Overview of research in operant and respondent conditioning (Proceedings)
Our popular understanding of animal behavior rests in two major areas of study – ethology and behavior analysis. Ethologists study the way animals behave in a natural habitat. They study body language, posturing and assorted ways that animals influence each other and their environment. Behavior analysts attempt to study the mechanism by which instinctive behaviors are changed and augmented in response to environmental influences. Added to these two major study areas is a lesser known, but firmly established study called respondent conditioning, also known as Classical or Pavlovian conditioning. Pavlovian conditioning is a pre-condition to and concurrent witho operant conditioning. Without the ability to connect initially neutral stimuli to external events and consequences, operant learning cannot take place.
Of these three concerns, ethology is the one best represented among current writers. When you hear people speak of "talking dog" or referring to dogs as "pack animals" you are hearing the ethologist's perspective on dog behavior. Though an ethologist may tell you that digging pot holes is a behavior common to dogs, knowing how to stop the behavior requires a behaviorist. The best the ethologist can do is suggest things like removing the opportunity to dig. Changing behavior requires influencing physiological functions that modify or eliminate an animal's "natural" behavior. This is the concern of operant and respondent conditioning.
Respondent Conditioning:The work of Ivan Pavlov began in about 1890 and stands are the first extensive scientific investigation into behavior. Pavlov started his research with a single thought. To paraphrase him, if a dog had to wait for the claws of a bear to grasp him before he responded, there would be no dogs. Contrary to popular assumptions, Pavlov wasn't looking for ways to trigger reflexes such as salivation and the Galvanic response. He was studying the mechanism by which animals make associations. His seminal work set the stage for the efficient connection of initially meaningless signals with internal physiological events. If this sounds like splitting hairs, just go to your waiting room and hear someone say "Sit, sit, sit" and you'll realize that Pavlov's goal has been forgotten. In a Pavlovian world, commands aren't shouted or repeated. They are subtle, specifically connected signals that trigger bodily reflexes or operant behaviors. In essence, respondent conditioning is the basis for all learning.
A typical respondent conditioning experiment includes several common factors. First, an unlearned or unconditioned stimulus is selected that will create a change in the animals physiological state. In the most familiar of Pavlov's experiments, food was the unconditioned stimulus that was associated with the sound of a bell — called the "conditional stimulus." (Note: In Russian, the words conditioned and conditional are the same. The title of Pavlov's book was translated into English as Conditioned Reflexes. The actual title of his book is "Conditional Reflexes" describing how the associations are formed and the limitations that are inherent in the process.)
Pavlov also taught dogs to discriminate between a black circle on a white background and a black ellipse on a white background. Other experiments included associating electric shock with food and a rudimentary tactile stimulation machine – a mechanical finger that pleasantly scratched the dog's skin. At each level of association, simple reflexes were tracked to see how well the unconditional reflex was associated with the conditional reflex. It is inherent in Pavlov's work to assume that there are limits to these associations. This point is rarely recognized in modern usage of the term conditioned reflex. Further, many people assume that because Pavlov could make an apparently perfect association between a bell and the presentation of food, the bell can be used as a surrogate for tangible reinforcement. Ironically, this assumption runs contrary to the theme of Pavlov's research. His extinction procedures clearly illustrate the weakening of "conditional reinforcers" when they are triggered repeatedly.
Operant conditioning describes "learning by consequences." If you bump your head on a low doorway, you will be likely to duck the next time you enter. If you find a $50.00 bill under a bush on a golf course, you will cast your eyes on that bush the next time you are playing that hole. Operant conditioning has little to do with thinking about events and more to do with simple reactions to things that help or hurt us. Most research in operant conditioning is called "discrete trial" learning. A single response is selected as the behavior to be observed. In traditional studies the operant is pressing a lever if the animal is a rat and pecking a key if the animal is a pigeon. The selection of these two simple behaviors as the standard for research has a major advantage and a serious drawback.
The advantage of this limited, standardized format is that studying single operants allows researchers to compare "apples to apples" in widely disparate studies. Additionally, the complexity of a given research question is not dramatically limited by the simplicity of the operant. Studies of operant variability require rats to press levers in specific "creative" sequences. Studies of this sort can be used as models of behaviors such as Attention Deficit Disorder.
The major limitation of current behavior analytic research is its focus on the rate of response of a single operant. These investigations have little to say about how animals access various behaviors within their normal repertoire. Additionally, the long-standing unscientific bias in favor of "positive" behavioral effects seriously weakens any reliance on this research. In practical terms, how often a rat presses a lever has nothing to do with whether a dog will bite a child. The practitioner is required to make sure that there is no rate of response of the operant behavior, biting. Likewise, complex studies of variable ratio schedules of positive reinforcement do not offer practical solutions for cat litter box problems or predict the reliability of a dog's tendency to sit on command. Until more complex studies are conducted that integrate more realistic variables into behavior analytic research they will be of limited use in veterinary practice.