Canine learning interference. A dog trainer’s perspective.

Models of canine teaching and learning can be additive and constructive – but if poorly understood can interfere with each other and confuse dog and trainer.

Introduction

Appreciating canine behaviours that “come naturally” - and distinguishing those from learned behaviours - is an important aspect of successful dog guardianship, management and training[1]. However, what are commonly overlooked are the relationships, counterpoints, interference and even conflicts in canine learning and behaviors which arise from the interplay between key learning and behavior paradigms.  How best to manage these issues in canine teaching and behavior modification is critical to training outcomes. A web of complexity can surround teaching stratagems in terms of developing or managing behavior, as models such as Experiential, Social, Cognitive and Constructivism (to list just a few) play into the learning, so, to simplify the understanding as much as possible, we will focus here on behaviors performed/developed through Instinctive influence (the Innate/ Instinctive/Genetic set[2]), Classically Conditioned behaviors and those learned through Operant Conditioning.  

Definitions

Before we consider how learning and behavior paradigms might overlap or even interfere and to keep the discussion on track, we need to first set down some straight forward definitions. We will attempt to keep the discussion at the level of canine training and avoid delving unnecessarily (for the purposes of this article) into the complex realms and interactions of phenotypes[3] epigenetics[4] and cognitive ethology[5].

1.     Instinctive Behavior set

·       Innate behaviours: Those expressed with no specific training or experience needed and do not require any learning to occur. Readily observable examples are the maternal “nesting” in preparation for the birth[6] and the suckle reflex of newborn puppies. A dog may have these innate reflexes conditioned to react to subtle signals. Over time, the dog learns to associate the signal with the event – hence sometimes referred to as “associative learning”- however this same term usually refers to both the Classical and Operant style of learning[7].

·       Instinctive behaviours: Inherited, but occur in response to particular stimuli and are breed-specific, usually with little variation in pattern – can be refined with experience. An example would be the hunting or herding instincts, as well as guarding or protective instincts, stronger in some breeds than others. Instinctive behaviors tend to be reflected in animal spontaneous play actions[8] (but these can be modified/developed with deprivation or practice interactions – so there is an element of learning). Play is prevalent in juveniles, but not uncommonly persisting into adolescence and less commonly adulthood (often following a “trigger” by a juvenile)[9].

·       Genetic (Encoded) behaviors: Animal behavior and the ability to learn is encoded in its genes. That behavior and genetics are closely linked is reinforced by the species evolution of morphology and capability, linked to preferred and practiced actions, under the influence of natural selection[10].

2.     Classical and Operant Learning

·       Classical learning: Classical Learning/Conditioning (or respondent behavior triggered by, or following, an eliciting stimulus) refers to a learning process where learning occurs by association and (generally) involuntary or reflex responses, sometimes undetectable as purely expressed within the animal’s physiology or homeostasis – that is, autonomic reactions. This is the “Pavlovian Conditioning” – the famous bell ring (or the entrance of laboratory technicians) signaling food delivery and hence the uncontrolled reflex salivating by the dog[11].

·       Operant learning: Operant Learning/Conditioning (or instrumental conditioning) is a theory of learning where behavior is influenced by its consequences – so might also be called consequential learning – i.e., behavioral responses reinforced by events which immediately follow. Skinner^[12] is regarded as the founder of Operant Conditioning based on the Behaviourist psychological model, following the works of Watson^[13] and Thorndike^[14] and demonstrated with various animals by the Brelands^[15]. Simplistically, Operant Learning comes from a stimulus or event, responded to by a particular behaviour/reaction, leading to a consequence, either a pleasant (appetitive/reinforcing) or unpleasant (aversive/punishing) outcome.  Behavior that is reinforced (rewarded) will likely be repeated, that which is not rewarded or punished is less likely to be repeated – the fundamental bases of the Behaviourist psychological model. For a detailed discussion of this important learning/teaching model – and how it is commonly misinterpreted, see the companion article on this website The Operant Conditioning Model Fallacy.

Now, although the separation of these learning and behavioural influences is critical[16], it would be a mistake to regard them – as might be convenient for certain discussions – as completely isolated, with no overlap or interaction. That is not the case. Many experiments have been performed and much has been written, for example, about the interplay between Classical and Operant Conditioning[17], identifying Classically Conditioned learning embodied in Operant canine training[18]. Equally important to appreciate is that attempting canine training which goes against (is opposed to) the animal’s innate/instinctive behaviors or encoded genetic reactions, will be extremely challenging, likely leading to guardian/trainer frustration and disappointment.

Also important to remember is the overlay (apart from breed characteristics) which can be undetected, that all learning is influenced by the animal’s physical health, emotional state, environment and previous learning experiences.

Fortunately, there are answers to overcome these training, learning and behavioral challenges, provided the trainer is aware of what the dog interprets a situation or stimulus/trigger dictates – that is, how the canine interprets the signals, as opposed to what the trainer hopes the dog understands.

Overriding all of this quite complex web of canine learning is the wonderful realization that learning engenders learning in canines. The more a dog is trained and experiences positive learning events and “socialisation” (referring here to not simply meeting other dogs, but experiencing different environments and learning opportunities – see companion article on this website Puppy socialisation and confidence building) the more mind expanding learning – apparent intelligence and decision making – will occur.

Behavior and Learning model interactions

To understand canine ethology and how the dog trainer can best interpret how the dog sees things and reacts – and thence position or modify the training to achieve the desired outcomes - we will consider some graphics to explain the interplay of the behavior paradigms and, importantly, what questions might help the trainer’s understanding of how the dog “sees things”[19] and, hopefully, better understand its reaction to a training regime.

Understanding the canine ethology from a biological perspective is key to appreciating behaviors and interactions of training ideologies. Tinbergen’s four questions[20] are commonly applied to observed instinctive behaviors, but can equally offer enlightenment in the context of Classical and Operant Conditioning. For example, focusing on Classical Conditioning,  appreciating what causes a particular stimulus-stimulus reaction, how has this evolved and developed over time and what is its function (for Pavlov’s dogs the answers might, for example, sheet home to making the canine a natural predator and so anticipate the hunt-kill-feed sequence). For Operant Conditioned behavior, the answers to the 4 questions all link to the ability to learn adaptive behaviors as part of survival, or, in modern company with humans, strengthen the synergies and bonds relating to the canine basic needs[21] (Fig.1).

Figure 1 – Model of Animal Ethology and Learning relationships: importance of Tinbergen’s 4 questions to analyse the causality and relationships of especially Instinctive behaviour patterns in order to understand observed stimuli and responses as comprehensively as possible. These four factors might also be collectively thought of (for simplicity) as “influences” on, or “reasons for”, a particular behaviour and its function – and how and why this might evolve over time[22] - and can apply to each of the three illustrated learning/behavior models.

The next important issue is that an attempt to Operantly Condition (train) a dog to behave or respond in a particular way (the “when I cue this, do that and this happens”) will be challenging – possibly thwarted – if the desired outcome/behavior is directly opposed to a genetically programmed instinctive or innate desire or reflex action/reaction (Fig. 2).

Figure 2 – The interplay of Instinctive, Classical and Operant Behaviors from a training or behavioral altering perspective. These can work in unison or can work against each other.

So, we have three concepts of behavior and training which need to be understood, not just  in terms of their cause, evolution, development and function relationships, but also their interference and conflicts (Fig. 3). We can then start to consider how to manage and balance some of the challenges.

Figure 3 – The training-learning-behavior paradigm: how animal behaviors and learning will be influenced by the canine’s genetics, evolution, development and life/environment “niche”- and then impacted by the canine trainer’s guidance and behavior outcome expectations.

Comparing the learning and behavior model interactions

Classical versus Operant Conditioning

Although not necessarily the simplest place to start, the models of Classical Conditioning and Operant Conditioning are certainly the most commonly referenced in the dog training and canine behavior modification world.

These two learning/reaction paradigms are most often separated simply because of the hurdle of understanding how the effects of Classical Conditioning interact with Operant Conditioning, if only because there is not an adequate performance model for the behavioral effects of Classical Conditioning[23]. The interaction effects a canine handler would like to detect can be a subtle change in canine bodily functions, outwardly undetectable “on the surface” from the dog’s body language.

In laboratory studies of Operant Conditioning of rats or pigeons, the experimenter commonly seeks to illicit the subject’s response to a particular stimulus and apply reinforcement (appetitive or removal/avoidance of aversive) causing a bilateral pair of contingencies – that of the stimulus generating the response (and anticipation of the reinforcer based on past experience) and that of the response generating the reinforcer (again based on past experience of receipt of a reward after completing the response). Typically, in these experiments, the stimulus might be a sound or light pulse and the response sought might be pressing a bar or pecking a button - followed by (say) the reinforcement of the release of a pellet of food, referred to in Operant Conditioning literature as positive reinforcement. 

This sequence has also been characterized as three phases (or trilateral contingencies)^[24] -

Firstly, the subject learns that specific appetitive reinforcers are available – this is called incentive learning.

Secondly, the subject learns that certain behaviors influence the availability of the incentive – this is strategy learning.

Thirdly, the strengthening of a specific response in a repertoire of behavioural or action options is characterized as response shaping.

Some experiments construct the stimuli and contingencies as cessation (or non-occurrence) of an active or anticipated aversive, such as an electric shock, in which case this is referred to as negative reinforcement.

Both negative and positive reinforcements (as well as punishments) are portions of the Behaviourist four quadrants (see the companion article on this website The Operant Conditioning Model Fallacy).

In the canine training world, the positive reinforcement sequence (which we think of as being a straight-line Operant Learning relationship of a single cue or direction responded to by the dog’s action) might be a verbal and/or visual cue delivered by the trainer, the response being a movement such as a “sit” or moving towards a “place”[25] and the reinforcement would be the delivery of a food treat (commonly separated by a “bridge” or “marker” the moment the desired action is performed, noting that the bridge/marker is also predictive that a reward is coming and possibly accompanied by positive verbal praise). In other words, the stimulus is predictive and the action produces reinforcement.  

In the case of negative reinforcement sequence, a canine training example is leash pressure which is relaxed once the dog sits or moves as required, which might also be referred to as “escape and avoidance” training. The simplest reference being teaching a dog to sit over several iterations by upward/backward leash pressure which is released (“negatived/ subtracted”) when the desired action is performed, so the pressure release is the reinforcement. The warning signal of leash movement or the handler ceasing forward motion becomes the Operant cue. However, this sequence might equally be seen as a Pavlovian fear elicitor which provides motivation and reward on removal or non-occurrence. Thus, exploring the learning/behavior through this lens, “Classical (Pavlovian) conditioning bears the primary responsibility for the Operant (Instrumental) avoidance behavior”[26] (see further below).

Hence the distinction between whether the learning is purely Operant or Classical (or both) becomes blurred – and cannot readily be separated. This has been referred to by Boakes (op. cit.) as “an additive interaction between the effects on behavior of classical and instrumental conditioning”[27].

Now, for the average canine trainer, does this blurred distinction really matter? We will answer this question in three parts:

Firstly, we must acknowledge that the whole subject of Classical-Operant Conditioning interactions is much debated, making simple and definitive statements on the fallout - as far as dog training goes - quite challenging.

Secondly, we can state that the interaction described does not matter too much as far as the outcome goes, but the knowledge can be used to consider the pathway of “cue-marker-reward” in terms of training efficacy, knowingly empowering the process with this “additive interaction” of Classical and Operant Conditioning. Conversely, the Operant (Instrumental) response may be attenuated – interfered with or diluted – by the Classically Conditioned emotional state, even overshadowing it (perversely creating overshadowing conflict or intensification of positive response, all in the same apparent cue that the trainer is delivering[28]). Clearly this can cause confusion for the dog and the trainer, so this understanding is important – but the dog doesn’t actually care whether the reinforcement/reward comes from Pavlovian or Operant logic and most often – but not always - the two influences act in unison.  So – let’s consider a common canine training example.

In training the “Place” exercise (in which the dog is directed to proceed forward to a particular spot, position or place some distance from the handler/trainer) the verbal “Place”, might not only alert the dog to the forthcoming action (forewarning or setting up a somatic reaction) but might in parallel engender a Classical involuntary  response (autonomic

reaction) such as a neurotransmitter release to a gland or muscle which might trigger an undetectable emotional response, muscle twitch, iris dilation or focus, or even salivation at the anticipated food reward on successful completion (all in a millisecond timescale and outwardly undetectable)[29]. That auditory cue would typically be accompanied by the trainer providing a physical pointing gesture in the direction of the desired end-position, followed by a bridge/marker (auditory “click” or “Yes”) when the dog moves to the target position, followed by the reward reinforcer. Teaching this sequence is commonly accomplished by AutoShaping or Shaping by Successive Approximations – which most commonly does engage both Classical and Operant Conditioning[30].

Does this knowledge that the Place cue might utilise both Classical and Operant Conditioning/Learning change what many modern dog trainers now practice? 

Probably not, but it is useful to think about this bilateral pair of contingencies when devising a new cue and (perhaps more complex) desired action/outcome, being sure that the pairing is indeed additive and does not detract (through interference) from the learning and desired result – the outcome or action. Further, the trainer needs to ensure that the dog is not confused in learning the new Operant cue because of an already ingrained Classically Conditioned response and, as we will discuss, because of an Instinctive Behavior, either or both of which can divert the Operant Learning and, at the extreme, confound it.

Thirdly, there is (of course) another highly debated potential confusion in the form of “Pavlovian second order conditioning”, which involves conditioning – via a second stimulus – often quoted in instances of fear-based avoidance behavior and referred to as a “Pavlovian reinforcer”. This second order stimulus and its impact on the first order stimulus – which may progressively extinguish – does also find relevance in the concept of instrumental overshadowing, where the second stimulus (or cue) effectively replaces the first[31] and in the worst case might confuse the animal under instruction as to what “Operant action/ response” is actually being cued.

So, what relevance can this “Pavlovian second order conditioning” possibly have in the canine training world? In simple terms, the second order conditioning is no more than a second stimulus (cue) applied in a linear fashion after the first stimulus, e.g., in caged rat experiments a bell noise followed by a light flash (or vice versa); in dog training, a hand gesture followed by a word cue - in this canine example, if both are affirmative and reinforcing stimuli with an appetitive outcome (positive reinforcement once the required action is performed) these two stimuli will promote the same desired response/action in the dog. That is, the saving grace is canine intelligence – with repeated events, in most cases, the dog figures out what action delivers the marker and reward and so, for example, a trained dog will sit whether the cue is the verbal “Sit” or the body language hand gesture. However, if the second delivered stimulus is counter to the desired outcome or, worse (and more common in dog training) engenders a poor or fearful reaction in the dog (such as sudden leash pressure/jerking or even as subtle as the handler showing anger or a bullying approach to training) and an aversive outcome ensues, then this “Pavlovian reinforcer” overshadows the first order stimulus and generates fear-based avoidance - and the first order stimulus is gradually extinguished (or rapidly extinguished - depending on the severity of the punisher and/or related fear reaction). Hence, how this sequence can play out in dog training is, quite simply, that applying an aversive Pavlovian reinforcer may well overshadow the dog’s responsiveness – its “willingness” – to train and comply with cues. This effect can be quite subtle, as in the case of handler tension under the stress of a competition, or simply frustration at the learning profile, such that the dog is slow – apparently reluctant – to respond. Canine handlers will commonly notice this effect when a dog is cued (in aversive terms “commanded”) to “sit” or “heel” and the dog responds reluctantly, slowly, with body language indicating unhappiness or fear, such as ears down, tail between legs, lip licking, eyes fearful, etc., that is, all the signs of fear-based avoidance behavior, despite understanding the action being cued.

 Something to contemplate – as mealtime approaches and the dog anticipates food – staring at the guardian and salivating, based on their internal clock and the conditioning which always delivers food at this time – then the dog sits whilst the food bowl is filled, based on the repeated request to sit nicely to receive the food – consider which parts of this sequence plays out as Classical Conditioning and which part as Operant Conditioning – and whether there is any innate/genetic role based on the need for sustenance -  and so how many stimuli are applied?

 Classical and Operant Conditioning versus Instinctive Behavior

What is the importance of recognizing the interaction of Instinctive Behavior (including “play behavior”, part of honing the genetic and necessary survival skills development) with the models of Classical Conditioning and Operant Conditioning? This “competition between motivations” is commonly observed in the non-training environment in which an animal is conflicted by shifting natural motivational priorities, such as whether to eat, drink, escape/avoid a threat, chase sighted prey, etc.[32]

At first blush the answer is quite simple – canine learning will be best facilitated if it parallels the dog’s natural instinctive behaviors encoded in its genes and conflicting motivational priorities are managed. Teaching can then amplify, hone and refine the desired behaviors and activate (or deactivate) them on cue through Operant Learning. What appear to be instinctive behaviors, which are regarded as species or breed specific stereotyped patterns, can be modified or even developed anew through “copy/ imitation learning” and practice[33].

This also means than attempting to train a dog to act entirely differently to its instincts will be extremely challenging – not impossible, but requiring thought and the application of stimulus and reinforcement protocols which will be most effective if they parallel the canine’s natural behavioral predilections.

So, let’s consider some commonly encountered examples.

Example 1A: High herding instinct: Learning to herd sheep will be most readily taught and come naturally to a Border Collie, but not to a Bulldog or Terrier breed. The challenge is to direct those natural instincts and drives into the desired outcomes.

Example 1B: Training a Kelpie not to stalk the ducks – against its natural instincts – will most likely be challenging and better included in the training plan to teach control and choices – such as herding the ducks to the safety of a pen/duck house at dusk and back to the duck pond at dawn, satisfying the dog’s natural “need to perform a herding task”[34].

Example 2: High prey drive: A common challenge with Terrier breeds is their constant desire to sniff, search and chase any small movement – a butterfly, a leaf blowing across their path or any insect or small animal they see.  

This is instinctive behavior bred into their DNA – encoded learning as part of the nature of their breeding with the aim of hunting and exterminating vermin. The challenge is to manage these urges with training which includes the allowance to undertake this highly satisfying activity when so directed or allowed, but not necessarily constantly without paying any attention to the guardian and cues to perform other activities. Many other breeds equally demonstrate prey drive – especially for fast moving prey like rabbits or squirrels, but also larger wild animals such as wild boar or kangaroos. This has been described in the companion articles on this website Should certain dog training devices be banned and Why use electric shock collars? A plea for enlightenment. The answer is not punishment, as some e-collar and balanced training pundits advocate – the referenced articles explain.

Example 3: Guardian breed dogs (Maremmas, Pyrenean Mountain Dogs, Bernese, Newfoundland, Asian Shepherds, certain Mastiff breeds and others) will commonly be alert to what they perceive as potential threats to their “wards”, scenting and visually searching for risks and dangers. These canine breeds commonly behave in a quiet, complacent manner, perhaps excited by friendly overtures but not apparently alert to danger, although depending on socialisation in upbringing can be “standoffish” with strangers. However, if a perceived threat arises – they will generally place themselves in the path of danger rather than attack (as “guard dog breeds” will do – see below) to “take the hit”.

In the companion article on this website  Do we unfairly manipulate dogs?  I describe a means of managing the training regime for basic manners (obedience) with such breeds and empathizing with their prevalent desires/behaviors to determine intrinsically positive experiences as the reinforcement/reward mechanism, especially when a dog doesn’t work for food rewards (nor a game with a toy) and is focused anywhere but on what the trainer is asking. Paraphrasing from the quoted article, although I appreciate that food deprivation to make treats more attractive is an option – it is preferable to firstly try to understand the breed, its natural instincts and drives, upbringing and personality - its age, of course - then try to find exactly what the dog is interested in – what is its motivation? Then link an exercise/activity the dog loves with exercises the guardian aspires to teaching and they feel will help the overall performance and dog satisfaction, as well as the relationship and bond between guardian and dog. The concept here is not that "you have to do this distasteful action first or you don't a reward!"[35] - but rather that sense of "whatever I ask you to do is going to be fun, beneficial and lead on to even better stuff – I know what you love and I will deliver - for both of us together enjoying life and developing your special skills".

Example 4: Guard breed dogs (Doberman, Belgian Malinois, German Shepherd, Cane Corso, Rottweiler, Giant Schnauzer) will (especially if from a “working line”) demonstrate strong natural instincts to protect their guardian and family, generally being constantly alert to any potential intruder or threat. Their first “line of defense” tends to be “prepare for attack” with posture and warning bark or growl (see examples in companion article on this website The Operant Conditioning Model Fallacy regarding a Doberman issuing clear – but self-controlled - warnings to perceived threats).

In all of these cases of managing and training in the context of strong instinctual drives, the key is firstly to assess and appreciate the canine situational/environmental drive, secondly to manage the environmental predicament which will typically trigger the canine drive (such as being off-lead in the presence of the likely targets of prey or herding) then thirdly parallel the instinct and reinforce the desirable choices the dog should make. Superimposing an aversive Classical or Operant escape/avoidance methodology is most unlikely to deliver a lasting solution, will not intensify the canine human understanding and bond, will not teach (or give the option for) the dog to make the right choices and will simply be unsustainable and unreliable in outcomes.

Two important observations

Firstly, as described above, much of the “scientific” evidence and discussion regarding the interplay or interference between Classical and Operant responses comes from experiments with rats and pigeons in boxes or cages. The question for the canine world will remain – is that all straight line applicable to dogs, or is it as flawed as many common assumptions dog training pundits make about the parallels between human emotions and reactions with those of the canine breeds? This is the question of release of innate instructive reactions which are not necessarily transferable across species. Lorenz (1937) introduced the concept of a "releaser," a stimulus that serves to "unlock" or release innate instinctive reactions in animals, deciding that behavior is a taxonomic character and therefore a diagnostic trait of a particular species, not synchronous nor transferrable across species. An overlay – or perhaps, refinement - on this insight was made by Pellis and Pellis (2009) in their conclusion that lumping observed behaviors into a single category may seem to be useful, but runs the risk of “pigeonholing behaviors that only really make sense when species are compared within a clade of related species”[36]. Tinbergen also acknowledged gene related “taxonomic characteristics”. So, a shocked rat experiment scientifically proves your dog will benefit from punishment? Not exactly firm logic, according to these observations.    

Secondly, the “Golden Age” of animal cage experiments in the early 20^th Century relied on strictly controlled learning confines and the determination of rigid models of behavioral learning. Such experiments, in which the test subjects were exposed to often unpleasant constraints and aversives and expected to conform with certain patterns of behavior, were aimed at formulating and testing psychological or behavioral hypotheses. This style of theory testing has waned somewhat, notwithstanding laying down a solid foundation of the Pavlovian and Behaviourism models. In the later part of the 20^th Century (fortunately for the chosen laboratory test subjects) this style of research has been replaced in favour of what might be referred to as a “biological paradigm”. This advanced framework of animal learning formulation recognizes not only differential learning experiences, patterns and animal abilities, but opens the window to evolutionary influences to suit the particular species’ life demands – that is, an element of learning instinctively (or instinctive developmental learning). This is an essential understanding for the canine ethologist and trainer and can be utilized as a powerful canine training paradigm – so some further context is briefly provided below.

Experiential Learning, Adaption and Evolution

The concept of animal experiential learning, memory and consequential behavioral adaption underpinned by evolutionary capabilities and capacities (or “instinctive developmental learning”) opens a whole world of complexities in associative learning interferences and linkages to animal drives and specialized development and application of senses – which we somewhat simplistically refer to as “instinctive and encoded behaviors”, gliding past the recognition of adaptive and predictive behaviors, cognitive planning and, ironically, even what appears to be (perhaps anthropomorphically) a sense of humour (or, at least, well developed playfulness) in many animal species. Although this is stretching the purview of this article, these ideas of genetic-drive linkages, overlapping with cognitive ethology, are certainly not new, but offer such fascinating concepts and canine training ramifications that they are worthy of a least a brief mention here, accompanied by some guidance on further research for those interested.

Hayes (1962)[37] explored the linkage between animal genes, drives and intellect which has been labelled “EPDs” – Experience-Producing Drives – linking behavioural genetics, learning theory motivation, intelligence and evolutionary theory. To go deeply into this subject would require a lengthy article on its own, so here we will just seed the idea that an outcome of evolution is that animals (and initially the concepts started with humans) will seek – and adapt to – circumstances in which they can thrive – or even excel. The “so what” of that is that genes influence behavioural response by what attracts an animal in their environment, how the surroundings/environment, events, objects or movements perceived are emotionally and cognitively interpreted and then responded to in motivated behavior patterns (i.e., the realm of cognitive ethology). The accumulation of these patterns then became recognised as animal “traits”, which canine handlers simply refer to as “typical behavior response of the breed”, which have, of course, over time, been manipulated as part of “managed evolution”[38].

Whilst the “meta-theory” of Experience-Producing Drives has been widely expanded and debated by such researchers as Roper (1983)[39] and Seligman and Hager (1972)[40], to cherry pick just two, the hope is that readers might judge the concept of genes and behavioral evolution worth exploring further. Importantly, one might start to wonder about the Border Collie propensity to herd and the Doberman and Malinois desire to protect – and how these “traits” were amplified through targeted breeding programs – effectively scaled, enhanced and accelerated evolution. Then consider – for these canine breed examples – the relevance and canine interpretation of Operant and Classical teaching in the context of Instinctive Drives, that is, how to train the breed, how such a dog might fit in the modern family home and how to keep it active and mentally stimulated, without developing undesirable habits or “bad behaviors”, purely through lack of cerebral stimulation and sensible challenges aligned with its genetic abilities[41].

Summary observations and conclusions

The canine training paradigms we like to work within – or believe we are working within – are helpful in their apparent simplicity. Classical Conditioning (“Pavlovian salivating”) and Operant Conditioning (“do this and this will happen”) - are easily understood and are commonly the first place to start in understanding or teaching canine training outcomes.

However, there are three important complexities to bear in mind.

Firstly, there are numerous learning and teaching models, such as Cognitivism, Constructivism, Experientialism and Socialism, to name a few better-known examples, which are not better nor worse than Behaviorism, but may operate synchronously and additively to Classical and Operant learning (and the intertwining of those two)[42].

Secondly, when we think we are teaching a dog to respond to a particular stimulus, we need to recognise not only the interlocking of innate, habituative (“desensitization” or “getting used to”) and associative (classical, operant and observational/experiential plus socialisation) learning processes, but the possibility of Pavlovian second order conditioning. Simply being aware of the overlap and possible interference of these influences can help the canine guardian or trainer appreciate why a dog appears stimulated by a particular signal or is confused by two or more inputs and whether or not innate, instinctive and encoded behaviors might in fact be influenced by associative learning events or by overshadowing and extinction. Each of these issues present challenges and possible answers to the canine trainer when faced with a dog which appears to not respond to the standard Operant Training techniques, or appears confused by the cues the trainer delivers, even if only occasionally.  

Thirdly, we need to be open to looking for “adaptive drive and sense driven behavior modifications”, as well as the canine’s general or specific preferences for company or solidarity in a particular situation - and how distracting any or all of these influences and those of the three major teaching paradigms discussed, can be in the training program responses.

Taking all these issues together, the pathway through the maze of teaching-learning paradigms and how or why a canine thinks and reacts the way it does, can be illuminated with the help of applying Tinbergen’s four analytical queries in the context of envisioning the interplay of Instinctive, Classical and Operant diverse behavior drives. For each teaching-learning paradigm, consider:

1)     What caused a behavior or reaction originally (in the canine genes, upbringing or training program accordingly)?

2)     How has such a behavior evolved “naturally” or due to changing training methodologies?

3)     Why/how is the behavior developed over time – again in the breed context (through deliberate and focused breeding to refine a desired set of senses and drives) but also in the training development?

4)     What is the function of these behaviors – in the natural setting (theoretically “in the wild”) – or as designed by human influence and in the norms of life cooperating and in harmony with human guardians.

Then, in facing a particular canine training or behavior modification challenge, a trainer can put the answers to the four Tinbergen questions in the next level context of:

1)     What is the situation in which the undesirable canine behavior is triggered?

2)     How can this situation be managed so that the dog is not triggered, is controllable or at least not distracted (so it can still think and respond to trainer cues)? Then,

3)     What training regime or protocol (including appreciating all the Instinctive, Operant and Classical conditioning influences) can be introduced to modify or create the desired canine response?    

Answers to these will assist in designing training programs and goals and overcome failures and frustrations caused by lack of appreciation of possible interference in the training and behavior response paradigms.

References

[1] As described in the companion article on this website [How Dogs Learn]

[2] Not replacing these relatively well understood terms with the less commonly used (in this context) label of “intrinsic”, which Coppinger, R. and Feinstein, M. (2015) How Dogs Work, Uni. Chicago Press, London, 243pp, preferred for both behavioral and physical “inherited” animal makeup.

[3] Phenotypes – observable characteristics or traits in an individual based on the genes (the genotype) – influenced by the environment.

[4] Epigenetics – how behavior and environment impact the way an animal’s system interprets its DNA – or how the genes work - without irreversible genetic changes.

[5] The canine mental experiences as it behaves throughout its life: Griffin, D, R, (1976) The Question of Animal Awareness - Evolutionary Continuity of Mental Experience; 2^nd ed. (revised/enlarged) 1981; New York: Rockefeller University Press; 209pp.

[6] Unmated bitches coming into season (“heat”) commonly show this behavior as well.

[7] Squire, L. R. et al, Eds. (2012/2014 – 4^th ed.) Fundamental Neuroscience: Academic Press: 1152pp (hardback).  Quote “Associative learning is a broad category that includes many of our daily learning activities that involve the formation of associations among stimuli and/or responses. It is usually subdivided into classical conditioning and instrumental conditioning”. 

[8] Fagen, R. (1981) Animal Play Behavior; NY Oxford Univ. Press; 684pp; provides a comprehensive treatise on play behaviors in mammals. Sivvy, S. M (2010) Play and Adversity: How the Playful Mammalian Brain Withstands Threats and Anxieties; American Journal of Play, 2:3, 297-314 PDf at https://share.google/L8dDtLpQ1pgPYKsNz draws linkages between stress, fear and play reaction – quoting experiments with rats - so again the caution about paralleling canine conclusions.

[9] This concept of adult play triggered by interaction with a juvenile is based on personal observation. Canine guardians commonly relate how a new pup introduced to an ageing dog gives the old dog a “new lease on life” as it plays with the puppy. For a very digestible analysis of animal play (including play fighting) relating to juvenility, brain size and differential brain development and how play establishes tolerance prior to feeding in physical proximity (p. 101) and courtship (pp.105-106), each of which appears to be associated with the regulation of animal stress response (pp.108-109), see Pellis, S. and Pellis, V. (2009; reprint 2017) The Playful Brain – Venturing to the Limits of Neuroscience; Oneworld Publ.; 258pp (reprint). Note, however, that Coppinger and Feinstein (op. cit.) postulate that animal play behavior does not have an obvious function, but is an “emergent consequence of the expression of a developing repertoire of motor patterns” (quote p.181) “keeping intrinsic motor patterns alive and functioning until the entire adult sequence is in place” (quote p.183) – although this does read a lot like “an obvious function”?

[10] Manning, A. and Dawkins M. S. (2012) An Introduction to Animal Behavior; Cambridge University Press; 6^th Ed (reprints 2015/2016/2018/2021); 458pp; Chapter 2 – The Development of Behavior: Genetics and Behavior.

[11] 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

[12] 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.

[13] 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.

[14] 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.

[15] 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.

[16] Despite the advice of Skinner, B. F. (1950) Are Theories of Learning Necessary? Psych. Rev. 57; pp.193-216; questioning the value of learning theory as only creating “a false sense of security, an unwanted satisfaction with the status quo” – advocating that “the most rapid progress towards an understanding of learning may be made by research that is not designed to test theories”. So, although we do list and categorize learning concepts – effectively identifying and separating learning theories – we will attempt to focus on examples, data and facts (and, unavoidably, inferences).

[17] Davis, H. and Hurwitz, H. M. B. (1977 – first publ.: 2021 reissue/update) Operant-Pavlovian Interactions. Routledge, UK (2021 reissue); 327 pp.

[18] The common example cited is that of teaching via “shaping” or “successive approximations”.

[19] This is different to the Umwelt, which refers to the animal’s perception of their entire environment. Here we are focused on the canine interpretation of and reaction to a particular stimulus.

[20] Tinbergen, N (1963) On aims and methods of ethology; Zeitschrift fur Tierpsychologie 20; pp 410-433. A possibly more accessible account of this work (English language) is summarised by Miklosi, A (2015) Dog Behaviour, Evolution and Cognition; Oxford University Press; Ch 2 pp 16 – 38.

[21] Compare the Five Domains model of animal welfare [e.g., Mellor, D.J. (2020) The 2020 Five Domains Model: Including Human–Animal Interactions in Assessments of Animal Welfare; Nat. Libr. Med.; https://pmc.ncbi.nlm.nih.gov/articles/PMC7602120/]. Somewhat parallel – but not exactly – to Maslow’s hierarchy of basic needs - see for example McLeod, S (2026 update) Maslow’s Hierarchy of Needs; Simple Psychology; https://www.simplypsychology.org/maslow.html for description and note Kenrick, D.T. et. al.  (2010) Renovating the Pyramid of Needs: Contemporary Extensions Built Upon Ancient Foundations; Nat. Libr. Med. https://pmc.ncbi.nlm.nih.gov/articles/PMC3161123/#ref-list1 publ. in Perspect Psychol Sci. 2010 May; 5(3):292–314:  original Maslow, A.H. (1943) Psychological Review. Vol. 50; A theory of human motivation; pp. 370–396: Griffin, K.E., et al. (2023) The Adaptation of Maslow’s Hierarchy of Needs to the Hierarchy of Dogs’ Needs Using a Consensus Building Approach; Nat. Libr. Med.; https://pmc.ncbi.nlm.nih.gov/articles/PMC10451230/.

[22] A fuller explanation appears in the companion Article on this website Canine ethology: anecdotes, experience and science

[23] This statement is close to the summary wording of Boakes, R.A., in Davis, H. and Hurwitz, H. M. B. (1977 – first publ.: 2021 reissue/update) Operant-Pavlovian Interactions. Routledge, UK (2021 reissue); 327 pp: at p. 68.

[24] Hurwitz, H. M. B. and Roberts, A. E., (1977) Aversively Controlled Behavior and the Analysis of Conditioned Suppression; in Davis, H. and Hurwitz, H. M. B. (op.cit.) at pp.189-231,

[25] Typically, in “shaping by successive approximations” training.

[26] Quote from Rescorla, R. A. Pavlovian Second-order Conditioning: Some implications for Instrumental Behavior; in Davis, H. and Hurwitz, H. M. B. (op.cit.) at pp. 133-164.

[27] Boakes (op. cit.) in discussing “AutoShaping” as a procedure and a phenomenon; at pp.70-78.

[28] Hurwitz, H. M. B. and Roberts, A. E., (op. cit.) discuss these complexities in considerable detail.

[29] Black, A. H. et al, A note on the Operant Conditioning of Autonomic Responses; in Davis, H. and Hurwitz, H. M. B. (op.cit.) at pp.27-45.

[30] Jenkins, H. M. (1977) Sensitivity of Different Response Systems to Stimulus-Reinforcer and Response-Reinforcer Relations; in Davis, H. and Hurwitz, H. M. B. (op.cit.) at 47-66: and Boakes (op. cit.).

[31] Rescorla, R. A. (op. cit.).

[32] Manning and Dawkins (op.cit.) Ch. 4; discuss these natural motivational conflicts in some detail for a variety of animal species.

[33] Manning and Dawkins (op.cit.) Ch. 2; “The characteristics of instinct and learning”.

[34] A client asked what to do about her two-year-old Kelpie which “acted crazy in her apartment (no outside yard), racing around the furniture barking”. The answer was simple – “Buy a sheep station”. That dog urgently needed a job.

[35] Effectively the Premack Principle, also called the relativity theory of reinforcement and the differential-probability hypothesis (and amusingly referred to as “Grandma’s Principle” – “eat your greens then you get dessert”), states that a more desirable activity can be used to reinforce a less desirable one. For a fuller general discussion refer URL https://www.simplypsychology.org/premack-principle.html.

[36] Konrad Lorenz (1937) The Companion in the Bird’s World: The Auk, Volume 54, Issue 3, 1 July 1937, Pages 245–273, https://doi.org/10.2307/4078077: Pellis, S. and Pellis, V. (op.cit.) entire Ch 6; but especially p.110.  

[37] Hayes, K. (1962) Genes, drives and intellect; Psychol. Rept. 10; pp.299-342. Also see review and discussion in Bouchard, T. J. and Johnson, W. (2020) “Keith Hayes” Experience-Producing Drives: An appreciation and extension, Personality and Individual differences; Univ. Edinburgh Research Explorer; https://doi.org/10.1016/j.paid.2020.110082: Although Hayes may have been instrumental in capturing these concepts, he was preceded by many researchers, notably Baldwin, J. M.  (1896) A new factor in evolution; The American Naturalist; 30; pp.441-451.

[38] This “managed evolution” lands on the periphery of the “Darwin vs Lamark” debate – viz; is evolution driven by non-random beneficial changes which are inheritable (Lamark) or by random undirected change sifted and reformed by selection (Darwin)? See for example, Koonin, E. V. and Wolf, Y. I. (2009) Is Evolution Darwinian or/and Lamarkian? Biology Direct; 4:42; 14pp.

[39] Roper, T. J. (1983) Learning as a biological phenomenon; Animal Behavior; V. 3; Genes, Development and Learning; eds. Slater, P. J. B. and Talliday, T. R.; Oxford; Blackwell Sci. Publ.; pp.178-212.

[40] Seligman, M. E. P. and Hager, J. I. (eds.) (1972) Biological Boundaries of Learning; Englewood Cliffs, NJ; Prentice-Hall.

[41] Failure to recognise these issues is why periodically so many Dalmatians (the Hollywood 101 Dalmatian Effect), Staffordshire Bull Terriers/Pit Bulls (possibly the “Once Upon a Time in Hollywood” effect) and, more recently, Malinois (the “John Wick 3 – Parabellum” effect), are consigned to canine shelters, because their enthusiastic past guardians had no concept of what they were bringing home and what cerebral and physical enrichment these breeds require to be fulfilled and well balanced.

[42] Refer companion Article on this website The Operant Conditioning Model Fallacy.