Intentional use of extinction procedures is a nuanced topic. In some situations it may be appropriate. In others, it’s not.
The principle itself is straightforward. If a behaviour is maintained by reinforcement and that reinforcement is removed, the behaviour will reduce over time.
Sound behavioural science.
What requires closer examination is how that principle is applied, particularly in food-based training with horses.
In the early stages of food-based training, so-called ‘unwanted’ behaviours are informative. Food has been introduced, but the learner has not yet fully understood the promise of how food will be delivered. We may see mugging, pocket checking, nose investigation around the food bag, increased vocalisation, facial tension, or dropping in males. These are quickly labelled as ‘pushy,’ lacking ‘manners,’ or having ‘no boundaries.’
But these responses are not failings on the part of the horse.
They are feedback. They tell us WE got it wrong. That the contingency between marker and food is not yet clear. Our bad!
When the reinforcement system lacks clarity, the learner experiments. This exploration is completely natural. The horse is information-seeking within an environment that doesn’t yet make sense.
This is where the conversation about extinction becomes more nuanced. Can you extinguish confusion?
Extinction is a procedure. It applies when a clearly defined behaviour is being maintained by reinforcement and that reinforcement is deliberately withdrawn. You might not realise that’s what you are doing but it is happening.
The common advice is to go ‘cold turkey’ and withhold food, ignore investigation, create physical barriers, wait out any escalation of the mugging behaviour.
If the horse does not yet understand where food appears, how it is delivered, or what precisely earns it, ignoring mugging does not automatically create understanding. Frustration increases behavioural intensity. Escalation is then interpreted as further evidence that extinction is necessary. Go harder.
The focus remains on stopping behaviour rather than examining our system design.
Contemporary welfare-focused training places an emphasis on reinforcement system design rather than behaviour suppression.
When antecedents are carefully designed, marker signals are clear, reinforcement placement is consistent and rates are appropriate, many of the behaviours people attempt to extinguish never establish in the first place.
Not because they were suppressed.
Because the learner understands the rules of the our environment.
Should escalation of frustration be a routine feature of thoughtful training? I would argue no. It is worth examining whether escalation is an unavoidable component of learning, or the predictable outcome of a preventable lack of clarity.
The shift is in the question.
Not ‘How do I stop this behaviour?’
But ‘What does this behaviour tell me about the clarity of the information I gave?’
Very low-level extinction may have a place in shaping responses once the learner understands the contingency. It is not a primary tool for teaching ‘manners’ around food.
Clarity builds understanding. Suppression builds tension. Suppressed behaviour nearly always comes back to bite and most definitely teaches horses to have no opinion or that their opinions don’t matter.
The difference is the kind of training we choose to practise and the kind of horse behaviour we choose to encourage.
Errorless learning with horses offers us a framework to support low stress training.
