Up to this point, we've been talking about conditioning as if it occurred in a vacuum -- that organisms come into the environment as a tabula rasa, capable of learning any one thing as easily as anything else. But organisms bring a rich genetic endowment to their interactions with the environment, and their inheritance influences what they are likely to do, when they are likely to do it, and their capacity for learning certain kinds of connections between their behavior and environmental contingencies.

When there is a conflict between learned and instinctual behavior, organisms sometimes do things that don't fit the simple model of conditioning first put forth by John B. Watson, and to a lesser extent B. F. Skinner. In the 1950's, some behaviorists came to call these conflicts between learned and instinctual behavior misbehavior. Misbehavior occurs when the contingencies of reinforcement alone can not explain the behavior of the organism. Muffy's story of Jason's days as a football star (Does Jason Count?) demonstrates a common form of misbehavior. Jason misbehaves because his operantly maintained behavior conflicts with some biologically constrained form of behavior -- the fundamental cause of all forms of misbehavior.

The term misbehavior comes from a 1961 American Psychologist article by Keller and Marion Breland entitled The Misbehavior of Organisms. The title was a playful jab at B. F. Skinner, their mentor, whose 1938 book The Behavior of Organisms is widely viewed as the seminal work in the development of the experimental analysis of behavior. In this article, the Breland's talked about how their conditioning of animals was more or less difficult depending on the species-specific behaviors of the animals they were training. For example, teaching a chicken to "dance" is pretty easy--as it waits for food, chickens naturally "scratch" at the ground, which resembles dancing. Trying to train a chicken to stand still to obtain food, however, is quite difficult. If you are a chicken, you scratch the ground when obtaining food is likely. Chickens don't stand around waiting for food. It's in the nature of the beast to scratch. You can try your best to reinforce a chicken for standing still, but it just won't do it. The closer it gets to the time when food will be presented, the more difficult the chicken finds it to stand still, even if standing still is what's being reinforced.

The reason why the Brelands considered this to be misbehavior is that it should be just as easy to shape a chicken for standing still as for scratching if in fact all operants are created equal. But there is a wired-in (biological) connection between scratching and food for a chicken.

How does misbehavior affect the general applicability of conditioning principles?

One goal of behaviorism has been to look for basic principles that underlie all behavior. The concept of misbehavior seems to make that goal less likely. But science advances when a theory can be broadened to incorporate data that previously did not fit. In Physics, Newtonian principles worked pretty well for most situations, but there were some anomalies (misbehavior of physical bodies, if you will). Einstein's theory of relativity accounted for Newtonian principles as well as the anomalies, and was therefore an advance. It didn't throw out Newtonian principles, just put them into a broader context. By the same token, misbehavior has turned out to be crucial in the development of a more inclusive theory of behavior, which we will be getting to shortly. But before we can get there, we need to understand the importance of misbehavior.

After discussing some laboratory examples of misbehavior, we will broaden our look at constrained behavior by discussing some examples of classically instinctual behavior.

The table below shows a rough approximation of where some forms of behavior may sit along the continuum of constraint. The notion of constraint, in this case, is how arbitrary the relationship is between the environment and the behavior typically seen in it. Two corollaries of this are that the more arbitrary the relationship, the more quickly and reliably the behavior occurs with little or no learning involved, and the less the behavior is modifiable by experience.

So, let's take a look at some examples, and see why we're considering them misbehavior.

Historically important examples of misbehavior.

Instinctive Drift-- The classic example of misbehavior described by Keller and Marion Breland. While training animals to perform "human" type actions for TV commercials, animals would perform well for a while, then their behavior would deteriorate for no apparent reason -- they were still being reinforced, etc., but the control exerted by the contingencies of reinforcement were no longer maintaining the behavior. But the misbehavior was nonrandom -- there was a characteristic pattern to the deterioration of the operantly maintained behavior. For example, if a pig was being conditioned to put a coin in a piggy bank (I remember seeing this ad as a youth growing up in Los Angeles, where the Brelands had their business), everything would go fine for a while, but then the pig would get into rooting the coin around the floor, and the reinforced chain of behaviors would break down. The Brelands saw this as misbehavior in that the principles of behavior would suggest that any behavior that does not lead to reinforcement would be dropped, yet this behavior, even though it delayed, or even prevented reinforcement, not only developed but became more elaborate. Why would the pig misbehave? Well, pigs will be pigs. Pigs, by their genetic nature, root their food. The coin is associated with food, so the pig starts rooting the coin. Racoons, in a similar training situation, began washing the coin, a typically raccoon thing to do. The significance of this misbehavior was not lost on other behaviorists, and the work of the Brelands lead to a revolution in the scientific analysis of behavior, the result of which we will explore in the final unit of the semester.

Taste Aversion Learning. We are wired (biologically prepared, in the terms of Martin Seligmann) to learn some things quickly -- showing strong disgust for anything we've eaten soon before getting physically ill is perhaps the clearest example. Taste Aversion Learning is a genetically predetermined form of Pavlovian conditioning -- we learn, in a single trial, to stay away from whatever we ate prior to getting sick, whether or not that food actually caused the illness. This phenomenon is also referred to as bait shyness.

What is remarkable about it is how quickly and strongly the conditioning occurs, how tolerant it is of long delays between the presentation of the CS and the UCS, how persistent it is, and how we are prevented from associating anything other than taste with getting sick (hence the term constraint). (See a synopsis of the study by Garcia for some data on the phenomenon.)

Keep in mind the adaptive value of this learning (and all constrained learning) -- learning what makes you sick -- what might kill you -- is something you want to learn very quickly. If you survive the first episode, you don't want to go through it again, and perhaps not be so lucky next time. The environment would tend to select those individuals who were more prepared, biologically, to learn Taste Aversion Learning quickly.

An example from last semester of taste Aversion learning.

Autoshaping. Normal shaping of a pigeon to peck a key, using successive approximations, takes about an hour to get a naive pigeon to peck the key reliably. Suppose, though, instead of the laborious process of successive approximations, we simply arrange some programming equipment to present a hungry pigeon with grain once a minute, turn on the key light ten seconds before each presentation of food, go get a cup of coffee, and come back later and see what happened? That's all autoshaping is. Keypecks, when they eventually occur, are reinforced with food, just like with regular shaping.

Yet, the pigeon does come to peck the key with this procedure, and in about half the time that it takes to condition it by successive approximations. As is true with many of the examples of misbehavior, autoshaping was viewed as a curiosity when experiments demonstrating it were first published. The curiosity came from the fact that, even though no specific response was being reinforced, autoshaping produced conditioning more rapidly than when a specific response was being shaped. People soon recognized the Pavlovian nature of the autoshaping procedure, and began talking about the elicitation of what is normally considered an operant response. The process is diagrammed below:

     UCS -------------------- UCR
    food                     pecking

      CS --------------------- CR
    light                    pecking

Omission effect. If autoshaping was viewed as a curiosity, the omission effect, or negative automaintenance as it is sometimes called, caused an immediate revolution. The omission effect is very similar to autoshaping, with one small change -- instead of the elicited keypeck producing food, it prevents the presentation of food that would have occurred following the ten seconds of light. In other words, key pecking is being negatively punished. So, no big deal, right? This seems like the pigeon equivalent of winning the lottery -- just don't peck the key, stupid, and you'll get food once a minute. But a pigeon can't do this -- it pecks the key when the key light comes on, and continues to do so, until eventually several minutes goes by without food, extinction sets in so it finally doesn't peck the lit key, food is presented, and a new round of keypecking begins. The pigeon will never learn to stop pecking the key -- it simply can't help itself. This is clearly misbehavior. The pigeon does the opposite of what the contingencies of reinforcement predict. And, the reason seems pretty clear -- keypecking is being elicited by a Pavlovian arrangement, and Pavlovian responding is largely involuntary. Keypecking for a pigeon is not like lever pressing for a rat. Pigeons peck, just like pigs root and raccoons wash. It's in their nature. So, we have an organism which is genetically prewired, or constrained, to peck in the presence of food.

Conditioned suppression. In a classic experiment by Estes and Skinner (1941), rats were reinforced for pressing a lever on a VI schedule of reinforcement. Every now and then, a tone would come on, stay on for five minutes, then be terminated along with the delivery of an unavoidable electric shock. Fairly quickly, the rats would suppress responding during the tone, even though lever pressing was still reinforced during this time. Skinner and Estes attributed the suppression to the generalized effects of punishment. Skinner used this experiment as evidence for why punishment is bad. But another way of looking at this experiment is to consider the five minutes of tone to be a CS, which evokes fear (CR), since it is reliably followed by shock (UCR). It is the Pavlovianly evoked CR that interferes with lever pressing, not the generalized effects of punishment. This may not sound like a real big difference from what Skinner and Estes were saying, until we change the situation slightly. Let's say that, instead of the five minutes of tone terminating with the delivery of an unavoidable shock, it is accompanied by several seconds access to water (the rat is both hungry and thirsty). This is exactly what Peter Harzem did. Conditioned suppression develops under these conditions as well. It seems inappropriate to talk about the fear-evoking effects of water. But, we still have a Pavlovian paradigm, and the CRs associated with drinking are also incompatible with lever pressing, so Conditioned Suppression is really an example of Pavlovian conditioning overwhelming operant, rather than the generalized effects of punishment.

If an antagonistic Pavlovian CS reduces an ongoing operant performance, what happens if a CS is presented during an operant performance that evokes a CR similar to the reinforcement? For example, let's say a rat is pressing a lever and food is the reinforcement. Now we present a tone for five minutes, and the tone is followed by free food. Under these conditions, the rat's operant performance will be enhanced. This is referred to as Conditioned Facilitation.

Instinctual behavior

Species-specific Defense Reactions. most organisms have characteristic ways of behaving when threatened. For example, there's an old saying about people when under attack -- fight or flight -- we either aggress back or escape the situation. We learn to do either of these pretty quickly. Therefore. we are constrained from learning other modes of behavior, even when they may be more adaptive. Say, for example, you are having a dispute with your mate. We have a tendency to escalate the argument, or just walk away from it, when reasoned negotiation would be better than either of these alternatives. But the fact is, reasoned negotiation is very difficult under these circumstances, because it's not what we are wired to do when threatened (which is how we respond to disputes), even with people we care very deeply about.

Ethology. Ethology is sometimes referred to as the Biology of Behavior, indicating the source of the perspective (biology), and the nature of ethologists' take on behavior (the instinctual behavior exhibited by all members of the species). This approach contrasts markedly with the approach of traditional behaviorists -- the acquired behavior of individuals. The contrast can be viewed as the classic nature vs. nurture dispute that has characterized (and diverted) psychology for decades. For example, consider how these two would differ in their interest in rats. A behaviorist would be interested in how the rat acquires new behavior, and how that process is similar to the behavior of other species, so would set up an environment devoid of most stimulation, except that which she wishes to manipulate -- the typical Skinner Box. The goal of this arrangement is to suppress those stimuli that would elicit characteristically rat-like behavior. An ethologist, on the other hand, would be interested in just those things that the behaviorist wishes to suppress, since she is interested in those actions that make a rat a rat. So her "lab" would likely be the natural environment of rats (e.g. -- Newark, NJ), or a reconstruction in the laboratory of a rat-like environment. In many respects, these two approaches anchor the two ends of the continuum of constraint, with behaviorists interested in unconstrained behavior, and ethologists interested in constrained behavior.

To understand the ethological approach better, let's look a couple of typical concepts from ethology, and how they are studied:

1. The Courtship of Sticklebacks.

   In the spring, male sticklebacks change physiologically (their eyes and underbelly turn red), mark out, and defend a territory. When a female enters the territory, the male begins a courtship dance, which the female follows. The dance leads to the nest that the male has prepared, she enters the nest, deposits her eggs, he chases her off, fertilizes them, then becomes dad until they hatch. Notice in this example that the behavior is automatic--sticklebacks instinctively know what to do when it comes to courtship and mating. In other words, it is not a product of the learning history of the individual. Konrad Lorenz, a Nobel Prize-winning ethologist, has proposed a hydraulic model of instinctual behavior to explain this form of behavior. His model breaks instinctual behavior down into four parts:

   Action Specific Energy (ASE)
   At times, determined by the characteristics of the species, there is a build-up of a drive (energy) to behave in certain ways (action specific)

   Sign Stimulus (SS)
   A pre-wired environmental event (SS), when encountered, will trigger a response in the organism.

   Innate Releasing Mechanism (IRM)

   An inborn neural mechanism that the individual has inherited (IRM) allows it to recognize the SS when it appears, given there has been sufficient build-up of ASE.

   Fixed Action Pattern (FAP)
   With the proper build-up of ASE, the right IRM, and the encounter of the SS, the individual will behave in well-defined, species specific ways (FAP). The male has the instinctual ability (IRM) to recognize the female (SS) when he is ready to mate (ASE) and court her (FAP).

2. The Imprinting of gosling. Early in life (around one day of age), newly-hatched goslings figure out who mom is. During this critical period, it's whoever, or whatever, is moving around in front of them for an extended period of time. Normally, that would be Mother Goose. But not always. In a classic experiment by Lorenz, he isolated goose eggs, and when they hatched, he was what was around during their first day of life. The goslings imprinted on Lorenz, and from then on, would follow him around as if he were their mother, ignoring other geese. It's important to keep in mind that, like other forms of constrained learning, the gosling learns, very rapidly, something very important -- who mother is. A gosling who doesn't learn this will not survive for long. The constraint with imprinting is that, if you don't learn this very important thing during the critical period, it becomes very difficult to learn it at a later date.

   Other forms of imprinting/critical periods.

   • Not only do babies imprint on who their parents are, but parents imprint on who their babies are as well. Ask any parent about how they felt when they first held their baby. It's a powerful experience. And, it makes sense -- parents must have a strong desire to care for their baby from the get-go, or it won't survive. My sister, who delivers babies for a living, tells me that lots of the ones she delivers are, quite frankly, pretty homely. But the parents invariably think what they're handed is the most beautiful baby in the world. Score one for Mother Nature!
     Brood parasites, such as Cowbirds and Cuckoos, depend on this parental imprinting for their survival. These birds lay their eggs in the nests of other species, and when they hatch, they dispose of their nestmates, the "true" offspring of the parents. But Mama and Papa Wren love whatever is in their nest, and raise the baby Cowbird as if it were their own.
   • Many linguists also believe that there is a critical period for language acquisition. If you don't encounter and learn a language during the first couple of years of life, language acquisition will be almost impossible.

   Other places to go for more information on constrained learning:

   • Sociobiology home page.

Evolution and Animal Intelligence

Before going on, be sure to Make Up Your Own Question!

Email Tom

Last modified on October 31, 1998.