Punishments are meant to decrease behaviour, yet in all studies of punishers used in training animals (or humans) it has been found to be very unreliable at changing behaviour at all. 

So why do we punish and do we even understand what punishment is?

If we look at the simplistic depiction of the ‘four consequences of operant conditioning’ used in most textbooks, reinforcers and punishers appear to have an equal effect on modifying behaviour. 
However, like most simplistic depictions this is far from true.

In fact, while this picture is suitable for introducing beginners to the concepts of operant conditioning, there is a whole section missing from this picture.  The opposite of any reinforcer is really more likely to be ‘lack of reinforcement’ and punishers, particularly positive punishments, are not only highly unreliable at decreasing behaviours but also highly likely to create alternate behaviours, i.e.  side-effects.

So first let’s clarify what punishment is*, recap on our negatives and positives in relation to punishment and then move on to looking at some of the problems that make it such an ineffective training tool.

A punisher could be described as anything that the animal doesn’t want and that it will change its behaviour to avoid in the future.  Now, you may already see that this definition also applies to negative reinforcement – a leg aid is something the horse doesn’t want that the horse will change it’s behaviour (by moving forward or sideways) to avoid in the future.  So the line between reinforcement and punishment can be a fine.  What makes the difference is timing and application.  

A positive punisher is the addition of something that the horse doesn’t want, such as a smack with the whip, after having done something we don’t like. A negative punishment is the removal of something the horse does want (which makes this something undesirable), such as removal of dinner or a timeout from play with us, after it has done something we don’t like.   Remember positive and negative don’t refer to good and bad, they are like the maths symbols and refer to add or subtract (remove).

Why is it so important that we really do understand what outcome or consequence we are using? Because there are different rules to using them successfully.  

This is particularly important with punishers because they are so very ineffective for training.  We must be able to identify punishers quickly and also recognise that often when we think we are using a punisher to discourage behaviour, we may not be, or we may be causing side-effects (completely new unwanted behaviours) and blaming the horse for this.  We must also be able to recognise when a negative reinforcement has slipped into punishment (for the behaviour we are wanting) because we are not applying it correctly, usually through bad timing and the lack of release (removal – the bit that makes it a NEGATIVE reinforcer).

So let’s take a look at the big problems with punishment – in particular positive punishments which are the ones most used by people with horses (and many animals) and the consequence that is most ineffective in changing behaviour.  The more we understand about punishment though, the more sparingly and correctly we tend to use it, so that instead of being something to ease our frustration or the tool we use when we don’t understand any better way to change behaviour, it can become an effective training tool. 

So here are the major reasons that punishments are ineffective:

Inappropriateness.  One of the biggest problems with punishment is that it can be so rewarding to the punisher.  We apply punishment, more often than not, not because we believe that it is the best way to train or because we have applied it in a thoughtful way but because it is satisfying to us at the time.  We are scared, frustrated, annoyed or angry at the horse for doing something and we lash out. 

Timing.  Just like rewards, punishers must be applied within a very short period of time of the behaviour occurring for them to have the desired effect.  We talk about a three second window in which to reinforce the horse for wanted behaviour or punish for unwanted behaviour.  Three seconds is actually quite a long time and you can still not ‘tell’ the horse what it did wrong.  Frequently punishers are applied too late to be associated with the behaviour.

Stress and Learning.  Learning cannot and does not happen effectively when the horse (any animal, including us!) is under stress.  This may be fear, anxiety, anger or other emotional states that cause stress hormones to release and interfere with the brain's ability to reason and importantly to lay down memory.  Punishment tends to cause stress.  Animals that are frequently punished will be in a high state of anxiety (stress) whenever interacting with the person that punishers - quite simply the more punishment applied the less likely it is that the animal is ever going to learn what is required.

Inconsistency.  Punishers are often applied inconsistently.  We like to ‘give the horse a chance’ and may only punish when the behaviour starts to annoy us.  Because we don’t apply the same punishment each and every time, the horse does not get sufficient information about the behaviour that causes the punishment.  Anything we use to modify behaviour relies on repetition in order for the horse to understand the connection between behaviour and consequence. 

Lack of Focus.  A punishment doesn’t tell a horse what to do instead.  This may not sound like a big problem but it usually is.  If the horse is stopped from doing “x” through punishment it may simply start doing “y” which is equally or even more annoying or destructive.  We then begin a cycle of punishment until the horse ‘finds’ something it can do.  During this the horse can become more and more frustrated and simply become averse to the trainer instead of learning what the right behaviour you want is.

Aversion to the Trainer. The cycle of punishment mentioned above or the type\level of punishment can cause the horse to dislike the trainer.  This is reasonably common; the horse makes the association of the punishment with a person rather than the behaviours.  This can make the horse wary of the person and while it may ‘behave’ in some instances, making the humans believe the punishment has worked, then the behaviour may reappear with other people or the horse may not trust the person anymore making it unreliable in its behaviour. 

Habituation\Desensitisation to the Punishment.  Another big problem with punishment is that the horse gets used to it.  In order to dissuade the horse you then have to up the level of punishment and the horse may then get used to that level and so on.  

Aggression or Increased Timidity.  A side effect of punishment can often be increased aggression or increased timidity.  If a horse perceives that it is being attacked, because the punishment is not clearly linked to the behaviour (delay in timing or inappropriateness) or the punishment is out of proportion to the behaviour or even because the horse has come to dislike or mistrust the person due to previous punishments, then this may lead to the horse defending itself – increased aggression.   A more naturally timid horse or one that has tried aggression and been ‘defeated’ may respond with increased timidity – head shyness, hard to catch, increased likelihood of bolting or escape behaviours. 

Learned Helplessness.  Learned Helplessness is a term often used for people (particularly humans that have been ‘desensitised’ using the controversial flooding technique) however it also applies to animals.  It is more commonly referred to as ‘shut down’ in animals.  It may also correspond to what was referred to as ‘broken spirited’.   This is where the animal simply gives in, submits, or shuts down to the point where it appears to do what is required but in truth the animal is no longer really participating or responding in a willing way.  Importantly, the submission tends to only last in familiar situations – the horse will suddenly freeze completely or react violently when in an unfamiliar setting or when pushed beyond its limits to cope.

Reward or Punishment? – The Trainer as a problem.  A consequence is only a reward or a punishment based on how the animal perceives it.  When judging whether ANY interaction we have with the horse is either a reinforcer or a punisher we should look at whether the behaviour increases or decreases. 

If the behaviour is being maintained (staying the same) or increasing then a reinforcer is at work.  
If the behaviour decreases then either a punisher or a lack of reinforcement is at work. 

Lack of understanding of how learning occurs in animals is a huge reason why punishment fails.  Simply using punishment infers that the trainer probably doesn’t understand what is causing a behaviour or how to teach the horse correctly.

Some common examples:

If a horse kicks and rattles the stall gate to gain attention then we may think that yelling at, smacking or throwing things at the horse are punishments because we are angry or unhappy when we do these things.   The horse on the other hand may view any or all of these actions as a reward because it is the attention that it is seeking in the first place.  So rather than decreasing, this behaviour increases in response to our ‘punishment’.

A horse that nips or threatens to bite, it is quite common for humans to flick the horse’s nose or use some other mild rebuke.   The horse may or may not see this as a punishment at all!  Even a fairly substantial thwack with a hand is often not terrible to a horse – it may get a surprise but if it is determined to perform the behaviour it will quickly become desensitised to such punishment.  It may even see this as an invitation to have some rough and tumble play – who hasn’t seen a pair of horses standing inches apart, biting and kicking at one another but not moving because they are ‘playing’?   

Finally, what about this common scenario? At the end of a work session the horse does some particularly nice work.  The rider immediately jumps off the horse and returns it to its stall. Is that a reward or a punishment?  Well, that depends on how the horse views this development.  If the horse views being worked as something unpleasant and being returned to its stall to rest as something pleasant, then this is a negative reinforcement – the removal of work.  Does it associate this with any particular behaviour – who knows?   If the rider’s timing is impeccable and he leaps off immediately following the required behaviour – maybe.  On the other hand if the horse enjoys the rider’s company and is happy doing the work then this could be a negative punishment – the removal of something it enjoys and being placed back in the stall where it is bored.

*
Strictly speaking all consequences are reinforcers.  Appetitive reinforcers (things the animal wants and will modify its behaviour to get) and aversive reinforcers (things the animal doesn’t want and will modify its behaviour to avoid).   We refer to aversive reinforcers as punishers to make life a little simpler for all concerned.  

© Horse Play Limited 2006