STIMULUS CONTROL OF COCAINE SELF-ADMINISTRATION

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR
2003, 79, 111–135
NUMBER 1 ( JANUARY)
STIMULUS CONTROL OF COCAINE SELF-ADMINISTRATION
STANLEY J. WEISS, DAVID N. KEARNS, AND SCOTT I. COHN
AMERICAN UNIVERSIT Y
AND
CHARLES W. SCHINDLER, AND LEIGH V. PANLILIO
NATIONAL INSTITUTE ON DRUG ABUSE
Environmental stimuli that set the occasion wherein drugs are acquired can ‘‘trigger’’ drug-related
behavior. Investigating the stimulus control of drug self-administration in laboratory animals should
help us better understand this aspect of human drug abuse. Stimulus control of cocaine self-admin-
istration was generated here for the ?rst time using multiple and chained schedules with short,
frequently-alternating components—like those typically used to study food-maintained responding.
The procedures and results are presented along with case histories to illustrate the strategies used
to produce this stimulus control. All these multicomponent schedules contained variable-interval
(VI) components as well as differential-reinforcement-of-other-behavior (DRO) or extinction com-
ponents. Schedule parameters and unit dose were adjusted for each rat to produce stable, moderate
rates in VI components, with minimal postreinforcement (infusion) pausing, and response cessation
in extinction and DRO components. Whole-body drug levels on terminal baselines calculated ret-
rospectively revealed that all rats maintained fairly stable drug levels (mean, 2.3 to 3.4 mg/kg) and
molar rates of intake (approximately 6.0 mg/kg/hr). Within this range, no relation between local
VI response rates and drug level was found. The stimulus control revealed in cumulative records was
indistinguishable from that achieved with food under these schedules, suggesting that common
mechanisms may underlie the control of cocaine- and food-maintained behavior.
Key words: Stimulus control, drug self-administration, multiple schedules, chained schedules, co-
caine, whole-body drug levels, rats
There is a growing consensus that environ-
ferentially associated with changes in the
mental stimuli that set the occasion wherein
probability of receiving nondrug reinforcers
drugs can be acquired, or are otherwise as-
such as food (Bindra, 1972; Rescorla & Sol-
sociated with the drug-taking experience, can
omon, 1967; Weiss, 1978; Weiss & Schindler,
come to act as stimuli that ‘‘trigger’’ drug-
1987, 1989).
related behavior. For example, it is well estab-
Shulman (1989) used a questionnaire to
lished that drug-related stimuli can elicit
study ‘‘cocaine triggers’’ in 200 addicts re-
drug craving in humans (Childress, Mc-
ceiving treatment. The most frequently cited
Lellan, Ehrman, & O’Brien, 1987; Ehrman,
category leading to drug seeking was people,
Robbins,
Childress,
&
O’Brien, 1992;
places, and things which included such cue de-
O’Brien, Childress, McLellan, & Ehrman,
scriptions as ‘‘places you copped,’’ ‘‘cocaine
1990). According to some current accounts
paraphernalia,’’ and ‘‘people using around
of drug dependence (Markou et al., 1993;
me.’’ Wallace (1989) found that environmental
Pert, 1994; Robinson & Berridge, 1993), such
stimuli were cited as the trigger for drug seek-
craving involves the same type of behavior-
ing by 34% of a patient population present-
increasing ‘‘incentive-motivation’’ mecha-
ing for treatment a second time. These stud-
nism as that operating when stimuli are dif-
ies, as well as those cited above, suggest that
environmental cues are critical in occasion-
ing drug-related behavior in many drug abus-
This research was supported by NIDA Grant DA-08651
awarded to Stanley J. Weiss and in part by NIDA Division
ers. In spite of this, Bickel & Kelly (1988) con-
of Intramural Research. The principles of laboratory an-
cluded that,
imal care as described in NIH publication No. 85-23, re-
. . . stimulus control is a behavioral process
vised 1985, were followed.
Correspondence concerning this article should be ad-
that has not received a great deal of scienti?c
dressed to Stanley J. Weiss, Department of Psychology,
attention, both in relation to the basic pro-
American University, Washington, DC 20016 (e-mail:
cesses involved and the application of that
sweiss@american.edu).
knowledge for the socially important prob-
111

---

112
STANLEY J. WEISS et al.
lems of substance abuse . . . . If we continue to
tomarily the case with more conventional
ignore stimulus control processes, our goal of
multiple schedules where many reinforcers
having a behavior analysis of substance abuse
can be earned per SD. To promote environ-
will be woefully incomplete. (p. 136).
mental stimuli gaining maximum control of
This observation was reiterated by Bickel &
behavior, in the present study multiple infu-
Kelly (1997).
sions were possible in an SD.
In most studies using multicomponent
For the most part, the study of stimulus
schedules of drug self-administration with
control has focused on behavior maintained
rats, responding was reinforced by drug in
by reinforcers, such as food or shock-avoid-
one component and food in another, typically
ance, on multicomponent schedules where
with 30-min or longer components, and there
an operant response is differentially rein-
were few components per session (Caine and
forced in the presence of frequently alternat-
Koob, 1994a, 1994b; Goeders & Guerin,
ing different stimuli. Much has been accom-
1994; Goeders, McNulty, & Guerin, 1993;
plished with this approach. (For some reviews
Shoaib, Swanner, Beyer, Goldberg, & Schin-
and research, see Dinsmoor [1995a, 1995b],
dler, 1998; Weissenborn, Yackey, Koob, &
Harrison [1991], Rilling [1977], Terrace
Weiss, 1995). For several reasons, long sched-
[1966], Weiss & Schindler [1987], and Weiss,
ule components may not be ideal for gener-
Thomas & Weissman [1996].) By investigat-
ating stimulus control and establishing con-
ing how stimulus control is produced and
ditioned effects of drug-related discriminative
maintained when behavior is reinforced by
stimuli. With long schedule components, the
pharmacological agents in animals, we may
delivery of a reinforcer may acquire discrim-
gain insight into the process by which envi-
inative properties that are potentially more
ronmental stimuli come to control drug
salient than the exteroceptive stimuli in sig-
abuse in humans.
naling the availability of further reinforce-
Goldberg and Kelleher (1976) studied a
ment. In addition, when only a few compo-
multiple (mult) ?xed-inter val (FI) 5-min
nents are presented per session, there are few
?xed-ratio (FR) 30 schedule for cocaine re-
‘‘trials’’ through which to establish condi-
inforcement in primates. Their components
tioned effects. Furthermore, some theories of
switched after every reinforcer and a 100-s
associative learning (e.g., scalar expectancy
timeout followed each injection. They found
theory [Gibbon, 1977; Gibbon & Balsam,
that responding in each component was ap-
1981]) postulate that excitatory properties
propriate to the multiple schedule, indicating
are most effectively conditioned to stimuli if
stimulus control. Also in primates, Balster
these stimuli are presented for a brief period
and Schuster (1973) studied a mult FI 9-min
of time relative to total session duration.
extinction (15-min timeout) schedule of co-
To ef?ciently and effectively generate stim-
caine reinforcement. Again, there was one in-
ulus control of drug self-administration, the
jection per FI component. As with the Gold-
SDs need to be presented many times per ses-
berg
and
Kelleher
study,
responding
sion, necessitating schedule components of
appropriate to the FI schedule occurred
short to moderate duration. Establishing such
when the stimulus for that component was
baselines can be challenging, and few re-
presented, and cumulative records also re-
searchers have used multicomponent intra-
vealed that responding was low in the extinc-
venous drug self-administration schedules
tion component.
with short components. This laboratory, how-
The multiple schedules used in the studies
ever, has a history of achieving stimulus con-
described above differ from those typically
trol with complex schedules wherein re-
used with conventional reinforcers in that (a)
sponding was maintained by food and/or
only one reinforcer could be earned per dis-
shock-avoidance (e.g., Weiss, 1964, 1969,
criminative stimulus (SD), and (b) the com-
1971; Weiss & Panlilio, 1999; Weiss, Panlilio,
ponent wherein reinforcement was earned al-
& Schindler, 1993a, 1993b; Weiss & Schindler,
ways terminated with reinforcer deliver y.
1989; Weiss, Thomas, & Weissman, 1996;
Therefore, reinforcer delivery itself could
Weiss & Van Ost, 1974). By applying strategies
have played a larger role in the stimulus con-
adapted from this previous work, it has been
trol generated in these primates than is cus-
possible to achieve stimulus control on com-

---

STIMULUS CONTROL OF COCAINE SELF-ADMINISTRATION
113
plex baselines of drug self-administration that
a whole-body drug level ‘‘satiety threshold’’ of
is virtually indistinguishable from that
approximately 1.7 mg/kg is surpassed; re-
achieved with food baselines.
sponding then resumes when cocaine levels
A description of how these techniques can
fall below this point. In order to evaluate
be used to achieve stimulus control of drug
whether this type of regulated intake or titra-
self-administration may be of value to other
tion of cocaine levels also occurs under com-
investigators. Therefore, the procedures and
plex schedules like those described in the
the resulting stimulus control established in
present paper, whole-body drug levels were
representative rats will be presented as de-
calculated for all of the sessions for which cu-
tailed case histories in an integrated, contin-
mulative records are presented. These analy-
uous tutorial style in the tradition of Ferster
ses were performed retrospectively, with drug
and Skinner (1957). Examples of the behav-
levels calculated only after all training pro-
ioral control established at various stages of
cedures had been completed and represen-
training under a variety of complex schedules
tative records selected. We examined whether
of drug self-administration will be provided in
drug level and response rate covaried in the
the form of cumulative records that are rep-
VI components of these schedules.
resentative of a subject’s behavior in early, in-
termediate, and terminal baseline sessions.
The terminal baseline stimulus control estab-
OVERVIEW OF OBJECTIVE
lished in other comparably trained rats is pre-
AND METHODS
sented in tables.
To determine whether drug reinforcement
One of the obstacles to establishing stimu-
can produce stimulus control comparable to
lus control with multicomponent schedules
that produced with food under similar contin-
of drug self-administration is the fact that
gencies, we examined several multicomponent
each intravenous drug infusion tends to be
schedules. To facilitate the comparison of
followed by an extended pause in respond-
drug- and food-maintained responding, con-
ing. These postreinforcement pauses could
tingencies and schedule parameters closely
potentially be due to satiation, aversiveness of
approximated those that we have used previ-
high cumulative doses, or motor effects that
ously with conventional reinforcers. In addi-
prevent operant responding (Katz, 1989;
tion, the same discriminative stimuli and stim-
Lynch & Carroll, 2001). Under the simplest
ulus/schedule combinations were employed.
schedule of reinforcement, where drug is de-
Stimulus control of cocaine self-administra-
livered each time a response is emitted (con-
tion was established in rats trained with mul-
tinuous reinforcement, crf), the postinfusion
tiple (mult) schedules—where schedule com-
pause is a direct function of unit dose
ponents lasted for speci?ed amounts of
(Dougherty & Pickens, 1973; Lynch, La-
time—and chained (chain) schedules—
Bounty, & Carroll, 1998; Pickens & Thomp-
where the operant response caused the com-
son, 1968). Thus, one objective of the train-
ponents to progress.
ing described below was to tailor for each rat
The three basic schedules studied here all
the infusion dose that would maintain a
produced comparable patterns of lever
steady, moderate response rate and have the
pressing by the end of training. In the pres-
rat return to lever pressing directly after an
ence of a discriminative stimulus (SD) [tone,
infusion.
light, or tone-plus-light (TL)], variable-inter-
Analyses of whole-body drug levels and oth-
val (VI) schedules were used to maintain re-
er drug effects (e.g., dopamine levels in the
sponding at stable, moderate rates, with min-
nucleus accumbens; Petit & Justice, 1989,
imal postinfusion pausing. In the absence of
1991; Wise, 1999; Wise et al., 1995) under
these stimuli (TL), extinction (EXT) or dif-
continuous reinforcement (crf) schedules
ferential-reinforcement-of-other-behavior
have led to suggestions that animals regulate
(DRO) operated, and responding essentially
their intake to maintain levels of drug or
ceased.1 The goal of training for all schedules
some speci?c drug effect within a narrow
range. For example, Tsibulsky and Norman
1 In our laboratory, we have used variable interval (VI)
(1999) concluded that cocaine self-adminis-
schedules that operated in tone, light, or tone-plus-light
tration under a crf schedule continues until
(TL) while in the absence of these stimuli (TL) extinc-

---

114
STANLEY J. WEISS et al.
was to produce response rates in VI compo-
the cumulative records. These changes usu-
nents that were at least seven times those in
ally occurred between sessions, but could oc-
EXT or DRO components, the criterion Pan-
cur several times over the course of even a
lilio, Weiss, and Schindler (1996) employed
single session. If responding became sporad-
in the original self-administration study re-
ic, the response requirement was reduced
ported from this laboratory. On the terminal
and/or the unit dose was increased to its pre-
baseline schedules, component durations
vious level. Although this meant that not ev-
were generally in the 1 to 3 min range.
ery rat received exactly the same training his-
Although the stimuli controlled compara-
tory, the essential contingencies were always
ble patterns of lever pressing in all three basic
consistent with the other rats in the same
schedules, the relation between lever pressing
group. To insure that the stimulus control re-
and the delivery of drug reinforcers differed
ported here can be unequivocally attributed
systematically over the three schedules. In the
to the cocaine contingencies, rats in the pres-
?rst schedule to be described (mult VI EXT),
ent experiment were not pretrained to press
all cocaine was received in the VI component
the lever for food, as is often done in drug
(tone, light, or TL), where lever pressing was
self-administration studies.
maintained. In the second schedule, mult VI
VI DRO, cocaine could be acquired in both
EXPERIMENT 1: MULTIPLE VI EXT
VI components (where responding was main-
AND MULTIPLE VI VI
tained) and in DRO components (where re-
EXT SCHEDULES
sponding ceased). Finally, in the third sched-
ule, a chain VI DRO, lever pressing in the VI
On a multiple (mult) schedule, different
component (tone, light, or TL) did not pro-
contingencies operate in the presence of spe-
duce cocaine, but caused the schedule to pro-
ci?c stimuli and components alternate auto-
gress to the DRO component (TL), where co-
matically. Under the two-component mult VI
caine was received contingent on response
EXT schedule of Experiment 1, a VI schedule
cessation. Therefore, under this chain VI
of cocaine self-administration operated in
DRO schedule, the drug was only received in
tone-plus-light (TL), and no cocaine was
the component where responding ceased.
available (EXT) in the absence of tone and
Thus, across the three respective schedules
light (TL). The procedures used and behav-
drug delivery was (a) explicitly paired with,
ioral progress for a representative rat (S-43)
(b) presented nondifferentially with respect
trained on this schedule are described in de-
to, or (c) explicitly unpaired with the SD oc-
tail. Terminal baseline data from nine addi-
casioning lever pressing. This variety of stim-
tional rats trained on this schedule are also
ulus-reinforcer and response-reinforcer rela-
presented.
tions
has
never
been
systematically
The three-component mult VI VI EXT
investigated with drug reinforcement.
schedule of Experiment 1 was like the two-
The goal of the present study was to deter-
component mult VI EXT schedule, but a VI
mine whether stimulus control of drug self-
cocaine contingency operated when either a
administration could be achieved under the
tone or a light was present. TL was still asso-
conditions described above as it has with food
ciated with extinction. The training history of
reinforcement. At all stages of training, pa-
a representative rat (S-16) trained on this
rameters were changed only after the effects
mult VI VI EXT schedule is described below,
of the previous change became apparent in
plus terminal baseline data from three addi-
tional rats trained on this schedule.
tion (EXT) or differential-reinforcement-of-other-behav-
ior (DRO) was programmed. Weiss (1969), Weiss and
METHOD
Emurian (1972), and Tsai and Weiss (1977), however,
Subjects
have shown that comparable control can be achieved
with other stimulus/schedule combinations. For exam-
Seven adult male Sprague-Dawley rats ( J-3,
ple, the VI schedule can operate in tone or in light while
J-4, S-9, S-16, S-110, SN-19, and SN-33) and 7
TL signals extinction, or the VI schedule can operate in
adult male Long-Evans rats (LD-3, LD-20, LD-
TL and in TL while tone or light signal extinction. Those
18, LF-16, LF-10, S-43, and S-44) were used.
studies were used to investigate the dynamics of Weiss’
(1972) composite-stimulus model of stimulus control, but
They were housed in individual cages in a col-
that is beyond the scope of the present report.
ony room with a 12-hr light/dark cycle (lights

---

STIMULUS CONTROL OF COCAINE SELF-ADMINISTRATION
115
on, 8 a.m. to 8 p.m.). Training sessions were
1996). Catheters consisted of approximately
conducted during the light-on cycle. Water
4 cm Silastic tubing (0.044 mm ID, 0.814 mm
was available continuously, except during the
OD) connected to vinyl tubing (Dural Plas-
4-hour training sessions. Weights were main-
tics, 0.5 mm ID, 1.0 mm OD). The vinyl por-
tained at approximately 80% of ad lib (348
tion of the catheter exited at the back of the
to 454 gm) with laboratory rat chow provided
neck and was obturated with a modi?ed 23-g
following training sessions. Training sessions
needle. A 20-mm plastic screw was cemented
were conducted 7 days a week.
with dental acrylic to 4 stainless steel jeweler’s
screws implanted in the skull during catheter
Apparatus
implantation surgery.
Six operant chambers (Weiss & Schindler,
Cocaine (National Institute on Drug
1989) were enclosed in sound-attenuation
Abuse, Bethesda, MD) in a saline vehicle at a
chests (Weiss, 1970). Experimental events
concentration of 2.56 mg/ml was delivered at
were controlled by a MED Associates (St. Al-
a rate of 3.19 ml/min through Tygon tubing
bans, VT) computer system from an adjacent
wrapped in a metal spring. The tubing was
room where cumulative recorders were also
suspended through the ceiling from a 22
situated. Each chamber measured 20 cm
gauge rodent single-channel ?uid swivel (Al-
high, 23 cm long, and 18 cm wide, and was
ice King Chatham Medical Arts, Hawthorne,
dimly lighted at all times by a shielded 7.5-W
CA). Drug infusions were delivered by a MED
houselight operated at 3 W. The level of il-
Associates or Harvard Apparatus model 22 sy-
lumination created by this houselight was
ringe pump, using a 10-ml syringe. Pumps
enough to make the rat barely discernible,
were situated outside the sound-attenuation
but did not activate a photometer (Simpson
chests. The spring was attached to the plastic
408-2).
screw mounted on the rat’s head, reducing
Each chamber contained a lever operan-
tension on the catheter.
dum and food trough (not used in current
experiment) on the front wall. A response on
Calculation of Drug Levels
the lever closed a Gerbrands microswitch, re-
Whole-body drug levels were calculated for
quiring a force of 0.14 to 0.18 N (15 to 20
all sessions from which cumulative records
gm). Ambient noise with the exhaust fan run-
are presented. Calculations were based on
ning was measured at 70 dB (Realistic SPL
the sequential interinfusion intervals and the
meter). An approximately 2000-Hz, 85-dB
established pharmacokinetic pro?le of co-
tone was generated by a BRS AO-201 audio
caine, using an elimination half-life of 18.1
oscillator, ampli?ed by a BRS AA-201 ampli-
minutes (Barbieri, Ferko, DiGregorio, &
?er, and presented through an 8-Ohm, 20-cm
Ruch, 1992). According to the formula, Bn
speaker mounted in an enclosure 21.5 cm
(Bn 1
D) exp KT, where T
time since last
above the training chamber. There were two
infusion, K
0.0383 min-1 based on a half-
15-cm, 25-W, 120-V tubular light bulbs 10 cm
life of 18.1 minutes for cocaine, D
unit
behind the two translucent side walls which
dose in mg/kg, Bn 1
calculated amount of
provided the visual stimulus. These lights
drug in the body from previous infusions,
were operated at 120 V and produced 130.2
and Bn
calculated amount of drug in the
cd/m at the center of a side wall. For rats on
body at the time of a given infusion. Al-
the three-component mult VI VI EXT sched-
though pharmacokinetics can be affected by
ule, the tone and the light described above
many factors, this type of model should pro-
each served as an SD. For rats on the two-
vide adequate estimates of drug levels for the
component mult VI EXT schedule, the SD
purposes of the analyses performed here
was a tone-plus-light (TL) compound com-
(Lau & Sun, 2002).
posed of a 79db 2000-HZ tone and a 0.55 log
ft-Lamberts illumination produced by oper-
Procedure
ating the bulbs at 74V.
Two-component mult VI EXT. Initially, 10 rats
Intravenous catheters were implanted un-
were trained with a crf schedule where each
der ketamine (60 mg/kg) and xylazine (10
lever press was immediately followed by a 1.0
mg/kg) anesthesia using procedures de-
mg/kg infusion of cocaine. There was no ad-
scribed earlier (Panlilio, Weiss, & Schindler,
ditional stimulus paired with the drug deliv-

---

116
STANLEY J. WEISS et al.
ery, and there was no timeout period follow-
for each rat to produce moderate response
ing an infusion. TL was present for the entire
rates in TL. The ?nal VI values ranged from
session. Once lever pressing developed on
VI 45-s to VI 90-s.
this crf schedule, the dose per injection was
Three-component mult VI VI EXT. Training on
decreased from 1.0 mg/kg to 0.75 mg/kg
the three-component mult VI VI EXT sched-
and then to 0.5 mg/kg. Then, the response
ule for 4 rats proceeded in a fashion similar
requirements were gradually increased to
to that described above for the two-compo-
?xed-ratio (FR) 5 or FR 10 as the unit dose
nent mult VI EXT schedule, with the excep-
was decreased to 0.25 to 0.32 mg/kg. Once a
tion that VI components were signaled by
rat was reliably responding on this FR sched-
tone and by light rather than by TL. At the
ule (as judged by response patterns on the
outset of training on the three-component
cumulative records), a variable-interval (VI)
schedule, tone components and light com-
schedule was introduced that allowed a rat to
ponents (x¯
10 min; range, 5 to 20 min)
acquire cocaine at approximately the same
alternated with TL components (x¯
60 s;
rate that it had under the FR schedule. The
range, 40 to 90 s). The duration of the tone
initial VI schedules used ranged from VI 30-
and the light components were gradually re-
s to VI 90-s.
duced to a mean of 60 s (range, 30 to 120 s)
Unit doses were gradually reduced over the
over the course of training. Tone or light com-
course of training. The ?rst two self-admin-
ponents were equally likely to follow TL,
istered infusions of every session, however, re-
with the restriction that no more than two
mained at 1.0 mg/kg to (a) accelerate the
consecutive VI components were signaled by
cocaine ‘‘loading phase,’’ (b) reduce variabil-
the same stimulus (tone or light). For both
ity in early session responding, and (c) avoid
two- and three component multiple sched-
reinforcing a response incompatible with le-
ules, training sessions were generally 4 hr
ver pressing, which might occur under the
long.
commonly used priming procedure, where
drug is infused independently of responding
RESULTS
at the beginning of a session.
Two-Component Mult VI EXT
Once stable patterns of responding oc-
curred on the VI schedule, discrimination
By the third session, S-43’s unit dose was
training began. Now, the VI schedule contin-
reduced from 1.0 to 0.75 mg/kg and it began
ued to operate in TL, while extinction (EXT)
responding approximately once ever y 20
operated in TL. The TL components alter-
min. Over the next ?ve sessions, the unit dose
nated with TL components (x¯
90 s; range,
was gradually reduced to 0.32 mg/kg, and the
45 to 180 s). To reduce responding in TL, a
response requirement was gradually in-
lever press within the last 10 s scheduled of a
creased from crf to FR 10. Response rates in-
TL component delayed the presentation of
creased with each progressive decrease in
TL until 10 s (the response correction value)
dose and increase in response requirements.
passed without a response. The response-cor-
On the eighth session, the FR 10 schedule
rection contingency may contribute to the
was changed to a VI 90-s schedule. Cumula-
formation of a discrimination between the VI
tive record a in Figure 1 is from this session,
and EXT components, because (a) it pre-
where S-43 responded at a steady rate (ap-
vents responses in TL from being immediate-
proximately 6 responses/min), and there
ly followed by presentation of TL (which
were no postreinforcement pauses or bursts
might adventitiously reinforce responding in
of responding.
TL if TL becomes a conditioned reinforcer due
After one more training session on this VI
to its pairing with drug delivery), and (b) it
90-s schedule, mult VI 60-s EXT training be-
increases the likelihood that a response incom-
gan. In Figure 1, the cumulative record la-
patible with lever pressing in TL may be rein-
beled b is from S-43’s 12th session, where the
forced by presentation of TL. The response
unit dose was reduced to 0.25 mg/kg. A dis-
correction was increased, if necessary, up to
crimination was developing, although rates in
30 s. During discrimination training the unit
TL were too high. Mean rates in TL and TL
dose was gradually decreased to approximate-
components were 15.9 and 7.8 responses/
ly 0.2 mg/kg. The VI value was then adjusted
min, respectively.

---

STIMULUS CONTROL OF COCAINE SELF-ADMINISTRATION
117
Fig. 1.
Cumulative records from rats with mult VI EXT terminal baseline schedules. One-hour cumulative record
segments are presented from an early (a), intermediate (b), and terminal baseline training session (c) for S-43.
Record d is from a terminal baseline session for S-9. Tone-plus-light (TL) was present when the response pen was in
the upper register and absent (TL) when the response pen was in the lower register. Downward slash marks of the
response pen indicate an infusion. The ?rst three records show the responding of S-43 on a: (a) VI 90-s schedule
with a 0.32 mg/kg unit dose where TL was present for the entire session (Session 8), (b) mult VI 60-s EXT schedule
and a 0.25 mg/kg unit dose (Session 12), and (c) mult VI 90-s EXT schedule and 0.2 mg/kg unit dose (Session 32).
Record d is the cumulative record from S-9’s 17th session where it was on a mult VI 45-s EXT schedule with a 0.2
mg/kg/infusion dose. Corresponding calculated whole-body drug levels for S-9 are shown below its cumulative record.
Numbers in record d identify components referred to in the text. Session time is shown in seconds.
Response rates in TL were still higher than
sessions, S-43’s discrimination improved sub-
desired after three more sessions, so the re-
stantially. The rat, however, would occasion-
sponse correction was increased from 10 s to
ally emit response ‘‘bursts’’ in TL that would
20 s on the 16th session and to 30 s on the
carry over into the next TL component. To
22nd session (when the unit dose was de-
reduce the rate in TL, the VI schedule value
creased to 0.2 mg/kg). Over the next nine
was increased from VI 60-s to VI 90-s during

---

118
STANLEY J. WEISS et al.
Table 1
For each of the 10 rats trained on the mult VI EXT schedule: Number of sessions, response
rates (responses/min) in TL and TL, reinforcers/min, dose per infusion (mg/kg), and rate
of cocaine intake (mg/kg/hr) (all averaged over the last three training sessions). For each
rat except S-43 and S-44, the schedule in TL was VI 45-s, while drug was not available in TL
(mult VI 45-s EXT). For S-43 and S-44, the schedule in TL was VI 90-s and VI 60-s, respectively.
For subjects J-4, LD-3, LD-20, LD-18, LF-16, and LF-10, the mean TL and TL component
durations were 60 s (TL range, 30 to 120 s; TL range, 40 to 90 s). For subjects J-3 and S-9,
TL and TL component durations averaged 90 s (TL range, 45 to 180 s; TL range, 60 to 135
s). For S-43 and S-44, TL component durations averaged 120 s (range, 60 to 240 s) and TL
component durations averaged 75 s (range, 50 to 112.5 s) and 60 s (range, 40 to 90 s),
respectively. The response corrections in TL were 30 s for all subjects except J-3 and S-9, for whom
it was 10 s.
Responses/min
Reinforcers/
Unit dose
Rate of intake
Subject
Sessions
TL
TL
min
(mg/kg)
(mg/kg/hr)
J-3
41
5.9
0.3
0.42
0.2
5.0
J-4
33
4.8
0.2
0.37
0.2
4.4
S-9
18
9.1
0.2
0.45
0.2
5.4
LD-3
30
8.8
0.8
0.39
0.28
6.6
LD-20
18
8.4
1.0
0.42
0.25
6.3
LD-18
21
8.8
0.6
0.43
0.25
6.5
LF-16
20
11.0
0.9
0.43
0.28
7.2
LF-10
15
8.2
0.6
0.38
0.28
6.4
S-43
36
18.6
1.1
0.31
0.2
3.7
S-44
32
20.1
1.7
0.48
0.2
5.8
MEAN
26.4
10.4
0.7
0.41
0.23
5.7
the 30th session. Record c in Figure 1 is from
periments 1 through 3 are presented after Ex-
S-43’s 32nd session, revealing excellent stim-
periment 3.
ulus control. Responding (a) began promptly
Stimulus control comparable to that de-
with the onset of each TL component, (b)
scribed above for S-43 and S-9 was also
was sustained throughout the component, (c)
achieved in the 8 other rats trained on this
ceased abruptly when the TL component was
mult VI EXT schedule. Table 1 presents ter-
entered, and (d) did not resume until TL was
minal baseline data for each of these 10 rats.
presented again. During this session, S-43’s
All were discriminating between TL and TL.
response rate in TL (21.9 responses/min)
The mean response rate in TL was 15 times
was more than 20 times its response rate in
that in TL, with TL controlling rates between
TL (1.0 responses/min).
4.8 and 20.1 responses/min. The mean molar
Cumulative record d in Figure 1 is that of
rate of drug intake was 5.7 mg/kg/hr, with 7
Rat S-9 which was trained on a two-compo-
of the 10 rats within 1.0 mg/kg/hr of the
nent mult VI EXT schedule and whose train-
mean. The parameters of the terminal base-
ing history was comparable to S-43’s. Like S-
line schedule on which each of these rats
43’s cumulative record in c, S-9’s record
were trained are presented in the caption of
shows that responding began at the start of
Table 1.
each TL component and continued until TL
was presented, when responding ceased al-
Three-Component Mult VI VI EXT
most completely. Rat S-9 responded more
Initial lever press acquisition training was
than 40 times faster in TL (10.1 responses/
similar to that described above for the two-
min) than in TL (0.2 responses/min). Below
component mult VI EXT schedule. Cocaine
S-9’s cumulative record (d) in Figure 1 is a
(1.0 mg/kg/infusion) was available on a crf
plot of its calculated whole-body drug levels
schedule during 10-min tone or light compo-
during the portion of the session from which
nents that alternated with 60-s TL components
the cumulative record is taken. Details of this
where no cocaine was available. Over the ?rst
plot and all of the drug-level plots from Ex-
seven sessions, the unit dose was gradually de-

---

STIMULUS CONTROL OF COCAINE SELF-ADMINISTRATION
119
creased from 1.0 mg/kg to 0.25 mg/kg and
ing began promptly with the onset of a tone
the response requirement in tone and in
or a light component and was sustained
light was increased from crf to FR 3. Cumu-
throughout the duration of the component.
lative record a in Figure 2 is from S-16’s
Response rates remained low (1.3 responses/
eighth training session, where responding
min) during TL components.
was maintained by a mult FR-3 FR-3 EXT
Including S-16 and S-110, 4 rats were
schedule with a dose of 0.25 mg/kg/infusion.
trained with the three-component mult VI VI
Responses were reinforced according to an
EXT schedule, and all exhibited stimulus
FR 3 schedule when tone or light was present,
control of cocaine self-administration. Table
and went unreinforced in TL. The rat self-ad-
2 presents the criterion session data for these
ministered cocaine on this schedule through-
4 rats. The parameters of the terminal base-
out the session, indicating that the lever press
line schedule used for these rats are present-
operant had been ?rmly established.
ed in the caption of Table 2. For 3 of the 4
Cumulative record b in Figure 2 is from S-
rats, response rates in tone and in light were
16’s 12th session, where the schedule was
within 1.2 responses/min of each other. Al-
mult VI 60-s VI 60-s EXT. By this session, stim-
though response rates in VI components in
ulus control was beginning to develop. Typi-
Table 2 varied less (6.2 to 10.9 responses/
cally, many responses were emitted during
min) than with those rats exposed to the two-
tone and light components and little re-
component mult VI EXT (a range of 4.8 to
sponding occurred during TL. There were
20.1 responses/min) shown in Table 1, the
still tone and light components where re-
rates in Table 2 overlap those of 80% of the
sponding was not steady, however, as well as
rats in Table 1. Molar rates of drug intake
some responding during TL components.
were consistent across rats within the three-
Over subsequent sessions, the mean length
component mult VI VI EXT group. All these
of the tone and light components was grad-
rats self-administered between 5.7 and 7.1
ually reduced from 10 min to 60 s (range, 30
mg/kg/hr, rates comparable to those of the
to 120 s) and a 10-s response correction con-
mult VI EXT rats in Table 1.
tingency was added to the TL component.
The response correction value was increased
to 20 s when response rates in TL continued
EXPERIMENT 2: MULTIPLE VI VI DRO
to be higher than required to meet the dis-
Like the mult VI VI EXT schedule just de-
crimination criterion of 7:1. Cumulative re-
scribed, the mult VI VI DRO schedule was
cord c in Figure 2 is from S-16’s 50th session
designed to produce moderate and stable
(see ?gure caption for schedule parameters).
rates of responding in tone and in light com-
Responding (a) began with the onset of each
ponents and response cessation in TL com-
tone or light component, (b) was sustained
ponents. On the mult VI VI DRO schedule,
at a moderately high rate for the duration of
however, reinforcers were received in all com-
the component, (c) abruptly ceased in TL,
ponents (tone, light, and TL), but on the
and (d) did not resume until the tone or light
mult VI VI EXT schedule, where VI and EXT
was presented. Rates in tone (6.6 responses/
components alternated, reinforcers were re-
min) and in light (6.6 responses/min) were
ceived in only half the components (tone and
13 times those in TL (0.5 responses/min).
light but not in TL). Therefore, the overall
Cumulative record d in Figure 2 shows the
rate of reinforcement (infusions/min) on the
performance of Rat S-110, trained on a mult
mult VI VI DRO schedule would be substan-
VI 45-s VI 45-s EXT schedule. This rat’s train-
tially higher than the overall rate of reinforce-
ing history was similar to S-16’s, with the pa-
ment on the mult VI VI EXT schedule. To
rameters of the schedule presented in record
compensate for this, the unit dose at each
d of Figure 2 identical to S-16’s, presented in
stage of training on the mult VI VI DRO
record c of Figure 2. Record d in Figure 2
schedule was lower than the unit dose used
reveals that in this session S-110 responded at
at corresponding stages of training on the
a faster rate (a mean of 10.2 and 16.0 re-
mult VI VI EXT schedule. For example, the
sponses/min in tone and light, respectively)
initial unit dose used with the mult VI VI
than S-16, but the quality of the stimulus con-
DRO schedule was 0.6 mg/kg while the initial
trol was comparable for the 2 rats. Respond-
dose used with the mult VI VI EXT schedule

---

120
STANLEY J. WEISS et al.
Fig. 2.
Cumulative records from rats with mult VI VI EXT terminal baseline schedules. One-hour cumulative
record segments are presented from an early (a), intermediate (b), and terminal baseline training session (c) for S-
16. Record d is from a terminal baseline session for S-110. Tone or light was present when the response pen was in
the upper register (tones are indicated by ?lled circles) and both were absent (TL) when the response pen was in
the lower register. Downward slash marks on the response pen indicate an infusion. The ?rst three records show the
responding of S-16 on a (a) mult FR-3 FR-3 EXT schedule with 0.25 mg/kg unit dose (Session 8), (b) mult VI 60-s
VI 60-s EXT schedule with a unit dose of 0.25 mg/kg (Session 12), and (c) mult VI 45-s VI 45-s EXT schedule with
a 0.25 mg/kg unit dose (Session 15). Record d is the cumulative record from S-110’s 66th session where it was on a
mult VI 45-s VI 45-s EXT schedule with a 0.25 mg/kg unit dose. Corresponding calculated whole-body drug levels
for S-110 are shown below its cumulative record. Letters in record d identify portions of the record referred to in
the text. Session time is shown in seconds.
was 1.0 mg/kg. On the terminal baseline, a
METHOD
0.08 mg/kg/infusion dose was used for rats
Subjects and Apparatus
trained on the mult VI VI DRO schedule
Two adult male Sprague-Dawley rats
while a 0.25 to 0.32 mg/kg/infusion dose was
(housed and deprived like the rats in Exper-
used on the terminal baselines for rats
iment 1) were used. Maintenance conditions
trained on the mult VI VI EXT schedule. Pro-
for the rats, surgical procedures, apparatus,
cedures effective in producing stimulus con-
and drug level calculations were also the
trol of cocaine self-administration on a mult
same as described in Experiment 1.
VI VI DRO schedule will be illustrated by the
training history of Rat S-11, with the terminal
Procedure
baseline stimulus control of S-21 also pre-
Initially, rats were trained on a mult crf crf
sented.
DRO 15-s schedule. Each lever press in tone

---