Canine ethology: anecdotes, experience and science

Science, anecdotes and dog guru’s stories in understanding canine behaviour and promoting training regimes.

Introduction

Dog trainers referencing personal experience with dogs to draw “proven conclusions” regarding canine behaviours and responses to stimuli, very often miss the middle premise. That is, they leap from one observation to a general conclusion without considering all the factors - environment, history, breed characteristics, uncontrolled contingencies, contaminating events or influences, etc. The next mistake is that, based on an established confirmation bias, they quote several other cases from their experience to prove their point of view, commonly just stating that “they have trained hundreds of dogs” and thus (we assume) they “know what they know”. Equally, science is often misquoted as irrefutable evidence of one understanding or another of canine ethology and even the “proven scientific” way to best train dogs. That direction is also plagued with assumptions, lack of solid or conflicting evidence and misrepresentations.

Observation data

Based on training hundreds of dogs in their work, many dog trainers launch a “statistical inference” as to how best to train dogs or alter undesirable canine behaviours. Impressive - but a fraction of the dog personalities and behaviour of tens of millions (or historically and globally, hundreds of millions) of dogs have actually been sampled across dog personalities, circumstances, environments and life experiences. So – how much credence can we put on this advice? That commonly comes down to a personal choice in the “student” hoping to learn something, but some guidelines might help the thought process.

There is very often a pretence that the expert observer “knows what is in the dog's mind”. This also assumes some special insight into the dog's current emotions, experience and forward intentions. Quite a leap. Now – although the Dunning-Kruger[1] effect may be playing a role here - are they always wrong in their conclusions? No, not necessarily. But it requires more thought and dissection than is commonly applied. Humans quite naturally tend to interpret the behaviours of animals in the context of their own metrics - viz., how they (the handler/trainer) think and “what they know”. The problem (obviously) is that this “knowledge” might well misalign with the reality - whether or not that “reality” is accurately observable and interpreted by an un-compromised, un-biased observer.

So, these pundits are not necessarily wrong, but certainly very often have incomplete information, poorly thought through and poorly explained. Such observers tend to draw qualitative conclusions based (usually) on anthropomorphic parallels. These are, at best, arbitrarily categorised into human-like behaviour or syndrome bundles, without the benefits of either a matrix in which to legitimately categorise observations and ethological interpretations, or even a full understanding of, not just the options, but also the influences on the observed behaviour (such as the dog’s history, training, environment, etc.). This is pseudo-science at its most dangerous, teetering on the precipice of totally invalid conclusion’s, because of unfortunately being believed by ardent followers of the particular pundit based on unrelated credentials - such as, for example, being a prizewinning dog trainer in some field totally unrelated to canine ethological study and scientific process, or the prolific uploader of YouTube videos demonstrating brutal techniques to (apparently) calm “savage” dogs, repeating the same flawed “fix” time after time[2].

Now, again, does this mean that every element of such demonstrations, observations and conclusions and, by extension, every act of training they promote, is completely worthless? Of course not, but the risk is that they are taken as fact and the best (fastest) way to train or to reprogram such unwanted canine behaviours by the laity, based on the red herring of “unrelated authority”. Rather than what they really are, simply a point of view, a thought - and maybe a possible guide for further analysis, study and future proof or, in the case of abusive/punishing animal training techniques, total avoidance.

In summary, single case examples – even when lumped with similar observations of other dogs by an “expert” dog handler - are all based on the handler’s experience (all too often a single experience repeated on unfortunate canine subjects over and over again), observation skills and pre-conceived ideas. This means that these examples are of value only in so far as they suggest avenues for follow-up studies to determine patterns of behaviour, if indeed patterns exist and are a useful guide to predicting canine behaviours and training techniques – or avoiding them where they are based on outmoded training paradigms and completely lacking in appreciation of canine body language.

Science and data validity

So, is the answer to only reference quantitative “scientific” data, rather the experienced dog handler’s qualitative general observations? Can either of these inputs be properly classified as empirical evidence-based data? Unfortunately, no.

Firstly, we need to appreciate the basis for any observation or experiment. This means defining and understanding the parameters on which controlled versus uncontrolled “scientific or general observations” differ in the planning and interpretation of outcomes.

·        Uncontrolled situations provide for observations with no interference, but with a particular question in mind the researcher/observer is attempting to answer. That is, the dog is in an “open chosen predicament”, but without influence on (control of) its reactions, so a high degree of spontaneity can be assumed with respect to the untrammelled reactions/behaviours observed.

·        Controlled situation observations, by comparison, involves an observer waiting for an act of an expected particular event under those specific “controlled circumstances”, in which case the design parameters in which the event and observation will occur, are defined and restricted to that particular control or constraint matrix.

Why is this controlled versus uncontrolled distinction important? Because it will influence the interpretation of the observations according to the knowledge of restricted options the subject has in reacting to a stimulus or behaving spontaneously in the open-ended environment. The conclusions may be wildly different. In its simplest form, just think about the situational parameters, the environment, the auditory and physical experiences of the subject under observation, how controlled or contrived these are (or are not) and what behaviour or reaction is left purely to spontaneity and ”innate or natural” responses without constraint.

Secondly, all experiments – qualitative or quantitative and whether labelled “scientific” or not – have inherent flaws built in to their design and data collection, which might be overcome by repeated trials on numerous subjects in varying conditions, but flaws arising due to preconceived outcomes as well as the “unknown unknowns” very often creep in.

Thirdly, specifically considering measured indicators of canine reactions (such as heart rate, body fluid chemistry – e.g., measuring serotonin) experiments also have inherent weaknesses relating to timing (the canine action or exposure to when the sample is taken - what is the correct time gap?) and basic relationship - what is the relationship between the elements measured and the experience of the canine – good or bad? To illustrate using a simple example, is the measured indicator showing “good stress” (a fun game or petting by the guardian or running a fun agility course) or “bad stress” (experiencing punishment, confronting the “hairy green monster”, or other fear induced stress)?

Equally, in considering the world of anecdotal qualitative data with laboratory controlled quantitative data, the latter lab-based observations commonly suffer from “unnatural” situations in a test run or, in the extreme, in a laboratory experiment, neither of which necessarily relate well to real world practical dog behaviour management or training situations[3]. Consider the information collected when a dog is in its “home environment” with its guardian present, compared to a staged experiment, in whatever environment.

So, overquoting “scientific evidence”, just as over quoting “personal observations”, is not absolutely the answer - and generally is not even helpful - in furthering the understanding of canine ethology, but rather all should be regarded as “point sources” of certain indicators, not conclusive proof that the pundits would often have us believe. If that discipline of simply noting information as “data and observations” would be maintained, then both the experimental and observational information has some value in constructing an idea, a concept, an hypothesis. We need to go a lot further in measuring, quoting and drawing conclusions from what we might regard as evidence – “scientific” or experience based “anecdotal”.

Key considerations in analysing behaviour

Tinbergen[4] described four key considerations in analysing the causality and relationships of behaviour patterns, in order to understand the observed ethology as comprehensively as possible. These factors (which might also be thought of, for simplicity, as “influences” on, or “reasons for”, a particular behaviour) are:

1. What is the function of the behaviour in terms of survival and fitness for the lifestyle requirements? This might readily be answered at first glance at a superficial level for the lion and antelope in the wild, but for the modern canine is a little harder to devise. As an example, perhaps functional fitness for life as a hunting dog might be considered, which in turn contributes to survival in terms of food and a reason to be maintained by a human cohort, somewhat similar to an ecological niche if thought of in the natural/wild state.

2. What is the canine’s evolution in terms of fitting into the anthropogenic niche? Think of the design development of dog breeds so that they develop the features to be “fit for purpose” desired by the human guardians. This evolution has impacted behaviour patterns – largely through the enhancement or suppression of targeted natural senses and drives as well as conformation. Recall that there are over 400 dog breeds – mostly targeting a specific niche in human society - so this question is the foundation for a lot of thought and analysis and needs to be considered in drawing ethological conclusions from tests or observations of a certain breed or mixed breed of dog.

3. What is the mechanism by which behavioural patterns are influenced (e.g., environmental or other signalling influences) and expressed (e.g., mental process/body language/behaviour)? This field of “cognitive ethology” is a complex area of evolving research.

4. What contributes to development of behaviours? In simple terms, think nature versus nurture, influences which for the modern canine become so interwoven that they are difficult to separate, based at least to some extent on the canine’s adaptability (referred to as development plasticity)[5].

Now, this all might seem to render the study of canine ethology – and the validity of assumptions and outcomes – highly complex. To some extent that is true, properly constructed observation and data analysis in ethology is certainly not simple and straight forward, but the aim here is to trigger the thought and understanding that both arenas of “scientific” laboratory or field experiments and “anecdotal” personal experiential observations are subject – generally – to some variant of failings in drawing conclusions, unless some considerable thought and account is given to these four foundations or influencing factors described by Tinbergen (op. cit.).

Umwelt - the canine view

Overlying all interpretations in canine ethological research is the factor often referred to in academic papers as “umwelt” – from the German word meaning environment/surroundings - which refers to the concept that different organisms may perceive their environment differently than do human observers – something that humans are not necessarily very good at appreciating. Based on the modern anthropogenic environment in which dogs exist (or, rather, co-exist) we humans readily jump to anthropomorphic conclusions. This not only brings into question the design of experiments on canine reactions but also interpretations of observed canine behaviours in “natural environments” compared to behaviours in domestic environments, as well as adding to three other important factors.

Firstly, canine reactions may not necessarily be manifested by observable physical changes (body language; eye movements; piloerection; etc.)[6].

Secondly, that in test situations, even ones that present new stimuli to the canine, the subject may already have a learned response to the test stimuli. Also, dogs can habituate very quickly and so reactions in sequential trials may be suppressed (not really surprising when one considers how trainable dogs are).

Thirdly, a subtle complexity is that in selecting canine examples to demonstrate particular behaviours or outcomes, the subjects (candidates) predictably are drawn from the anthropogenic environment in which they were bred and raised and so, inevitably, may lack the development of the natural abilities the experimenter is hoping to demonstrate[7].

This all begs the question as to whether the experimental/observational set-up somehow mirrors - even remotely to the human frame of reference - something the dog has previously encountered and so the desired “novelty factor” is absent, or alternatively the “natural state” the researcher/observer aimed to focus on is not really what the canine is actually experiencing.

Conclusion

Drawing conclusions regarding all aspects of canine ethology, in terms of predicting situational outcomes and future behaviours, especially relating to training and behaviour modification, primarily requires an appreciation of the basis of the conclusions – are they “anecdotal”, “scientific” or “experience” based and then, what is the best combination of those scenario observations? Added to this, a certain depth of understanding of the canine character is required – without (and this is the hard part) confirmation bias leading to quick conclusions. So, to counter this and attempt to place all observations and conclusions in a constructive matrix, what is helpful is a mind open to at least the following questions:

·        “What if that particular conclusion is not true; what alternative explanations are there?”

·        “What unseen factors could be impacting the observed/construed behavioural outcome?”

·        “What advances in understanding have come to light in the last few decades which could help explain the observations?” And then: “What other evidence is available to support the conclusion drawn and how has it been verified?”

·        “How is the canine observing the environment, predicament and events?” That is: “Is the way the dog sees and comprehends the situation (umwelt) being properly appreciated – does the dog’s body language align with this interpretation?”

·        “What state of mind is the dog in when it shows the observed reaction – the thinking brain or the emotional brain?”[8]

Only with this type of mindset will false and outmoded assumptions no longer be promoted, and perhaps real advances will ensue in the appreciation of canine ethology. Importantly, in the application to training pet, working (including police and military) and sport dogs, these same elements are critical to avoid false logic, wrong conclusions and poor outcomes in canine welfare and positively influencing the canine-human bond.

References

In addition to the references specifically footnoted, the book by Miklosi, A (2015) Dog Behaviour, Evolution and Cognition; Oxford University Press; was particularly instructive and I have drawn from a number of concepts discussed in that treatise to add to my own thoughts in building this article, noting that this does not mean I concur with every element of that work. Any misinterpretations or wrong conclusions are, of course, my responsibility.

[1] Dunning, D (2011) Chapter five - The Dunning–Kruger Effect: On Being Ignorant of One's Own Ignorance; Advances in Experimental Social Psychology, V 44; pp. 247-296.

[2] See Training articles this website “The fallacy of dog behaviour quick-fixes” and “Managing reactive and predatory dog behaviour” for insights as to how such training should be approached.

[3] A simple example is that in a real training environment, the savvy trainer will modify their techniques, as they observe the dog’s body language and judge the dog’s understanding and abilities, not keep applying the same criteria to look for the desired outcome.

[4] 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.

[5] Scott, J. P. and Fuller, J. L. (1965) Genetics and the social behaviour of the dog; Univ. Chicago Press; 506pp; [paperback published 1998]: provide the definitive foundation study on this subject.

[6] That is, not all environmental incidents a canine might observe - and which result in some neural activity - will show up in some outward change in a dog’s stance or body language.

[7] This aspect also provides food for interesting thought regarding a pet dog’s social skills and reactions to stimuli according the historic influences of its dyadic interactions with human guardians or canine companions. This can be viewed especially in terms of social competence, self-confidence in meeting assumed unfamiliar scenarios and compliance with cues.

[8] The thinking brain (frontal cortex) is responsible for problem solving, memory, language, judgment, impulse control, and reasoning. The primal brain (hindbrain and medulla) is responsible for survival, drive, emotions and instinct. When the primal brain is engaged, the thinking brain is not fully functional.