Classical conditionning
A horse that paws at his stable door, or neighs when the wheelbarrow containing his feed is moved in the stable, a horse attempting to establish contact and moving his upper lip at the sight of a brush, a horse who responds to the sound of a clicker… All these behavioural reactions are due to classical conditioning. The horses unconsciously learns to react to a stimulus which would normally not trigger a reaction on his part. This is a very useful learning process. But what rules apply, and how should it be used appropriately?
- A few reminders about learning processes
- The main principles involved in classical conditioning
- How should one use classical conditioning?
- What if this does not work?
- To remember
A few reminders about learning processes
A learning process is a durable change in an individual’s behaviour pattern resulting from past experience, expressed by an animal when exposed to environmental stimuli*. The animal will reproduce the new behaviour pattern when faced with the same stimulus again. Generally speaking, animals will alter their behaviour in a way which is most favourable to them.
Classical conditioning : a learning process through association
Classical conditioning is an « associative » learning process where the horse is induced to perform an action by creating an involuntary association between two stimuli : one which initally does not provoke any specific reation in the animal, and another which naturally leads to a reaction or a strong emeotion for the horse.
The main principles involved in classical conditioning
Classical conditioning, also known as respondant conditioning, or pavlovian conditioning, named after the famous 20th century scientist Ivan Pavlov (1849-1936) who brought it to light. Pavlov’s studies conducted on dogs, showed that by associating food to the sound of a bell, in the long term just the sound of the bell created a drooling reaction in the dog (without the presence of food). He called this a « conditioned reflex ». In the same way, horses will react strongmy to the noise of the feed wheelbarrow, or when they see a bucket used to provide cereal. What process is behind this type of learning?
The horse involutarily associates two stimuli
Classical conditioning is a learning process where the horse will involuntarily create an association between:
- An unconditioned stimulus, which when there is no prior conditioning, has a naturally strong valency for the horse and provokes a reflex reaction (non conditioned response) on his part (food, social companion…).
- A neutral stimulus which would not normally trigger any particular response on the horse’s part.
This learning process can be divided into three main steps which we will illustrate with the example of providing feed in a bucket.
Prior to conditioning
Before conditioning, the unconditioned stimulus creates an involuntary reflex reaction on the animal’s part, whatever the conditions. The neutral stimulus on the other hand, does not trigger the reflex reaction. In the example, providing feed is the unconditioned stimulus : pleasant for the horse, it automatically induces a positive emotion and excitement for the animal. It is easy enough to note that at feed time, any horse will neigh, get restless in the stable, or start pawing at the door to the stable. On the other hand, when a bucket is first presented, it is then only a plastic object with no particular meaning. It does not cause any reaction in the horse : it is a neutral stimulus.
During conditioning
Simultaneously presenting the two stimuli repeatedly conditions the horse to associate the unconditioned stimulus to the neutral stimulus. In the example, if the horse is systematically fed in the same bucket, he will begin to associate his feed to the bucket, and get excited by the idea of getting fed, as soon as he sees the bucket.
After conditioning
By associating the unconditioned stimulus to the neutral stimulus, the latter becomes a conditioned stimulus. This means that it now induces the same reflex reactionon its own. However, as it is the result of conditioning, it is said to be a « conditioned reaction » (and no longer a « non-conditioned reaction »). In the same example, the horse will now become excited just when he sees the bucket, even if there is no feed inside.
Neutral stimulus, secondary reinforcement and the horse’s motivationtimulus
Once the association between the neutral stimulus and the unconditioned stimulus is established, the conditioned stimulus can be used as a reinforcement in the operant conditioning process. In this case it is known as secondary reinforcement, not to be mistaken for primary reinforcement which is naturally effective. To understand this an example is probably much better than a long explanation:
- A feed reward (carrot for example) is something pleasant which the horse will seek naturally : this is primary reinforcement.
A word or a pat do not mean anything to the horse, but when associated to a feed reward, they can little by little be used as a reward instead of the titbit, thanks to classical conditioning : this is known as secondary reinforcement.
In brief, secondary reinforcement is not reinforcement to begin with. It is a neutral stimulus which becomes reinforcement once associated to a primary reinforcement, thanks to classical conditioning. We will demonstrate below that here again there are two types of reinforcement:
- Positive reinforcement (R+) which consists in providing something pleasant as soon as the horse gives the right response.
- And negative reinforcement (R-) which consists in stopping any unpleasant pressure as soon as the horse gives the right response.
When does classical conditioning apply?
Classical conditioning can be beneficial in numerous everyday situations. E.g : giving the horse a reward (a handful of feed) everytime he takes the bit himself, or lifts his feet properly… Associating these situations to feed makes them attractive to the horse, who will then seek to perform them, rather than remaining indifferent. It is also an excellent way to get the horse to accept a stimulus or a situation which naturally creates anxiety, such as clipping, loading into a trailer, going through water, washing down… It will help get the horse used to the situation without fear.
When riding, the rider also often uses classical conditioning without necessarily realising he is doing so. This is the case for example when he associates a neutral stimulus, such as slight contact of the calf on the flanks, to a more unpleasant stimulus like pressure from the heel or the spur. In a novice horse, contact of the calf will not cause a reaction, it is the neutral stimulus. However, once the association is established, it will have the same effect as pressure from the heels or spurs : it becomes the conditioned stimulus. The aim is to get the horse to react to a slight pressure from the lower leg, rather than having to systematically touch him with heels or spurs
In the same vein, classical conditioning can be used to associate a neutral stimulus, such as a word or a sound or a pat, with food. The aim is to use this neutral stimulus instead of food.
How should one use classical conditioning?
Below is a step by step method to help implement secondary reinforcement thanks to the classical conditioning process.
Step 1: choose the neutral stimulus which will become a secondary reinforcement
Generally speaking, the more precise the stimulus, the easier the horse will be able to identify it, and the quicker the learning process will be established. The choice of the stimulus is therefore essential.
Type of reinforcement | Positive reinforcement(R+) | Négative reinforcement (R-) |
Neutral stimulus | Gesture, touch, word…. | Word pronounced loudly and distinctly, moving the crop (without necessarily using it), pressure on the horses flanks with the lower leg, slightly raising the lunge whip... |
If a word is used, it should always be the same word, with the same tone, so that the horse distinguishes it (« yes », « good », good boy »…). If touch or gestures are used, check that they are not similar to those used on a regular basis.
Step 2: choosing the unconditioned stimulus
The unconditioned stimulus should be really meaningful to the horse, i.e. it should be:
- Really pleasant when we’re talking about positive reinforcement, as the aim is for the horse to seek it out.
- Or unpleasant enough when talking about negative reinforcement, as the aim is to startle the horse
This is a primary reinforcement which induces, on its own, a reaction from the animal.
Type of reinforcement | Positive reinforcement(R+) | Négative reinforcement (R-) |
Unconditional stimulus = primary reinforcement | Food reward, scratching an area of the body which the horse is partial to (withers for example), presence of another horse…. | Sudden presure from the rider’s heels, contact from the spurs, tap with the crop, touching with the lunge whip… |
Step 3 : associating the neutral and unconditioned stimuli
So that the horse asoociates the two stimuli, the neutral stimulus should be given immediately prior to the unconditioned stimulus. The effectiveness of conditioning depends on the clarity of the combination between the two stimuli. The association may take some time before it is established, and should therefore be repeated several times.
The more easily identifiable the neutral stimulus is to the horse, the better it will catch his attention (distinctive sound…) and the quicker he responds to conditioning. The unconditioned stimulus should also induce a strong enough reaction or emotion in the horse (food reward…). Imagine a thirsty horse who is presented with water in a bucket : the association betwen the bucket and the water will probably be established straight away. On the other hand if you try to associate a word which is difficult to identify, with a handful of grass, which the horse already has free access to, conditioning will most likely take much longer to establish...
Step 4 : use this secondary reinforcement within the framework of operant conditioning
Once the association between the neutral and unconditioned stimuli has been established, the neutral stimulus, which has become a conditioned stimulus, may then, at least in theory, be used directly as a reinforcement instead of the unconditioned stimulus. In practice, instead of givng a food reward as soon as the horse responds to your request, you can use a pat, or a vocal reward as positive reinforcement. In the same way when riding, just a slight touch from the lower leg will gradually replace pressure from the heels or contact with the spurs or crop, when using negative reinforcement. However using conditioned stimuli alone will gradually extinguish the response ; the horse no longer reacts to it. It is therefore necessary to continue to introduce the unconditioned stimulus on a regular basis.
What if this does not work?
- Does the unconditioned stimulus induce a strong enough reaction or emotion on the horse’s part?
- Can the horses easily identify the neutral stimulus?
- Has the association between the neutral and unconditioned stimulus been repeated enough?
- Are the two stimuli used in quick enough succession?
- Has the conditioning process become extinguished?
To remember
This type of learning process is complementary to operant conditioning and is very useful to fine tune the horse’s training and dressage, so as to move towards finer riding, with more subtle codes. It is also a way to dedramatize and/or make worrying situations for the horse more attractive. It enables you to establish reinforcements either positive or negative, which have no specific meaning to the horse to begin with. In any case, a good trainer will seek to reduce the pressure necessary, and will favour light codes.
Training a horse is a long process, which takes place over several years. It is gradually, by combining different learning processes, that the horse can be brought to perform complex actions such as dressage figures.
Know more about our authors
- Translated from french by : Karen DUFFY Translator
- Nelly GENOUX Development engineer IFCE
- Christine BRIANT Veterinarian - development engineer IFCE
- Marianne VIDAMENT Veterinarian - development engineer IFCE
- Léa LANSADE Research engineer in ethology IFCE-INRAE
Bibliography
- BARAGLI P., PADALINO B. et TELATIN A. (2015). The role of associative and non-associative learning in the training of horses and implications for the welfare (a review). Annali dell'Istituto superiore di sanita, 51, pages 40-51.
- IFCE (2015). Travailler son cheval selon les principes de l’apprentissage. Editions Haras Nationaux, 77 pages.
- ISES (International Society of Equitation Science) - Principes fondamentaux de l’entraînement du cheval → Poster en français
- ROCHE H. (2013). Motiver son cheval, clicker training et récompenses. Editions Belin.
- SANKEY C., HENRY S., GORECKA-BRUZDA A., RICHARD-YRIS M.A. et HAUSBERGER M. (2011). Aliment ou grattage : quelle récompense pour le cheval ? 37ème Journée de la Recherche Équine, Paris, pages 179-182.
- SANKEY C., HENRY S., RICHARD-YRIS M.A. et HAUSBERGER M. (2009). Le renforcement comme médiateur de la relation homme/cheval. 35ème Journée de la Recherche Équine, pages 89-100.
- VALENCHON M., LEVY F. et LANSADE L. (2013). Influence du tempérament sur les performances d'apprentissage et de mémoire du cheval : bilan de trois années de travail de thèse. 39ème Journée de la Recherche Équine, Paris, pages 147-150.