such as “The Canine Paradigm” Ep 342; She blinded me with dog science: https://open.spotify.com/episode/7E41V81eMJQIgCt4jfl5hC?si=IsbLxi_MT7aBIfj3bnw7JQ. Note: Dr Uhde education/experience is Bachelor’s & Master’s Biology; Ph.D. Natural Sciences: Postdoctoral Research; Training companion dogs/Mondioring; Observation of village dogs-Thailand & Jamaica: Certified Canine Athlete Specialist.

[39] Skinner, B. F. (1976) About Behaviourism; Vintage books (Random House) NY; 291pp: Aversive stimuli which generate a host of bodily conditions are the stimuli which function as reinforcers when they are reduced or terminated. Negatively reinforced behaviour may be strengthened to the point that the subject acts compulsively or aggressively, or may move to escape. When behaviour is punished, the punished behaviour is then displaced by incompatible behaviour conditioned as escape or avoidance. A punished person remains inclined to behave in a punishable way, but avoids punishment by doing something else instead, possibly just doing nothing (at p. 68). Skinner is also credited with the training technique of “shaping by successive approximations”, which very successfully utilises positive reinforcement to teach complex actions or sequences. Note that even this has been criticized as frustrating a dog by denying rewards – this is addressed in the Article AVSAB vs Balabanov analysis on this website under the heading Aversive training unnecessary vs “If there was a better way….” ( Footnote 18 in that article).

[40] Some trainers prefer to deliver the verbal cue first, before the lure action. The order of events presented here (physical luring gesture then later/sequentially adding the verbal cue) is based on the understanding that dogs learn and follow body language (i.e., gestures) more readily than learning words. We do find in practice that dogs learn to associate – and differentiate – the verbal and body language cues and with training will respond when used separately or together.

[41] Easily practiced in the home or in a safe, fenced yard.

[42] Note that we are not describing “dominance” here – refer companion article on this website

Canine dominance – is it a real thing?

[43] Iron Age/Roman era Goddess of victory, both of war and death. Andraste (alternate spelling) can be asked to assist those in a fight, or those who have passed, to help grant them victory and peace in life and death.

[44] Goddess in Greek mythology known for representing balance, justice, and vengeance. Her name translates to "she who distributes or deals out," highlighting her role in ensuring fairness and retribution; sometimes takes the form of a griffin. In common parlance, the meaning translates to being the enemy of ill-fate or wrong doing, setting the scales right (like saving someone from the fate of drowning).

[45] Demonstrated in YouTube video https://youtu.be/LqDMdwH2Vls 

[46] YouTube water rescue training videos of Nemesis, my Landseer Newfoundland: https://youtu.be/rggKtg8Oub4: https://youtu.be/AticPtFbe_A: https://youtu.be/sN7CJMiTO70: https://youtu.be/m1qyheoiCEA : and see the video at https://youtu.be/LqDMdwH2Vls to see Nemi’s powerful swimming action and the use of her “tail rudder” for steerage. 

[47] Department of Defence (USA) (2019) U.S. Military’s Dog Training Handbook: Official Guide for Training Military Working Dogs; US Air Force; Rowman & Littlefield | The Lyons Press; 304pp.

[48] Crippen L (2017) Military Working Dogs: Guardians of the Night; TRADOC; URL https://www.army.mil/article/56965/military_working_dogs_guardians_of_the_night#. A few extracts from this article will help illustrate key points on dog temperament and training techniques: “The suitability [of dogs for the service] rate runs around 50 percent …… to produce 100 serviceable dogs per year, the program will attempt to train about 200”: “All of the dog training is based on positive reward or feedback”: “When dogs get assigned to field units, they want a dog that's motivated, really wants to do the task and is really happy to do it - all through reward-based systems. Of course, all these rewards must be instinctive prior to their certification. They have to be able to do this without any reinforcement - other than the handler's praise and affection": “Every military working dog is an NCO - in tradition at least. Some say the custom was to prevent handlers from mistreating their dogs; hence, a dog is always one rank higher than its handler. That's out of respect.": Also Palman (op. cit.). 

[49] Haverbeke, A. et al  (2008)  Training methods of military dog handlers and their effects on the team's performances; Applied Animal Behaviour Science Volume 113, Issues 1–3, September 2008, Pages 110-122, describe harsh training methods in the Belgian military environment and observe the “depressed/fearful” (this author’s words) body language of the dogs after training – and suggest a more positive reinforcement regime would work more effectively.