Bono f Beeler – BCCSDip.Bhv.Prac
SpeakingDog K9 Services
Christchurch, New Zealand
10th May 2020
“Discussions about predictive coding which occupy my mind”
The dominant hypothesis, especially within animal sciences, still views behaviours as reactions to the world, in turn, learning reflects the world being referenced in the brain, by the world. [Stimuli -> Response / Stimuli -> Cognition -> Response]
Our individual human perception causes to anthropomorphize a many ideas onto our animals behaviours; some positive but as many negative.
The predictive nature of the brain is nothing new. Its not a new concept that has arisen in the last few decades. Even the so called father of psychology, William James, stipulated on predictive properties of the brain, there were many others, but somehow theses hypotheses never really gained popularity like models such as classical conditioning and behaviorism in animal sciences.
And so, this short paper, as others I wrote before, is a reflection of what occupies my mind taking into account the fact that each of our individual perceptions are layered with biases of our past experiences, priors, which influence what we read, how we study and research, present and teach others. The problem: our stipulation are ours, and very unlikely the perceptions of the very animals we observe, study and work with. To understand the brain as a statistics gathering predictive organ leads me to profoundly question the inferences we have made of animal behaiours but also the methods used to study animals; specifically, in this article: Domestic Dogs (canine lupus familiaris)
Behaviour: inside -> out
As we think of behaviour as reactive actions to the world, we seem to miss to ask, how does a brain and the system it controls, its body, react, in a timely fashion by awaiting a stimuli as not to be delayed in the responsive action? How does a brain perceive by vision, audition, olfaction, touch, taste to reference the world outside? How does the brain perceive and reference the body itself to be able to control every system down to the nanometer and move it purposefully at every level of every system within the body?
Are we asking these questions enough within the animal sciences especially the animal training professions?
The very nature of matching “word descriptions” to behaviours, brain regions, brain circuits and brain activity brings with it linguistic and cultural biases. Take for example the words “Calming Signals”- used by Turid Rugaas to describe behavioral interactions between dogs to communicate politeness, discomfort, stress and also producing a physical calming effect by a dogs system for itself or to calm another dog- is translated into German to “Beschwichtigungs-Signale”.
The very word: “Beschwichtigung” however, translates into English as “appeasement”.
“Appeasement” as a definition as: “giving something to an aggressive power to ensure the peace”, is however rather a submissive action that may lead to more discomfort for the appeasing dog. For example, if one dog gives up food to keep the peace, it may be calming in our perception for the lack of further aggressive action by the intimidating dog, however for the appeasing dog, who is giving up the food its system needs for survival, it becomes at the very least uncomfortable due to hunger and in prolonged circumstances stressful to the body and its brain for the lack of energy intake. From what I understand of the hypothesis of how dogs use Calming Signals; they are suggested to be purposeful communicative tools resting on cooperative abilities for successful social interactions in dog groups [packs].
The very translation -Appeasing- may imply to some something different compared to the original meaning the Author intended Calming Signals to be perceived. This has consequences; what we infer enforced by our cultural biases, and in turn I suggest, may have little to do with what the dog perceives, the world he moves in to be.
The so called “displacement behaviours” are another good example of misinterpretations. Originally Nikolaas Tinberg and Konrad Lorenz refered to “Übersprungsverhalten” -> then translated into English as: “Displacement Behaviours”- as behaviours which were perceived by the observer as unexpected irrelevant actions between 2 instinctive behaviours. These substitute activities were perceived as such because they did not fit a situation in any context and seemed to show a conflict between two instinctive behaviours, thus, the animal presenting an irrelevant third action. In later behavioural research however, “Übersprungsverhalten” – “Displacement Behaviours” seemed to conform more to social signaling. Thus while these actions may seemed to be irrelevant and out of context to the observer, in the animals perceptions of the environment its brain and its body moves in, these actions have profound meanings. (Consciously and unconsciously experienced) The observer is simply experiencially blind to the meaning of these actions- to the observer these are just noise.
Hannah-Maria Zippelius and Peter Sevenster pointed out in different publications their view, that none of the hypotheses by Lorenz and Tinberg, which supposedly explain displacement behaviours [Übersprungsverhalten], are empirically measurable with methods of continued neuronal and postured behavioural changes of the actions leading to substitute movements [displacement behaviours] within innate behaviours.
Sure, we can write the ethogram of a dog from continuous observations made with inferences we have made. As stated above; these inferences are influenced by prior experiences; what we learnt from others, agreed on with others, our cultural upbringings, believes, social needs and so on. Priors influence what we see, hear, feel, smell and taste this feel and emote.
Is a hen that is running up and down next to food she cannot get to, showing irrelavant behaviour or trying to communicate to the feeder that she cannot get the food? The latter would suggest the behaviours have context and are very relevant in that situation to the hen in the environment she moves in.
The examples above also present us with another insight. In all examples inferences were made. These guesses are predictions which the observers made of an animal in specific situation. They then went out researching, measuring their predictions to look for statistical reoccurrences which are then presented as evidence suggesting an accurate prediction.
The very nature of any system has a predictive rather then reactive nature; from the smallest chemicals moving by diffusive and electrostatic forces across neurons membrane through ion channels leading to various neuronal actions, making it possible for the sensors to sense. Action has to occur to move they eye to be able to process light. To do that, the brain has to have an idea what it is sensing. We cant see light with our ears, so there must be inference leading to actions to move the right sensor, the eyes. It means, every action has a purpose.
We may be displeased with some behaviours our dogs display. We call these problem behaviours. We go out of our way trying to change these behaviours to accustom the dogs to our life styles; how we envisage them to live with us. For that, many pay a lot of money to trainers and behaviourists to tell them how to change the very nature of a behaviour, to “inprint” the dogs brain with our ideals and all the while, the brain of “the mans best friend” is predicting possible outcomes, updating the prediction errors in milliseconds just to be pulled along on a short leash as reprimand for wanting to sniff and we infer from those actions, proof that the dog learnt through operant principles.
We can make the above statement with a positive reinforcer as well. How many of us have made a dog sit and look at us to “not react” to something uncomfortable and then stipulated this as proof that the dog is now calm and “ok” with the “stimuli it reacted upon”? Through the repetitive nature of training these behaviours increase and we can argue that the dogs brain has made the concept of looking at you during a situation it would have acted and felt differently before. But the predictive nature of the brain had already made inferences about the handlers actions (you) and the probabilistic outcome of the food by the action of sitting while also inferring the actions and/or movements of the uncomfortable stimuli (another dog) leading to the same outcome in affective mood. The dog will still be aroused and feel unpleasant if that’s how he felt before he was taught the action of sit in that specific environment. The movement has changed the prediction of the feeling has not because there is no new experience with the uncomfortable stimuli (the other dog) to create prediction errors and update the inference of: “the other dog is a threat”. What we do by looking at behaviours from the outside is the equivalent of treating a symptom rather then the cause.
Many will condemn me for the above suggestions but maybe some will be open to spend the energy to learn more about the predictive nature of living systems to create a future of animal training which teaches the very nature of our dogs to owners, to have deeper and richer bonds where dogs can be dogs and owners wonder how their furry friends perceive the world rather than sculpt them to fit into our world only to later band aid symptoms rather then treating physical and mental causes.
“An evidence-based decision assistance model for predicting training outcomes in juvenile guide dogs”
Naomi D. Harvey, Peter J. Craigon, Simon A. Blythe, Gary C. W. England, Lucy Asher
●Great Expectations: Is there Evidence for Predictive Coding in Auditory Cortex?
lMicha Heilbron Maria Chaitc
Département de Biologie, École Normale Supérieure, Paris 75005, France
Université Pierre et Marie Curie P6, Paris 75005, France
Ear Institute, University College London, London WC1X 8EE, United Kingdom
●Predictive coding in the visual cortex: A functional interpretation of some extra-classical receptive-field effects(Article)
Rao, R.P.N., Ballard, D.H.
View Corr2espondence (jump link)
●”Does Subjective Rating Reflect Behavioural Coding? Personality in 2 month old dog puppies: an open-field test and adjective-based questionnaire “
Shanis Barnard, Sarah Marshall-Pescini
●CALMING SIGNALS – The Art of Survival
●Derived activities: their causation, biological significance, origin and emancipation during evolution
Nikolaas Tinberg, The Quarterly Review of Biology. Band 27, 1952, S. 25.
● Die vermessene Theorie. Eine kritische Auseinandersetzung mit der Instinkttheorie von Konrad Lorenz und verhaltenskundlicher Forschungspraxis. Braunschweig: Vieweg 1992, S. 260,
●Hanna-Maria Zippelius, Die vermessene Theorie, S. 261.
●the free energy principle, karl friston
●the theory of constructed emotions, lisa feldman barrett, 2017
●Principle of neural design, stirling & laughling, 2015