BEHAVIORAL AND PHARMACOLOGICAL VARIABLES AFFECTING RISKY CHOICE IN RATS

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR
2001, 75, 275–297
NUMBER 3 (MAY)
BEHAVIORAL AND PHARMACOLOGICAL VARIABLES
AFFECTING RISKY CHOICE IN RATS
BARBARA J. KAMINSKI AND NANCY A. ATOR
JOHNS HOPKINS UNIVERSIT Y SCHOOL OF MEDICINE
The effects of manipulations of response requirement, intertrial interval (ITI), and psychoactive
drugs (ethanol, phencyclidine, and d-amphetamine) on lever choice under concurrent ?xed-ratio
schedules were investigated in rats. Responding on the ‘‘certain’’ lever produced three 45-mg pellets,
whereas responding on the ‘‘risky’’ lever produced either 15 pellets (p
.33) or no pellets (p
.67). Rats earned all food during the session, which ended after 12 forced trials and 93 choice trials
or 90 min, whichever occurred ?rst. When the response requirement was increased from 1 to 16
and the ITI was 20 s, percentage of risky choice was inversely related to ?xed-ratio value. When only
a single response was required but the ITI was manipulated between 20 and 120 s (with maximum
session duration held constant), percentage of risky choice was directly related to length of the ITI.
The effects of the drugs were investigated ?rst at an ITI of 20 s, when risky choice was low for most
rats, and then at an ITI of 80 s, when risky choice was higher for most rats. Ethanol usually decreased
risky choice. Phencyclidine did not usually affect risky choice when the ITI was 20 s but decreased
it in half the rats when the ITI was 80 s. For d-amphetamine, the effects appeared to be related to
baseline probability of risky choice; that is, low probabilities were increased and high probabilities
were decreased. Although increase in risky choice as a function of the ITI is at variance with previous
ITI data, it is consistent with foraging data showing that risk aversion decreases as food availability
decreases. The pharmacological manipulations showed that drug effects on risky choice may be
in?uenced by the baseline probability of risky choice, just as drug effects can be a function of baseline
response rate.
Key words: drug effects on risky choice, intertrial interval and risky choice, ?xed-ratio schedule and
risky choice, foraging, risk taking, lever press, rat
When choosing between two alternatives
should be risk averse. Under conditions in
differing in amount and probability of rein-
which reinforcement is scarce, animals
forcement, an organism is said to be risk
should be risk prone. For the most part, an-
prone if it prefers a larger, probabilistic re-
imal research that has studied the effects of
inforcer and risk averse if it prefers a smaller,
the daily energy budget rule has supported
certain reinforcer (Mazur, 1988). According
this prediction when amount of food avail-
to risk-sensitive foraging theories, a major,
able from each source was manipulated (Ka-
perhaps the most important, determinant of
celnik & Bateson, 1996), but studies of risky
risky choice is reinforcer availability (Mazur,
choice in birds have supported the prediction
1988). In particular, under conditions in
more consistently than have those in rats (re-
which reinforcement is plentiful, risk-sensi-
view in Kacelnik & Bateson). Data inconsis-
tive foraging theories predict that animals
tent with the prediction were reported for
rats by Hastjarjo, Silberberg, and Hursh
This research was supported by Grant DA04133 award-
(1990).
ed by the National Institute on Drug Abuse. The studies
Hastjarjo et al. (1990) adapted the proce-
were conducted according to the principles in the Guide
dures used in studies of risk taking in hu-
for Care and Use of Laboratory Animals, NIH Publication 85-
mans, namely repeated-gambles experiments
23, and were approved by the Johns Hopkins School of
Medicine Animal Care and Use Committee. Portions of
(Silberberg, Murray, Christensen, & Asano,
these ?ndings were reported at the meetings of the As-
1988). Rats were presented with concurrent
sociation for Behavior Analysis in 1995 (Washington, DC)
?xed-ratio (FR) 1 FR 1 schedules. They
and 1997 (Chicago). The expert assistance of Donna Pyle
earned all of their daily food ration during
in data collection, Susan James in ?gure preparation, and
Jennifer Martin in manuscript preparation is gratefully
the experimental session. In Experiment 1,
acknowledged.
each response on one lever produced three
Requests for reprints should be sent to Nancy A. Ator,
pellets and a response on the other lever pro-
Behavioral Biology Research Center, Johns Hopkins Bay-
duced either 15 pellets (p
.33) or no pellets
view Campus, 5510 Nathan Shock Drive, Suite 3000, Bal-
timore, Maryland 21224-6823 (E-mail: ator@mail.
(p
.67). Overall reinforcement was more
jhmi.edu).
plentiful for consistent responding on the
275

---

276
BARBARA J. KAMINSKI and NANCY A. ATOR
risky lever, because the mean number of pel-
next with respect to the pharmacological ma-
lets available from the risky alternative was
nipulations conducted under different base-
greater (?ve pellets per trial) than the num-
line conditions.
ber of pellets available from the certain alter-
The ?rst behavioral variable manipulated
native (three pellets per trial). To study the
was response requirement under an FR
effects of reinforcer availability, Hastjarjo et
schedule. When the animal has to do more
al. manipulated the number of trials per ses-
work to obtain reinforcers (i.e., to obtain ac-
sion. As the number of trials per session de-
cess to each ‘‘patch’’ in foraging terminolo-
creased, the rats’ choice behavior became
gy), it may be that food effectively becomes
more risk averse (i.e., they were less likely to
more scarce, which might decrease risk aver-
choose the risky lever), and the rats lost
sion. Previous research with birds that manip-
weight. This result was not consistent with the
ulated number of responses under concur-
prediction that animals should be risk prone
rent schedules of reinforcement (i.e., not
when the available food supply is insuf?cient
discrete-trials procedures) found behavior to
to meet caloric needs, and suggested that oth-
be risk prone, regardless of response require-
er variables may play a role in in?uencing
ment (Ha, 1991; Ha, Lehner, & Farley, 1990).
risky choice. Hastjarjo et al. speculated that
The second behavioral variable manipulat-
increased day-to-day variability in number of
ed was intertrial interval (ITI). Because we
pellets earned when the number of trials per
held session duration constant, increasing the
session decreased produced risk aversion,
ITI effectively decreased trials per session, as
which overcame the risk proneness to be ex-
in Hastjarjo et al. (1990), but it also de-
pected from the reduction in the total num-
creased reinforcement density (per unit
ber of reinforcers received. In a second ex-
time) and thus increased delay to reinforce-
periment, they held total trials per session
ment. Experiments that have manipulated
and relative probability of reinforcement on
variability in delay to reinforcement have
the two levers constant but manipulated total
found almost uniformly that animals showed
amount of food available in each session.
risk-prone behavior (Kacelnik & Bateson,
They found that risk aversion was inversely
1996). Kacelnik and Bateson’s analysis of the
related to the total amount of food available.
contribution of various time components of a
The Hastjarjo et al. (1990) data exemplify
laboratory model of foraging suggested to
a point made by Kacelnik and Bateson
them that delay to reinforcement once the
(1996), which is that the occurrence and di-
response requirement has been met is a
rection of risk-sensitive choice can be in?u-
enced by apparently small changes in exper-
greater determinant of the effect of variability
imental procedures. Thus behavior generated
in delay to reinforcer delivery than is ITI. Few
under the procedure introduced by Hastjarjo
studies have manipulated ITI systematically,
et al. seemed likely to provide a sensitive base-
however, and none appear to have done so
line for studying other variables that might
in the rat (see review by Kacelnik & Bateson).
in?uence risky choice. We initially were inter-
In the pharmacological manipulations, we
ested in employing this behavioral baseline in
investigated the effects of the sedative drug
the course of our behavioral pharmacology
ethanol, the dissociative anesthetic phencycli-
research to determine whether psychoactive
dine (PCP), and the stimulant d-amphet-
drugs that have been considered to increase
amine, all of which have been considered to
the probability of risky behavior in humans
promote risk-taking behavior under some cir-
would increase the probability of risky choice
cumstances in humans (Goodman, Rall, Nies,
in this rat model. After we failed to ?nd
& Taylor, 1990). The effects of these drugs
changes in risky choice following drug ad-
on schedule-controlled responding in rats
ministration, we made systematic behavioral
and other species have been studied exten-
manipulations with the hope of changing the
sively (Dews & Wenger, 1977; McMillan & Le-
baseline probability of risky choice. Because
ander, 1976) but, to our knowledge, they
the results of the behavioral manipulations
have not been studied on behavior generated
were themselves relevant to risk-sensitive for-
under a risky choice procedure. Dose–effect
aging theory, we present the data ?rst with
relations were studied during two ITI condi-
respect to the behavioral manipulations and
tions, 20 and 80 s, which produced lower and

---

VARIABLES AFFECTING RISKY CHOICE
277
higher probabilities of risky choice, respec-
Procedure
tively, in most rats.
After pressing both levers was shaped using
the method of successive approximations, ex-
METHOD
perimental sessions for all 6 rats were con-
ducted at the same time of the afternoon
Subjects
Monday through Friday. The rats were placed
Six experimentally naive, Long-Evans
in the experimental chambers for a 10-min
hooded male rats (Harlan Sprague-Dawley),
presession timeout during which the cham-
received at 6 weeks of age, served as subjects.
ber was dark and responses on the levers had
They were housed individually under a 12:12
no effect. At the end of the presession time-
hr nonreversed light/dark cycle with contin-
out, the light over one lever was illuminated
uous access to water and with food freely
and the ?rst of 12 forced-choice trials began.
available. At the beginning of habituation to
During forced-choice trials, only a response
the chambers, weights ranged from 241 to
(FR 1) on the lever below the illuminated
279 g. Feeding was restricted during shaping
jewel light could produce food pellets. For
of the lever press, but weight gain still was
each rat, one lever was designated the certain
permitted. By the beginning of the risky
lever and the other lever the risky lever. A
choice procedure itself, weights ranged from
response on the certain lever always pro-
324 to 347 g, which is near the range (330 to
duced three pellets, and a response on the
340 g) at which we typically maintain body
risky lever produced either 15 pellets (p
weights for adult male rats of this strain dur-
.33) or no pellets (p
.67). The response
ing behavioral studies (Ator, 1991). Under
turned off the jewel light and initiated the
the risky choice procedure, rats earned all
delivery of food pellets, at a rate of 1 per sec-
food (45-mg Noyes Precision rat pellets) dur-
ond, if they were available on that trial. Re-
ing the experimental sessions. When sessions
sponses on the other lever had no conse-
were not conducted, the rats were fed 15 g
quence but were counted. A 20-s ITI began
of commercial laboratory rat chow, which typ-
following the delivery of the last pellet or fol-
ically results in weight stability or weight gain
lowing the response if no pellet was sched-
on the day the rat next is weighed.
uled to be delivered on that trial. During the
ITI, the jewel lights were dark; responses on
Apparatus
the levers had no effect but were counted. In
Six experimental chambers (27.7 cm by
the forced-choice trials, the probability of a
30.3 cm by 53.2 cm) were used, and each rat
risky trial or a certain trial was .5, with the
was assigned to one of the chambers for the
restriction that six trials had to be with the
duration of the study. The apparatus is shown
risky lever and six with the certain lever. The
in Ator (1991, p. 29). Two Gerbrands rodent
subsequent trials all were free-choice trials,
levers (G6312) were mounted 13 cm apart
on which the jewel lights above both levers
and 5 cm above the ?oor; identically colored
were lit and a concurrent FR 1 FR 1 schedule
jewel lights were mounted 5 cm above the le-
was in effect. The session ended after 93 free-
vers, one over each. A food cup was centered
choice trials or 90 min, whichever occurred
on the opposite wall 2.5 cm above the ?oor,
?rst.
into which a Gerbrands pellet feeder (G5100)
Before the ?rst experimental manipula-
delivered the 45-mg food pellets. The exper-
tion, 20 baseline sessions were conducted.
imental enclosure was housed inside a larger
The ?rst manipulation was to reverse the con-
sound-attenuating chamber. A ventilation fan
tingencies across the two levers to determine
and white noise, delivered through a speaker
whether each rat’s preference for either the
in the enclosure, masked extraneous sounds.
risky or the certain contingency would track
Experimental control and data collection
the change in lever assignment. After there
were accomplished via an IBM-compatible
were no increasing or decreasing trends in
386SX microcomputer programmed in the
percentage of risky choice for at least ?ve ses-
MedState Notation Language and interfaced
sions, the contingencies again were reversed.
to the chamber with Med Associates solid
The other experimental manipulations were
state components.
carried out in the following order: effects of

---

278
BARBARA J. KAMINSKI and NANCY A. ATOR
drugs at ITI 20 s; FR manipulations; ITI ma-
and the beginning of the session (see below).
nipulations; effects of drugs at ITI 80 s. The
Drug testing occurred on Tuesdays and Fri-
behavioral manipulations are grouped to-
days, provided that the percentage of risky
gether ?rst and the pharamacological sec-
choices in the preceding session fell within
ond, however, for ease of presentation and
the range of percentages of risky choices in
discussion.
the last ?ve sessions of the baseline before
FR manipulations. To investigate the effects
drug testing began, which usually occurred,
of work requirement, the FR values for both
so that testing did not have to be delayed.
levers were increased in tandem across ses-
The order in which the drugs were studied
sions. All rats were exposed to FR values of 1,
during the 20-s ITI condition was ethanol, d-
2, 8, and 16, in ascending order, followed by
amphetamine, and PCP; during the 80-s ITI
replication of FR 1 and then FR 16. Three of
condition, it was d-amphetamine, PCP, and
the 6 rats, which were ahead of the other 3
ethanol. Each dose usually was studied once
in moving through the conditions, also were
in the 20-s ITI condition, because there was
exposed to FR 4 and FR 12; their data did
little general effect of the drugs. Doses often
not indicate value in studying FR 4 and 12
were repeated in the 80-s ITI condition be-
with the other rats. Each FR condition lasted
cause more effects were observed.
for at least three sessions and until there were
Ethanol was administered intragastrically
no increasing or decreasing trends in the per-
15 min before the session, by use of a curved
centage of trials on which the risky lever was
20-gauge, 3-in., 2.25-mm tip, stainless-steel
chosen. The number of sessions until this cri-
feeding needle. An ethanol stock solution was
terion was met was usually three to six; at FR
prepared by adding distilled water to 95%
2, it was seven for Rats 36-3 and 36-5 and nine
weight/volume (w/v) ethanol (obtained
for Rat 36-3. The median number of sesssions
from the Johns Hopkins Hospital pharmacy)
in the FR conditions was 4.5.
to make a 15% w/v solution. Volumes of in-
ITI manipulations. Following the FR manip-
jection ranged between 1 and 5 ml, depend-
ulations, all rats were exposed to ITI lengths
ing upon the amount needed to achieve the
ranging from 20 s to 120 s in ascending order.
desired total dose. A 20% ethanol solution
Then the 20-s and the 80-s ITI conditions
was used if the total volume of ?uid would
were replicated. Each ITI condition lasted at
have exceeded 5 ml, which occurred for Rats
least three sessions and until there were no
36-2 and 36-4 in the 20-s ITI condition and
increasing or decreasing trends in the per-
for Rat 36-6 in the 80-s ITI condition. Vehicle
centage of trials on which the risky lever was
was injected at both 1- and 5-ml volumes in
chosen. The number of sessions until this cri-
the 20-s ITI condition to control for the
terion was met in each ITI condition ranged
range of volumes of injection to which the
from a low of 3, 4, or 5 to a high of 10, 11,
rats might be exposed. Phencyclidine hydro-
13, or 14, except that it was 7 to 16 for ITI
chloride (National Institute on Drug Abuse)
60 s and 8 to 18 for the replication of ITI 20
and d-amphetamine sulfate (Sigma) were ad-
s. The median number of sessions in the ITI
ministered by intraperitoneal injection (27-
conditions was 7.5.
gauge 0.5-in. needle) 1 min and 15 min be-
Pharmacological manipulations. Pharmaco-
fore the session, respectively. Both drugs were
logical manipulations were conducted under
dissolved in 0.9% sodium chloride solution at
the concurrent FR 1 FR 1 condition at two
concentrations appropriate for injection of
ITI lengths (20 s and 80 s) to determine drug
each dose in a volume of 1 ml/kg. The pla-
effects on different baseline rates of risky
cebo controls for the latter two drugs were 1
choice. Study of drug effects did not begin
ml/kg of the vehicle.
under either ITI condition until after at least
The percentages of risky choice were cal-
10 baseline sessions and no increasing or de-
culated for each session by dividing the total
creasing trend in percentage of risky choice
number of choice trials completed on the le-
over the last ?ve. On test days, the drug was
ver that provided food pellets on a probabi-
given and the rat was placed in the chamber
listic basis by the total number of choice trials
for the presession period, which was length-
completed in that session. The rate of trial
ened or shortened to correspond to the in-
completion in each session was calculated by
terval desired between drug administration
dividing the total number of trials completed

---

VARIABLES AFFECTING RISKY CHOICE
279
Fig. 1.
Percentage of total lever choices that were for the lever on which responding produced 15 or 0 food
pellets per reinforcement with a probability of .33 and .67, respectively, versus the lever on which an equal number
of responses produced three pellets with a probability of 1.0. Responses on the former lever are characterized as the
risky choice. For the 3 rats for which data are presented in the left set of panels, the risky choice initially was the left
(L) lever; it was the right (R) lever for the 3 rats for which data are presented in the right set of panels. The
contingencies were then reversed (middle section of each panel) and ?nally were returned to the original condition
for each rat.
(including the 12 forced-choice trials) by the
baseline condition, each rat’s preference for
total session time minus the time spent in the
the risky or the certain lever tracked the con-
ITI.
tingency when the contingencies were re-
versed between levers and again when they
changed back to the original levers. As seen
RESULTS
in Figure 1, 5 of the 6 rats responded pre-
Lever-Reversal Condition
dominantly on the certain lever across the 20
In the lever-reversal condition carried out
sessions before the contingencies were re-
after training under the FR 1 and ITI 20-s
versed (i.e., risky choice was less than 50% for

---

280
BARBARA J. KAMINSKI and NANCY A. ATOR
Rats 36-1 to 36-5 and usually was less than
had 20% or more risky choices at FR 1, but
30%). When the risky lever changed from be-
percentages were not as high as they had
ing the left to the right lever for Rats 36-1,
been in the previous FR 1 condition. When
36-3, and 36-5 and from the right to the left
the FR 16 condition was replicated, none of
lever for Rats 36-2 and 36-4, the former 3 rats
the 6 rats made the risky choice.
persisted in responding on the right lever
Body weights across the last three sessions
and the latter 2 rats persisted in responding
of each FR condition are given in Table 1.
on the left lever, even though this resulted in
There was a slight tendency for weight to de-
fewer pellets, but this behavior occurred only
crease at the higher FR values, as the per-
in the ?rst one or two sessions for most rats
centage of risky choice decreased, for Rats 36-
before risky choice decreased. The rat (36-6)
2, 36-4, 36-5, and 36-6, but the decreases were
that had a high probability of risky choice be-
small. The largest percentage decrease was
fore the reversal of contingencies persisted in
3%.
responding on the right lever after the rever-
Session duration, choice trials completed
sal, which dramatically reduced baseline per-
(if less than the total available), and rate of
centage of risky choice from about 80% to
trial completion are provided in Table 2. Dur-
about 20%. When the contingencies were re-
ing the ?rst FR 1 condition and at FR 2, half
turned to the former lever assignments, all
the rats (36-1, 36-2, and 36-6) failed to com-
rats persisted in responding predominantly
plete all 93 available choice trials within the
on the certain lever, which increased per-
90-min maximum session duration in one or
centage of risky choice; but this time, perse-
two of the three sessions at each FR shown in
verance on the risky lever continued for more
Figure 2, but they completed 53 to 91 trials.
sessions than previously for most rats, and
Otherwise, sessions were completed in 45 to
risky choice remained higher than in the pre-
82 min (medians were 52 to 54 min). As the
vious conditions for Rats 36-1, 36-2, and 36-6.
FR requirement increased, session duration
Throughout the remainder of the experi-
did not systematically increase for all rats;
mental conditions, to be described below, the
they usually completed all 93 choice trials
?gures will show that percentages of risky
within 60 to 80 min.
choice under FR 1 ITI 20 s tended to remain
in the same relative range as shown in the
ITI Manipulations
?nal condition of Figure 1 during the phar-
Figure 3 shows the percentage of choices
macological manipulations under this set of
of the risky lever for the last three sessions at
parameters for all rats. By the behavioral ma-
each ITI duration. When the ITI was the
nipulations though, as will be seen, the range
shortest (20 s), all rats preferred the certain
shifted down for Rat 36-1 and up for Rat 36-5.
lever. That is, they chose the risky alternative
on less than 50% of the trials in most sessions,
FR Manipulations
and 3 of the rats (36-1, 36-3, and 36-4) chose
Figure 2 shows the percentage of choices
the risky alternative on less than 20% of the
of the risky lever when FR value was manip-
trials. At most, responses were distributed
ulated. In the last three sessions of the ?rst
roughly equally between levers in some ses-
FR 1 condition, all rats usually chose the risky
sions.
alternative on fewer than 50% of the trials,
As the ITI increased from 20 to 60 s, per-
and 4 of the 6 never chose the risky lever on
centage of risky choice increased for 3 rats,
more than 30% of the trials. Only Rat 36-5
decreased for 1 rat (36-5), and remained less
chose the risky alternative approximately
than 20% for 2 rats (36-1 and 36-3). As the
equally as often as the certain lever. As FR
ITI increased from 60 to 120 s, however, per-
value increased, the relative frequency of
centage of risky choice increased for 5 of the
choosing the risky lever decreased. At FR 8
6 rats (albeit the percentage for Rat 36-3 nev-
and FR 16, the percentage of risky choice in
er was above 10%); it remained between 80%
the three sessions shown was usually zero but
and 100% for the 6th rat (36-6). When the
was below 10% for all 6 rats.
ITI was 120 s, 5 of the 6 rats chose the risky
When the FR value returned to 1, percent-
alternative on more than 50% of the trials,
age of choices of the risky lever increased for
and for 3 of the 5, risky choice was above
those 3 rats (36-2, 36-4, and 36-5) that initially
80%.

---

VARIABLES AFFECTING RISKY CHOICE
281
Fig. 2.
Percentage of risky choices under a concurrent FR FR schedule of food delivery as a function of FR value.
Each panel shows data for 1 of 6 rats across the last three sessions at the FR value at the top of each column. The y
axis has been truncated at 60%. The FR values were studied in the order given across the top of the ?gure. Data are
missing in some columns because not all rats were studied under each FR value.

---

282
BARBARA J. KAMINSKI and NANCY A. ATOR
Table 1
percentages in the initial 20-s ITI condition.
Mean body weight (in grams) in the last three sessions
When the 80-s ITI condition was replicated,
of sequential FR conditions.
percentage of risky choice increased to a
range that was either approximately the same
Rat
as or higher than the range shown for the
FR
36-1
36-2
36-3
36-4
36-5
36-6
previous 80-s condition.
1
346
374
374
416
374
374
Number of choice trials completed, session
2
347
372
371
406
373
375
duration (if less than 90 min), and rate of
4
345
405
372
trial completion are provided in Table 3. In
8
343
367
375
404
367
365
12
345
404
365
the ?rst 20-s ITI condition, almost all rats
16
349
364
374
406
368
364
completed the 93 free-choice trials that could
1
349
365
381
405
372
373
be completed in each session. This was also
16
347
363
381
407
371
348
true for most sessions at ITI 40 s. The num-
ber of choice trials completed per session
necessarily decreased as the ITI increased fur-
When the ITI was shortened to 20 s again,
ther (although rats usually completed the
the percentage of risky choice decreased. All
maximum possible), and the number of pel-
rats chose the risky alternative on fewer than
lets earned also decreased. Figure 4 shows the
50% of trials. For 5 of the 6 rats, the per-
mean number of pellets earned during the
centages in the replication of the 20-s ITI
last three sessions at each ITI. The rat that
condition were the same as or close to the
did not show greater than 10% risky choice
Table 2
Fixed-ratio (FR) value manipulations: Session durations (in minutes) and rate of trial com-
pletion (trials per minute excluding time in intertrial interval) for the last three sessions at
each FR value for each rat for the sessions shown in Figure 2.
Rat
36-1
36-2
36-3
36-4
36-5
36-6
FR
Min
Rate
Min
Rate
Min
Rate
Min
Rate
Min
Rate
Min
Rate
1
90a
1.6
54
5.5
47
8.8
54
5.5
90b
1.7
90
ND
57
4.8
59
4.4
51
6.6
59
4.4
80
2.3
62
3.9
90c
1.9
90d
1.0
51
6.6
51
6.6
90e
1.2
47
8.8
2
90f
1.6
90g
1.7
45
10.5
58
4.6
82
2.2
90
ND
46
9.5
52
6.2
46
9.5
55
5.2
90g
1.7
49
7.5
90h
1.8
63
3.8
67
3.3
53
5.8
90i
1.8
52
6.2
4
53
5.8
54
5.5
55
5.2
90j
1.3
53
5.8
54
5.5
51
6.6
52
6.2
63
3.8
8
90g
1.7
55
5.2
73
2.8
53
5.8
65
3.5
83
2.2
78
2.4
51
6.6
69
3.1
50
7.0
84
2.1
71
2.9
54
5.5
48
8.1
57
4.8
48
8.1
65
3.5
72
2.8
12
70
3.0
51
6.6
90h
1.8
62
3.9
52
6.2
90k
1.9
76
2.6
52
6.2
89
1.9
16
79
2.4
56
5.0
72
2.8
55
5.2
78
2.4
88
2.0
70
3.0
61
4.0
72
2.8
54
5.5
73
2.8
85
2.1
67
3.3
54
5.5
70
3.0
56
5.0
64
3.6
90l
1.5
1
56
5.0
68
3.2
51
6.6
55
5.2
79
2.4
82
2.2
50
7.0
63
3.8
43
13.1
44
11.7
63
3.8
61
4.0
52
6.2
77
2.5
44
11.7
47
8.8
53
5.8
53
5.8
16
67
3.3
53
5.8
61
4.0
52
6.2
64
3.6
80
2.3
63
3.8
54
5.5
65
3.5
54
5.5
64
3.6
85
2.1
64
3.6
53
5.8
68
3.2
56
5.0
71
2.9
90m
1.6
Note. ND
no data (session-duration data lost, so rate could not be calculated). The number of free-choice trials
was 93 unless otherwise indicated by a footnote, as follows: a 80, b 87, c 91, d 53, e 64, f 82, g 85, h 89, i 88, j 70, k 92,
l 78, m 83.

---

VARIABLES AFFECTING RISKY CHOICE
283
Fig. 3.
Percentage of risky choices under a concurrent FR 1 FR 1 schedule of food delivery as a function of ITI
duration. Each panel shows data for 1 of 6 rats across the last three sessions at the ITI value shown at the top of each
column. The ITI values were studied in the order shown across the top of the ?gure.
(Rat 36-3) experienced the greatest decrease
body weight did not decrease until the 80-s
in mean number of pellets earned per session
ITI condition.
at ITI values of 80 to 120 s.
Figure 5 shows mean body weights during
Ethanol
the last three sessions at each ITI value. As
Figure 6 shows that during the 20-s ITI con-
ITI increased, body weight decreased for all
dition, baseline percentage of risky choice
rats. Decreases in body weight ?rst appeared
was less than 20% for 2 rats (36-3 and 36-4)
at ITI 60 s for Rats 36-1, 36-3, and 36-5 and
and was between 15% and 65% for the other
at ITI 80 s for Rats 36-2, 36-4, and 36-6. They
4 rats. Ethanol decreased risky choice below
increased upon return to the 20-s ITI for all
the baseline range and below the vehicle con-
6 rats and then decreased again, if only slight-
trol in the 4 rats for which it was possible to
ly, for 5 of the 6 rats during replication of the
detect decreases in this measure. The highest
80-s ITI condition. The ITI at which body-
ethanol dose (3 g/kg) decreased risky choice
weight changes ?rst occurred for each rat did
to zero or virtually zero in all 4 of them; for
not correspond with the ITI at which increas-
2 rats (36-1 and 36-5), risky choice decreased
es in risky choice ?rst occurred, and the or-
also at lower ethanol doses. Only Rat 36-6
der of these changes was not the same across
showed an increase, albeit slight, after etha-
rats. For Rat 36-1, for example, body weight
nol administration.
decreased in the 60-s ITI and risky choice in-
Use of the 80-s ITI resulted in a higher
creased in the 80-s ITI; but for Rat 36-2, risky
baseline range of percentages of risky choice
choice increased in the 40-s ITI condition but
than the 20-s ITI for 4 rats (36-2, 36-4, 36-5,

---

284
BARBARA J. KAMINSKI and NANCY A. ATOR
Table 3
Intertrial-interval (ITI) manipulations: Total number of free-choice trials and overall rate of
trial completion (total free- and forced-choice trials per minute excluding time in ITI) for
the last three sessions at each ITI for each rat for the sessions shown in Figure 3.
Rat
36-1
36-2
36-3
36-4
36-5
36-6
ITI (s)
No.
Rate
No.
Rate
No.
Rate
No.
Rate
No.
Rate
No.
Rate
20
93a
5.2
93b
4.6
93c
10.5
93d
17.5
78
1.5
90
1.8
93e
7.0
84
1.7
93f
13.1
93g
15.0
93h
2.3
93i
4.2
93j
2.1
93k
6.2
93c
10.5
93c
10.5
93l
2.0
93i
4.2
40
93m
9.5
93m
9.5
93n
15.0
93o
7.5
92
5.0
86
4.0
93p
13.1
93q
5.5
93n
15.0
93r
5.8
93q
5.5
91
4.8
87
4.1
67
2.1
93n
15.0
89
4.5
93o
7.5
93l
6.2
60
70
10.2
64
5.4
71
11.9
68
8.0
67
7.2
58
3.5
69
9.0
69
9.0
72
14.0
71
11.9
71
11.9
61
4.3
70
10.2
67
7.2
71
11.9
67
7.2
71
11.9
56
3.1
80
50
8.5
50
8.5
51
10.5
50
8.5
49
7.0
45
4.1
50
8.5
50
8.5
50
8.5
50
8.5
51
10.5
41
2.7
47
5.2
50
8.5
51
10.5
50
8.5
38
2.1
41
2.7
120
27
3.2
30
7.0
30
7.0
30
7.0
30
7.0
26
2.7
29
5.1
29
5.1
30
7.0
30
7.0
29
5.1
25
2.3
29
5.1
29
5.1
30
7.0
30
7.0
30
7.0
25
2.3
20
92
1.9
93s
3.3
93k
6.2
93t
3.4
77
1.5
93u
3.9
93v
3.8
93n
2.5
93s
3.3
93w
5.0
90
1.8
93x
4.8
84
1.7
77
1.5
93k
6.2
93y
5.5
93z
2.2
93aa
3.1
80
48
6.0
50
8.5
51
10.5
50
8.5
42
3.0
40
2.5
47
5.2
50
8.5
51
10.5
50
8.5
42
3.0
43
3.3
46
4.6
45
4.1
51
10.5
49
7.0
48
6.0
40
2.5
Note. Session duration was 90 min unless otherwise indicated by a footnote, as follows: a 55, b 58, c 45, d 41, e 50,
f 43, g 42, h 80, i 60, j 86, k 52, l 87, m 81, n 77, o 84, p 78, q 89, r 88, s 67, t 66, u 62, v 63, w 56, x 57, y 54, z 83, aa 69.
and 36-6). Ethanol decreased risky choice be-
nol (for Rat 36-6), and one of two rate de-
low baseline and below the vehicle controls
creases (for Rat 36-5).
for 3 rats (i.e., 36-2, 36-4, and 36-5). Rat 36-6
Tables 4 and 5 present total number of tri-
had the highest baseline percentage of risky
als completed, session duration, and rate of
choice at the 80-s ITI, but did not show any
trial completion for the test sessions shown in
decrease after ethanol administration, which
Figure 6 for the 20- and 80-s ITI conditions,
was in contrast to the results for this rat at the
respectively. Rate of trial completion was de-
20-s ITI. This rat did again show an increase
creased for all 6 rats by 2.5 or 3.0 g/kg eth-
in risky choice at the lower doses of ethanol,
anol and by lower doses in 4 rats, but only 1
however.
rat failed to complete any trials at a dose and
For the other 2 rats (36-1 and 36-3), the
most rats always completed 60 to 105 trials.
baseline range was not above zero, so decreas-
In the 80-s ITI condition, the rats completed
es in risky choice could not be observed. This
fewer trials after vehicle compared to the 20-
baseline range at the 80-s ITI was decreased
s ITI condition (i.e., 38 to 62 trials in 90 min
for Rat 36-1 compared to the 20-s ITI condi-
compared to 72 to 105 in 44 to 90 min at the
20-s ITI). In contrast to the 20-s ITI condi-
tion. For Rat 36-3, the upper end of the range
tion, rate of trial completion remained within
was even lower than under the 20-s ITI. These
the range for vehicle in all except two tests
2 rats showed increases in risky choice after
in which a rat failed to complete even the
vehicle administration, and Rat 36-1 showed
forced-choice trials (at 2.0 or 3.0 g/kg for 2
an increase at 1 g/kg ethanol.
rats).
Repeating one or more doses in the 80-s
ITI condition (Figure 6) con?rmed two of
Phencyclidine
the four effects of vehicle (for Rats 36-1 and
Figure 7 shows that during the 20-s ITI con-
36-2), one of three rate increases after etha-
dition, phencyclidine produced an effect dif-

---