Evaluating Animal Behavior
The techniques of behavioral pharmacology allow scientists to evaluate the relationship between an experimental manipulation such as a lesion or drug administration and changes in behavior. In a well-designed experiment, it is necessary to compare the behavior of the experimental treatment group with that of placebo control subjects. The neurobiological techniques (such as selective lesioning and intracerebral drug administration) described earlier tell us very little unless we have an objective measure of the behavioral consequences. Behavioral measures are crucial for (1) understanding the neurochemical basis of behavior as well as drug-induced changes in that behavior; (2) developing animal models of psychiatric disorders; and (3) screening the large number of newly designed and synthesized drug molecules in preclinical pharmaceutical settings.
Animal testing needs to be valid and reliable to produce useful information
Animal studies clearly have several advantages over studies using human subjects. The most obvious advantage is the use of rigorous controls. The living conditions (e.g., diet, exercise, room temperature, exposure to stress, day-night cycle) of animal subjects can be regulated far more precisely than those of humans. In addition, the histories of animal subjects are well known and the genetic backgrounds of a group of animals are very similar and well characterized. Finally, animals are the most appropriate subjects for the study of mechanisms of drug action because an understanding of the electophysiological and neurochemical bases of drug effects often requires invasive techniques that are obviously unethical with human subjects. Consider, for example, the valuable information gained from transgenically manipulated animals. In addition, drugs can be administered to animal subjects in ways not generally appropriate for humans, for example, over long periods of time to determine toxic effects or the potential for addiction. Finally, the brains and behavior of nonhuman mammals and humans are similar enough to allow generalization across species. For example, lesions of the central nucleus of the amygdala of rats produce profound changes in the animals’ conditioned emotional response. Likewise, tumors, strokes, or surgical procedures that damage the human amygdaloid complex produce profound changes in fearfulness, anxiety, and emotional memory.
The impact of animal testing in biomedical research on the quality of human life and its alternatives is discussed in a thought-provoking manner by Hollinger (1996) . The need for animal experimentation is best seen under conditions when research is impossible using human subjects, as when testing the effects of alcohol on fetal development. Ethical constraints prohibit researchers from administering varying doses of alcohol to groups of pregnant women to evaluate the effects on their newborns. Instead, data collected on alcohol consumption during pregnancy and the occurrence of fetal alcohol syndrome (FAS) suggests a relationship that tells us that the more alcohol a pregnant female consumes, the more likely it is that her infant will show signs of FAS. Although we know that infants of mothers who consume alcohol are more likely to show fetal abnormalities, the type of study described shows only a correlational relationship; we cannot assume alcohol causes FAS since other factors may be responsible for both. For example, poverty, poor diet, or other drug use may both lead to increased alcohol consumption and cause developmental defects in the fetus. Therefore, to learn more about how alcohol affects fetal development, we need to perform animal studies. Since animal testing remains an important part of new drug development and evaluation, strict animal care guidelines have been developed to ensure proper treatment of subjects. The animal-testing stage provides an important step between basic science and the treatment of human conditions.
The Health Extension Act of 1985 provides strict guidelines for the care of animals used in biomedical and behavioral research. The goal of the legislation is humane animal maintenance and experimentation that limits both the use of animals and animal distress. Each research institution must have an animal care committee that reviews each scientific protocol with three considerations in mind: (1) the research should be relevant to human or animal health, advancement of knowledge, or the good of society; (2) alternative methods such as computer simulations that do not require animal subjects must be considered; and (3) procedures must avoid or minimize discomfort, distress, and pain. Periodic inspections of living conditions assure that they are appropriate for their species and contribute to health and comfort: size, temperature, lighting, cleanliness, access to food and water, sanitation, and medical care are ensured. Animal care and use committees have the ability to veto any studies that they feel do not meet all the predetermined criteria.
Some animal tests used to evaluate drug effects on physiological measures such as blood pressure or body temperature closely resemble the test used for humans. These tests have high face validity. However, for many psychiatric disorders the symptoms are described in typically human terms, such as a certain facial expression, altered mood, or disordered thinking. In these cases a correlated, quantifiable measure in an animal is substituted for a more cognitive human behavior for testing purposes. When the correlation is strong, a drug that modifies rat behavior in a specific way can be expected to predictably alter a particular human behavior, even though the two behaviors seem unrelated. For instance, if a new drug were to reduce apomorphine-induced hyperactivity in rats, tests on humans might show it to be effective in treating schizophrenia. Tests such as these have low face validity. However, if the drug effects in the laboratory test closely parallel or predict the clinical effect, the tests may be said to demonstrate construct validity, or empirical validity. To be optimal, an animal behavioral test should also:
1. Be specific for the class of drug being screened. If antide-pressants, for example, produce a consistent response in a behavioral test, we would probably not want to see analgesic drugs producing the same effect.
2. Be sensitive so that the doses used are in a normal therapeutic range and show a dose-response relationship.
3. Demonstrate the same rank order of potency (i.e., ranking drugs according to the dose that is effective) as the drugs’ order of potency in therapeutic action.
In addition, good behavioral measures have high reliability, meaning that the same results will be recorded each time the test is used (Treit, 1985). Valid and reliable animal tests are an important component of the preclinical trials for new drug development.
A wide variety of behaviors are evaluated by psychopharmacologists
There are many behavioral tests used by psychopharmacologists and they vary considerably in complexity, time needed to be carried out, and cost, as well as validity and reliability. In this next section we will describe just a few of the available procedures, many of which will be referred to in subsequent posts.
Simple behavioral observation Many simple observations of untrained behaviors require little or no instrumentation. Among the observations made are measures of tremors, ptosis (drooping eyelids), salivation, defecation, catalepsy, reflexes, response to tail pinch, and changes in eating or drinking. Animals demonstrating catalepsy are still and immobile and will sometimes remain in an unusual posture if positioned by the experimenter. The time it takes for the animal to return to normal posture gives an indication of the extent of catalepsy. The use of catalepsy as a test to identify antipsychotic drugs that produce motor side effects demonstrates the usefulness of screening tests that are not clearly related to human behavior.
Measures of motor activity These measures identify drugs that produce sleep, sedation, or loss of coordination or, in contrast, drugs that stimulate activity. Spontaneous activity can be measured in a variety of ways. One popular method counts the number of times infrared light beams (invisible to rodents) directed across a designated space are broken. Automated video tracking with computerized analysis is a second method. A third, less automated technique (open field test) involves placing the animal in a prescribed area that is divided into squares so the investigator can record the number of squares traversed in a unit of time. It is also possible to count the number of fecal droppings and to observe the amount of time an animal spends along the walls of the chamber rather than venturing toward the open space. High fecal counts and low activity that is primarily at the perimeter of the cage are common indicators of anxiety.
Measures of analgesia Analgesia is the reduction of perceived pain without loss of consciousness. Analgesia testing with human subjects is difficult because the response to experimentally induced pain is quite different than that to chronic or pathological pain, in which anxiety and the anticipation of more pain influence the individual’s response. Of course, we cannot know whether an animal “feels pain” in the same way that a human does, but we can measure the animal’s avoidance of a noxious stimulus. One simple test is the tail-flick test, in which heat produced by a beam of light (the intensity of which is controlled by a rheostat) is focused on a portion of a rat’s tail. The latency between onset of the stimulus and the animal’s removal of its tail from the beam of light is assumed to be correlated with pain intensity.
Tests of learning and memory Objective measures of learning and memory, accompanied by careful interpretation of the results, are important whether you are using animal or human subjects. Keep in mind that these tests very often do not determine whether altered responses are due to drug- induced changes in attention or motivation, consolidation or retrieval of the memory, or other factors contributing to overall performance. Unless these other factors are considered, tests of learning are open to misinterpretation. Despite the challenges posed, finding new ways to manipulate the neurotransmitters involved in these functions will be central to developing drugs that are useful in treating memory deficits due to normal aging or neurological injuries or diseases such as Alzheimer’s and other dementias. There are a wide variety of tests available that depend on the presentation of information (training stage) followed by a delay and then the opportunity for performance (test stage). Higher cognitive processes can be evaluated by creating situations in which reorganization of the information presented is necessary before the appropriate response can be made.
Mazes Although the size and complexity of mazes can vary dramatically, what they have in common is a start box at the beginning of an alley with one (T-maze) or more (multiple T- maze) choice points that lead to the final goal box, which contains a small piece of food or other reward. A hungry rat is initially given an opportunity to explore the maze and find the food goal. On subsequent trials, learning is evaluated based on the number of errors at choice points and/or the time taken to reach the goal box. Careful evaluation of results is needed because drug-induced changes in behavior may be due to a change in either learning or motivation (e.g., does the drug make the animal more or less hungry? sedated? disoriented?).
Spatial learning tasks help us investigate the role of specific brain areas and neurotransmitters, such as acetylcholine, in forming memories for the relative locations of objects in the environment. One special type of maze, the radial arm maze, is made up of multiple arms radiating away from a central choice point with a small piece of food at the end of each arm. With very little experience, normal rats learn to forage efficiently by visiting each arm only once on a given day, indicating effective spatial memory for that particular episode. The task can be made more complex by blocking some arms on the initial trial before the animal is returned to the central choice point. The animal is expected to remember which arms have already been entered and move down only those that still contain food. For a normal rat the task is not complex, since it mimics the foraging behavior of animals in the wild, where they must remember where food has been found. But animals with selective lesions in the hippocampus (and other areas) as well as those injected with a cholinergic-blocking drug show significant impairment. Low doses of alcohol also interfere with spatial memory. Because the arms are identical, the animals must use cues in the environment to orient themselves in the maze, hence the need for spatial memory. The task is similar to our daily activity of driving home from work.
Not only do we need to recognize each of the landmarks along our route, but we must learn the relative locations of the objects with respect to each other. As we move along our route, our perceptions of the objects and their relative locations to us tell us where we are and where we should be going. Failure in this complex cognitive process is characteristic of Alzheimer’s patients who wander away and fail to find their way home.
A second test of spatial learning, the Morris water maze, uses a large circular pool of water that has been made opaque by the addition of milk or a dye. Animals placed in the pool must swim until they find
the escape platform that is hidden from view just below the surface of the water. The subject demonstrates that it has learned the spatial location of the submerged platform by navigating from different starting positions to the platform. Since there are no local cues to direct the escape behavior, successful escape requires the learning of the spatial position of the platform relative to landmarks outside the pool. When curtains surrounding the pool are drawn to block external visual cues, performance falls to chance levels, demonstrating the importance of visuospatial cues. As a laboratory technique, the water maze has several advantages. No extensive pretraining is required, and testing can be carried out over short periods of time. Escape from water motivates without the use of food or water deprivation or electric shock; this makes interpretation of drug studies easier, since drug-induced changes in motivation are less likely. One disadvantage, however, is that water immersion may cause endocrine or other stress effects that can interact with the drugs administered.
Delayed-response test This design assesses the type of memory often impaired by damage to the prefrontal cortex in humans. It is similar to tasks included in the Wechsler Memory Scale, which is used to evaluate working memory deficits in humans. In this task, an animal watches the experimenter put a piece of food in one of the food boxes in front of it. The boxes are then closed, and a sliding screen is placed between the monkey and the boxes for a few seconds or minutes (the delay). At the end of the delay, the screen is removed and the animal has the opportunity to recall under which of the covers food is available.
Visual short-term memory can be tested by slightly modifying the procedure. At the beginning of the trial, an object or other stimulus is presented as the sample. After a short delay, during which the sample stimulus is removed, the animal is given a choice between two or more visual stimuli, one of which is the same as the sample. If the animal chooses the pattern that matches the sample, it is given a food reward; an incorrect response yields no reward (see the section on operant conditioning techniques later in the post). To make the correct choice after the interval, the animal must “remember” the initial stimulus.
Measures of anxiety There are many biobehavioral measures available to identify novel antianxiety compounds and evaluate the neurochemical basis of anxiety. Most use induced fear as an analogy to human anxiety. Some use unconditioned animal reactions such as a tendency to avoid brightly lit places or heights, while others depend on traditional learning designs (see the conflict test described in the section on operant conditioning later in the post). The light-dark crossing task involves a two-compartment box with one side brightly lit (normally avoided by rodents) and the other side dark. Measures include the number of crossings between the bright and dark sections and the amount of time spent on each side, as well as total motor activity. Anxiety-reducing drugs produce a dose-dependent increase in the number of crossings and in overall activity while also increasing the amount of time spent in the light. The elevated plus-maze is a cross-shaped maze raised 50 cm off the floor that has two open arms (normally avoided due to aversion to heights) and two arms with enclosed sides. This quick and simple test shows a selective increase in open-arm exploration following treatment with antianxiety drugs and a reduction following treatment with caffeine and amphetamine, drugs considered to increase anxiety.
Measures of fear In contrast to the spontaneous-behavior models described so far, tests based on learned behaviors require a certain amount of training and hence are generally more costly and time-consuming. The conditioned emotional response depends on presentation of a signal (a light or tone) followed by an unavoidable electric shock to form a classically conditioned association. When the warning signal is presented during ongoing behavior, the behavior is suppressed (i.e., “freezing” occurs). Although this method has not always produced consistent results when used to screen antianxiety drugs, it has become an important tool in understanding the role of the amygdala and its neurochemistry in the conditioned fear response.
A second method is fear-potentiated startle, which refers to the enhancement of the basic startle response when the stimulus is preceded by the presentation of a conditioned fear stimulus. For example, if a light has been previously paired with a foot shock, the presentation of that light normally increases the magnitude of the startle response to a novel stimulus, such as a loud clap.
Measures of reward Although several popular measures to evaluate the rewarding and reinforcing effects of drugs are operant techniques (described in the next section), a method called conditioned place preference relies on a classically conditioned association between drug effect and environment. During conditioning trials over several days, the animal is injected with either drug or saline and consistently placed in one compartment or the other so that it associates the environment with the drug state. The rewarding or aversive effect of the drug is determined in a test session in which the animal has access to both compartments and the amount of time spent in each is monitored. If the drug is rewarding, the animal spends much more time in the compartment associated with the drug. If the drug is aversive, the animal prefers the compartment associated with saline injection. Additionally, researchers may study the biological basis for the rewarding effects by pretreating animals with selected receptor antagonists or neurotoxins to modify the place preference. Stolerman (1992) reviews several behavioral principles and methods related to drug reward and reinforcement.
Operant conditioning techniques provide a sensitive measure of drug effects
Operant conditioning has also made contributions to the study of drug effects on behavior. The underlying principle of operant conditioning is that consequences control behavior. An animal performs because it is reinforced for doing so. Animals learn to respond to obtain rewards and avoid punishment.
Although it is possible to train many types of operant responses, depending on the species of animal used, experiments are typically carried out in an operant chamber (Skinner box). An operant chamber is a soundproof box with a grid floor that can be electrified for shock delivery, a food or water dispenser for rewards, lights or loudspeaker for stimulus cue presentation, and levers that the animal can press. Computerized stimulus presentation and data collection provide the opportunity to measure the total number of responses per unit time. In addition, the technique records response rates and interresponse times, which provide a stable and sensitive measure of continuous behavior.
In a brief training session, the animal learns to press the lever to receive a food reinforcer. Once the behavior is established, the requirements for reinforcement can be altered according to a predetermined schedule (schedule of reinforcement). The rate and pattern of the animal’s behavior is controlled by the schedule, and it allows us to examine the effect of a drug on the pattern of behavior. For instance, on a fixed-ratio (FR) schedule, reinforcement is delivered after a fixed number of responses. Thus, an FR-3 schedule means that the animal must press the lever 3 times to receive 1 food pellet. Changing the fixed ratio from 3 to 20 or 45 will tell us how hard the animal is willing to work for the reinforcement. Interval schedules also are commonly used and are characterized by the availability of reinforcement after a certain amount of time has elapsed (rather than the number of bar presses). Thus, on an FI-2 schedule (fixed interval of 2 minutes), reinforcement follows the first response an animal makes after 2 minutes have passed since the last reinforcement. Responses made during the 2-minute interval are “wasted,” that is, they elicit no reinforcement. This schedule produces a pattern of responding that includes a pause after each reinforcement and a gradual increase in the rate of responding as the interval ends. For a description of other variations in schedules and their use in drug testing see Carlton (1983).
Measuring anxiety One classic method used to evaluate anxiety in animals is the conflict test, originally designed by Geller and Seifter. The animals are first trained to press a lever in the operant chamber for a standard (food or water) reinforcer. Once the behavior is established, the test sessions involve two stages. In the first, the animals press the lever for the reinforcer. After 10 or 15 minutes, a tone signals a change in the procedure: at this point lever pressing produces a reinforcer (approach) that is accompanied by a foot shock (avoidance), producing an approach-avoidance “conflict” for the subject. This situation is considered analogous to human anxiety experienced in approach-avoidance situations. As you would expect, lever pressing is steady during the reinforced situation but is reduced and variable during the conflict procedure. Antianxiety drugs have no effect on the reinforced schedule but increase the lever pressing during the conflict procedure, indicating that punishing situations are less inhibiting than normal. Naturally, one must be sure that the drugs being tested are not analgesics, which might also be expected to increase responding during the conflict session.
Methods of assessing drug reward and reinforcement The simple FR schedule has been used very effectively in identifying drugs that have abuse potential—that is, drugs that are capable of inducing dependence. We assume that if an animal will press a lever in order to receive an injection of drug into the blood or into the brain, the drug must have reinforcing properties. The drug self-administration method used with rodents is a very accurate indicator of abuse potential in humans. For instance, animals will readily self-administer morphine, cocaine, and amphetamine, drugs that we know are readily abused by humans. In contrast, drugs like aspirin, antidepressants, and antipsychotic drugs are neither self-administered by animals nor abused by humans. Table 1 lists some of the drugs that are reinforcing in rhesus monkeys. Compare this list with what you know about substances abused by humans.
Furthermore, we can ask the animal which of several drugs it prefers by placing two levers in the operant chamber and training the animal to press lever A for one drug and lever B for the alternative. Given free access to the levers, the animal’s choice will be readily apparent. An additional question we can pose regards how much the animal really “likes” a particular drug. By varying the schedule of reinforcement from FR-10 to FR-40 or -65, we can tell how reinforcing the drug is by how hard the animal works for the injection. The point at which the effort required exceeds the reinforcing value is called the breaking point. The higher the breaking point, the greater the reinforcement of the drug and presumably the greater the abuse potential in humans. Drugs like cocaine sustain incredibly high rates of responding: animals will lever- press for drug reinforcement until exhaustion.
|TABLE 1 Drugs That Act as Reinforcers in the Rhesus Monkey
A modification of the method allows the animal to self- administer a weak electric current to discrete brain areas via an indwelling electrode (electrical self-stimulation). The underlying assumption is that certain brain areas constitute “reward” pathways. It is assumed that when the animal works to stimulate a particular cluster of neurons, the electrical activation causes the release of neurotransmitters from the nerve terminals in the region, which in turn mediate a rewarding effect. The fact that pretreatment with certain drugs, such as morphine or heroin, increases the responding for even low levels of electrical stimulation indicates that the drugs enhance the brain reward mechanism (Esposito et al. 1989). In combination with mapping techniques, this method provides an excellent understanding of the neural mechanisms of reward and the effects of psychoactive drugs on those pathways.
Drugs as discriminative stimuli A discriminative stimulus is any stimulus that signals reinforcement for a subject in an operant task. For example, “light on” in the chamber may signal that reinforcement is available following lever pressing, while “light out” signals that no reinforcement is available regardless of the animal’s response. An animal that learns to press a lever in the presence of “light on” but not during the “light-out” period can discriminate between the two conditions. Although discriminative stimuli are usually changes in the physical environment, internal cues can also be discriminated. Thus an animal can learn to press the lever for reinforcement when it experiences the internal cues associated with a particular drug state (like the “light on”) and to withhold responding in a nondrugged or different drug state (like the “light off”). The animal’s response depends on its discrim-inating among internal cues produced by the drug. For example, if an animal has been trained to lever-press after receiving morphine, other opiates can be substituted for the internal cue and signal to the animal that reinforcement is present. Heroin or methadone are experienced like morphine. In contrast, drugs like amphetamine or marijuana, which apparently produce subjective effects very different from those of morphine, are treated by the animals as a nonreinforced cue. In this way novel drugs can be characterized according to how similar their internal cues are to those of the known drug. The same technique can be used to identify the neurochemical basis for a given drug cue. The drug cue can be challenged with increasing doses of a suspected antagonist until the cue has lost its effect. Likewise, neurotransmitter agonists can be substituted to find which more closely resembles the trained drug cue. Goudie and Leathley (1993) provide an excellent description of the basic methodology of drug discrimination as well as an assessment of potential pitfalls.
Negative reinforcement A variation on the FR schedule utilizes negative reinforcement, which increases the probability of a response that terminates an averse condition. This technique can be easily applied to operant analgesia testing. First, the animal is trained to turn off an unpleasant foot shock by pressing the lever. In the test phase, the researcher administers increasing amounts of foot shock up to the point at which the animal responds by pressing the lever. The lowest shock intensity at which the animal first presses is considered the aversive threshold. Analgesic drugs would be expected to raise the threshold of electric shock. The method is very sensitive even to mild analgesics such as aspirin. However, an independent measure of sedation is necessary to distinguish between failure to respond due to analgesia and failure to respond due to behavioral depression.
Although clinical depression is typically a human condition, an animal model utilizing negative reinforcement called learned helplessness provides some fascinating insights. In this design, the subjects in each of two groups are exposed to aversive events (e.g., repetitive foot shocks). The difference between the two groups is that the control group has the opportunity to make a response (e.g., press a lever) that turns the shock off for both groups, while the experimental group has no control over the shock. Hence, although the experimental group cannot modify the shock, it receives the same amount of shock as the control group. The question to be asked is how the experimental group will behave in a new situation that provides them with the opportunity to help themselves. When the animals are placed in a situation in which they can run from an electrified shock chamber to a non-electrified chamber, the control group learns to escape very quickly, while the experimental group shows signs of anxiety but makes no appropriate response. Apparently, faced with their earlier experience in which their behavior had no effect on their environment, they have learned to be helpless and to make no attempt to cope. Human depression often follows a personal catastrophe over which the individual has had no control, such as death of a loved one, physical disability, disease, or rejection, and these individuals express feelings of hopelessness and the belief that nothing they do has an effect. This sense that they are passive victims of circumstance provides the theoretical framework for learned helplessness as a model for depression. The effectiveness of traditional antidepressant drugs in reversing the helpless behavior in the animals further validates the model.
Techniques in behavioral pharmacology provide a means to quantify animal behavior for drug testing, developing models for psychiatric disorders, and evaluating the neurochemical basis of behavior. The advantages of animal testing include having a subject population with similar genetic background and history, maintaining highly controlled living environments, and being able to use invasive neurobio- logical techniques.
Animal testing includes a wide range of measures varying not only in validity and reliability but also in complexity, time needed for completion, and cost. Some measures use simple quantitative observation of behaviors such as motor activity and response to noxious stimuli. Other methods assess more complex behaviors such as learning and memory using a variety of techniques such as the classic T-maze as well as mazes modified to target spatial learning: the radial arm maze and Morris water maze. Animal models of anxiety, depression, addiction, and response to pain provide the means to assess human conditions and examine the drugs that modify those responses. Operant conditioning has a special place in pharmacology and is the basis for many tests of addiction potential, anxiety, and analgesia. Each method has benefits and limitations and must be rigorously evaluated to provide data that produce nonbiased and valid conclusions.