Getting Started with Positive Reinforcement
Positive reinforcement = adding an appetitive stimulus after the behaviour has been performed ( the behaviour can be formed using target training, capturing the movement or using a mild tactile stimulus and a bridging signal is used to mark the exact time of the wanted behaviour.) E.G a verbal signal or a clicker. So how do we start? First we have to condition the signal we want to use to indicate a desired behaviour has been performed.
An unconditioned stimulus (UCS) results in an unconditioned response (UR), when the UCS is paired with a neutral stimulus (NS) this stimulus may become a conditioned stimulus.(CS). Therefore the CS triggers the response which then becomes a conditioned response (CR).
Charging the clicker or verbal bridge signal. The verbal bridge signal needs to be something not normally used in everyday conversation, so a foreign word like “si”, “bon” or a high pitched “good” etc.
The bridge signal is the neutral stimulus and the food or scratches is the unconditioned stimulus which elicits an unconditioned response (they want more of the reinforcer).
Pair the clicker/bridge with the appetitive reinforcer ( food/scratches or anything else the horse wants more of) and the clicker/bridge signal becomes a conditioned stimulus and the reinforcer elicits the conditioned response – anticipation of the reinforcer, so they will stop when clicked and wait for the reward.
This is best done in protective contact so we teach manners around the food delivery at the same time. Only hand feed when the horse is calm, not seeking the food or mugging. Make sure his head is straight and not oriented towards you.
Also not hand feeding when he is soliciting, or when not in a training session helps them learn the food only arrives after the bridge signal. It is also good to vary the reinforcement so they don’t always expect food, so scratches (for itchy horses) or anything else the horse finds reinforcing. I had a horse who liked doing side passes so I put that on a cue and cued it as a secondary reinforcer.
Horses can get frustrated during training if not taught correctly to take food, or if there is no signal to say “yes that is what I want” but just a random treat arriving for seemingly no reason.
We should be able to walk in a field full of horses with food in our pockets and not be mugged. So we need to teach an alternative behaviour – ignore them unless they are acting like a normal horse – grazing etc. Only then reinforce that normal behaviour with a scratch or attention or pick them some grass.
It is important too to have start and finish signals for training sessions. Usually the first click or bridge signal indicates the start and a handful of food and a hand/verbal signal as a finished sign.
We can also wear a particular waist bag or gilet when training so that can become the start signal. If we do free shaping sessions it can be useful to wear a different item of clothing so they know the difference between a more structured training session and free shaping.
Getting Started with Target Training
Once the bridge signal is in place we can introduce a hand held target. This can be any novel object – not one that has previously been associated with aversive stimuli e.g not a whip or carrot stick (unless so disguised as to not elicit a conditioned emotional response.
Targets can be used to form behaviours, so we can ask the horse to follow and once reliably follow can introduce a verbal cue “walk”. Can also be used to send the horse to another person to capture trot. Targets can be used to teach back up, or hips over or towards and any number of things.
Once the behaviours are solid and reliably on a cue we can fade out the target and just ask for the behaviour using a verbal or visual cue.
To start with we use a continuous rate of reinforcement i.e bridge and reinforce for each step of the behaviour.
Once the behaviour is on a cue we can use a variable rate of reinforcement as well as using different reinforcers. So scratches and less appetitive food or even a behaviour the horse likes doing.
It is also possible to chain behaviours so that behaviour 1 is reinforced by the cue for behaviour 2, so that cue becomes a secondary reinforcer.
If we stay too long with continual reinforcers then the can get frustrated if the reinforcement rate is dropped, we need to build resilience.
The horse will work harder to figure out how to get reinforced, so we do need to be careful and watch for any over arousal of the SEEKING system. This can tip over into RAGE if the horse gets frustrated.
Stimulus control in positive reinforcement
As soon as you are getting the behaviour reliably, teach a cue, otherwise the horse will start to offer it whenever they aren’t getting any other reinforcement / when it is the thing with the greatest history.
To get it on cue introduce the cue AS the behaviour is happening – the horse has committed to it, or immediately before you cause it to happen with the target.
At this point fade out the target and only re-introduce if the behaviour falls apart. Only reinforce the behaviour you want.
1) if behaviour A is offered without the cue, do not reinforce it
2) if you cue behaviour A and you get behaviour B, do NOT reinforce that .
3) if you cue behaviour B and you get behaviour A do NOT reinforce that.
4) if you cue behaviour A and the horse does nothing, check for competing motivations. Check the environment for distractions. Check he is physically capable of performing the behaviour. Check you haven’t asked too much too quickly.
The only way to get behaviour on cue and under stimulus control is to be true to your cue. One cue = one behaviour – do not use a cue previously taught with negative reinforcement this can cause poisoned cues.
Form the behaviour, reinforce and repeat until it happens reliably, introduce a cue, fade out the mechanism you used to form it, get it under stimulus control.
Stimulus is not equal to the perception of the stimulus.
So whilst the stimulus applied to a horse can be the same it depends on how that animal perceives it and what is happening at the time (background stimuli).
A different horse may perceives the same stimulus in a different way.
This is why we need to be mindful of the context and the emotional and physical state of any animal we with which we interact.