Skinner and Pavlov

Skinner - 1904-1990

• Skinner's box - conditioning chamber.
• Instrumental/ operant conditioning - type of learning where an animal has to produce a response in order to receive some kind of outcome, a reinforcer (reward) in this case.
• He would shape the response of animals in order to get them to do things like press on levers.

Thorndike's Law of Effect

• Thorndike's law of effect says that if a response leads to a satisfying outcome, then that response is strengthened.
• The problem some people like Skinner and Watson had with Thorndike's law of effect is that what you mean by satisfying outcome, that implies you can determine something about the animal's own personal motivation.
• The assumption with the stimulus response association theory is that all the animal is learning is the strengthening or weakening of the connection between the stimulus (in this case the level), and the response (pressing on the level). The likelihood of them doing that depends on the likelihood of the outcome occurring.

Colwill and Rescorla (1985) p1

• Trained to do 2 things.
• When the lever is present, pressing the lever will lead to some food.
• If the lever is removed, there is a chain that can be dropped down, and when the rats pull on the chain they will get some sucrose from the same hole.
• the rats like sucrose and they like food, so they will learn to both press the lever and pull the chain.
• If stimulus response learning is responsible for that, then whenever they see the lever, they will press the lever, and whenever they see the chain, they will pull the chain, no matter what you do in stage 2.

Colwill and Rescorla (1985) p2

• In stage 2, they devalue one of those outcomes.
• they let the rat eat some food and pair that with illness (injecting the food with lithium chloride).
• So it will reject that food in the future because it has learnt that food makes them ill.
• Then they let them drink the sucrose but don't make them feel ill.
• So if you give them the 2 things they will drink the sucrose but leave the food.
• Then they test them with a lever and chain, both present at the same time - but they don't give them any food or sucrose as a result of pulling on that chain or pressing on that lever - done in extinction.

Colwill and Rescorla (1985) p3

• The stimulus response association that is formed in the first stage, if that is what the animals have learned, then what you have done to the food in stage 2 makes no difference to that association because the lever has never resulted in food that makes them ill.
• The only way you can change the likelihood that the animal is going to press that lever, according to stimulus response association, is by changing the likelihood that it gets food. They haven't done that. All they have done is given it the opportunity to press the lever and pull the chain.
• What happens in extinction is, the animals much prefer to press the chain than to pull the lever. This increases over time because they are not getting any payoff, there is no outcome being given there.

Skinner schedules of reinforcement

• He had a laboratory and he gave the rats food for pressing on the lever.
• He then started to ration the food and found that the animals would still continue to produce these behaviours.

Pavlov - Pavlovian/ classical conditioning

• Behaviours animal were producing like the saliva were conditional upon some other event - so conditional responses.
• Light/ buzzer - neutral stimuli - they don't have an important biological consequence to the dog.
• Becomes a conditioned stimulus because it has to be learned about in order for the animal to produce a response to it - paired with food.
• So the dog receives some food (unconditioned stimulus because the dog doesn't need to learn about that), and they begin to salivate to the presence of the food.
• After time pairing the light and food together (US and CS) the dog starts to salivate during the presence of the light at the anticipation of the food being delivered.

Difference between two forms of learning

• Operant/ instrumental conditioning:
• Where the animal has to produce a response to affect the change upon its environment, to get some food.
• Classical conditioning/ Pavlovian conditioning:
• Where the animal doesn't have to do anything. There is a stimulus, like a bell, and that is paired with an outcome, food, and the animal starts to learn the connection between those two things.

Slide 9

• 
• First when the animal doesn't know anything about the significance of the CS (the bell), the dog doesn't produce any saliva in response to being presented with the bell.
• Fairly quickly it starts to learn the significance of the bell and it starts to produce a conditioned response, which is too salivate.
• Doesn't go up forever, has a negatively accelerated curve, it goes up quickly and then levels off - as it levels off we call this asymptote which is the maximum level of responding that the unconditioned stimulus can support.

Slide 10

• 
• Extinction is where we take away the biological significant event.
• So the animal has learnt the connection between the bell and the food, but then we stop delivering the food after the bell has been played.
• At first the animal keeps responding as if it is going to get the food, but then it learns quickly that it is not going ti get the food, and stops producing that conditioned response.
• Extinction - animal stops producing conditioned response.

Slide 11

• 
• Spontaneous recovery - you train the animal to go through acquisition, then you put it through extinction, and then once it has extinguished its response, you leave it for 24 hours and give it the conditioned stimulus on its own again (ring the bell and see if it has produced saliva). An animal that has forgotten the link between the CS and the US would behave as if it has never heard the bell before.
• However, instead it spontaneously recovers its conditioned response - it responds as if it has learnt that connection and produces the conditioned response again.
• You can do it after another 24 hours and then it produces a smaller response.

Slide 12

• 
• What we think might be happening there is that animals are maybe learning 2 associations.
• First association is that the bell leads to food.
• The second association during extinction is that the bell leads to no food.
• So it has two competing associations that it is learning at the same time. So that one wins out at first, but after a 24 hour gap, it has got both of those memories in long term memory and so it responds partly as if the bell doesn’t signal food (like the extinction faze), and partly as if it does signal food, so halfway between the 2. But it goes through lots of extinction before finally towards this idea that the association that the bell doesn’t lead to food wins out.

Present stimuli but varying properties of CS

• 
• What happens when you present the stimuli but you vary some properties of the CS?
• The CS might be tones (tones of a particular frequency can vary - pitch/ intensity/higher/ lower).
• This is what happens when you train a dog to learn that a tone of a particular frequency (1200HZ) signals food.
• The peak of its responding is when a stimulus is presented.
• If you change the tone higher, the animal will continue to respond to the stimulus.
• Once you change the properties quite a lot e.g. much higher, the animal stops responding.
• We call this the ability to generalise from the trained stimulus, to other stimuli, known as stimulus generalisation.
• The curve where we change lots of different properties is known as generalisation gradient.
• The difference between its peak responding and its responding it does to other stimuli is known as generalisation decrement.
• Caption: Rachman, S J (1966) Psychological Record, 16, 293-296