One of the foundational observations in the field of psychology is that animals are capable of learning associations between events in the environment. This ability underlies many complex behaviors and is what allows individual animals to adapt to changing conditions instead of relying on much slower processes of evolution. In this module, we won’t be studying learning directly, but we will be taking advantage of it to study how sensory information is processed and categorized. This page will provide you with a primer on the major concepts of associative learning and how they are measured in the lab.
As the name implies, associative learning involves the formation of an association or connection between two events that would otherwise be unrelated. In other words, animals are able to learn that one event predicts another and are able to use this information to perform actions that enhance their potential for survival. The emergence of this learned behavior is called conditioning. A major conceptual distinction in associative learning is based on the nature of the predictive event. If the predictive event is something that happens out in the environment, it’s called a stimulus, and the conditioning is called classical or Pavlovian. In contrast, if the predictive event is an action that the animal itself makes, then it’s called a response, and the conditioning is called operant or instrumental learning. Operant and classical conditioning are similar in many respects, including the underlying cellular and molecular mechanisms, but they are different enough in an operational sense that we’ll consider them each in turn.
In classical conditioning, a neutral stimulus becomes associated with a stimulus that has an innate value. For historical reasons, the neutral stimulus is called the conditioned stimulus (CS) and the innately valued stimulus is called the unconditioned stimulus (US). The value, or valence of the US can be positive or negative. Positive USs include food, water, and opportunities to mate or socialize. Negative USs include pain and nausea. You know a stimulus is a US if there is an innate, or unconditioned response (UR) to it. Animals will approach a positive US and avoid a negative US. Classical conditioning occurs when the CS and US occur in such a way that the CS helps to predict the US. In general, this means that the CS happens before the US with some degree of reliability. Over repeated experiences of a CS-US association, animals will start to respond to the CS, even when the US is not present. This conditioned response (CR) often resembles the UR, but more generally is some sort of behavior that allows the animal to prepare for the US. The classic example comes from Pavlov, who discovered that a dog will salivate (the CR) when it hears a metronome or other sound (CS) that predicts the arrival of food (US). Similarly, if a mouse learns that a tone CS predicts that the floor of the cage will electrify and shock it (US), the CR will be to freeze or to try to get to a part of the cage where the shock doesn’t occur. Several important observations to note:
In operant conditioning, the animal associates its own responses with events in the environment, which in this context are called reinforcers and punishments. Animals tend to increase the rate of responses that lead to reinforcers (food, water, opportunities to mate and socialize, etc) and decrease the rate of responses that lead to punishments (pain, nausea, etc). Contrary to common usage, positive and negative in this context do not refer to the valence of the outcome, but to whether something is added or taken away. Negative reinforcement is the removal of a undesired feature of the environment, and it also causes the response rate to increase. Similarly to classical conditioning, operant responses can be extinguished when the response is no longer reinforced, or even if the frequency of reinforcement decreases.
The value of a reinforcer may fluctuate in accordance with the animal’s behavioral or physiological state. A hungry animal is much more motivated by food than one that has just finished a big meal. Interestingly, a hungry animal is more likely to engage in a wide variety of exploratory behaviors, which increase the chances that it will do something that leads to food. In contrast, a sated animal is less active. Operant behavioral experiments almost always start by putting the animal in a state where it is motivated by the reinforcer that will be offered, usually by depriving the animal of it for an hour or so.
Operant responses can be almost anything in an animal’s behavioral repertoire. Pigeons have been trained to turn in circles, play ping-pong, bow, and engage in “conversations” with each other through operant conditioning. In most situations, though, the experimenter wants to be able to unequivocally (and preferably automatically) measure responses, so animals are usually conditioned to operant manipulanda like levers or keys.
Animals are exquisitely sensitive to the state of the environment when they are learning operant responses and will quickly discover whether a response is only reinforced or punished under certain conditions. These conditions are known as discriminative stimuli. For example, a pigeon might learn that pecking a key is rewarded when the color of a light in the room is green, but not when the light is red or off. Discriminative stimuli are extremely useful because they allow us to ask how an animal perceives, or internally represents, a stimulus. This facet of operant learning is what we will use to study how starlings interpret the songs of other individuals.
We will have a brief quiz at the beginning of class to assess your knowledge of the major concepts and terms in this page. You should also be prepared to give some examples of operant and classical conditioning in humans and to identify the CS, US, CR, UR, response, reinforcer, and punishment in each as they apply. There are a plethora of online videos and resources you can consult if you’re unclear on any of the above. You will need a laptop with at least one free USB port to connect to the computer that will control the operant apparatus.