Should You Ever Punish Your Horse?

We’ve either done it, or seen it done, a horse refuses a fence and gets a whack, a horse moves in too close to its handler and gets a slap on the nose, a horse bucks and gets some smacks with the whip, horse refuses to load onto a float and gets a leadrope aimed at their rump, horse kicks out at a child and gets a hiding from an angry parent, a horse dumps its rider so is put away without its usual ration of feed to teach it a lesson. Is this ever justified and what in effect are we actually doing when we “punish” a horse in this way?

Many people punish their horses for doing something “wrong” because they believe that the horse knows it did the wrong thing and therefore has to pay a price for making a wrong choice. That is, suffer consequences in the same way that a child who lies about nicking a biscuit from the jar experiences unpleasant consequences such as getting a smack or being denied dessert. Or they believe that the horse is dissing their human and needs to be shown who is the “real” boss. The question is, do horses really see us as their leaders and so understand the consequences of doing the “wrong thing” in the same way that we do? This is critical because if horses don’t make these kind of connections, when we punish them, what “lesson” have they really learnt? To answer this question we need to consider how animals learn from their environment and how we manipulate this for our own ends. We will start with considering the concept of what we do when we train a horse and what our choices are when it goes “wrong”. What Is “Training”?
When we train an animal what we are doing is putting something they already know how to do under our stimulus or cue control, that is, they do that something when and where and for how long we signal to them to do so. For example, we don’t need to teach a young horse how to canter, it can do that within hours of birth, what we do is teach it to canter when signalled and to keep on cantering until we signal it to do something else. Broadly speaking then when putting an animal’s behaviours under our stimulus control we have two goals, either to get the animal to do more of something or less of something. In general, our training focus is on getting more of a behaviour and it is only when the animal does something we don’t appreciate that we attempt to train it do less of that behaviour. For example, we want more of a calm transition from trot to canter and less of a pigroot when we apply the aid. We consciously set up the conditions to ask for the canter transition but only deal with the pigroot if we need to. The vast majority of our training is focussed on this first part of the equation, which means we practice the means to get more of the good stuff and only resort to stopping the bad stuff as a last resort. Punishers And Reinforcers
So how do we get more of the good stuff and less of the bad? By using punishers and reinforcers. Huh? Hang on for a bit while I unpack this. Something is either a punisher or reinforcer depending on what effect it has on the target behaviour. A reinforcer will motivate the animal to do more of a behaviour and a punisher will motivate the animal to do less of a behaviour. At its most basic, animals are motivated to do more of a behaviour that brings good consequences and less of behaviours that bring unpleasant consequences. This makes sense for survival- behaviours that bring food, rest, companionship, protection from predators and the opportunity to mate are worth learning and then repeating. Behaviours the cause pain, add stress, make you more vulnerable to being eaten, less likely to eat yourself or prevent you from reproducing are not going to help you (or your genes) survive so are not worth repeating. So to get more of a particular behaviour we need give the animal good consequences from it performing that behaviour and to get it to perform less of a behaviour we have to deliver bad consequences. The effect that these consequences have on the target behaviour tells us whether we have reinforced (strengthened) or punished (reduced) it. Thus, a reinforcer is anything that strengthens or makes a behaviour more likely. There are two ways we can reinforce a behaviour, either positively (adding something like food, sex, liberty etc) or negatively, (taking something away such as pressure). In each case, the animal values the addition of something or the removal of something it finds unpleasant highly enough to repeat the behaviour that it believes led to that outcome. It’s the law of effect, whatever behaviour immediately precedes the consequence that the animal values will be repeated. A punisher is the opposite of a reinforcer, it makes a behaviour weaker, that is it suppresses a behaviour. Just like reinforcement the law of effect applies- whatever behaviour preceded the punishing consequence is the behaviour that will be suppressed, that is the animal should be less likely to repeat that behaviour. Just like reinforcement there are two types of punishment, positive punishment and negative punishment. Que? Positive punishment simply means adding something to the animal’s world to suppress a behaviour whereas negative punishment means removing something from the animal’s world to suppress a behaviour. Here Are Some Examples: Positive Punishment:
Horse goes to bite its handler and is smacked on the nose and doesn’t bite- the smack has been added to its world and the biting behaviour is suppressed or punished. Negative Punishment:
The horse mugs you for treats, you withhold access to treats, so the horse stops trying to mug you- the treats have been withdrawn from the horse’s world, the mugging stops, so the withdrawal of the treats has punished the mugging behaviour. In horse training, we mostly rely on positive punishment, usually in the form of whip strikes, strong rein or leg aids, hitting with leadropes, shouting at or scaring the horse and so on. We use negative punishment less because its much harder to link the consequences of removing something to the particular behaviour we are aiming to suppress. In dogs trained with positive reinforcement (play or toys), negative punishment (withholding the toy or food reward if the dog responds incorrectly to a cue) is very effective in motivating them to try the correct response next time the cue is given. Causes And Effects- Measuring The Success Of Your Punishment
Whether something has actually reinforced or punished a behaviour can only be gauged by considering its effect on the behaviour. If as a result of the unpleasant stimulus you have applied to your horse, it stops performing the annoying behaviour, you have successfully punished your horse. However, if for example, you repeatedly strike a horse for refusing a fence but it keeps on refusing your whip strike has not actually punished the refusing- the refusing hasn’t been suppressed. The same is true of reinforcement, if you keep pulling on the reins but the horse doesn’t stop then your rein pull isn’t reinforcing- it has not strengthened the stopping behaviour. Back to punishment, for it to be effective, it has to suppress a behaviour which means the animal has to be able to make an association between its behaviour and the likely consequence. It has know which particular behaviour will result in the unpleasant consequence and then choose not to repeat it and thus avoid the unwanted experience. The application of unpleasant or painful stimuli to a horse (or any animal) that doesn’t allow that animal to make the clear connection and thus modify its behaviour in the future so it avoids the unpleasant stimuli is not punishment, but it could be abuse. How Effective Is Punishment As A Training Tool Really?
Punishment is much harder to effectively deliver than reinforcement. Here’s why:

1. In many of the situations in which punishment is delivered it is not clear to the animal which of its responses it should cease performing. Take for example hitting a horse for refusing a fence- its it the stopping, the cantering up to the fence, or the change in its head position as it judges the height of the fence that caused it to be hit? How would it know? What’s more likely to happen is that it makes an association between the fence, the surrounding context and getting a smack and becomes more fearful and thus either rushes at the fence and leaves a leg or starts refusing earlier as it tries to remove itself from the possibility of another whack. In the latter case it could be argued that what the whack has actually punished (suppressed) is the horse approaching the fence at all, rather than the coming to halt in right front of it. How many of us have seen horses get eliminated by refusing the same fence three times in a row, with each refusal accompanied by a whack and the horse veering away from the fence earlier and earlier each time it is pointed at it? 2. In many cases the punishment is not contiguous with the behaviour being suppressed, that is there is too large a gap between the unwanted behaviour and the punishment so it is not clear to the horse which of its responses caused the addition of the unpleasant stimuli. Taking the refusing horse, the horse refuses, the rider turns him back to the fence, gives him a whack and then asks him to jump the fence. What behaviour is being suppressed? The refusing, the turning, or the re-asked for jumping effort? Similarly, remounting a horse that’s just bucked you off and then giving it a few smacks is not punishment- the horse does not have mental abilities to make the association between its bucking behaviour, and the pain it is now feeling from the whip. The gap between the buck and the smack is too long, and the fact that it will have done several things in between the buck and the smack (run away, had a shake, been mounted) means it cannot make the association. The smacks will not be what suppresses future bucks. What might be suppressed/punished is its standing still when mounted behaviour and next time a rider attempts to mount, it might not stand still but fidget or walk away, a form of fleeing, because last time it stood still to be mounted it got a hiding immediately after. 3. Punishment doesn’t tell the animal what response it should give, it only suppresses the response that immediately came before the punishing stimuli. In effect it only tells it what not to do, not what it should to do. Our refusing horse isn’t being told what it should do by the whip strikes, so there is no reinforcement of the correct response to an obstacle- to jump over it. Consequently the whip strike might make it less likely to refuse, but not necessarily more likely to jump cleanly and calmly which is after all what we are aiming for with a showjumper. 4. Punishment can either be too weak or too strong. If we take the rule that we judge the effectiveness of our reinforcement or punishment by analysing its effect, if a punishing stimuli is too weak and it doesn’t suppress the behaviour then it hasn’t punished anything and its likely our horse will simply habituate to it, that is, stop reacting to it whilst still experiencing its negative effects potentially leading to misdirected conflict behaviours. If the punishing stimulus is too strong the horse may attempt to completely flee the situation or react strongly to the punishment itself and thus not learn anything about what behaviour it needs to stop performing to avoid experiencing it in the future. 5. Punishment is often administered at full volume and without warning, giving the animal no time to change its behaviour before experiencing its full effect. In negative reinforcement, we can start with a very light pressure and if it is not strong enough to motivate the animal to respond correctly, we can incrementally increase the pressure until it reaches a point whereby the animal chooses to respond because it wants to terminate the effects of that pressure. Thus the animal is in control of its responses. A strong and sudden punishment, delivered out of the blue doesn’t give the animal that choice, with the consequence that it will react to the pain, but not make the association between the incorrect behaviour and how to avoid it in the future. 6. Punishments often cause the animal pain and fear, and research in rats, dogs and cats has shown that fearful, stressed animals are less likely to trial novel responses to problems, that is, their fear interferes with their ability to learn and thus work out how to avoid getting punished in future. So whacking your horse for refusing may make him less likely to attempt to jump again because of an association of fear/pain and the jump, or cause him to rush over the fence in a panic as a way of fleeing the whack he thinks is coming and so fail to learn how to manage the tricky striding between elements of a triple. A horse which is yanked in the mouth for not performing a flying change on cue is going to be even less able to respond the cues next time because its focus is on avoiding the yank rather than performing the flying change. 7. The animal may make an association between the person giving the punishment and the pain and discomfort and thus react fearfully next time they encounter that person or find themselves in a similar context. This is called sensitisation, and it means that the horse will respond more quickly and often more strongly to a lower volume of the same stimulus- meaning it will be more reactive, more fearful, more likely to trial a strong flight response. If it associates you with the pain it may attempt to flee from you before you have even begun to apply the stimulus. 8. If the punishing stimulus is strong enough to motivate the horse to trial a severe flight response- such as rearing and spinning, bucking etc and the horse succeeds in removing itself from that situation what has actually happened is that the strong flight response is negatively reinforced (eg, it removes the horse from the unpleasant stimulus which is rewarding for the horse). It is then likely that its behaviour will quickly escalate from mild to strong in a short space of time when confronted with the same situation in the future because that’s what worked last time. These “lessons” can be learned in one go and can be very resistant to extinction. Horses which have been hit with a whip and become whip shy demonstrate these last two points, they react extremely strongly to any hint of a whip. We had a pony that attempted to attack a person who was holding its leadrope who also had a whip in their hand. (The retreat of that person negatively reinforced the aggression). We don’t know what behaviour its previous owner had attempted to suppress with the whip, but its reaction to being in the vicinity of one revealed it was highly sensitised to it. The whip and the human holding it had become a predictor of pain and fear for this pony to which it reacted precipitately and violently despite the whip being held a good metre away from his body at the time and the fact that individual who was holding him had never hit him with it. For punishment to be considered effective it has to result in a reduction of the unwanted behaviour so that the animal has learned what to do to avoid being punished in the future. If the behaviour yields rewards that are more valuable to the horse than avoiding the effects of the punishment it is still very likely to keep on with the unwanted behaviour in spite of the punishment. We see this when we smack a horse for trying to sneak into a feed shed, pretty soon its back with its head through the door because the rewards of getting into the shed are greater than the pain from the smack. If the punishing stimulus does not lead to a reduction in the unwanted behaviour then punishment has not occurred and it is likely that repeated application of the stimulus is simply abuse- applying pain or discomfort from which the animal has no means of or does not understand how to escape from. So are there contexts in which it is effective to punish a horse? Very few. Where the horse’s safety or the handler’s safety is at risk; where the benefit is obviously higher than the cost. For example if you yell at and hit a horse that has gotten loose and is about to walk out onto a busy road and by doing so cause the horse to stop walking and turn back, you have punished the walking forward behaviour and in so prevented the horse from getting injured. Likewise, if a horse goes to kick you or a child and you hit the horse and it doesn’t kick you have punished the kicking and prevented an injury. But what you haven’t necessarily done is trained the horse what it should do when it feels a child brushing its hind legs. You can only judge your success if the horse doesn’t attempt to kick again. {sidebar id=3}More Reinforcement, Less Punishment
In the case of the kicking, what is likely to be more useful for the long term is to train the horse to do less kicking by training it do more standing still and not reacting to sensations on its hind legs and then having that behaviour under your command. In the case of the refusing horse, the key to training it to not refuse is not to focus on getting less refusing but on getting more of reliable go- that is by training a light, immediate and responsive go response to leg aids, coupled with good turn and slow responses as well as the appropriate fitness, so that when presented at a fence it finds a bit scary, the horse still responds correctly to the go aid and clears it. The vast majority of horse behaviour problems can be solved by focussing on training the horse to offer the correct response and setting up the conditions whereby it is easy for the horse to do so. Relying on positive punishment of horses is fraught and its very hard to do it humanely. As noted above with dogs, negative punishment can be an effective training tool, provided the relationship between the unwanted behaviour and the punishment is clear to the animal. We use negative punishment a lot when clicker training our horses- rewards are withheld if the horse gives an incorrect response. In this scenario the horse is in control of what happens to it- if it chooses the correct response it gets reinforced, if it chooses or trials an incorrect response it loses out on the food reinforcer but otherwise does not suffer any painful or aversive experiences. Because its highly motivated to receive food it quickly stops offering behaviours that don’t lead to food. If you use positive punishment, make sure that it is very clear to the horse which of its behaviours result in it experiencing the punishing stimulus. You can do this by:
• only punishing one behaviour at a time,
• delivering the punishing stimulus either during or immediately after the unwanted behaviour- you have a 3 to 7 second gap, after which the horse will not know what behaviour caused the unpleasant consequence,
• using the smallest amount of stimulus needed to cause the horse to stop the behaviour,
• only applying the stimulus for a very brief period of time (3 seconds max) and removing the stimulus immediately the horse ceases the unwanted behaviour,
• immediately stopping the stimulus if it does not result in a quick cessation of the unwanted behaviour- if you keep applying the stimulus and the horse doesn’t reduce the unwanted behaviour your stimulus is not punishing but abusive and you need to change what you are doing. If your punishment has not led to a reduction in the unwanted behaviour- for example the horse continues with the behaviour or starts it up again a short time later you may have to accept that you have missed this particular opportunity to punish that behaviour or that your approach is not successful and requires a change in focus from you. With many problem behaviours, reinforcing a horse for behaving in a way that is incompatible with the problem behaviour is far more effective than trying to punish the unwanted behaviour. For example, reinforcing a head down response to a cue in a hard to bridle horse instead of punishing it for throwing its head up should quickly lead to the horse giving you more of the behaviour you want-lowered head and less of the behaviour you don’t want- head up, but you haven’t had to expose the horse to unpleasant stimuli to achieve your training goal. In all horse training, the most humane option when confronted with unwanted behaviours is to consider what is reinforcing or rewarding that behaviour (causing the horse to offer it) and then change what you are doing so its no longer worth the horse’s while to repeat the behaviours you don’t value and much more profitable for it to repeat the behaviours that you do value. In this way your horse gives you more of the good stuff, less of the bad and avoids getting punished at all.