Conditioning

Much of animal learning is captured by the conditioning paradigm. In its simplest form (probably what actually happens under field conditions) an association is formed between an action and a reward. Associative learning allows birds to efficiently find bugs under rocks and bees to find nectar in specific flowers. This simple type of learning allows animals to behave efficiently, seeking resources where they have been found before, or collecting them in ways that have worked previously.

Learned associations can be built using normal stimulus-response patterns. A normal stimulus, such as the odor of food, elicits salivation in dogs. These are termed the unconditioned stimulus (US) and unconditioned response (UCR). When Pavlov rang the bell at the same time he presented food (a US), the bell became what we can a conditioned stimulus (CS). Though the bell is irrelevant to normal feeding in a dog, the dog associates the bell with the food (the US with the CS) and eventually responds by salivating when the bell is run (CS) even if no food is present.

Taking this one step further, many animals are able to build an association with a reward between seemingly irrelevant actions and stimuli, if the stimulus is given at the time the reward is received. You can teach your dog to "shake" hands when you present your hand, based on the dogs expectation of receiving a food reward after shaking. Once the association between your hand signal and the dogs' extension of its paw is built, then you can stop giving the food reward. Such behavior is built on positive reinforcement--association of the desired action with the receipt of a reward.

Not all animals can learn something as irrelevant to their biology as shaking hands. Ease of conditioned learning is often dictated by the importance of the stimulus and response to the animal species' evolutionary history and ecological conditions.

You might think that negative reinforcement would be effective in the same way as positive reinforcement. Indeed, many animals can be taught boundaries based on painful experiences, such as electric fences or shock collars. Most training using negative reinforcement, though, is less effective than positive reinforcement. This is because when being trained using positive reinforcement, an animal is encouraged to associate an act with the reward because the trainer is planning, in advance, to elicit the desired response. Going back to the example of dog shaking hands, it is likely that you would start the training by showing the dog the food. This would get her interest, and helps to build the association between the desired behavior and subsequent reward.

Negative reinforcement, on the other hand, generally comes after an act which the trainer cannot predict. If your dog urinates on the floor, and you then hit the dog, will it associate the punishment with the bad behavior? Likely not, because it was unaware that punishment was likely before it urinated--and even if the association is made, how does the dog know what alternative behavior will receive a positive reward? Urinary retention is a more common result that learning to go outside to urinate. Moving the urinating dog to an appropriate location, combined with positive reinforcement, is much more likely to achieve the desired result. Understanding this critical difference between positive and negative reinforcement in training is an important step in learning how to train domestic animals.

In studies of conditioning, negative reinforcement can be used in "extinction", or inhibition experiments, to determine if a previously learned association can be masked by later experience.

Operant conditioning is an extension of classical conditioning, in which the animal learns to operate an environmental feature to receive a reward. We most commonly think of rats and pigeons pressing levers in boxes (the "Skinner box") in operant conditioning; the hand-shaking dog has probably been operantly conditioned, as well.

Much insight animal learning and memory has been gained from the study of conditioning. Ethologists and behavioral ecologists often dismiss laboratory tests of conditioning as being so far removed from the animal's biology as to be irrelevant to "natural" biology. Observations of animals in the field, though, suggest that the trial-and-error learning which often is used to gain experience with the environment is, in fact, one and the same as conditioning.

 

 

Return to beginning of Chapter 6

Return to Table of Contents

page 6-1
copyright ©2001 Michael D. Breed, all rights reserved