What Behavioural Science Do I Need To Know?
Also known as Respondent Conditioning or Pavlovian Conditioning, classical conditioning is the subconscious linking of a previously meaningless stimulus with a meaningful stimulus. Classical Conditioning involves involuntary/reflexive responses such as the blink of an eye, food anticipation or a fear response.
Example:In the famous experiment by Pavlov, dogs learnt that a bell was rung just before the arrival of food. The bell became a predictor of food and the dogs started salivating, an automatic or reflexive response, when they heard the bell. In the same way horses make automatic associations, e.g. if a saddle causes pain and the horse responds by "fidgeting" due to the pain, this can become an automatic response and the horse might become apprehensive just upon seeing the saddle, even if the fit is improved and no longer causes pain.
A factor which causes a behavioural response in an animal.
Example: Pavlov's dogs salivated in response to the ringing bell, the conditioned stimulus, as a result of it having been paired with the food, the unconditioned stimulus.
Also known as "instrumental learning", "stimulus-response learning" or "trial and error learning", operant conditioning is when an animal learns that his/her actions are linked to consequences and, as such, modifies his/her behaviour. The possible responses, or consequences, are positive reinforcement, negative reinforcement, positive punishment and negative punishment.
Example: a rat in a "Skinner box" chooses to push a lever, thereby operating on his environment. The outcome of this action might be a food reward, an electric shock etc which has a direct affect on how the rat will choose to behave in the future.
The addition (hence "positive") of a stimulus which the horse finds pleasant and, therefore, has the effect of increasing the likelihood (i.e. "reinforcing") that a behaviour will occur in the future. Whether or not a reward is salient depends on the horse's perception, rather than the trainer's "intent", and genuine positive reinforcement will trigger the "pleasure" circuitry of the brain. Food rewards or scratches are often used successfully with horses as rewards. Patting is often considered to be a reward but is not necessarily a pleasant experience for the horse and, given that it typically has no impact on future behaviour, cannot be considered positive reinforcement.
Example: a horse browsing through a hedgerow will be positively reinforced for doing so when he finds a wider variety of plants for him to eat. See also articles and FAQ on clicker training for details on how to incorporate positive reinforcement into training.
The cessation/removal (hence "negative") of a stimulus which the horse finds aversive and, therefore, has the effect of increasing the likelihood (i.e. "reinforcing") that a behaviour will occur in the future. In traditional horse-training, this is typically the release of some form of pressure at the instance of a desired behaviour. This applies equally to very mild pressures as extreme ones. While the release of pressure is indeed desirable to the horse, it is not a reward; a reward is positive reinforcement and activates different brain circuitry.
Example: pressure applied, via the reins, to the bit of a moving horse is released at the moment the horse stands still. Stopping has been negatively reinforced. Similarly standing in the sun can eventually become uncomfortably hot and moving indoors can provide a release of that discomfort - moving indoors is negatively reinforced.
The addition (hence "positive") of an aversive stimulus which has the effect of decreasing the likelihood that a behaviour will occur in the future. Note that attempted punishment very often does not affect future behaviour and therefore is considered a non-contingent (or gratuitous) aversive stimulus, rather than punishment. By its nature, punishment does not consider the reasons behind any perceived unwanted behaviours and thus can constitute a welfare issue.
Example: a horse who bites may be hit on the nose and, as a result, biting becomes less frequent in the future.
The removal (hence "negative", as in "taking away") of something desirable in response to an unwanted behaviour. There are limited uses in horse training due to the difficulty in well-timed removal of something desirable. By its nature, punishment does not consider the reasons behind any perceived unwanted behaviours and thus can constitute a welfare issue, particularly if food, water or companions are removed.
Example: a horse who becomes pushy in requesting scratches may become less pushy in future if the trainer makes scratches unavailable in these instances.
An instruction which predicts a consequence if a specific behaviour is performed. The behaviour will then be (to some degree) "on-cue" or "under stimulus-control", offered only in response to that cue and not in the absence of the cue. Note that any positive or negative associations with the behaviour will equally be linked with the cue via classical conditioning. Cues involving body language rather than voice tend to be more successful with horses.
Example: a horse may be trained to touch a target. Once the behaviour is well-established, the trainer may use a command such as pointing to the target to indicate the desired behaviour.
A marker which indicates that a reinforcement is on its way. Typically used in clicker training, the noise of the click forms the bridge between the behaviour and the moment of the actual reward. Through classical conditioning, the click becomes a secondary reinforcer; its real power is that it aids operant conditioning by helping the horse make the link between the desired behaviour and the food reward. Bridges may be "terminal", indicating that the behaviour is completed and the final reward is in its way, or "intermediate", taking the role of a "keep going signal". Note that the bridges chosen for use by the trainer are not necessarily the ones observed by the horse.
Example: A horse touches a target and simultaneously hears a "click". He then sees the trainer's hand going into the treat bag to obtain a treat. Both the click and the hand movement serve as bridges.
Something which will innately act to increase the likelihood of a behaviour reoccurring in future.
Example: Food, sleep, sex, water, increased comfort
Something which will act to increase the likelihood of a behaviour reoccurring in future only when paired with a primary reinforcer via classical conditioning. The secondary reinforcer "takes on" the properties of the primary reinforcer.
Example: from experience the horse learns that the noise of the clicker comes before a food reward and so the clicker "takes on" the properties of the food and becomes rewarding in itself. Similarly money is a powerful secondary reinforcer which we associate with our true desires.
Continuous Schedule of Reinforcement
The reinforcement is given every time a behaviour is performed. While the behaviour continues to be offered, performance sometimes tends to deteriorate as the animal (humans included) offers the bare minimum to obtain the reward and less frequently as the animal habituates to the reinforcer.
Example: a horse is given a treat every time he is caught in the field
Variable Schedule of Reinforcement
The reinforcement is given increasingly less frequently and less predictably. The uncertainty helps to maintain a higher frequency or duration of the behaviour and greater performance than in a continuous schedule of reinforcement, since the delay in reinforcement creates the beginnings of an extinction burst. When an animal performs a behaviour which has previously been rewarded, an absence of the reward causes the animal to try again, as though checking. If the delay in reward is too great then the behaviour will tend to extinguish.
Example: a gambler is more likely to continue playing for unpredictable high stakes, than consistent and predictable low stakes. A horse is more likely to come when called if rewarded only sometimes.
If a conditioned stimulus ceases to be paired with the unconditioned stimulus, the conditioned stimulus will cease to ellicit a conditioned response; the response has been extinguished. Similarly a behaviour that is not reinforced in any way will tend to extinguish and cease to be offered. If the animal is expecting to be rewarded then the lack of reinforcement will initially lead to increased incidence and heightened performance of the behaviour. This is known as an extinction burst and, depending on the behaviour can be dangerous to work through. Following the extinction burst the behaviour will typically die away. This is a gradual process and there will be episodes of spontaneous recovery "just in case". It is crucial that reinforcement is not given at the peak of the extinction burst or the extreme form of the behaviour will be reinforced and become very resistant to future extinction.
Example: a horse who tends to mug the owner's pockets for treats can (if safe to do so) be permitted to mug repeatedly. While the mugging will get worse initially, the incidence will then reduce as the horse learns that mugging does not achieve anything.
A behaviour which has been extinguished will sometimes reoccur in the form of a mini extinction burst. This may be triggered by some event, some inadvertent reinforcement or random chance. They can be treated as for extinction bursts and, again, great care should be taken not to reinforce so as to avoid reinforcing the extreme versions of the behaviour.
Example: a horse who has previously had his mugging behaviour extinguished, smells treats in the pocket of the owner so starts to sniff and gently nibble the pockets again
An animal's response to a stimulus gradually diminishes through many repetitions and the animal "gets used to" the stimulus. Note that if the stimulus is fear-provoking and the animal is unable to escape when necessary, flooding is more likely to take place than habituation. Habituation is context specific and subject to spontaneous recovery
Example: horses who live in a field alongside a railway line may resort to flight the first few times they see a train pass. But gradually the flight distance will reduce and eventually they will not feel the need to flee. Habituation in one context doesn't mean that the horse will be habituated to trains in all contexts and if there were no trains running for a period of time, the original response might return (spontaneous recovery).
An increase in reflexive response to a stimulus due to a high-intensity reaction to that stimulus.
Example: if a horse is handled aggressively during a new experience, such as applying fly spray, it may sensitise him to the application of fly spray in the future.
The gradual process of reducing a horse's fear of a stimulus through repeated but tolerable/minimal exposure to that stimulus. As the horse relaxes and ceases to respond to each stage, the degree of exposure can be increased slightly.
Example: a horse who dislikes being touched in the girth area might be stroked along the back until he is relaxed and unconcerned (perhaps using multiple sessions). He is then stroked in the shoulder area until relaxed and unconcerned. Then the sides, gradually working downwards momentarily and returning to areas of no concern. Gradually the strokes extend further into the girth area and for longer intervals before returning to areas of no concern. Finally the girth area can be stroked and the horse remain relaxed.
The replacing of a reflexive response to a stimulus with an alternative, (usually) more pleasurable response. Typically this will be the gradual process of reducing a horse's fear of the stimulus through repeated pleasurable experiences in the vicinity of that stimulus. Gradually the fear of the stimulus is replaced by the newly associated pleasurable anticipation.
Example: a horse who is afraid of being left behind by a field-mate might be given high-value food (e.g. haylage) when the field-mate is taken out on a ride. The feelings of fear are replaced by the positive expectations of receiving the food. A horse who has previously learnt to be fearful of a hose can be taught instead to associate treats with, and thus form a positive association with, the hose
The continued exposure to a fear-inducing stimulus with no opportunity for escape until the horse ceases to produce any active behavioural response. Importantly no defensive behaviour, e.g. flight, provides escape from the situation because this is prevented. Flooding may result in a decrease in fearfulness to the stimulus in future, but since only active behaviour is inhibited the fearful emotional state may not be, and fear related behaviour may return in response to the frightening stimulus at a later date. There are significant downsides to flooding - sensitisation may occur, and/or the horse may develop a conditioned fear of the trainer or training environment.
Example: a horse afraid of a horse-box is forced to enter within a short space of time via techniques perceived by the horse to be worse than the box itself. He appears to relax and stand quietly in the horse-box, seemingly "cured". A horse fearful of being hosed down is tied up and hosed down without the opportunity to escape; he appears to be less fearful of the hose in future because he doesn't try to run away.
A phenomenon that occurs when a horse learns that their behaviour does not provide any escape from an aversive situation. When presented with other aversive situations in future the horse makes no attempt to escape those situations. Horses are most likely to be at risk of this problem when they have never had any opportunity to solve such problems, or when subject to prolonged and multiple aversive, inescapable situations. Such horses appear to be shut down and unable to learn new things because they are unwilling to experiment with their behaviour.
Example: A very young foal that is restrained on the floor and subjected to a variety of potentially aversive procedures with no possibility of escape. The foal is very tractable subsequently because they inhibit any active behaviour when people apply the same procedures to them.
When horses cease all on-going reward-seeking (appetitive) behaviour when they are presented with an stimulus that has previously been associated with a painful and or frightening event via classical conditioning. Conditioned suppression has been associated with learnt helplessness because the stimuli that predict the inescapable situation that caused it may provoke behavioural inhibition during the recovery from learned helplessness.
Example: A horse stops eating his hay net when he hears a tractor running because he previously was frightened by a tractor while out riding on the road. The engine noise predicts that the tractor may come.
When a behaviour has been trained in a specific location we need to remember that it has not been trained to take place elsewhere. Training in different contexts is an important part of shaping a new behaviour so that the horse can extrapolate the training to include all locations, weather conditions etc. We call this process of extrapolation generalisation. Horses are not typically very good at generalisation, hence the need for allowing it in our training. Note that generalisation also applies in classical conditioning and habituation.
Example: a horse who has been trained to load in a horse box may then be taken to a show and, at the end of the day, refuse to load for the homewards journey. He has not been able to generalise the loading behaviour. Training a horse to load needs to includ loading in multiple locations so that the horse can generalise and ultimately be able to load wherever he is.
Reinforcers and punishers need to be meaningful to the horse in order for them to be effective. If you opt to use punishment then the punishment will only work if it is sufficiently aversive to overcome other stimuli in the environment. Similarly, rewards need to be of high value to the horse so that he will choose to work for them. This "value" that the horse assigns to each reinforcer or punisher is known as saliency. Note that this also applies to cues and classical conditioning
Example: A horse grazing in a field may choose to participate in a clicker training session in order to earn pieces of carrot but continue grazing if only grass cubes are on offer. We say that the carrot is more salient, or of higher saliency, than the grass.
Often confused with positive reinforcement, bribery or "luring" is used to prompt behaviour. It is successful when the horse is given the food bribe or lure directly after they perform the desired behaviour, in which case the bribe functions as positive reinforcement. When the reward is produced, but kept out of reach, in advance of the behaviour being offered, the horse tends to fixate on obtaining the reward through any means rather than performing the behaviour.
Example: a reluctant loader stands at the bottom of the ramp, stretching his head in an attempt to reach the bucket of feed held at the top by the desperate owner. The horse tries his utmost to avoid moving his feet!
Training a behaviour and/or overcoming a fear is more effective and least distressing to the horse if it takes place via shaping, or successive approximation of that behaviour. The behaviour is broken down into a series of tiny steps, each step well-defined and easy to achieve. Each step is trained separately to competency and then gradually combined together with lots of reinforcement at every step of the way. By the time the finished behaviour can be performed it is merely a final tiny step from the previous piece of training. We strongly recommend reading the shaping resources in the recommended reading list.
Example: to pick up a young horse's hoof start with a behaviour that he already finds easy, such as stroking his shoulder. Gradually extend down the shoulder to the foreleg a few times, repeat on the other foreleg. Gradually extend the stroking down towards the knee, again repeating each side. Gradually extend towards the hoof, still making no effort to pick up the hoof. Subsequent stages, to be repeated on each leg might include hold the hoof for longer, start to apply some pressure for lifting, wait for the horse to shift his weight slightly, start to lift the hoof, hold the hoof higher, go back a few steps but repeat for longer.
Stereotypies are defined as unvarying, repetitive behaviour patterns that have no obvious goal or function (Mason 1993) and which cannot be easily interrupted. They often start when the horse is undergoing some stress, and is motivated to perform the behaviour to reach a naturally desired goal, but is somehow prevented from performing that behaviour, and achieving the goal. The resulting distress then prompts the horse into this aberrant behaviour instead, and hence stereotypies are often referred to as "coping" mechanisms. The horse tries to fulfill an unmet need. As these behaviours are practised, and rewarded internally, the more they are repeated, often with far less stimulation than in the original eliciting situation. Eventually they become fully emancipated from that original cause whenever the horse's sympathetic nervous system becomes aroused beyond a threshold point individual to the particular horse. The nature of the stereotypy matches the unmet need - e.g. box-walking is usually caused by restricted movement. It is important that the cause of the stereotypy is addressed rather than the behaviour prevented - for example cribbing collars and anti-weave grids do not address the reason for the behaviour and such measures remove the horse's ability to attempt to cope with their situation.
Example: windsucking, cribbing, door-chewing, weaving, box-walking.....
When an animal is conflicted about performing a given behaviour, he may instead - or as well - perform an additional behaviour. This additional behaviour may have no relevance to the desired behaviour but it is likely to provide some form of the comfort to the animal.
Example: The horse begins to fidget in anxious anticipation, instead of standing still, when he sees the owner carrying his badly-fitting saddle
What Behavioural Science Should My Trainer Know?
This section is a "work in progress" and will be extended gradually.......
Trainers have their own specialist disciplines and not all have had time to study in depth the science of equine behaviour. However EBTA feels that whatever discipline you are teaching - be it dressage, show-jumping, eventing or natural horsemanship - a certain knowledge of the basics of equine science is absolutely necessary in order to know why a horse is behaving as he is. All unwanted behaviours are the horse's way of telling you something is wrong. Understanding basic learning theory and the science of behaviour will help trainers to give clients constructive advice in a positive way which, ultimately, will enhance performance.