In our studies, we view the context as encompassing all environmental stimuli, experimental history and social status. For this paper, we will limit the context to a brief description of models used to assess cocaine reinforcement and to non-human primate social behaviour. When describing the models of drug self-administration with respect to the schedule of reinforcement, an important distinction should be made between reinforcing ‘effects’ and reinforcing ‘strength’. A reinforcing effect simply means that responding leading to drug presentation occurs at higher rates than responding leading to vehicle presentation. For every drug that has reinforcing effects, the shape of the dose–response curve approximates an inverted U-shape. That is, there is an ascending limb characterized by dose-dependent increases in responding, a dose that results in peak rates of responding and a descending limb in which increases in dose result in lower rates of responding (see
Zernig et al. 2004). Because several factors influence the shape of the curve, it is impossible to compare dose–response curves from different drugs and make statements related to which drug is ‘more reinforcing’ (
Woolverton & Nader 1990). However, other schedules can be used to make assessments related to reinforcing strength; these will be described in more detail below. The main point to highlight is that different schedules of reinforcement have features that render them suitable for answering different questions about the behavioural effects of cocaine. For example, questions related to the relative importance of drug seeking (i.e. simply self-administering cocaine) versus total cocaine intake in producing changes in the brain can be assessed by studying different schedules of cocaine self-administration. Such a distinction is of clear relevance when considering treatment options for drug abuse—does it matter how much drug a patient has taken or how long (s)he has been abusing the drug?
Volkow et al. (1999) found that levels of DA-D
2 receptor availability as measured with PET were more dependent on the duration of cocaine use than on the amount of drug used prior to the study. This finding suggested that the behaviours leading to drug procurement, independent of the pharmacology of cocaine, could contribute to the reported changes in DA receptor availability in cocaine abusers and supported the hypothesis that the environment can have profound effects on the brain.
The use of non-human primates, PET imaging and different schedules of reinforcement provided an opportunity to directly assess the importance of drug seeking versus total cocaine intake (
Czoty et al. 2007a,
b). To directly test this hypothesis, 12 experimentally naive rhesus monkeys received baseline PET scans using the D
2 receptor ligand [
18F]FCP. Six of these monkeys were then trained to self-administer cocaine under a second-order schedule, a very lean schedule of reinforcement in which drug seeking was maintained by the presentation of conditioned stimuli throughout the 60 min session until cocaine was finally administered (
Katz 1980). Under the final schedule parameters, the first response after 3
min (fixed interval; FI 3
min) produced a stimulus change (S) associated with cocaine reinforcement and the tenth completed FI (i.e. fixed ratio 10) resulted in cocaine presentation (designated FR 10 [FI 3
min:S]). Sessions ended after two cocaine injections (0.1
mg
kg
−1injection
−1). Thus, these animals had an extensive drug-seeking history, but very low levels of cocaine intake. The second group of six monkeys was trained to respond under an FR 30 schedule of cocaine presentation. Conditions for this group were arranged to model ‘binge’ access—monkeys could receive up to 30 injections of 0.3
mg
kg
−1 cocaine twice per day, 2 days per week. Thus, relative to the other group of monkeys, this set of subjects received much more cocaine but drug seeking was only 2 days per week. We found that binge access to cocaine resulted in significant reductions in D
2 receptor availability at every time point, while ‘drug seeking’ under the second-order schedule did not significantly affect D
2 receptor availability over 1 year. These findings suggest that the reductions in D
2 availability seen in humans were primarily due to the direct effects of cocaine on DA receptor levels.
(i) Organism×environment interactions: part 1
Acquisition of drug reinforcement is influenced by characteristics of the individual (i.e. trait variables) as well as by features of the environment (e.g. state variables). One of the first studies of the relationship of trait variables to sensitivity to drug reinforcement was provided by
Piazza et al. (1989) in which two groups of rats were differentiated based on locomotor activity in an open-field apparatus as high responders (HR) or low responders (LR). Rats were implanted with indwelling intravenous catheters and given access to low doses of
d-amphetamine under an FR schedule. HR rats acquired
d-amphetamine self-administration at lower doses than LR rats. The use of this simple schedule allowed for characterization of vulnerability based on an inherent behavioural characteristic, namely, locomotor activity in an open field.
More recently, laboratory animal studies have examined behaviours related to ‘impulsivity’, a trait shown to be high in cocaine abusers (
Moeller et al. 2002). Rats characterized as more impulsive acquired cocaine self-administration more rapidly than less impulsive rats (
Dalley et al. 2007).
Perry et al. (2005) addressed whether impulsivity precedes drug abuse. In that study, rats were trained on a delay discounting procedure in which responding on one lever under an FR 1 contingency resulted in the immediate delivery of one food pellet, while responding on another lever under an FR 1 contingency resulted in the delivery of three food pellets after a variable delay. If the rat chose the immediate option, the delay value decreased on the next trial for the alternative; if the delay option was chosen, the delay value increased on the next trial. A mean adjusted delay (MAD) value was calculated for each rat by averaging all delay values across trials. As described by
Perry et al. (2005), the MAD served as a quantitative measure of the extent to which each rat discounted delayed food reinforcers. Higher MAD values, representing longer delays, were indicative of low impulsivity, while smaller MAD values indicated more impulsive behaviour. The rats were divided into two groups, high and low impulsiveness (HiI and LoI, respectively) based on MAD values. When cocaine acquisition was studied, HiI animals acquired self-administration more rapidly and at higher levels than LoI rats. Taken together, these findings support the hypothesis that there are behavioural traits that predispose individuals to drug abuse and these can be examined using animal models.
Our group has studied trait variables and gene–environment interactions in relation to drug abuse in non-human primates for over a decade. Much of our research has been conducted in cocaine-naive monkeys prior to being exposed to cocaine in order to address gaps in the clinical data—questions that cannot be answered in humans due to ethical concerns. For example, as described above, cocaine abusers have lower levels of D
2 receptor availability than control subjects (Volkow
et al.
1990,
1993;
Martinez et al. 2004) and non-drug abusers with lower basal levels of D
2 receptor availability found methylphenidate more reinforcing (
Volkow et al. 1999). It is not known whether low D
2 levels were the result of cocaine use or a pre-existing feature that conferred vulnerability to the reinforcing effect of cocaine. The question is whether D
2 receptor availability is a trait marker for vulnerability to cocaine abuse. We have directly addressed this question in two ways. First, we correlated basal D
2 receptor availability in cocaine-naive monkeys with subsequent rates of cocaine self-administration. Second, we studied the changes in D
2 receptor availability in cocaine-naive monkeys over 1 year of access to determine whether cocaine reinforcement decreased these levels (
Nader et al. 2006). A summary of the findings is shown in . Initially, cocaine-naive monkeys were scanned with the D
2 receptor ligand [
18F]FCP and then trained to respond under an FI 3
min schedule of food presentation. When responding was stable, each monkey was surgically implanted with an indwelling venous catheter, a dose of cocaine (0.2
mg
kg
−1injection
−1) was substituted for food and response rates were recorded. An important point is that there was no training under the cocaine self-administration paradigm—the monkeys were simply exposed to the drug and response rates were recorded. We found an inverse relationship between baseline D
2 receptor availability and the rates of cocaine self-administration (
a). Monkeys with low D
2 receptor levels self-administered cocaine at higher rates compared with monkeys with high D
2 receptor availability. These findings are very similar to the observations by
Volkow et al. (1999) using non-drug abusers and methylphenidate. We also found that, over a 1-year period in which cocaine intake increased steadily, D
2 receptor availability decreased irrespective of what the initial levels of D
2 receptor availability were for each monkey (
b). Thus, it appears that low D
2 receptor availability makes an individual more vulnerable to cocaine reinforcement and continued exposure to cocaine further decreases those levels (Nader
et al.
2002,
2006).
The above findings clearly support the idea that there are biological trait variables, in this case D
2 receptor availability, that influence vulnerability to cocaine abuse. We have also examined the impact of environmental variables on D
2 receptor availability and whether these effects influenced vulnerability to cocaine reinforcement. Earlier work from our group demonstrated a relationship between D
2 receptor availability and social rank in female monkeys, such that subordinate monkeys had lower D
2 receptor levels than dominant monkeys (
Grant et al. 1998). We next assessed whether D
2 receptor availability was a trait variable that predicted social rank. For these studies, we used 20 experimentally naive and individually housed male cynomolgus monkeys. After baseline PET scans using [
18F]FCP were conducted, monkeys were placed in social groups of four monkeys per pen and after three months were rescanned with [
18F]FCP (
Morgan et al. 2002). D
2 receptor availability was not a trait marker for eventual social rank. After three months of social housing, we observed the same effect that was reported by
Grant et al. (1998) in female monkeys that had been living together for over 3 years—subordinate monkeys had lower D
2 receptor availability compared with dominant monkeys. However, it came about in a manner opposite to what we had expected. We had hypothesized that the lower D
2 receptor levels in subordinate monkeys compared with dominant monkeys arose as the result of chronic social stress that is unequivocally experienced by subordinate monkeys (
Kaplan et al. 1982;
Shively & Kaplan 1984). However, the over 20% difference between dominant and subordinate monkeys in our study was due to a significant
increase in D
2 receptor availability in dominant monkeys whereas subordinates, on average, did not change. These increases in D
2 measures were in the same direction as reported in rodent studies demonstrating the influence of environmental enrichment on DA function—including increased D
2 receptor densities (e.g.
Bowling et al. 1993;
Rilke et al. 1995;
Hall et al. 1998). Based on these rodent studies and on our findings that there was an inverse relationship between D
2 receptor availability and cocaine self-administration, we hypothesized that the subordinate monkeys would self-administer more cocaine than the dominant monkeys. Our hypothesis was borne out (
Morgan et al. 2002). In fact, cocaine was not a reinforcer in the dominant monkeys when assessed under an FR 50 schedule of reinforcement (see
Nader & Czoty 2005 for additional discussion).
We also examined other behaviours that we hypothesized could be trait variables predictive of social rank. In our initial study (
Morgan et al. 2000), locomotor activity predicted eventual social rank in that eventual subordinate monkeys had higher locomotor scores compared with eventual dominant monkeys; interestingly, this was not extended to female monkeys (
Riddick et al. submitted). Most recently, we have extended our measures to include behaviours deemed to assess impulsivity in an effort to extend more recent work in rodents (e.g.
Perry et al. 2005;
Dalley et al. 2007). In a group of experimentally naive and individually housed female cynomolgus monkeys, we used a measure of novel object reactivity to assess impulsivity in each animal prior to being socially housed (
Riddick et al. submitted). Monkeys that would eventually become subordinate had shorter latencies to approach the novel object compared with eventual dominant female monkeys. Shorter latency is hypothesized to represent greater impulsivity. Whether the more impulsive monkeys are also more vulnerable to self-administer cocaine as was reported in rodents by
Perry et al. (2005) and
Dalley et al. (2007) is currently being evaluated.
(ii) Organism×environment interactions: part 2
In our socially housed male monkeys, we have extended earlier work in an effort to further enhance our homologous model of the human condition. These experiments primarily focus on changing environmental conditions. For example, we found that the protective effect associated with being the dominant monkey can be attenuated by continual exposure to cocaine (
Czoty et al. 2004). That is, while there were differences in rates of self-administration when initially exposed to the FR 50 schedule (
Morgan et al. 2002), repeated exposure to cocaine over a 1-year period resulted in cocaine becoming a reinforcer in dominant monkeys (see e.g.
a). After several months to years of cocaine self-administration, neither response rates nor D
2 receptor availability were different in dominant compared with subordinate monkeys (
Czoty et al. 2004). As mentioned above, simple schedules do not provide information related to reinforcing strength. Thus, we examined whether there would be differences between social ranks under conditions in which cocaine was available in the context of an alternative, non-drug reinforcer (
Czoty et al. 2005). We found that subordinate monkeys were significantly more sensitive to the reinforcing effects of cocaine using this procedure, such that they would choose a lower dose of cocaine over food compared with dominant monkeys (
b). These findings highlight several important facets of organismal and environmental interactions. These data support the observations that measures of reinforcing strength provide different information related to cocaine self-administration than measures of reinforcing effects. In addition, these findings indicate that after years of living in these stable groups the influence of the social context was still apparent.
A question that is frequently asked is ‘what if circumstances change and a dominant monkey becomes subordinate and a subordinate monkey becomes dominant?’ To address this question, we rearranged groups such that one pen consisted of four previously dominant (first-ranked) monkeys and another pen was made up of four previously subordinate (fourth-ranked) monkeys. Additional pens were composed of intermediate (second- and third-ranked) monkeys and experimentally naive monkeys (
Czoty et al. in preparation). After three months of social housing under these new conditions, PET studies were conducted and cocaine self-administration was examined under the concurrent schedule of reinforcement with food as the alternative. The relationship between new social rank and D
2 receptor availability was not evident—that is, the newly dominant monkeys did not have significantly higher levels of D
2 receptor availability compared with newly subordinate monkeys. (Note, some of the dominant monkeys were previously subordinate and some of the subordinate monkeys were once dominant.) Also, there were no differences in cocaine choice between the monkeys. Additional studies using other measures, including novel object reactivity, noted that previous rank was more predictive of outcome than current rank. There is a long and extensive literature on behavioural and pharmacological history influencing behaviour and drug effects (e.g.
Barrett et al. 1989) and these studies extend those findings to include a history of social interactions.
Another example of organism×environment interaction involves the use of socially housed monkeys to examine drug-induced changes in social behaviour and the consequence of those effects on subsequent cocaine self-administration. There is an extensive literature on the interaction of social rank with drug effects in non-human primates (e.g.
Smith & Byrd 1985;
Martin et al. 1990; reviewed by
Miczek et al. 2004). For example, Miczek and colleagues (e.g.
Miczek & Yoshimura 1982;
Miczek & Gold 1983a) have shown that the effects of alcohol, amphetamine or cocaine can be influenced by social rank and environmental context. In one study (
Winslow & Miczek 1985), low to intermediate doses of alcohol produced increases in aggression by dominant monkeys, but no effect on aggression by subordinate animals. However, co-administration of alcohol and testosterone to subordinate monkeys resulted in increases in aggression. Crowley
et al. (
1974,
1992) examined the effects of a number of abused drugs on the social behaviour of macaques. Methamphetamine produced pronounced increases in locomotion and stereotypes, and declines in food-foraging behaviour and aggression. In a low-ranking monkey, high doses of methamphetamine produced such profound increases in submissive behaviours that the amount of aggression directed from the (untreated) dominant monkeys towards the drug-treated animal increased. Of all the studies examining the effects of drugs on social behaviour, this result is one of the few descriptions of the behaviour of the untreated monkeys. In our socially housed monkeys, we tested the hypothesis that if reinforcing doses of cocaine resulted in increased aggression and changes in social rank, then the frequency of cocaine self-administration in that monkey would increase in subsequent experimental sessions.
Monkeys lived in stable social groups of three and social rank was determined in each pen as described above. For these studies, only one monkey in the social group was given access to cocaine (saline, 0.01–0.1
mg
kg
−1injection
−1) under an FR 50 schedule of reinforcement, while the remaining monkeys in the pen had access to food presentation under an FR 50 schedule; conditions remained in effect for five consecutive sessions. When the session was completed, monkeys were returned to their social groups and agonistic and submissive behaviours were recorded over a 15
min period. All monkeys (dominant, intermediate and subordinate) were studied at all cocaine doses. Social interactions did not affect response rates or cocaine intake for any monkey. However, cocaine-induced changes in social behaviour were dependent on the rank of the monkey. Irrespective of which animal in the pen self-administered cocaine, the first- and second-ranked monkeys showed increases in aggression; the subordinate monkey never demonstrated any aggression during the course of the study. These data indicate that social rank is the most important determinant of cocaine-induced changes in social behaviour. One possible reason that self-administration was insensitive to the consequences of social behaviour is that cocaine access was not scheduled until approximately 24 hours after the social interaction. Current studies are examining the consequences of cocaine-induced changes in social behaviour on cocaine self-administration that are more closely associated in time.
This article has been 
