Abstract
The effects of methamphetamine, phencyclidine and Δ9-tetrahydrocannabinol on responding under differential reinforcement of low rate schedules (DRL schedules) were studied under three different DRL time requirements. Under the DRL schedules studied, rats were required to space responses at least a minimum, but not more than a maximum, time interval apart. The time intervals between responses (interresponse times, or IRTs), when plotted as a frequency distribution, were usually a normal distribution with the peak at or near the minimum IRT required for delivery of the reinforcer. Methamphetamine flattened the IRT distribution and increased the frequency of long pauses under the DRL 1–1.3 sec schedule, but shifted the IRT distribution toward shorter IRTs under the DRL 4–5.2 and 10–13 sec schedules. Under the DRL 1–1.3 sec schedule, phencyclidine also increased long pauses. Under the DRL 4–5.2 sec and 10–13 sec schedules, phencyclidine produced dual effects on the IRT relative frequency distributions producing increases in the proportion of short IRTs similar to methamphetamine at low doses, but higher doses increased long pauses as well. Δ9-Tetrahydrocannabinol had little effect on responding under the DRL 1–1.3 sec and DRL 4–5.2 sec schedules, but it greatly increased the relative frequency of short IRTs under the DRL 10–13 sec schedule. Thus the effects of drugs on responding under these DRL schedules depended on the drug, the dose and the time requirements of the schedule, which suggests that a simple description of the effects of drugs on timing behavior or time perception is inadequate.
Temporally spaced responding schedules (DRL schedules) with a wide range of time-duration requirements have been used to assess timing behaviors (Sidman, 1955;Weiss and Laties, 1964; McMillan, 1969; Ando, 1975). Under a DRL schedule, responses after an interval of T seconds without a response are reinforced (Ferster and Skinner, 1957). Sidman (1955)suggested that not only is the DRL schedule an effective schedule for producing and measuring timing behavior in experimental animals, but also that this schedule is particularly effective for investigating the effects of drugs on temporally dependent responding in minimally restrained animals. He enumerated such advantages as the production of stable behavior over long periods of time, allowance of automated recording and production of orderly responses to drugs.
The DRL schedule that was used in the present investigation required that responses occur at least T seconds after the preceding response, or the initiation of a discriminative stimulus, but not more than a maximum time interval of T + T′ sec to obtain the reinforcer. This has been classified by some as “differential reinforcement of low rates of responding with pacing” (DRP) (Ferster and Skinner, 1957). It has been alternatively referred to as a “differential reinforcement of low rates of responding with a limited hold” (DRL LH) (Kelleher et al., 1959, 1961). For consistency in this study, and because the terminology is somewhat incongruous in the literature, the use of a DRL schedule with upper and lower time limits will simply be referred to as a DRL schedule.
One important measure of performance under a DRL schedule is the frequency distribution of IRT or the time intervals between responses. Under DRL schedules, the IRT frequency distribution is usually a normal distribution with the peak at or near the minimum IRT required for delivery of the reinforcer. Stimulants such as amphetamine, MAP, methylphenidate, nicotine, cocaine and caffeine generally cause animals to respond with shorter IRTs. This causes the relative frequency distributions of IRTs to shift leftward, and produces increases in the mean rate of responding (Stretch and Dalrymple, 1968; Pradhan and Dutta, 1970; Robbins and Iversen, 1973; Ando, 1975; Woolverton et al., 1978). However, this leftward shift in the IRT distribution is not a consistent effect observed with all central nervous system stimulants. There have been reports of increases in long IRTs caused by stimulants and even shifts of the relative frequency distribution rightward because of these longer IRTs (McMillan and Campbell, 1970;Balster and Baird, 1979; Levine et al., 1980).
Central nervous system depressants have also produced variable effects on IRT patterns. Animals injected with morphine produced both shorter and longer IRTs that flattened the IRT distributions in most studies (Ford and Balster, 1976; Adam-Carriere et al., 1978; Wenger and Wright, 1990). Pentobarbital both flattened IRT distributions (Balster and Baird, 1979) in a manner similar to morphine and shifted IRT distributions leftward with predominantly shorter IRTs (Stretchet al., 1967; Stretch and Dalrymple, 1968), similar to the effects of some of the stimulants. Alcohol, diazepam, chlordiazepoxide and meprobamate have produced differing effects on behavior under schedules with different time-duration requirements. In some cases, animals responded with IRTs that were longer than the required IRT for reinforcement (Sidman, 1955; McMillan, 1970; McMillan and Campbell, 1970; Ando, 1975), resulting in rightward shifts or flattening of the relative frequency distributions. In other instances, the IRTs were too short to produce the reinforcer resulting in leftward shifts (Kelleheret al., 1961; Sanger et al., 1974) in the relative frequency distributions. Δ9-THC increased the frequency of shorter IRTs causing moderate leftward shifts in IRT distributions (Conrad et al., 1972; Ferraro, 1972; Ferraro and Billings, 1972; Elsmore and Manning, 1974). However, there are also reports of Δ9-THC flattening IRT distributions, which is an effect similar to that seen with morphine (Manning, 1973).
Studies of responding under DRL schedules with drugs with mixed stimulant-depressant properties, such as PCP, and various noncompetitive NMDA antagonists, such as NANM, memantine and dizocilpine (MK-801), are few in number, and the results have been variable. In some cases, these drugs have produced only leftward shifts in the relative frequency distributions (Sanger and Jackson, 1989). In others, they have produced bimodal IRT distribution patterns with increases in both extremely short IRTs and extremely long IRTs (Balster and Baird, 1979; Freeman et al., 1984).
Antidepressants usually produce flattened IRT distributions under DRL schedules and show a much greater increase in the frequency of long IRTs than in shorter ones (O’Donnell and Seiden, 1983; Seiden et al., 1985). Such increases in the frequency of predominantly longer IRTs are also seen after many hallucinogens such as LSD, mescaline, DOM, the 5-hydroxytryptamine antagonist, ketanserin and methysergide (Appel, 1971; Marek and Seiden, 1988).
These conflicting results raise questions about the effects of drugs on timing behavior under DRL schedules. The species used, the time requirements of the schedule, the drug vehicle, routes of administration, the failure to show relative-frequency distributions and procedural differences make it difficult to systematically compare drug effects on the behavior under DRL schedules.
Therefore, a comparison of the behavior of rats under three DRL schedules with different time-duration requirements was conducted. MAP, PCP and Δ9-THC were selected as representative abused drugs from different pharmacological classes for comparison of drug effects under these schedules. Altered time perception has been reported with PCP and Δ9-THC, based on human anecdotal reports (Hollister and Gillespie, 1970; Tinklenberg et al., 1976; Yesavage et al., 1978), and with MAP using animal temporal discrimination studies (Stubbs and Thomas, 1974; Maricqet al., 1981). The DRL schedules studied were DRL 1–1.3 sec, DRL 4–5.2 sec and DRL 10–13 sec. The range of time between the upper and lower limits under all schedule parameters was proportional. These time parameters were chosen so that we could compare drug effects on responding under DRL schedules with drug effects on responding under TRD schedules (McClure et al., 1997). The TRD schedules in that study and the DRL schedules used in this study shared identical temporal requirements which allowed a direct comparison of the timing behaviors generated under these two types of schedules. Under a DRL schedule, the animal is reinforced for withholding responding for a specified duration. However, under a TRD schedule the animal must emit a continuous response of specified duration. By using the same timing requirements under DRL schedules as those of another study which used the TRD schedules (McClure et al., 1997), a consistent, methodical comparison of drug effects on behavior of rats under DRL and TRD schedules at several parameter values was achieved.
Methods
Subjects.
Twelve male, Sprague-Dawley rats weighing 300 to 372 g at 80% of their free-feeding weights were used. Animals were individually housed in standard Plexiglas rat cages in a colony room maintained at 70–74°F with a 12-hr light/dark cycle (lights on at 6:00 a.m. to 6:00 p.m.). The rats were approximately 10 months of age at the beginning, and 16 months of age at the end of these experiments. All other conditions were the same as those reported in McClure et al. (1997).
Apparatus.
Rats were trained and tested in four standard two-lever chambers (model G6312, Gerbrands Corp., Arlington, MA) enclosed in sound-attenuating Gerbrands enclosures (model G7210). A feeder mounted between the levers delivered 97-mg food pellets (Noyes Corp., Lancaster, NH). A houselight and a stimulus light consisting of 28-V DC bulbs were mounted in the ceiling of the experimental chamber and above the right lever. A continuous frequency tone (Sonalert, model SC628H in series with a 15 K-Ohm resistor) provided sound stimuli to the chamber. Events in the chambers were controlled and monitored by use of a Firestar 386 computer with Med Associates Interface (St. Albans, VT) housed in a separate, but adjoining room to minimize external noise which might disrupt animal activity.
Drugs and testing.
Doses of MAP (Sigma Chemical Co., St. Louis, MO), PCP and Δ9-THC (National Institutes on Drug Abuse, Bethesda, MD) were given in ascending or descending dose order with random assignment of the rats to a dose order. Both PCP and MAP (0.3, 1.0, 1.7, 3.0 or 5.6 mg/kg) were dissolved in saline and administered intraperitoneally 10 min before sessions in a volume of 1 ml/kg. A Δ9-THC solution in ethanol was evaporated to dryness under nitrogen and redissolved in DMSO (Sigma Chemical Co., St. Louis, MO ). Δ9-THC (0.3, 1.0, 1.7, 3.0, 4.2 or 5.6 mg/kg) was administered intraperitoneally 1 hr before behavioral-test sessions in a volume of 0.5 ml/kg. All dose levels are reported as the salts, except Δ9-THC, which is reported as the free base. Control vehicle injections consisted of saline during the MAP and PCP drug series, and DMSO during the Δ9-THC drug series.
Procedure.
After all rats learned to lever press, they were trained to respond under DRL schedules by differential reinforcement as described previously (Sidman, 1955; Kelleher et al., 1961). Rats were randomly assigned to one of the three DRL training schedules. Rats were trained under the DRL 1–1.3 sec schedule for a 6-month period, and for 3 to 3.5 months under the DRL 4–5.2 sec and 10–13 sec schedules. Under the DRL 1–1.3 sec schedule, the minimum IRT necessary for reinforcement was first lengthened to 1.0 sec in 0.3-sec increments. Once responding under the minimum IRT was established and stable, the maximum IRT that permitted the reinforcer to be obtained was decreased from 10 sec to 1.3 sec by gradually reducing the upper limit maximum by 0.5-sec intervals. Rats consistently performed below the 40% accuracy level when upper limits were reduced below 3 sec. Therefore, they received additional training with no lower time limit, but with a maximum upper limit which started at 5 sec and was decreased by 0.3-sec intervals. Once rats were pressing within the 1.3-sec upper limit, the lower limit was gradually added again.
Under the DRL 4–5.2 sec schedule, both progressive 0.5-sec increases followed by progressive 0.5-sec decrements (decreased from 10 sec), with a final 0.3-sec decrement, were used to achieve a reinforced IRT of at least 4.0 sec, but less than 5.2 sec. Under the DRL 10–13 sec schedule, progressive 0.5-sec increments, followed by progressive 0.5-sec decrements (beginning with 15 sec), were used to achieve the final IRT of at least 10 sec but less than 13 sec.
Initially the criterion for changing the minimum and maximum IRTs was an accuracy level in which 40% or more of the lever presses were reinforced, but this was reduced to 30% during the second phase of training under the DRL 1–1.3 sec schedule and with the initiation of the upper limit under the DRL 4–5.2 sec schedule. Later a 25% accuracy level was adopted under the DRL 1–1.3 sec schedule.
Illumination of a houselight mounted in the ceiling of the experimental chamber, a stimulus light above the right lever and the onset of the Sonalert tone signaled the beginning of the session. Depression of the lever turned off the tone and the stimulus light. If the response did not follow the onset of the tone and light by more than the minimum, but less than the maximum time duration required to produce the reinforcer (reinforcer window), the tone and the light resumed after 100 msec. If the response was within the reinforcer window and a food pellet was delivered, a 5-sec time-out occurred. During the time-out, the tone and the stimulus lights remained off. Responses during time-out had no programmed consequences and were not recorded. The tone and light came on again at the end of the time-out, signaling another trial. Although this was a signaled DRL schedule, in that reinforcer delivery was contingent on responses that followed the onset of the tone by a prescribed time interval, we will use the traditional terminology for spaced-responding schedules (i.e., DRL and IRT) throughout the manuscript. Sessions were conducted Monday through Friday between 6:45 and 7:30 a.m. (DRL 1–1.3 sec), 8:30 and 9:30 a.m. (DRL 4–5.2 sec) and 10:30 and 11:30a.m. (DRL 10–13 sec). All other conditions were the same as those reported in McClure et al. (1997).
Data analysis.
The total number of responses emitted, response rate (responses/sec), accuracy (% reinforced responses) and mean interresponse time were collected as performance indicators during every session for each rat. The mean values and standard errors for animals under each schedule were calculated for each of these performance measures. ANOVA was used to determine whether there were significant differences (P value was less than .05) in base-line values across schedules. ANOVA was also used to determine whether drugs produced effects that were significantly different from the average of 10 control sessions conducted during the three drug series for each group of rats. If animals failed to respond at least 25 times, the data were used to calculate only response rates.
IRTs were sorted into 24 consecutive time bins. A time bin is defined in this study as a range of time intervals of predetermined lengths into which IRTs of similar duration are accumulated. The time bins were proportional to the time-duration requirements for the specific DRL schedules and the relative frequencies, cumulative frequencies and significance were calculated as reported in McClure et al.(1997). The relative frequency of responses in each time bin (the number of IRTs per bin divided by the total number of IRTs made during the session) was calculated for each individual rat for each session. The DRL 1–1.3 sec schedule used 0.1-sec bins, with all IRTs shorter than 0.100 sec collected in bin 1 and all those greater than 2.300 sec collected in the last bin (bin 24). The DRL 4–5.2 sec schedule used 0.4-sec bins, with any IRTs shorter than 0.400 sec collected in bin 1 and all IRTs greater than 9.200 sec collected in the last bin (bin 24). The DRL 10–13 sec schedule used 1.0-sec bins with any IRT shorter than 1.0 sec collected in bin 1 and the final time bin collecting all IRTs 23.000 sec or greater. Bins 11, 12 and 13 collected IRTs that were reinforced under each schedule (1.0 sec to 1.3 sec, 4.0 sec to 5.2 sec or 10.0 sec to 13.0 sec).
The percentage of IRTs in each of the 24 time bins was used to construct relative-frequency histograms. These relative frequencies of IRTs under each schedule also were summed to calculate cumulative IRT frequencies. These cumulative IRT frequencies were plotted as sigmoidal curves. The portion of these curves corresponding to data collected in bins 7 through 15, which was relatively linear, was subjected to regression analysis of the slopes of these lines. Changes in these slopes caused by drug treatments could then be examined. AnF statistic was calculated by use of the formula:
Results
Base-line performance.
Animals under the DRL 1–1.3 sec, DRL 4–5.2 sec and DRL 10–13 sec schedules reached the stability criterion within 120, 65 and 70 sessions, respectively. The stability criterion for each rat was defined as consistent accuracy level with daily variability of no more than ± 5% over the 10 consecutive sessions, after completion of training procedures. At the beginning of the experiments, all rats under DRL 1–1.3 sec, 4–5.2 sec and 10–13 sec schedules had accuracy levels equal to or greater than 25%, 40% or 50%, respectively. Base-line levels of accuracy tended to increase gradually under the DRL 4–5.2 sec and DRL 10–13 sec schedules after the stability criterion was reached. Before each drug series was administered, accuracy levels for the previous 5 base-line days for each rat were assessed.
The mean accuracy, mean IRT and mean response rate of rats under all reinforcement schedules for a total of 10 base-line sessions are shown as control values in table 1. The 10 base-line sessions were selected from sessions before and after the MAP, PCP and Δ9-THC drug series, along with 4 Thursday-training days during the drug series with MAP and PCP, but not Δ9-THC. Thursday-training days during the Δ9-THC administrations were excluded because of concern that its long half-life might produce residual effects on control performance. This selection process for base-line sessions was done to include trials representative of the entire drug period. The overall rate of responding decreased with increasing time requirements under the DRL schedules, whereas accuracy and mean IRT increased. ANOVA indicated that significant differences occurred across the three DRL schedules for the base-line accuracies, mean interresponse times and response rates.
Figure 1 shows base-line relative frequency distributions and cumulative frequency curves under all reinforcement schedules. Although the relative frequency distribution (fig.1, upper frame) shows that the peak of the IRT distribution occurred at an IRT slightly shorter than the lower limit of the reinforcer window, the mean IRT under the DRL 1–1.3 sec schedule was longer than the upper limit for reinforcement because of the high frequency of long IRTs (bin 24). Responding under all schedules produced IRT histograms resembling normal distributions. Response burst, a high frequency of short IRTs (Sidman, 1956; Campbell and Seiden, 1973; Sanger et al., 1974), was apparent to a slight degree under the DRL 4–5.2 sec schedule (fig.1, middle frame) and to a lesser extent under DRL 10–13 sec schedule (fig.1, bottom frame).
General performance measures after drug administration.
The effects of MAP, PCP and Δ9-THC on the various measures of performance under all DRL schedules are shown in table 1. MAP caused a decrease in accuracy at doses of 1.0 mg/kg and higher under the DRL 4–5.2 sec schedule and with all doses under the 10–13 sec schedule. Only the 1.0 and 3.0 mg/kg doses of MAP produced significant effects on accuracy under the DRL 1–1.3 sec schedule. MAP produced statistically significant decreases in the mean IRT at 1.0 mg/kg and higher under the DRL 4–5.2 and at the 1.7 and 3.0 mg/kg doses under the DRL 10–13 sec schedule. MAP did not produce significant response rate effects under the DRL 1–1.3 and 4–5.2 sec schedules until dose levels were reached that eliminated responding of two of the four animals in each group. Under the DRL 10–13 sec schedule significant rate effects occurred at the 1.7 mg/kg dose only.
Under all schedules except the DRL 1–1.3 sec schedule, PCP decreased accuracy across a range of doses. PCP significantly decreased the mean interresponse time with all doses under the DRL 10–13 sec schedule. Under the DRL 1–1.3 sec and the DRL 4–5.2 sec schedules, PCP significantly increased the mean IRT at the highest dose. Low doses of PCP produced slight increases in response rates under all schedules. Under the DRL 4–5.2 sec schedule, however, PCP rate effects were statistically significant only at the 3.0 mg/kg dose where a decrease occurred. Under the DRL 10–13 sec schedule, rate increases were significant at doses of 1 mg/kg and higher.
Δ9-THC produced a dose-dependent decrease in accuracy under the DRL 10–13 sec schedule, but had little effect on the accuracy or variable results under the DRL 1–1.3 sec or the DRL 4–5.2 sec schedules. Long pauses tended to increase mean IRTs under the DRL schedules, but under the DRL 10–13 sec schedule, these pauses were counterbalanced by shorter IRTs (figs. 8, 9, 10), producing a mean IRT with little change. Only at the lowest doses was the IRT significantly effected under the DRL 10–13 sec schedule. Response rates were decreased significantly at the higher doses under the DRL 1–1.3 and 4–5.2 sec schedules. However, under the DRL 10–13 there were few changes in the response rates at any dose.
MAP effects on IRTs under the DRL 1–1.3 sec schedule.
Relative frequency distributions of the IRTs indicated little effect of MAP other than to lower the peak and slightly flatten the distribution (fig. 2 D). At the lower doses, there was a slight increase in the proportion of shorter IRTs (fig. 2B), which caused the cumulative IRT curve (fig. 2, curve B) to shift slightly left. However, this effect diminished with higher doses. Cumulative frequency curves in figure 2 showed small shifts to the right in the IRT distributions with higher doses and a slower rise of the curve with a decrease in magnitude of the slope. The relative frequency distribution showed an increase in the proportion of IRTs in bins to the right of the reinforcer window at the two highest MAP doses, which indicated an increase in the proportion of longer IRTs. The linear portions of the MAP curves were significantly different from the linear portion of the control curve at 1.0, 1.7 and 3.0 mg/kg doses. The 0.3 mg/kg dose did not produce a statistically significant difference from the control data.
MAP effects on IRTs under the DRL 4–5.2 sec schedule.
The cumulative frequency curves (fig. 3) showed a shift to the left and became more linear with increasing doses. MAP had little effect on the asymptotes of the curves. However, with increasing doses, the IRT relative frequency distribution shifted to the left of the reinforcer window (fig. 3, inserts). The proportion of IRTs in bins 4 to 8 increased slightly at the 1.0 mg/kg dose shifting the mode from bin 11 to bin 10. The secondary peak of the IRT distribution (0.1–0.2 sec) gradually disappeared at higher doses because of a more uniform distribution of IRTs shorter than the reinforcer window. The regression analysis indicated that the linear portions of the MAP curves were significantly different from the linear portion of the control curve at the 1.0, 1.7 and 3.0 mg/kg doses.
MAP effects on IRTs under the DRL 10–13 sec schedule.
Figure4 shows the effects of MAP on the cumulative frequency distribution of IRTs for rats under a DRL 10–13 sec schedule. The cumulative frequency distributions shifted to the left of the control curve with increasing doses because of the increased relative frequency of shorter IRTs, as shown in the inserts. At the 1.0 mg/kg dose, there was an increasing percentage of shorter IRTs (insert B), which became more pronounced with increasing doses. At the highest dose (insert D), the relative frequencies of responses shorter than the reinforcer window were fairly evenly distributed across bins, and the relative number of IRTs falling within the reinforcer window was greatly diminished. At the highest dose more than 50% of the IRTs were in the first six bins. The linear portions of the MAP curves for all dose levels were significantly different from the linear portion of the control curve.
PCP effects on IRTs under the DRL 1–1.3 sec schedule.
The effect of PCP on the relative frequency distributions were slight (fig.5). At low doses, there was a small shift of the cumulative curve up and to the left. The curves for the 1.7 and 3.0 mg/kg doses shifted down and to the right because of an increase in longer IRTs. The large decrease in the asymptote corresponded with the large increase in responses in bin 24 (insert D) of the relative frequency distribution, which indicated long pauses. The Fstatistic on the linear portions of the PCP curves showed a significant difference from the linear portion of the control curve at the 1.0 mg/kg, 1.7 mg/kg and 3.0 mg/kg doses.
PCP effects on IRTs under the DRL 4–5.2 sec schedule.
The cumulative frequency curves (fig. 6) show two distinct effects. There was an increase in the early peak of the relative frequency distribution at doses of 0.3 to 1.7 mg/kg because of an increase in short IRTs (fig. 6, A–C). At these doses the IRT cumulative distribution shifted to the left and remained fairly parallel to the control curve. At the 3.0 mg/kg dose, the slope was altered with the early portion of the cumulative curve again shifting to the left because of the relatively even distribution of responses with IRTs less than those needed to produce the reinforcer. Also at this dose, there was a decrease in the asymptote from the control curve because the increase in relative frequency of IRTs in bin 24 (insert D) caused the latter portion of the cumulative PCP curve to intersect the control curve (fig. 6, curve D). The linear portions of the PCP curves were significantly different from the linear portion of the control curve at all dose levels.
PCP effects on IRTs under the DRL 10–13 sec schedule.
The cumulative frequency response curves (fig. 7) shifted to the left at all doses because of the increased relative frequency of shorter IRTs as shown in the inserts. At the 3.0 mg/kg dose, the peak of the IRT distribution decreased to 6 to 7 sec (insert D). The linear portions of the PCP curves were significantly different from the linear portion of the control curve for all dose levels.
Δ9-THC effects on IRTs under the DRL 1–1.3 sec schedule.
Little effect occurred in the relative frequency distributions of IRTs with Δ9-THC at increasing doses. A decrease in the peak height of the IRT relative frequency distribution occurred at the 3.0 and the 4.2 mg/kg doses (fig.8, inserts D and E) because of the corresponding increases in bin 24 (long pauses). The cumulative frequency IRT curves (fig. 8) showed two effects. There was a small leftward shift after the 1.7 mg/kg dose, whereas at the 3.0 and 4.2 mg/kg doses the rightward shift of the cumulative curves was characteristic of longer IRTs. The linear portions of the Δ9-THC curves were significantly different from the linear portion of the control curve at 1.7, 3.0 and 4.2 mg/kg.
Δ9-THC effects on IRTs under the DRL 4–5.2 sec schedule.
At the lowest dose there was a slight increase in short IRTs which caused a slight upward shifting of the early portion of the cumulative frequency curve (fig.9). The predominant effect of Δ9-THC on the cumulative frequency response curves was a delay in reaching asymptote (longer IRTs) at all doses that produced effects. This effect was quite large at the 3.0 mg/kg dose, causing a large decrease in the asymptote of the curve because of a large number of long IRTs. The large increase in longer responses caused a flattening of the overall relative frequency distribution at the 3.0 mg/kg dose as well. However, the effect decreased at the 4.2 mg/kg dose (fig. 9, inserts D and E). The linear portions of the Δ9-THC curves were significantly different from the linear portion of the control curve at the 0.3, 1.0, 3.0 and 4.2 mg/kg doses.
Δ9-THC effects on IRTs under the DRL 10–13 sec schedule.
The effect of Δ9-THC on the relative frequency of IRTs under a DRL 10–13 sec schedule caused an increase in IRTs that were shorter than the reinforcer window even at the lowest doses. The mode shifted to the left at doses of 1.0 mg/kg and higher (fig. 10, inserts B–E). In addition to the increase in short IRTs, at the 3.0 and 4.2 mg/kg doses, there was an increase in long pauses (bin 24). Δ9-THC effects differed somewhat from those of MAP and PCP, because although all drugs increased the frequency of short IRTs, the distinct bell shape of a normal distribution was retained with Δ9-THC (fig. 10E). Figure 10 shows that the cumulative frequency curves shifted toward shorter IRTs at all doses. With high doses of Δ9-THC, the cumulative frequency curves retained their sigmoidal shapes characteristic of cumulative representation of a bell-shaped relative frequency histogram; however, the modes were now in earlier bins. The cumulative curves of the highest doses of Δ9-THC (3.0 and 4.2 mg/kg) additionally intersected the control curve and asymptote was reached in a later bin, which is characteristic of an increased proportion of long IRTs. The linear portions of the Δ9-THC curves were significantly different from the linear portion of the control curve for all doses.
Discussion
DRL schedules have been used to measure drug effects on timing behavior, but little consideration has been given to the interaction between drugs and the schedule parameters that control timing behavior. Although all of the drugs in the present study have been reported to alter time perception (Hicks et al., 1984; Karniol et al., 1975; Walker et al., 1981; Meck, 1983), it is apparent that they produce differential effects on behavior under DRL schedules. The drug, the dose, the time-duration requirements of the schedule and the use of a limited hold may all influence these behavioral effects.
Clearly, different drugs produced different effects on responding under the same DRL schedule. For example, under the DRL 4–5.2 sec schedule, MAP and PCP increased the relative frequency of short IRTs at low doses and produced a flattening of the relative frequency distribution at higher doses. In fact, the effects of MAP and PCP on timing behavior were fairly similar under all of these DRL schedules. In contrast the main effect of Δ9-THC on behavior under the DRL 4–5.2 schedule was to slightly flatten the relative IRT distribution by increasing the proportion of rather long IRTs. Clearly, the effects of Δ9-THC were different from the effects of the other two drugs on responding under this same schedule.
Long pauses, such as those produced by Δ9-THC, were a prominent effect of high doses of all of the drugs on behavior under the DRL 1–1.3 sec schedule. These long pauses lowered the asymptotes of the cumulative frequency distribution curves and gave the impression of large effects of the drugs on behavior. However, the occurrence of these long pauses under DRL schedules may not reflect direct effects of drugs on timing behavior. Instead they may reflect drug-induced suppression of responding, or increases in adjunctive behaviors (Falk, 1971) that compete with behaviors mediating timing (Laties et al., 1965, 1969; Hodos et al., 1962). Under schedules with short time-duration requirements and narrow reinforcer windows, such as the DRL 1–1.3 sec schedule, even brief occurrences of competing behaviors, or other mechanisms of drug-induced suppression of timed responding, could have large consequences on the number of reinforcers earned and the shape of the cumulative frequency distribution of IRTs. This may occur without having large effects on the peak associated with reinforcement in relative frequency distribution of IRTs.
Other investigators have also observed long-duration IRTs under DRL schedules after the administration of PCP (Sanger, 1992; Freemanet al., 1984; Balster and Baird, 1979), Δ9-THC (Manning, 1973; Ferraro et al., 1971) and amphetamines (Balster and Baird, 1979). The most common feature in all of these studies is the lack of an upper limit on the reinforcer window. The lack of an upper limit on the reinforcer window results in the delivery of the reinforcer when long IRTs are terminated and may adventitiously reinforce behaviors that compete with and interfere with timing behavior. When an upper limit on the reinforcer window is in effect, these long pauses in responding do not terminate with delivery of the reinforcer. The use of an upper limit on the reinforcer window thereby decreases the reinforcement of possible nontiming behaviors that could compete with timing behavior. Thus, the use of upper limits on the reinforcer window may be an important determinant of drug effects on responding under DRL schedules.
The time-duration requirements of DRL schedules also appear to be important determinants of the effects of drugs on responding under DRL schedules. Under the 1–1.3 sec schedule, the predominant effect of all of the drugs was to increase the proportion of long IRTs. Under the 4–5.2 sec schedule, both MAP and PCP increased the proportion of short IRTs, but Δ9-THC only produced an increase in the proportion of long IRTs. Under the 10–13 sec schedule, all of the drugs increased the relative frequency of short IRTs. Thus, the effects of these drugs not only depended on the drug and the dose, but the effects also depended on the schedule parameters. At the shortest IRT duration required for reinforcement (DRL 1–1.3 sec), the drugs increased the proportion of long IRTs. As longer duration IRTs were required to produce the reinforcer, there was a tendency for all drugs to increase the relative frequency of short IRTs and decrease the relative frequency of long IRTs. Clearly the parameters of the schedule influenced the effects of each of these drugs.
Most experiments reported in the literature have used DRL schedules where IRTs of 10 sec or longer were required to produce the reinforcer. Our data under the DRL 10–13 sec schedule are most directly comparable with these studies. The effects of MAP observed under the DRL 10–13 sec schedule were similar to the effects other investigators have reported for amphetamines under DRL schedules requiring IRTs of 10 sec or longer. In both rats and mice, doses of about 3 mg/kg amphetamine produced increases in short IRTs (Wenger and Wright, 1990; Balster and Baird, 1979). At much higher doses, an increase in longer IRTs was seen. This was similar to the effects of MAP in the present study. The leftward shift in the IRT distribution after amphetamines has also been found by several other investigators (Adam-Carriere et al., 1978; Sidman, 1955; Pradhan and Dutta, 1970).
In our studies, PCP also produced an increase in the relative frequency of short IRTs and flattened the relative frequency distribution of IRTs. Sanger (1992) reported similar effects of PCP in rats responding under a DRL 15-sec schedule. Sanger found a fairly flat distribution of short IRTs after both 4 and 8 mg/kg doses and an increase in longer pauses after the 8 mg/kg dose. In mice, Freeman et al.(1984), and Balster and Baird (1979) also found a fairly flat distribution of short IRTs under a DRL 10-sec schedule after administration of PCP. Thus, the effects that we found with PCP under the DRL 10–13 sec schedule were similar to those reported by others.
Δ9-THC had little effect on responding under the DRL 1–1.3 and 4–5.2 sec schedules, but its effects on responding under the DRL 10–13 sec schedule were similar to the effects of Δ9-THC in other studies requiring that IRTs be longer than 10 sec to produce the reinforcer. Increases in the frequency of shorter IRTs have been found in chimpanzees under a DRL 10-sec schedule (Ferraro et al., 1971), squirrel monkeys responding under a DRL 28-sec schedule (Galbicka et al., 1980) and monkeys responding under a DRL 60-sec schedule (Manning, 1973). In all of these studies the IRT distribution retained the bell-shaped distribution characteristic of a normal curve, even as the peak of that distribution shifted toward shorter IRTs. Although relative frequency histograms were not shown, increases in slightly shorter IRTs have also been reported in time estimation studies of 10 sec or longer which incorporated a motor response (Kopell et al., 1978;Fernendez-Guardiola et al., 1976) similar to that used under DRL schedules. In these studies, human volunteers were required to press a lever when an estimated 10 sec had elapsed.
Although there are some differences among these drugs in their effects under the reported DRL schedules with different time-duration requirements, all three drugs resulted in what can be described as an overestimation of the rate of passage of time (subjects responded before enough time had elapsed to allow the response to fall within the reinforcer window). Human time estimation studies requiring the subject to respond when 30, 60 or 120 sec have passed have shown that amphetamines (Hollister and Gillespie, 1970), cannabinoids (Tinklenberget al., 1976; Hicks et al., 1984; Musty et al., 1976; Bachman et al., 1979) and dissociative anesthetics (Krystal et al., 1994) caused a significant overestimation of time intervals, which suggested a perception that time was passing more quickly. Furthermore, there appears to be a direct correlation between urine PCP levels and the degree of overestimation of time in time production tasks performed by humans requiring hospitalization because of PCP usage (Yesavage et al., 1978; Yesavage and Freeman, 1978; Walker et al., 1981).
Animal studies with temporal discrimination paradigms also have shown that MAP produces overestimation of time (Maricq et al., 1981; Maricq and Church, 1983; Meck, 1983; Meck and Church, 1983). However, the effects of Δ9-THC were equivocal in the same studies. In experiments in which subjects are trained to discriminate between two timed stimuli presented for different durations, Δ9-THC produced a greater decrease in accuracy on long (8-sec) duration trials than on short (4-sec) duration trials (Daniel and Thompson, 1980). In contrast, although Elsmore (1972) found that Δ9-THC decreased accuracy, the monkeys responded on the lever that was appropriate for a 100-sec duration stimulus when, in fact, the stimulus was shorter. This would indicate an underestimation of the duration of the stimulus.
A major purpose of the present study was to compare the effects of drugs on responding under DRL schedules with the effects of these same drugs on responding under TRD schedules (McClure et al., 1997). In both studies the same doses were studied in the same species under the same schedule parameters. Both DRL and TRD schedules require time production. But whereas DRL schedules require the animals to respond after a pause of specified duration, TRD schedules require the animal to make a continuous response, in this case, a lever press of a specified duration.
PCP produced similar effects on responding under TRD and DRL schedules with time-duration requirements of 4–5.2 and 10–13 sec. That is, it increased the relative frequency of short IRTs and flattened the IRT distribution (McClure et al., 1997). However, PCP had different effects on responding under the DRL and TRD 1–1.3 sec schedules. Under the DRL 1–1.3 sec schedule, PCP’s primary effect was to increase the percentage of long IRTs with minimal effects on the rest of the relative frequency distribution. In contrast, under the TRD schedule, 3 mg/kg PCP, and to some extent 1.7 mg/kg PCP, shifted the response-duration distribution to the left and flattened it.
Similar to PCP, Δ9-THC increased the relative frequency of longer IRTs under the DRL 1–1.3 sec schedule. Unlike PCP, Δ9-THC had little effect on responding under the TRD 1–1.3 sec, the TRD 4–5.2 sec or the DRL 4–5.2 sec schedule, other than some long pauses.
MAP produced similar effects on timed responding under both the TRD and the DRL 1–1.3 sec schedules at both low doses and high doses. At low doses, MAP produced small shifts toward an increase in the percentage of shorter response durations and IRTs. At higher doses of MAP, increases in both shorter and longer response durations or IRTs occurred under both schedules. The difference in base-line performance under the two schedules may have contributed to the differential effects with the other two drugs but did not seem to influence MAP as much. Thus, these drugs produced differential effects on responding under DRL and TRD schedules.
Under the DRL 4–5.2 and 10–13 sec schedules, MAP shifted the relative cumulative IRT distribution to the left with a fairly flat distribution across bins short of the reinforcer window at the 1.7 and 3.0 mg/kg doses. Under TRD 4–5.2 sec and 10–13 sec schedules, MAP also shifted the distribution toward shorter response durations, but the effect was much more pronounced with most of the response durations falling into the first two bins at these doses. PCP had similar effects on responding under DRL and TRD 4–5.2 sec schedules, in that it increased the relative frequency of both short IRTs and response durations with the distribution of responses in each bin short of the reinforcer window becoming fairly constant at the higher doses under both schedules, although the effects were more pronounced under the TRD schedule. Under the DRL 10–13 sec schedule, both drugs shifted responding toward shorter IRTs, but under the DRL schedule there was a greater tendency for the distribution to hold the shape of a normal distribution. The effects of Δ9-THC were small under DRL and TRD 4–5.2 sec schedules, with an increase in long IRTs at most doses of Δ9-THC for animals responding under the DRL 4–5.2 sec schedule, but with little effect on responding under the TRD 4–5.2 sec schedule. Under DRL and TRD 10–13 sec schedules, Δ9-THC shifted the relative frequency distributions toward shorter IRTs and response durations, but again under the DRL schedule there was a greater tendency for the distribution to retain the shape of a normal distribution, and the effects were much smaller than under the TRD schedule.
It has been suggested that drug effects on DRL responding might be rate-dependent. For example, amphetamines increase low rates of responding and decrease high rates (Dews, 1958, 1969; Dews and Wenger, 1977). Under both DRL and TRD schedules, the base-line rates of responding decrease as the time-duration requirement of the schedule increase. It might be predicted that shifts toward shorter IRTs and response durations would be observed under the schedules with the longer time-duration requirements. This appears to describe what happened under DRL schedules. As the time-duration requirements of the DRL schedule increased, MAP produced larger increases in the rate of responding (table 1). This was not true under TRD schedules in which increases in response rate after MAP were small and not statistically significant (McClure et al., 1997). The reason for this difference may relate to the requirements of the two schedules. Under DRL schedules, rate of responding depends almost completely on the IRT duration, because the animals do not hold the lever down for extended periods of time. If a drug increases or decreases IRT durations, the rate goes down or up, respectively. In contrast, under TRD schedules, both response durations and IRTs determine the rate of responding. For example, a drug that shortened response durations, but increased pausing between responses, might show little overall change in response rate under TRD schedules. Under DRL schedules, this unopposed decrease in IRTs would increase overall response rates. This suggests that the effects of drugs on responding under TRD schedules is less dependent on overall rates of responding than responding under DRL schedules.
Under both TRD and DRL schedules, as the drug effect disrupts timed responding, the frequency of reinforcer delivery decreases. This decrease in reinforcer delivery might cause a further disruption in timing behavior, or it might activate compensatory mechanisms which would counteract the drug effect. In an attempt to answer this question, McMillan et al. (1994) expanded the size of the reinforcer window on days when drugs were given to maintain the frequency of reinforcer delivery. This manipulation had little effect on the relative frequency distribution of response durations. When the frequency of reinforcer delivery was decreased under control conditions by reinforcing only 50% of the response durations that fell within the reinforcer window, there was a slight increase in the percentage of responses that fell within the reinforcer window. Similar experiments have not been performed with DRL schedules. However, Schuster and Zimmerman (1961) have shown that the repeated administration of 1.0 mg/kg dl-amphetamine administered before each daily session to rats responding under a DRL 17.5-sec schedule at first shifted the IRT distribution toward shorter IRTs, but with repeated administration the IRT distribution returned toward the control IRT distribution. This suggests that animals may learn to compensate for decreased reinforcement under drug with repeated drug experience.
In summary, responding under DRL schedules, like that under TRD schedules, depends on the drug, the dose and the time-duration requirements of the schedule. However, the effects of drugs on timing behavior often differ under TRD and DRL schedules, even when the time-duration requirements for the two schedules are identical. These data suggest that DRL and TRD schedules may be measuring different aspects of timing behavior.
Acknowledgments
The authors thank W. C. Hardwick and Dean W. Wright for skilled technical assistance, and the National Institute on Drug Abuse for providing the PCP and Δ9-THC used in this study. Additionally, the authors thank Dr. Galen Wenger for his suggestion of cumulative curves for analysis of drug effects and for suggestions in preparing the manuscript.
Footnotes
-
Send reprint requests to: D. E. McMillan, Dept. Pharmacology and Toxicology, University of Arkansas for Medical Sciences, 4301 W. Markham Street, Little Rock, AR 72205.
-
↵1 This work was supported National Institute on Drug Abuse Grant DA 02257.
- Abbreviations:
- Δ9-THC
- delta 9-tetrahydrocannabinol
- DMSO
- dimethylsulfoxide
- DOM
- 2,5-dimethoxy-4-methylamphetamine
- DRL
- differential reinforcement of low rates of responding
- DRL LH
- differential reinforcement of low rates of responding with a limited hold
- DRP
- differential reinforcement of low rates of responding with pacing
- IRT
- interresponse time
- LSD
- lysergic acid diethylamide
- MAP
- methamphetamine
- MK-801
- dizocilpine
- NANM
- N-allylnormetazocine
- NMDA
- N-methyl-d-aspartate
- PCP
- phencyclidine
- T,a time variable
- TRD, temporal responses differentiation
- ANOVA
- analysis of variance
- Received February 7, 1996.
- Accepted February 28, 1997.
- The American Society for Pharmacology and Experimental Therapeutics