If there is a Setter pic here, then this is a Setters only page
The Effect of Anticipation
Introducing Anticipatory Behaviour
first published July 2018
– outline
Most of my experience in this wonderful, challenging, irritating world of training has been in the practical sphere, of training my own dogs and teaching others. Occasionally I get a quick peek at greener pastures such as neuroscience, behaviour analysis, computing, but inevitably the dogs find me out.
Anticipation surrounds us all the time. All behaviour cycles A B C are using anticipation of the consequences.
Are we making the reinforcement process too vanilla, too convenient and not adding sufficient anticipatory build?
What are we not realising? What can we explore further?
Anticipatory Power
Author: KL
We work with well-established principles, happily contest their findings by exploding them into tiny fractions to examine more carefully. When a laboratory study finds its way to our pasture it will get elaborately tested by the practioners. Often our practical experience brings different light to the theory. The example that springs to mind is the “three second rule”, which suggests that the reinforcer should be given within 3 seconds of the click, or marker. This has resulted in a cul-de-sac of limited application and it was probably suggested as a simplistic explanation for beginners. When the term “click AND treat” became written into lore, it prevented many beginner trainers exploring outside a treat following a click as opposed to a reinforcement process following the click.
Simplistic explanations have their usefulness but should be declared and remembered as such.
Anticipatory Behaviours:
Stalking – anticipation of kill, enables movement towards the prey to reduce proximity, minimise chase and use of energy, below speed of movement that would alert the prey, extremely tight focus on prey. Usually involves a reduced breathing sound, closed mouth, very slow carefully placed movements. Often masks discomfort or other responses.
Collie outrun, leave the shepherd, long distance run to position behind the sheep in anticipation of moving the sheep towards the shepherd. Often morphs into stalk once the optimum position is found. Both behaviours have a tight focus, often results in the dog running into trees or each other. The anticipatory state excludes irrelevant information or responses.
Fade-in. When beginning the process of fade-in we introduce the disrupting event (DE) during the phase closest to the capture of the reinforcer. For me this is after the click and as we travel to the reinforcement station, during the reinforcement process.
The dog is asked to station on the platform, marked when successful and then released to accompany me to the treat box waiting on the side unit. During the accompanying travel I am verbally inducing an increase in anticipation by demonstrating my own anticipation excitement and for most of my dogs this will also include either a chase or catch of the treat. These are predatory types of behaviours that, for these dogs, increase the excitement level in anticipation of the food delivery.
I can recognise the dog is in a high anticipatory state with the increased animation after the release to accompany me, and the predatory-like closing of the mouth as I get closer to the treat pot.
During the anticipatory travel the disrupting event (DE) is noted but minimally interrupting: we may see a flick of ear or visual acknowledgement. The DE may be a person standing near-by, walking towards us, an object being placed to the floor etc. Events that would under non-training conditions be attractive to the dog. At no time do we use disrupting events that would cause an avoidance response.
Training then progresses with the same behaviour and anticipatory reinforcement travel to reinforcement station and the DE is moved to occur:
2: between the cue and the behaviour
3: just before the cue
4: during the behaviour
The point most likely to be disrupted during the cycle is after the reinforcer is delivered and before the cue is given, also the behaviour is vulnerable to being weakened and is so left until the last point of disruption.
The key to using the DE during the strongest part of the cycle: anticipatory travel, is that the anticipatory state is likely to exclude other opportunities, focus the dog towards the know reinforcers and associate with the focus of predatory anticipation. It is the point most likely to be successful.
Because this is successful and repeated several times, the DE can become a cue to anticipate the anticipatory travel and therefore strengthen focus during other parts of the cycle.
Long duration before the anticipatory travel can be taught gradually and successfully.
The cycle must be clearly taught, practised and reinforced with the anticipatory travel.
Cue seeking > strong, stable behaviour > anticipatory travel > reinforcement process
Without this cycle being well practised, experienced and reliable, frustration, rage and aggression is likely to be the response to the DE.
When anticipation goes badly wrong
Travelling to the park for a walk with a dog in the back of the car screaming, barking.
Barking in a crate on anticipation of release.
Biting the treat delivery hand.
Pulling on lead
These behaviours occur during the absence of teaching the dog what behaviour causes the consequences to occur.
Often labelled as “lacking impulse control”, but more likely “lacking any teaching or learning” and a result of single, positive experience that accidentally linked the cause with a result.
These behaviour will need to be exposed to brokens patterns to prevent the anticipation as well as learning what behaviour results on the anticipatory event. This learning of anticipatory behaviours will need to be clear, clean and solid.
Untapped reinforcement processes
All behaviour cycles A B C are using anticipation of the consequences.
But are we making the reinforcement process too vanilla and not adding sufficient anticipatory build? The process of fast delivery with very minimal anticipation, (I describe this protocol as “click and shove” ) can have the effect of reducing the pleasures that can be enjoyed in the reinforcement processes. The purpose seems to be to prevent additional behaviours from interrupting the cycle, and avoid the dog making choices.
Connect and Share
sounds like a biscuit …….
Comment?
Comments made in this section are only on view to other Setters.
Community
Setters are folk who have passion, experience, knowledge and dogs.
When people with the same passion get together learning explodes.
More than a comment?
If you would like to add more than a comment, links to videos, your own article etc, please contact us direct – use the Get In Touch at the bottom of the page
Love this!
Really love to see the anticipation when playing/training my own dogs
I mostly see this in the anticipation of a ‘back hand shunt across the floor’ ‘ bowl’ or ‘catch’ of a treat when training my own dogs, just that extra pause when poised just before the treat is delivered and therefore the ‘freeze-stalk’
Need to get better at teaching my pet dog owners this as this is when you really see the magic !
Disclaimer: I have no dog training experience
This article made me think about how I can use this with the horses. Horse are generally enjoying table service, chasing a treat is not in our repertoire (grass doesn’t run away and we don’t want them to eat a sand-covered apple slice). Shopping -and the linked anticipation- is something I use with my senior horse. It is not that I consciously decided to use it a training approach but was borne out of the necessity of keeping the very much appreciated water melon slices somewhere else than in my pockets. Now I use the “water melon bucket” as shopping-style food delivery and anticipation is expressed as an energetic trot towards the feeding location as I invite him in, accompanied by a nicker.
Thinking about a disrupting event DE and how to use the Fade-in protocol, I thought that for my horses, I am more concerned about “startling events” SE or, more severely BYBSE, “break your bones event” in case I sit on the horse as he startles. My training area is in the middle of a forest with lots of sounds, many concerning to the horses (in their defense, I have to add that they are surrounded by lots of wildlife including very scary wild boar and wolves). So a breaking branch is often categorised as BYBSE.
I can’t control these sounds, but I could simulate the sounds and fade them in, e.g. play a recording initially at a very low volume at the strongest point of the cycle, i.e. during the (short) anticipatory travel. If that becomes irrelevant, I could then either move the moment of DE to other points in the cycle that are weaker or increase the intensity of the startling characteristic (sound volume) at the strongest moment before moving. Using different sounds from the start should make the type of sound irrelevant.
Some dogs often are fearful of fireworks, aren’t they? Would such an approach work to help them in such situations? Say you use the sound of fireworks at very low intensity as a DE for an easy behavior at the strongest moment and build that up carefully. In the unescapable event of madness around end-of-the-year festivities, could targeting with that background noise help them to recover? What do you think?
As I see this I want to try to engineer two processes, teach them individually, but then combine them.
Firstly, a strong anticipatory thread associated with specific behaviours, where “shopping” is the usual, drawn out, powerful interactive reinforcement / rewarding experience. I want that strongly conditioned to certain behaviours / patterns.
Secondly, experience of surrounding events that are noticed (we see the dogs’ ears twitch to a sound, eye flick to a movement) but not of sufficient attraction to override the anticipatory experience the dog is expecting. I have only engineered this with other opportunities for reinforcement, not events that may trigger a fearful reaction. But even these opportunities we design that we expect to be attractive often pose a suspicious response – possibly the start of fearful events.
What seems to happen is that once this event is filed away of “no relevance or benefit” then it does not appear to disrupt other behaviours until it becomes significantly invasive.
Then the answer is “we don’t know” until we have tried it, seen the effect ….. but I am sure if we give them these learning opportunities they become more flexible at self-management during unexpected disruptions. I regard it as a similar lay down of experiences during development, that may not match those in later life, but have set up solutions that the youngster is experienced of finding answers.
We will need video of course …. and melon.
At a very low intensity, such DE may not be emit a fearful response at all, just a flick of the ear initially. If the DE is introduced very carefully, it may be possible to trigger curiosity instead of fear which would bring us back to it being a competing reinforcer. Given careful introduction of the DE, and it being followed by the treats and generally surrounded by the reinforcement process as it is initially placed between the marker and the food consumption, it may help to change the association with a (sudden) sound under these conditions – and then maybe others as well.
Willing to try (and video). I just need to find a technical solution for such a project and do some dress rehearsals as to be able to push some button to play a sound before fetching the water melon from the bucket without getting pushed over by a horse demanding his well-deserved slice of fruit.
I like to set up something as a constant, in the distance, and then gradually take my circuit towards it. Could you use a CD player or phone with speakers? Radio shows where the music / voice / adverts are not predictable?
It would really be a benefit if the animals first instinct to novel sounds was curiosity, not avoidance.