- Location
- Andorra and Spain
When getting into behaviour geekery, you will come across the terms "classical conditioning" and "operant conditioning" quite regularly. These terms can be a bit intimidating for people who aren't familiar with them, so I'll try here to present them in a low-key way that's easy to understand. Maybe later, we'll go into a little more detail, but for now, let me present "Conditioning 101".
Classical conditioning
It's almost impossible to have a discussion about classical conditioning without mentioning Pavlov and his experiments. In fact, classical conditioning is often called "Pavlovian conditioning". The essence of classical conditioning is that we are pairing something that previously caused no reaction in the dog with a predictable outcome that causes an uncontrollable reflex in the dog. The most famous of these experiments involved the ringing of a bell causing the dog to salivate. In this example, the bell is the "neutral stimulus"; that is, something that the dog has no innate reaction to. Food, in this context is an "unconditioned stimulus"; that is, something that the dog will react to without having had to learn that reaction. The drooling is the "unconditioned response"; an unlearned response the dog has to the food.
Pavlov noticed that, when the bell was rung immediately before food was presented, the dog learned to associate the delivery of food with the bell and the ringing of the bell alone would cause the dog to salivate.
Before conditioning:
Food = salivation
Bell = no response
During conditioning:
Bell + food = salivation
After conditioning:
Bell = salivation
The key point in classical conditioning is that you are pairing something that was previously neutral (causes no reaction) with something that causes a reflexive reaction, and that reflexive reaction is passed on to the previously neutral stimulus.
Examples are all around us. Consider the person who, as a teenager, gets horribly drunk on cider. Yes, I'm using my husband's experience for this one. Shhhh, don't tell him! Prior to "conditioning", he had no reflexive response upon tasting cider. During conditioning, he drank so much he became nauseated. Since conditioning, the taste of cider makes him feel nauseated.
We can use classical conditioning to make either positive associations or negative associations with a previously neutral stimulus; the process itself doesn't care which it is. The dog who flinches away from a raised hand has been classically conditioned to fear that hand. The dog who gets excited by the sight of his collar has been classically conditioned that the collar means a fun walk.
Neutral stimulus? But what about counter conditioning?
Classical counter-conditioning uses the same principles of classical conditioning above, but rather than starting with a neutral stimulus (something that the dog doesn't have a reflexive reaction to), we are starting from a point where he does, and we are trying to counter that reaction. In the world of dog training, this is normally taking something the dog has a negative association with (the sound of thunder, the presence of strangers, travelling in the car etc) and replacing those feelings with a positive response. The practicalities of this are outside the scope of this post, but suffice to say this is a far harder task than starting with a neutral stimulus, so as dog trainers, it makes sense that we should anticipate anything that we introduce to our dogs that has the possibility to cause a negative reaction and do our utmost to classically condition it to a positive reaction from the very start. As soon as our dog makes a negative association between two things, our job of getting them to accept it becomes so much harder.
Operant conditioning
Whereas classical conditioning works with reflexes, operant conditioning works with behaviours. That is, the dog performs a behaviour which results in an outcome, be that either good or bad for the dog. In essence, this is about the dog learning that their actions have reliable consequences.
These can be bundled into two categories; consequences that increase the chance of the behaviour happening again in the future are "reinforcers", and consequences that decrease the chance of the behaviour happening again in the future are "punishers". It is important to note that the only one who can decide if something is reinforcing or punishing is the learner. I might perform a behaviour if I knew there was a nice cold glass of cider waiting for me at the end of it. Poor J would avoid performing the same behaviour because he can't abide the smell of cider. It is also important to note that reinforcers and punishers can be fluid in their definition. If it's a really hot day and I'm thirsty, that cider would be a high value reinforcer. If it's a cold, wet day, it's far less appealing. If I have a hangover, it might even be punishing.
The important thing to remember is that we cannot say, in the moment, whether something is reinforcing or punishing because that is a function of the behaviour; only if it actually makes the behaviour more likely to happen in future can we say it's a reinforcer, and only if it actually makes the behaviour less likely to happen in the future can we say it is a punisher. Of course, we can have a pretty good guess in broad terms; giving your dog food for doing something you like will generally act as a reinforcer. Using physical violence against your dog will generally act as a punisher. But, until it has been shown that the behaviour increases or decreases in likelihood in the future, we cannot say for sure.
I'll talk more about the quadrants of operant conditioning in another post.
In summary
Classical conditioning pairs an involuntary response with a stimulus.
Operant conditioning pairs a consequence with a voluntary behaviour.
Classical conditioning
It's almost impossible to have a discussion about classical conditioning without mentioning Pavlov and his experiments. In fact, classical conditioning is often called "Pavlovian conditioning". The essence of classical conditioning is that we are pairing something that previously caused no reaction in the dog with a predictable outcome that causes an uncontrollable reflex in the dog. The most famous of these experiments involved the ringing of a bell causing the dog to salivate. In this example, the bell is the "neutral stimulus"; that is, something that the dog has no innate reaction to. Food, in this context is an "unconditioned stimulus"; that is, something that the dog will react to without having had to learn that reaction. The drooling is the "unconditioned response"; an unlearned response the dog has to the food.
Pavlov noticed that, when the bell was rung immediately before food was presented, the dog learned to associate the delivery of food with the bell and the ringing of the bell alone would cause the dog to salivate.
Before conditioning:
Food = salivation
Bell = no response
During conditioning:
Bell + food = salivation
After conditioning:
Bell = salivation
The key point in classical conditioning is that you are pairing something that was previously neutral (causes no reaction) with something that causes a reflexive reaction, and that reflexive reaction is passed on to the previously neutral stimulus.
Examples are all around us. Consider the person who, as a teenager, gets horribly drunk on cider. Yes, I'm using my husband's experience for this one. Shhhh, don't tell him! Prior to "conditioning", he had no reflexive response upon tasting cider. During conditioning, he drank so much he became nauseated. Since conditioning, the taste of cider makes him feel nauseated.
We can use classical conditioning to make either positive associations or negative associations with a previously neutral stimulus; the process itself doesn't care which it is. The dog who flinches away from a raised hand has been classically conditioned to fear that hand. The dog who gets excited by the sight of his collar has been classically conditioned that the collar means a fun walk.
Neutral stimulus? But what about counter conditioning?
Classical counter-conditioning uses the same principles of classical conditioning above, but rather than starting with a neutral stimulus (something that the dog doesn't have a reflexive reaction to), we are starting from a point where he does, and we are trying to counter that reaction. In the world of dog training, this is normally taking something the dog has a negative association with (the sound of thunder, the presence of strangers, travelling in the car etc) and replacing those feelings with a positive response. The practicalities of this are outside the scope of this post, but suffice to say this is a far harder task than starting with a neutral stimulus, so as dog trainers, it makes sense that we should anticipate anything that we introduce to our dogs that has the possibility to cause a negative reaction and do our utmost to classically condition it to a positive reaction from the very start. As soon as our dog makes a negative association between two things, our job of getting them to accept it becomes so much harder.
Operant conditioning
Whereas classical conditioning works with reflexes, operant conditioning works with behaviours. That is, the dog performs a behaviour which results in an outcome, be that either good or bad for the dog. In essence, this is about the dog learning that their actions have reliable consequences.
These can be bundled into two categories; consequences that increase the chance of the behaviour happening again in the future are "reinforcers", and consequences that decrease the chance of the behaviour happening again in the future are "punishers". It is important to note that the only one who can decide if something is reinforcing or punishing is the learner. I might perform a behaviour if I knew there was a nice cold glass of cider waiting for me at the end of it. Poor J would avoid performing the same behaviour because he can't abide the smell of cider. It is also important to note that reinforcers and punishers can be fluid in their definition. If it's a really hot day and I'm thirsty, that cider would be a high value reinforcer. If it's a cold, wet day, it's far less appealing. If I have a hangover, it might even be punishing.
The important thing to remember is that we cannot say, in the moment, whether something is reinforcing or punishing because that is a function of the behaviour; only if it actually makes the behaviour more likely to happen in future can we say it's a reinforcer, and only if it actually makes the behaviour less likely to happen in the future can we say it is a punisher. Of course, we can have a pretty good guess in broad terms; giving your dog food for doing something you like will generally act as a reinforcer. Using physical violence against your dog will generally act as a punisher. But, until it has been shown that the behaviour increases or decreases in likelihood in the future, we cannot say for sure.
I'll talk more about the quadrants of operant conditioning in another post.
In summary
Classical conditioning pairs an involuntary response with a stimulus.
Operant conditioning pairs a consequence with a voluntary behaviour.
