Like it or not, there is no getting away from a little bit of behavioural science. Almost every time we interact with our horses we are, sometimes unwittingly, invoking classical or operant conditioning - very often both. This article addresses the definitions of common ways we train our horses, both intentionally and inadvertantly. There is much more to training than these definitions but this is a good place to start. It is important to remember, however, that whatever means of training is being implemented, Karen Pryor's Rules of Shaping (see her book "Don't Shoot the Dog") are paramount.
Classical and Operant Conditioning
Classical, or Pavlovian, conditioning is a process through which we create a subconscious and reflexive pairing between two previously unrelated events. The famous example is of Pavlov's dogs - having been fed after the ringing of a bell, the dogs started to salivate on hearing a bell. They had learnt subconsciously that the bell signalled food. Similarly a horse who has been successfully clicker-trained will experience a feeling of pleasure when hearing a click. But classical conditioning can work against us too - a "man smelling of chemicals" may trigger a fear response, causing a horse to become very wary of vet visits.
Operant conditioning is different in that instead of an association being formed between two random events, the horse learns through trial and error that he can control his environment. He learns that there are consequences to his actions - sometimes good, sometimes bad. If something good happens, the behaviour resulting in that outcome is likely to reoccur, i.e. it is reinforced. If something bad (or aversive) happens then the behaviour is less likely to reoccur, i.e. it is punished.
Reinforcement and punishment can be positive or negative. These terms refer to whether the stimulus resulting in the change of behaviour was added or subtracted - positive does not mean good and negative does not mean bad. Therefore we have:
- Positive Reinforcement
- The addition of something pleasant results in an increase of occurence of a behaviour. Typically a reward is given, although praise or increase in "feel-good factor" can also be very positively-reinforcing. For example, we might give our horse a treat or scratch in a favourite itchy spot in return for picking up a foot
- Negative Reinforcement
- The "cessation of an aversive stimulus", i.e. the removal of something aversive results in an increase of occurrence of a behaviour. Typically this is a release of pressure, either physical or psychological. For example, the release of rein-pressure on the bit as the horse stops is a means of telling the horse that he has done the right thing. N.B. the pressure used might be very gentle but it is still pressure - this is not necessarily a bad thing!
- Positive Punishment
- The addition of something unpleasant results in an decrease of occurence of a behaviour. Typically some form of hit or kick might be applied. Or a fine or rebuke or criticism.
- Negative Punishment (also Response Cost)
- The removal of something pleasant results in an decrease of occurence of a behaviour. There are not many practical applications of this in horse-training, with the possible exception of removing your attention from a horse who is being demanding or mugging you. It is more affective with, for example, children - you might confiscate a favourite toy to punish some inappropriate behaviour. It might be used inadvertantly with horses if you fail to reward some behaviour that the horse has offered.
Don't forget that it is the subject, rather than the trainer, who "decides" whether something is reinforcing or punishing. A treat is only reinforcing if the behaviour increases. A smack is only punishing if the behaviour stops. Just because some reinforcer/punisher works, doesn't mean it is appropriate or humane. Behaviour is the only way our horses can communicate with us - if we use operant conditioning to change behaviour then we might be losing vital information as to how our horse is feeling, for example using treats and/or aversives to force a horse to accept a badly-fitting saddle on his back. Very often it is our inadvertant use of reinforcers and punishers that result in undesirable behaviour - this is one of the main reasons why it is so important to understand both our horse's behaviour and our own behaviour.
Other Ways to Change Behaviour
Habituation takes place when a stimulus is repeated so frequently that the horse no longer reacts - the riding-school horse who ignores the repeated kicking of novice riders is the obvious example. If you have a "lazy" horse, you might want to think a little more about this.
Extinction is an additional way of decreasing the likelihood of a behaviour reoccuring, not through the use of an aversive (as in punishment) but by simply ignoring the behaviour. It is perhaps most effective when used in conjunction with some other means of changing behaviour, for example, mugging for treats is more likely to extinguish if the horse is taught that he can instead earn a treat for touching a target - a typical novice clicker training exercise. A problem with extinction is that the behaviour will get worse before it gets better in an event known as an extinction burst. Think of a broken vending machine as an analogy - we put our money in, nothing happens. So we fiddle with the buttons more and more frantically, our behaviour escalates and eventually we might give it a big kick (you never know, very occasionally this works!) before finally giving up in disgust. Horses also need to go through this process, so that they too can learn that the behaviour really does not work. It is vital that the behaviour at the peak of the extinction burst is not reinforced - it will be the escalated version of the behaviour that will reoccur in future! Once a behaviour has extinguished, it is likely that you will still see occasional reappearances of the behaviour - spontaneous recovery. These episodes are "mini extinction bursts" which need to be worked through - and not reinforced - just as with the original extinction burst. Depending on the behaviour that you are hoping will extinguish, it may be advisable to find professional help - working through extinction bursts can be dangerous, particularly if inadvertantly reinforced.
Extinction bursts take place because of the nature of variable schedules of reinforcement. The early behavioural scientists discovered that if reinforcement was presented only intermittently, as opposed to every time a behaviour was displayed, the rate of reoccurence of the behaviour actually increased. A gambling addict is a classic example - it is the unpredictability and potential for large rewards which maintains the gambling behaviour. Effective use of variable schedules is key to good positive reinforcement training - insufficient reinforcement will just lead to extinction of the behaviours you want.
Finally, as we all know, horses tend not to be very brave. This is what has made them such a successful species - they have been running away from scary things for 60 million years! Forcing them to confront fears can work, in the sense that the horse might allow scary things to take place, but this is a process known as flooding, i.e. the repeated exposure to some scary situation until the subject no longer responds. The results of flooding have not been formally documented in horses but it's my belief that it can lead to variations on the theme of conditioned suppression and learnt helplessness - i.e. the horse obeys because he has no choice, he does not become less afraid but is just unable to resist.
Thankfully, there are alternatives to flooding when we wish to teach our horse that it doesn't need to be terrified of some object or experience. Slow, progressive introduction to the stimulus via systematic desensitisation can be effective - the horse should be allowed to relax at each stage of the process and harder stages should not be introduced until the horse is genuinely comfortable and relaxed with earlier stages. If the process takes place too quickly then it becomes "slow flooding" which I believe can result in a heightened sensitivity to the stimulus. Counter-conditioning is a similar slow, progressive process but at each stage the horse is given the opportunity to make a new association between the "scary" stimulus and some sort of reward, for example, a horse who is scared of being shut in a stable could be fed a favourite food in the stable, gradually becoming happy to stay in the stable for progressively longer periods before receiving the food. The stable becomes associated with a good thing rather than the bad.
I shall save discussion of ethical use of these training tools and the "feel" and art that are needed in combination with the science for other articles. Both science and art are critical to good horse-training, but you need to learn the science first.
Copyright Catherine Bell 2006