3.3.2.1. Percent Drug Lever Responding

An animal’s degree of progress in learning the DD is determined by an evaluation of its distribution of presses on the two levers. In particular, an animal’s learning of the discrimination can be evaluated either prior to, or up to, the delivery of the first reinforcement. Thus, when an FR 10 schedule of reinforcement is used, DD learning can be assessed for each subject by dividing the number of responses that occurred on the drug-designated lever by the total number of responses that occurred on both levers up to the delivery of the first reinforcement; percent of responses on the drug-appropriate lever is then obtained by multiplying the value by 100. For instance, assume that a rat has the right-side lever designated as the diazepam-appropriate lever. On a Monday, the animal is injected with 3 mg/kg of diazepam, placed in its assigned operant chamber, and proceeds to press the left-side lever 9 times and the right-side lever 10 times; food reward (in this example) could be presented after the 10th right-side lever press. For this day, discriminative control would be assessed at 53% diazepam-appropriate responding (i.e., × 100). On Tuesday, this same rat is injected with vehicle, placed into its designated chamber, and presses the right-side lever 4 times and the left-side lever 10 times; food is presented after the 10th left-side press. On this day, discriminative control would be assessed at 29% diazepam-appropriate responding (i.e., × 100). Alternatively, if the VI schedule of reinforcement is programmed, then discrimination performance is evaluated during a short period (e.g., 2.5 min) of non-reinforced responding (referred to as extinction) at the beginning of a session; extinction sessions usually occur once or twice per week. Each animal’s distribution of presses on the two levers is then evaluated in the same manner as it is under the FR schedule of reinforcement. As might be expected, the administration of drug or vehicle during initial training sessions under either FR or VI schedules of reinforcement usually results in the animals dividing their responses equally (e.g., 50% diazepam-appropriate responding after injection of drug or saline) between the two levers (Figure 3.1). However, as training sessions progress with drug and vehicle, the animals gradually learn to respond on the drug-designated lever (i.e., percent of responses on the drug-designated lever is high and percent of responses on the vehicle-designated lever is low) when given drug, and on the vehicle-designated lever (i.e., percent of responses on the drug-designated lever is low and percent of responses on the vehicle-designated lever is high) when given vehicle. In other words, the learning of a DD occurs gradually over time (Figure 3.1). A generally accepted guideline is that after 6 to 9 wk of training, animals (individually and, consequently, as a group mean) consistently make ≥ 80% of their responses on the drug-appropriate lever after administration of drug (e.g., 3 mg/kg of diazepam) and ≤ 20% of their responses on the same lever after administration of vehicle (Figure 3.1).

FIGURE 3.1

Learning curve results of rats trained to discriminate the stimulus effect of 3 mg/kg (IP) of diazepam (closed squares) from 1 mL/kg of vehicle (open squares). Ordinate: Mean (n = 12) percent (± SEM) of responses on the diazepam-appropriate lever (more...)

3.3.2.2. Response Rate

In addition to the animals’ distribution of responses on the two levers under FR or VI schedules of reinforcement, their response rate data (i.e., total number of responses on both levers expressed as responses per second or minute) also can be calculated. For example, animals’ (individual and/or group mean) response rate can be calculated under the FR schedule of reinforcement for a behavioral session (e.g., 15 min). Alternatively, under the VI schedule of reinforcement, the total number of responses made during the 2.5 min extinction session (or the entire session) can be recorded. The animals’ response rate can be viewed as another indicator of the effect(s) of a drug on behavior. In some cases, the animals’ response rate after the training dose of the training drug is suppressed when compared to that of the vehicle. In other instances, response rate data can assist the experimenter in the selection of (1) an appropriate training dose of the training drug, and/or (2) a range of doses to be examined for test drugs. Further, animals’ response rate can be an ancillary measure in cases where a test drug may, or may not, affect the main dependent variable (i.e., percent responding on the drug-designated lever). Finally, this parameter can be used in conjunction with an evaluation of the subjects’ general behavioral condition (e.g., sedated, incapacitated, overly excited).

3.4.2.1. Dose Response

A very important consideration in tests of stimulus generalization is the necessity of a thorough dose-response investigation. An extensive literature review of DD tests of stimulus generalization reveals that certain agents produce saline-like effects at particular doses (usually relatively low doses), and disruption of behavior at some higher doses. While an initial conclusion to the results of such a study may be that there is a lack of stimulus generalization, it has been found in a number of instances that a careful evaluation of additional doses (i.e., doses between the highest dose that resulted in vehicle-like responding and the lowest dose that produced disruption of behavior) ultimately resulted in stimulus generalization. This has even been observed with agents where the difference in vehicle-like and disruptive doses has been quite small. Thus, several instances have been reported where doses of a challenge drug, administered in a logarithmic progression (e.g., 0.1, 0.3, 1, 3, 10 mg/kg), resulted in saline-like responding at the lower doses and disruption of behavior at the highest doses. However, an examination of doses between, for example, 3 mg/kg and 10 mg/kg resulted in stimulus generalization. As such, these types of situations appear to emphasize (1) that stimulus generalization to a test agent may occur within a “narrow window of doses,” and (2) the sensitive and specific nature of the drug-induced stimulus.

3.4.2.2. Comparison of Results of Test Agents

A preferred tactic is to evaluate doses of a challenge drug in drug-trained subjects until either stimulus generalization or disruption of behavior (i.e., no responding) occurs. If, for example, the highest test dose (e.g., dose X) of a challenge drug elicits 50% drug-appropriate responding, and, for some reason, the evaluation of higher doses is precluded, it is not appropriate to conclude that the challenge drug is half as potent as the training drug when, in fact, there has been no demonstration that the two agents can produce a common effect. In this situation, comparisons can only be made in a qualitative sense. That is, an appropriate conclusion that could be stated is that the challenge agent is less effective than the training drug in producing a training-drug–like effect (or, correspondingly, that it is less effective than some other challenge drug which, at a dose below dose X, produced training-drug–like effects). Likewise, if two challenge drugs produce partial generalization (e.g., 40% and 60% training-drug–appropriate responding) at dose X, it should not be stated with certainty that the second challenge drug is more potent than the first because the possibility exists that one (or both) agent(s) may not exert a stimulus effect that is common (i.e., complete generalization) to that of the training drug.

3.4.3.1. Statistical Analysis

The use of statistical analysis (e.g., analysis of variance, Fischer’s exact probability, t-tests, etc.) with DD data can be very problematic. One major concern is the failure of statistical procedures to account for behavioral disruption (i.e., no responding) into the analysis. In particular, an animal that fails to press a lever after being administered a dose of test agent in a stimulus generalization test cannot be assigned a percent score; 0% drug-appropriate responding cannot be assigned because it has a different meaning (i.e., the animal pressed the saline-appropriate lever). Some investigators argue that statistical analysis is robust enough to account for such missing data. However, the percent drug-appropriate lever responding data are not missing because the animals failed to appear for a test appointment. The data are lacking and unavailable because the drug interfered with the ability of the animal to respond. A more palpable description of such data is that the effect should be characterized as “disruption.” Statistically, one could (and some investigators do) ignore the disruptive effect of a dose of drug, use only the data from very few animals (sometimes n = 1 or 2 out of 6 or 8 subjects that were tested) that respond (usually those few subjects have responded to a high degree on the drug-designated lever), and statistically conclude the occurrence of stimulus generalization. However, in such cases, it would seem more appropriate and meaningful to characterize the effect as disruption rather than to promote a statistical conclusion that may be misleading. In any case, the most prudent approach to the presentation of stimulus generalization data in DD studies is to account fully, by description of the effects on subjects and/or statistically for the behavioral effect of the test agent in all subjects, individually and/or as a group.

3.4.3.2. Examples of Complete, Partial, and No Substitution

In rats trained to discriminate the benzodiazepine diazepam at 3 mg/kg from vehicle at 1 mL/kg, the administration of lower doses of diazepam (i.e., construction of a dose-response function for diazepam) led to progressively less responding on the diazepam-appropriate lever (Figure 3.2); furthermore, an ED₅₀ value (ED₅₀ = 1.2 mg/kg) was calculated. Moreover, several metabolites of diazepam were evaluated in tests of stimulus generalization. Figure 3.2 shows that the diazepam stimulus generalized in a dose-related manner to oxazepam (ED₅₀ = 1.4 mg/kg), temazepam (ED₅₀ = 1.4 mg/kg), and desmethyldiazepam (ED₅₀ = 2.3 mg/kg). The latter results indicate that, in comparison with diazepam, the metabolites are relatively potent behaviorally, and indicate the distinct possibility that the metabolites may contribute to the stimulus effect of diazepam. Figure 3.2 also reveals that the diazepam stimulus generalized to the barbiturate anxiolytic/sedative pentobarbital (ED₅₀ = 4.5 mg/kg), which illustrates the idea that animals trained to discriminate a dose of a training drug can display stimulus generalization to an agent that exerts a similar behavioral effect, although not necessarily through an identical mechanism of action; diazepam and pentobarbital do not share the same mechanisms of action (see antagonism tests below). In comparison, the administration of buspirone (0.3–3.0 mg/kg), a serotonin 5-HT1A receptor (partial) agonist anxiolytic agent that is structurally unrelated to diazepam, produced only partial diazepam-appropriate lever responding (i.e., maximal 43% diazepam-appropriate lever responding), while the administration of doses between 4 mg/kg and 10 mg/kg produced disruption of behavior (Figure 3.2; disruption data not shown). Thus, the diazepam stimulus may partially generalize to buspirone because there may be some degree of pharmacological effects that is common to both diazepam and buspirone at low doses. Complete stimulus generalization does not occur, however, because the overlap of effects is incomplete. Lastly, the administration of S(+)-amphetamine (0.1–1.5 mg/kg), a CNS stimulant, to the diazepam-trained animals produced vehicle-appropriate responding (i.e., maximal 18% diazepam-appropriate lever responding; data not shown in Figure 3.2), while the administration of doses of 2–3 mg/kg produced disruption of behavior (i.e., no responding; data not depicted in Figure 3.2). Since percent diazepam-appropriate lever responding is fairly low (i.e., S(+)-amphetamine–induced responding on the vehicle-designated lever), it can be stated that the stimulus effect produced by 3.0 mg/kg of diazepam is quite different from that produced by S(+)-amphetamine. However, the fact that S(+)-amphetamine, and for that matter buspirone, can serve as training drugs indicates that these are not inert substances. Thus, animals trained in a DD task respond on the drug-designated lever only when administered a test agent that produces some degree of effect that is similar to the training dose of the training drug. If the test agent produces an effect that is “inert” or unlike that of the training drug, then responding will occur on the vehicle-designated lever until doses of the test agent are administered that disrupt lever response behavior by the animal (i.e., little or no responding). In a final comment, it is noted that results from stimulus generalization studies of test drugs under particular training agents have been consistent across different species trained under different schedules of reinforcement.

FIGURE 3.2

Results of stimulus generalization tests with diazepam (closed squares), desmethyldiazepam (closed triangle), temazepam (closed inverted triangle), oxazepam (closed diamond), pentobarbital (closed circle), and buspirone (open square) in rats trained to (more...)

3.4.3.3. Time Course

Once a dose of a drug has been established as a discriminative stimulus, tests can be performed to determine its time course of actions. Such tests investigate the effects of changing the PSII of the training dose of drug and the beginning of a test session. In particular, the time course of any drug stimulus can be characterized by its latency of onset of action, peak activity, and duration of effect. The latency of onset of action refers to the time (or interval) between the administration of the training drug and the first indications of a marked effects on drug-appropriate responding. The peak activity of the drug refers to the time (or interval) that the drug exerts maximal percent drug-appropriate responding (i.e., ~80%–100% drug-appropriate responding). Lastly, the duration of action refers to the interval of time between onset of action and the point of time (or interval) that the drug no longer exerts percent drug-appropriate responding that is notable (i.e., ~20% drug-appropriate responding). In the present example, 3.0 mg/kg of diazepam was established as a discriminative stimulus with a 15 min PSII; PSII intervals of 5, 10, 30, 45, 90, 120, 180, and 240 min also were examined (Figure 3.3). The results indicated that the onset of effect of the diazepam stimulus occurred between the PSIIs of 5 min and 15 min; peak activity was exhibited from PSIIs of 10 min to approximately 90 min; and duration of action occurred between PSIIs of 10 min and ~180 min. In addition, Figure 3.3 illustrates the time course effects of the major metabolites of diazepam. These studies were conducted with the dose of each metabolite that produced stimulus generalization in the 3 mg/kg diazepam-trained animals: desmethyldiazepam (6 mg/kg), oxazepam (3 mg/kg), and temazepam (3 mg/kg). An important difference among benzodiazepines is their pharmacokinetic properties. For example, in humans, diazepam is considered to have a relatively rapid onset of action and a relatively long half-life. In comparison, oxazepam and temazepam are considered to have slower onsets of action and much shorter half-lives, relative to diazepam. The time course studies of the stimulus effects of these agents in the diazepam-trained animals are not inconsistent with the human data. In any case, it is clear that familiarity with the time-course of action of the training drug or challenge compounds in tests of stimulus generalization and/or stimulus antagonism is of great importance; drug responses should not be measured too long, or short, after drug administration.

FIGURE 3.3

Results of time course studies (i.e., stimulus generalization tests with various pre-session injection intervals) with diazepam (3 mg/kg; closed squares), desmethyldiazepam (6 mg/kg; open squares), temazepam (3 mg/kg; open triangle), and oxazepam (3 mg/kg; (more...)

3.4.3.4. Stimulus Antagonism

An effective strategy to determine the mechanisms of action of psychoactive agents is to study drugs that block their effects. In DD studies, the rationale of such an approach is that the training dose of the training agent will only be blocked by receptor antagonists that interfere with the mechanism of action of the drug. The results of antagonism tests, as with generalization tests, typically fall into one of three categories: (1) complete antagonism (i.e., saline-appropriate responding); (2) partial antagonism (i.e., ~ 40%–70% drug-appropriate responding); and (3) no antagonism (i.e., ≥ 80% drug-appropriate responding). Three strategies to study stimulus antagonism can be employed. One approach can determine whether the stimulus effect of the training dose of the training drug can be attenuated when various doses of an appropriate receptor antagonist are combined with the training dose of the training drug. Thus, the rats trained to 3.0 mg/kg of diazepam were administered various doses of the benzodiazepine receptor antagonist flumazenil prior to the administration of their training dose of training drug. If a drug is an effective antagonist, then a dose-related antagonism of the animals’ percent drug-appropriate responding should occur. Figure 3.4 shows that the administration of various doses of flumazenil (3–12 mg/kg) prior to the injection of the 3 mg/kg training dose of diazepam was sufficient to produce antagonism (i.e., responding ultimately occurred on the vehicle-designated lever). In contrast, the administration of various doses of flumazenil prior to the injection of the dose of pentobarbital (10 mg/kg) that produced complete stimulus generalization in these animals (see above) failed to produce antagonism (i.e., responding occurred on the diazepam-designated lever). Lastly, when the animals were administered doses of flumazenil (3–40 mg/kg) in control tests, they failed to respond on the diazepam-designated lever. Taken together, these data support the idea that diazepam and pentobarbital can induce a similar stimulus effect but the mechanism of action can be differentiated by flumazenil, a benzodiazepine receptor antagonist.

FIGURE 3.4

The effect of flumazenil administered alone (open squares), in combination with 3 mg/kg of diazepam (DZP; open triangles), or in combination with 10 mg/kg of pentobarbital (PB; open circles). The administration of various doses of flumazenil prior to (more...)

In a second approach, the dose response of the training drug is determined in both the presence and absence of a constant dose of the antagonist. If the antagonism is competitive, then the dose response of the training drug (in the presence of the constant dose of the antagonist) should shift in a rightward and parallel manner. Figure 3.5 (top figure) shows the dose-response effect of diazepam in the absence (i.e., left dose-response function) and the presence (right dose-response effect) of flumazenil (5 mg/kg). As can be seen, pretreatment of the animals with flumazenil induced a rightward shift of the dose-response function of diazepam. In a third approach, various doses of the training drug can be combined with various doses of the receptor antagonist. This approach will generate a series of training-drug/antagonist dose-response curves and probably provide the most comprehensive or detailed picture of the interaction between the agents. Figure 3.5 (bottom figure) shows the dose-response effect of diazepam in the absence (i.e., left dose-response function) and the presence (middle and far right dose-response effects) of flumazenil (5 mg/kg and 12 mg/kg, respectively). Clearly, the dose-response functions of the discriminative stimulus effect of diazepam were shifted rightward and these data strongly indicate the presence of competitive antagonism.

FIGURE 3.5

The effect of various doses of diazepam alone (DZP; closed squares) or in combination with 5 mg/kg of flumazenil (open squares) in rats trained to discriminate 3 mg/kg of diazepam from vehicle (top figure). The bottom figure depicts the effects of various (more...)