Behavioural Processes 84 (2010) 573–580

Contents lists available at ScienceDirect

Behavioural Processes
journal homepage: www.elsevier.com/locate/behavproc

Operant behavior in dwarf hamsters (Phodopus campbelli): Effects of rate of

reinforcement and reinforcer flavor variety
Gwen Lupfer-Johnson ∗ , Eric S. Murphy, Linda C. Blackwell, Jennifer L. LaCasse, Sarah Drummond
Department of Psychology, University of Alaska Anchorage, 3211 Providence Drive, Anchorage, AK 99508-4614, USA

a r t i c l e i n f o a b s t r a c t

Article history: We investigated operant behavior in a novel species, the dwarf hamster (Phodopus campbelli). In two
Received 19 November 2009 experiments, hamsters were trained to lever-press for food reinforcement. In Experiment 1, rate of rein-
Received in revised form 14 January 2010 forcement was manipulated across conditions using four variable-interval schedules of reinforcement
Accepted 18 February 2010
(delivering one to eight reinforcers per min). As predicted, within-session decreases in responding were
steepest on the richest schedule. In Experiment 2, lever-pressing was reinforced by either a constant
Keywords:
or a variety of flavored food pellets. Within-session decreases in responding were steeper when the
Habituation
reinforcer flavor remained constant than when it was varied within the session. In both experiments,
Within-session changes
Dwarf hamsters
subjects hoarded most reinforcers in their cheek pouches rather than consuming them in the operant
Variety effects chambers. These results are incompatible with post-ingestive satiety variables as explanations for within-
Stimulus rate session decreases in operant responding and suggest that habituation to repeatedly presented reinforcers
Hoarding best accounts for subjects’ response patterns. Additionally, a mathematical model that describes behav-
Phodopus ior undergoing habituation also described the present results, thus strengthening the conclusion that
habituation mediates the reinforcing efficacy of food.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction P. campbelli, subjects readily learned to approach and contact

Dwarf hamsters (Phodopus campbelli) are mouse-sized rodents
found in sandy, arid regions of Siberia. Within the genus Phodopus
(Latin for “hairy-footed hamster”) are three species: P. sungorus,
P. roborovskii, and P. campbelli. P. campbelli hamsters are often
studied for their parental behavior. Compared to most rodent
species, including other hamsters, P. campbelli males are highly
paternal (Wynne-Edwards, 2003). They serve as midwives during
the birth process, using their incisors and paws to help remove
pups from the birth canal; they then lick their newborns’ nos-
trils, facilitating the pups’ first breaths (Jones and Wynne-Edwards,
2001). Fathers also retrieve experimentally displaced pups quickly
(often within 10 s) and bring food to their offspring in their cheek
pouches, sometimes after the mother has moved on to invest
in a subsequent litter. Even under favorable laboratory condi-
tions, pups rely on biparental care to survive (Wynne-Edwards,
2003).
The paternal behaviors of P. campbelli are well documented. Less
is known about their learning abilities, although some recent stud-
ies have documented classical conditioning and foraging behaviors
in this species. In an investigation of conditioned responding using
∗ Corresponding author. Tel.: +1 907 786 1722; fax: +1 907 786 4898.
E-mail address: afgjl@uaa.alaska.edu (G. Lupfer-Johnson).
either a small cardboard square or a restrained conspecific sig-
nal for food. Conditioned responses directed toward the cardboard
involved mainly gnawing, whereas social behaviors such as head
and anogenital sniffs were elicited by conspecific signals for food
(Lupfer-Johnson, 2008). Similarly, male dwarf hamsters attended
to and sniffed their recently fed mates and learned to use which
food their mates had eaten as a discriminative stimulus for which
food was available in an open-field foraging task (Lupfer-Johnson
et al., 2009).
Very little is known about P. campbelli’s operant behavior.
Related Syrian hamsters (Mesocricetus auratus) have occasionally
served as subjects in traditional operant paradigms (e.g., Anderson
and Shettleworth, 1977; DiBattista, 1999), but to our knowl-
edge, only one unpublished investigation of operant behavior in
P. campbelli has been documented (Wertz, 2007). In Wertz’s the-
sis, P. campbelli subjects were housed in operant chambers in a
closed economy, so that all food was earned by lever-pressing on
fixed-ratio (FR) schedules of reinforcement. Male and female ham-
sters performed similarly in terms of response rates and hoarding
behavior. A tendency was noted in which subjects emitted more
responses if the experimenter interfered with the accumulation
of a hoard (i.e., by removing the hoard manually or by use of a
wire mesh floor), but this tendency did not reach statistical signifi-
cance. Finally, subjects increased their response rates as their ratio
requirement increased from an FR 1 to FR 5, essentially earning the

0376-6357/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.beproc.2010.02.016
574 G. Lupfer-Johnson et al. / Behavioural Processes 84 (2010) 573–580
same number of pellets regardless of the schedule in effect (Wertz
and Frieman, 2005; Wertz, 2007).
Wertz’s (2007) study also reported that lever-pressing occurred
in bouts over a 24 h period. Although large amounts of between-
subject variability in the frequency and duration of lever-pressing
bouts occurred, one pattern emerged: the size of the bout was
directly related to the ratio requirement. That is, longer bouts of
lever-pressing occurred on the larger ratios, and smaller bouts
occurred when the ratio requirement was low. These results sug-
gest that rate of reinforcement is an important variable governing
operant behavior in dwarf hamsters. Although finding a relation-
ship between rate of reinforcement and operant responding is
not new (e.g., Herrnstein, 1970), it is the first time that the rela-
tionship has been reported with P. campbelli. What is not known,
however, is how operant lever-pressing was temporally organized
within the bouts. In traditional laboratory animals (e.g., rats and
pigeons), it is well established that rate of operant responding
often varies within an experimental session, particularly at high
rates of reinforcement (e.g., McSweeney et al., 1996). Specifically,
rate of operant responding increases, increases then decreases, or
decreases within a standard operant conditioning session. Perhaps
the most important variable governing the within-session pattern
is rate of reinforcement. For example, higher rates of reinforce-
ment produce steeper within-session decreases in responding than
leaner rates of reinforcement (e.g., McSweeney, 1992).
The goal of the present project is to further explore bouts of
operant behavior in P. campbelli. They are an appropriate species
for the current research for several reasons. First, they readily for-
age for palatable reinforcers without food deprivation in an open
field (Lupfer-Johnson et al., 2009). In addition, they store food in
their cheek pouches. The food can be easily removed and quanti-
fied by an experimenter (Lupfer et al., 2003), and subjects typically
do not consume reinforcers until back in their home cages (Lupfer-
Johnson et al., 2009). This makes them an interesting species for
studying theories of within-session changes in response rates. For
example, one hypothesis suggests that within-session decreases
in operant responding are related to post-ingestive satiety vari-
ables, such as cholecystokinin release, blood glucose, or stomach
distension (Bizo et al., 1998). An opposing hypothesis suggests that
habituation, defined as a decrease in responsiveness to a repeat-
edly presented stimulus (e.g., Thompson and Spencer, 1966), to
the reinforcer modulates the within-session decrease in respond-
ing (McSweeney, 2004). Quantifying the number of food pellets
hoarded in cheek pouches versus those consumed in a session
enables investigation of the extent to which within-session changes
in operant behavior may be related to post-ingestive satiety vari-
ables, or to a process that does not depend on ingestion, such as
habituation.
In particular, the present study separated the predictions
of habituation from the satiation explanation of within-session
decreases in responding. The presence of habituation was estab-
lished by testing for two of its fundamental empirical properties.
Empirical tests were used because there is no generally accepted
theory of habituation. Although many theories have been pro-
posed (Sokolov, 1963; Wagner, 1976), none has been generally
accepted (Mackintosh, 1987; Staddon and Higa, 1996). The present
study tested for two fundamental properties of behavior under-
going habituation: stimulus rate and variety effects. The stimulus
rate property of habituation refers to the finding that faster rates
of stimulus presentation yield faster and more pronounced habit-
uation compared to slower rates (Thompson and Spencer, 1966).
To test the stimulus rate property of habituation, the rate of food
presentation was delivered on four interreinforcer intervals: 7.5 s,
15 s, 30 s, and 60 s. It was predicted that within-session decreases
in responding should be larger for the higher (e.g., 7.5 s) than the
lower (e.g., 60 s) rates of food reinforcement.
A variety effect occurs when habituation is slower to stimuli
that are presented in a variable, rather than a constant, manner. For
example, Broster and Rankin (1994) showed that habituation of the
tap withdrawal response in Caenorhabditis elegans proceeded more
quickly when the taps were presented on fixed, rather than vari-
able, interstimulus intervals. In the present study, variety effects
were tested by keeping the flavor of the food reinforcer constant
(i.e., bacon or chocolate), or by unpredictably changing the flavor
of the food pellet, within an operant conditioning session. If habit-
uation governs within-session decreases in responding, then these
decreases should be larger when the flavor is held constant, than
when the flavor is varied within the session. If satiation governs
within-session decreases in responding, there should be no differ-
ence in the within-session pattern because the food pellets were
identical in size (i.e., 20 mg) and nutritional content.
2. Experiment 1
The protocol to test the above questions was approved by the
University of Alaska Anchorage Institutional Animal Care and Use
Committee and conformed to the guidelines of the care and use of
laboratory animals as required by the National Institutes of Health
(National Research Council, 1996).
Experiment 1 was conducted to examine the within-session
changes in subjects’ rates of responding, relate those changes to rate
of reinforcement, and measure the proportion of earned reinforcers
which were consumed versus stored in subjects’ cheek pouches.
We hypothesized that dwarf hamsters would exhibit larger within-
session decreases in response rates on richer (e.g., 7.5 s) than on
leaner (e.g., 60 s) variable-interval (VI) schedules.
2.1. Method
2.1.1. Subjects
The subjects were 8 adult male dwarf hamsters (P. campbelli)
bred from a colony acquired from Queen’s University (Kingston,
Ontario). The Queen’s University breeding colony was created
from wild-caught hamsters in Siberia (Wynne-Edwards and Lisk,
1987). Two 8-month-old subjects had previously served in an
open-field foraging experiment (Lupfer-Johnson et al., 2009) and
6 1-month-old dwarf hamsters were experimentally naïve. At
the beginning of the study, the mean weight of the subjects
was 28.63 g (SEM = 3.17). Subjects were housed individually in
26.67 cm × 48.26 cm × 20.32 cm plastic cages in a vivarium main-
tained at a temperature of 21 ± 1 ◦ C. While in their home cages,
subjects had free access to Mazuri® Rodent Pellets (PMI Nutri-
tion International, LLC, Brentwood, MO) and water. Environmental
enrichment (i.e., small cardboard boxes or paper bags) was added
to each cage 1–2 times per week. The subjects were exposed to a
12:12 h light/dark cycle (lights on from 7:00 a.m. to 7:00 p.m.). To
be consistent with previous work in our laboratory (e.g., Lupfer-
Johnson et al., 2009), all sessions were conducted during the light
portion of the cycle. Additionally, Wertz (2007) reported that oper-
ant lever-pressing in dwarf hamsters occurred during both the light
and dark portions of the cycle.
2.1.2. Apparatus
The apparatus was a side-loading MED Associates (St. Albans,
VT) two-lever operant conditioning chamber for mice, measuring
21.6 cm × 17.8 cm × 12.7 cm. A 2.5-cm × 2.0-cm food tray, which
allowed access to chocolate flavored 20-mg dustless precision pel-
lets (Bio-Serv® , Frenchtown, NJ), was centered on the front panel,
0.5 cm above the floor. Two 1.6-cm × 1.0-cm levers appeared 4.0 cm
from this opening, one on each side. The levers, which required
approximately 0.02 N for operation, were 2.2 cm above the floor
and extended 1.0 cm into the enclosure. A 0.5-cm diameter yellow
 G. Lupfer-Johnson et al. / Behavioural Processes 84 (2010) 573–580 575

LED light was located 5.0 cm above each lever. The chamber was
illuminated by a 2.0-cm 100 mA houselight, centered on the back
wall, 8.0 cm above the grid floor. The apparatus was enclosed in
a sound-attenuating chamber and an exhaust fan masked noises
from outside. Experimental events were presented and data were
recorded by MED Associates software run by an IBM-compatible
computer.
2.1.3. Procedure
We trained subjects to press the right lever by placing them
in the chamber until they had emitted 50 presses, each of which
was reinforced by one 20-mg chocolate pellet. After shaping,
subjects responded in each of the following four conditions of rein-
forcement: VI 7.5-, VI 15-, VI 30-, and VI 60-s. These schedules
were chosen to provide a wide range of rates of food reinforce-
ment. Interreinforcer intervals were calculated according to a
25-interval Fleshler and Hoffman (1962) series. A Latin-square
design (Campbell and Stanley, 1963) was used to determine the
order of conditions for each subject.
Each condition was conducted to stability with the requirement
that the condition was in effect for a minimum of 20 sessions.
Responding was considered stable when rates of responding during
the last 5 sessions of a condition fell within the range of responding
for the entire condition. If this criterion was not met, more sessions
were conducted until responding was deemed stable. All sessions
were 30 min long and were conducted daily, 5–6 times per week.
After reaching stability on a particular schedule of reinforce-
ment, each subject experienced one “pellet count” in which the
percentage of earned reinforcers hoarded versus consumed was
quantified. The procedure involved removing the subject from the
operant chamber immediately following a session. Subjects were
then restrained by grasping the skin behind their heads. The con-
tents of each subject’s cheek pouches were then removed by gently
massaging the subject’s cheeks using a forward motion. After the
pellet count, the extracted food pellets were returned to the sub-
ject’s home cage. Each “pellet count” was conducted after a subject’s
final session on a particular schedule of reinforcement and before
the subject progressed to another condition.
2.1.4. Data analyses
With the exception of the pellet count results, the data were
averaged over the last 5 sessions for which each condition was
in effect. Within-session changes in responding were determined
by dividing the 30-min session into ten 3-min intervals. Rates of
responding in each 3-min interval were calculated by dividing
the number of responses emitted in that 3-min interval by 3 min.
Rates of responding per 3-min interval were converted into pro-
portions. Proportions were calculated by dividing the number of
responses emitted during each 3-min interval by the total number
of responses emitted during the session. To avoid violating the nor-
mal distribution assumption of the analysis of variance (ANOVA),
proportions were converted using the arcsine transformation sug-
gested by Cohen and Cohen (1983). The transformed proportions
were then compared across conditions by a 4 (Schedule) × 10 (3-
min interval) repeated-measures ANOVA. The percentage of food
pellets hoarded, and the absolute number of food pellets consumed,
were analyzed by separate one-way repeated-measures ANOVAs.
the omega squared statistic, these guidelines also apply to the inter-
pretation of partial eta squared values (Olejnik and Algina, 2003).
As suggested by Keppel (1991), partial eta squared values were
reported for all treatment effects, not just those that were signifi-
cant.
Data were additionally analyzed using the quantitative model
proposed by McSweeney et al. (1996). Eq. (1) describes data from
the habituation literature and accounts for approximately 90% of
the variance in the temporal changes in several forms of motivated
behavior (feeding, drinking, exploration, and escape; McSweeney
and Swindell, 1999):
b c
P= − (1)
eaT c+T
P is the predicted proportion of the total responses that should
occur during successive time intervals (T). T is the ordinal num-
ber of time interval, e is the base of the natural logarithm, and
a, b, and c are free parameters. The exponential component of
Eq. (1) describes a decreasing process identified as habituation;
the hyperbolic component describes an increasing process iden-
tified as sensitization. Thus, a and b govern habituation, and c
applies to sensitization. Larger values of a and b denote larger
late-session decreases in responding (i.e., greater habituation), and
larger values of c indicate larger early-session increases in respond-
ing (i.e., greater sensitization). Eq. (1) has described within-session
data from habituation (McSweeney et al., 1996) and motivation
(McSweeney and Swindell, 1999; Murphy et al., 2003) experiments
well, accounting for approximately 90% of the variance in the data.
To reduce the number of free parameters of Eq. (1), a simpli-
fied version has been applied to behavior that primarily shows a
decrease in responsiveness to the reinforcer (i.e., behaviors that
lack sensitization, c):
b
P= (2)
eaT
Eq. (2) has accounted for approximately 90% of the variance in
the data on within-session changes in wheel running (Aoyama and
McSweeney, 2001), alcohol self-administration in rats (Murphy et
al., 2007), and sucrose licking in marmosets (Ishii and Watanabe,
in press). Eq. (2), rather than Eq. (1), was fit to the data because
response rates primarily decreased within sessions. The parameters
of Eq. (2) were estimated by the nonlinear regression procedure in
SPSS 11.0 (Chicago, IL).
3. Results and discussion
The mean number of sessions required to reach stability was
25.0 (SEM = 2.72), 20.88 (SEM = 0.87), 23.0 (SEM = 1.35), and 22.0
(SEM = 0.98) for the VI 7.5-, VI 15-, VI 30-, and VI 60-s sched-
ules, respectively. Table 1 contains the mean rate of lever-pressing,
averaged over the session, obtained rates of reinforcement, and
percentage of food pellets hoarded for each condition.
Table 1
Mean rates of responding per min, obtained reinforcers per session, and the percent
of food pellets hoarded.

Results were considered significant when p < 0.05. Condition Responses/min Obtained reinforcers % Hoarded
Because dwarf hamsters are new to the operant literature, we
M SEM M SEM M SEM
also reported effect sizes for treatment effects as this information
may be helpful to researchers who use this species in the future. VI 7.5 s 7.04 2.29 62.35 13.64 71.74 7.53
VI 15 s 5.84 1.61 41.28 8.77 62.91 10.29
Partial eta squared (2p ) is an appropriate measure of effect size VI 30 s 6.28 1.79 29.45 4.59 69.06 8.66
for studies without a blocking variable (Olejnik and Algina, 2003). VI 60 s 6.49 1.61 20.05 2.25 76.66 6.42
According to Cohen’s (1977) criteria for interpreting effect sizes,
Note: Responses/min and obtained reinforcers were calculated by averaging sub-
values of 0.01, 0.06, 0.15 are considered small, medium, and large, jects’ final 5 sessions of operant responding for each condition. Percentages of food
respectively. Although Cohen’s criteria were developed for use with pellets hoarded were based on a single measurement per subject per condition.
576 G. Lupfer-Johnson et al. / Behavioural Processes 84 (2010) 573–580
Fig. 1. Rates of responding (A) and proportion of responses (B) during the VI 7.5-
(closed circles), VI 15- (open circles), VI 30- (closed triangles), and VI 60-s (open
triangles) schedules as a function of 3-min interval. Each function is the mean of
all subjects’ (N = 8) responding during the last five sessions of each condition. Error
bars represent ±1 standard error of the mean.
3.1. Within-session responding
Fig. 1 presents rates of responding (Fig. 1A) and the pro-
portion of responses (Fig. 1B) emitted during successive 3-min
intervals for each schedule of reinforcement. Each function
is the mean of all subjects. Fig. 1 indicates that rates of
responding changed within the session for each condition, and
decreased steeply during the VI 7.5-s schedule. The main effect
of schedule of reinforcement was not significant, F(3, 189) = 1.01,
p = 0.41, 2p = .13. However, the main effect of time interval
was statistically significant F(9, 189) = 3.96, p = 0.000, 2p = .36,
indicating that responding changed within the session. The
Schedule × 3-Min Interval interaction was also statistically sig-
nificant F(27, 189) = 1.63, p = 0.032, 2p = .19, suggesting that the
within-session pattern changed as a function of the rate of
reinforcement.
The parameter estimates of Eq. (2) (McSweeney et al., 1996) are
presented in Table 2. Estimates for both a and b were largest for the
VI 7.5-s schedule, as would be expected if the faster rate of rein-
forcement produced greater habituation than the leaner schedules.
The proportions of variance explained by the equation were good,
but lower than previous reports on within-session decreases in
responding. However, our results are consistent with McSweeney
et al. who reported higher R-squared values for higher, than for
lower, rates of stimulus presentation.
Table 2
Parameter estimates of Eq. (2) as a function of rate of reinforcement.
Condition a b R2
M SEM M SEM M SEM
VI 7.5 s 0.18 0.06 0.27 0.05 0.67 0.08
VI 15 s 0.09 0.05 0.18 0.03 0.60 0.06
VI 30 s 0.10 0.06 0.15 0.04 0.48 0.13
VI 60 s 0.05 0.06 0.17 0.06 0.35 0.08
Note: Eq. (2) was fit to the mean proportion of total-session responses during succes-
sive 3-min intervals in the session. The variance accounted for by Eq. (2) is indicated
by the R2 values.
3.2. Obtained rates of reinforcement
On average, subjects hoarded 70.1% (SEM = 6.23%) of all rein-
forcers earned in the operant chambers and consumed 8.56
(SEM = 0.94) of the 20-mg pellets (see Table 1). A repeated-
measures ANOVA was conducted on the percentages of hoarded
pellets on the four schedules of reinforcement. The result was non-
significant F(3, 21) = 0.79, p = 0.56, indicating that the percentage
of reinforcers hoarded did not differ by schedule of reinforcement.
However, the absolute number of pellets consumed in the oper-
ant chambers differed by schedule of reinforcement F(3, 21) = 4.02,
p = 0.02. Subjects consumed more reinforcers on the richer sched-
ules of reinforcement. The mean number of pellets consumed by
subjects on the VI 7.5-, 15-, 30-, and 60-s schedules of reinforce-
ment were 12.63 (SEM = 2.66), 9.00 (SEM = 1.26), 7.50 (SEM = 0.66),
and 5.13 (SEM = 1.37), respectively.
The finding that subjects decreased their within-session
responding most steeply on the richest schedule of reinforcement
was predicted because faster rates of stimulus presentation pro-
duce more habituation (Thompson, 2009). However, subjects also
consumed more food on the richest schedule of reinforcement.
Although the total amount consumed was quite small (approxi-
mately .25-g on average on the VI 7.5-s schedule), post-ingestive
factors cannot be completely ruled as contributing to the observed
within-session decreases.
4. Experiment 2
In Experiment 1, P. campbelli decreased their rates of responding
within sessions, particularly on the VI 7.5-s schedule. In Experiment
2, the subjects responded on a fixed-interval (FI) 8-s schedule of
reinforcement. An FI, rather than a VI, schedule was used because
previous research indicates that more habituation accrues to stim-
uli that are presented in a fixed, rather than a variable, manner (for
a review, see McSweeney and Murphy, 2009; Murphy et al., 2007).
The goal of the present experiment was to determine whether
increasing reinforcer flavor variety (e.g., Melville et al., 1997) would
decrease rate of habituation in dwarf hamsters. We hypothesized
that smaller within-session decreases (i.e., less habituation) would
occur when the sensory properties of the reinforcer (e.g., smell,
taste, and color) varied compared to when they remained constant.
Although most reinforcers were hoarded in Experiment 1, approx-
imately 30% of earned food pellets were consumed; therefore, we
anticipated an effect of varying reinforcer flavor despite the fact
that hamsters tended to hoard the food reinforcers.
4.1. Method
4.1.1. Subjects and apparatus
The subjects in Experiment 2 consisted of an additional 8 adult
dwarf hamsters bred from subjects acquired from Queen’s Uni-
versity (Kingston, Ontario). All 8 of these subjects (4 male and 4
female) were experimentally naïve at the beginning of the study,
 G. Lupfer-Johnson et al. / Behavioural Processes 84 (2010) 573–580 577

were 45.75 (SEM = 5.59) days old, and weighed 28.25 g (SEM = 1.18). Table 3
Mean rates of responding per min, obtained reinforcers per session, and the percent
Subjects were housed and maintained as in Experiment 1. Two
of food pellets hoarded for the constant and variety conditions.
identical MED Associates operant chambers for mice with the
same dimensions as the one described in Experiment 1 were Condition Responses/min Obtained reinforcers % Hoarded
used in the present experiment. Reinforcers consisted of choco- M SEM M SEM M SEM
late flavored 20-mg dustless precision pellets (as in Experiment 1)
Constant 2.19 0.52 34.21 3.84 64.46 6.77
and/or bacon flavored 20-mg pellets, also obtained from Bio-Serv® Variety 2.58 0.65 37.84 4.21 70.86 3.19
(Frenchtown, NJ). The two flavored pellets had similar nutritional
Note: Responses/min and obtained reinforcers were calculated by averaging sub-
content; both types of pellets were made by adding either bacon jects’ final 5 sessions of operant responding for each condition. Percentages of food
or chocolate flavoring to the same nutritionally complete formula pellets hoarded were based on a single measurement per subject per condition.
(containing 18.8% protein, 5% fat, 5% fiber, 61.5% carbohydrates, and
3.66 kcal/g).
within-session decreases were steeper when the flavor of the
4.1.2. Procedure reinforcer was held constant. As in Experiment 1, data were con-
The 8 subjects were randomly assigned to one of two groups verted to proportions and subjected to an arcsine transformation
(1 or 2) with each group containing 4 subjects. All subjects were before analysis. The main effect of reinforcer flavor variety was not
trained to press a lever for food reinforcers as in Experiment 1 statistically significant, F(1, 63) = 4.97, p = 0.06, 2p = .42, although
before proceeding to Phase 1. Subjects in Group 1 then received responding tended to be higher when reinforcer flavor varied (see
chocolate flavored pellets on an FI 8-s schedule of reinforce- Table 3). The main effect of time interval was statistically signifi-
ment until stable responding (as assessed in Experiment 1) had cant F(9, 63) = 11.48, p = 0.000, 2p = .62, indicating that responding
been reached by all subjects. During Phase 2, subjects in Group 1 changed within the session. The Condition × 3-Min Interval inter-
alternated between the following two conditions: an FI 8-s sched- action was also statistically significant, F(9, 63) = 2.92, p = 0.006,
ule of reinforcement in which reinforcers were always chocolate 2p = .29, indicating that the within-session patterns differed for
pellets and an FI 8-s schedule in which there was a 0.50 prob- the two conditions.
ability of receiving a chocolate or a bacon reinforcer. Subjects Eq. (2) parameter estimates and proportions of variance
experienced 10 days of each condition, and alternated between explained for the constant and variety conditions are presented
conditions according to a pre-determined random sequence. Sub- in Table 4. Both a and b estimates are larger for sessions in which
jects in Group 2 proceeded through the same two phases but
the flavor they earned was bacon in Phase 1. Following com-
pletion of Phase 2, all subjects completed the two phases again,
but earning the opposite flavor during Phase 1 and the constant
days of Phase 2. Phase 1 was conducted in order to allow operant
responding to stabilize; response rates during Phase 1 were not
analyzed.
Each subject experienced four pellet counts, performed as in
Experiment 1. Two of these occurred during the constant condition
(one with each flavored reinforcer), and two occurred during the
variable reinforcer flavor condition. These counts were performed
after subjects had completed Phase 2.

4.1.3. Data analyses

Data from Phase 2 were averaged across the 10 days of each
condition for each subject, for a total of 20 constant days (repre-
senting responding for both chocolate and bacon reinforcers) and
20 variety days per subject. The analyses of within-session changes
in responding were the same as in Experiment 1. The percentage
of food pellets hoarded, and the absolute number of food pellets
consumed, were analyzed by separate dependent samples t-tests.
Results were considered significant when p < 0.05.

5. Results and discussion

The mean numbers of sessions required to reach stability (Phase

1) were 21.0 (SEM = 0.68) and 21.13 (SEM = 0.48) for the chocolate
and bacon baseline conditions, respectively. Table 3 contains the
mean rate of lever-pressing, averaged over the session, obtained
rates of reinforcement, and percentage of food pellets hoarded for
each condition during Phase 2.

5.1. Within-session responding

Fig. 2 presents rates of responding (Fig. 2A) and the proportion

of responses (Fig. 2B) emitted during successive 3-min intervals
for the constant and variety conditions. Each function is the mean
of all subjects. Fig. 2 indicates that rates of responding primar-
ily decreased within the session for each condition; however,
Fig. 2. Rates of responding (A) and proportion of responses (B) during the constant
(closed circles) and variety (open circles) conditions of reinforcement as a function
of 3-min interval. Each function is the mean of all subjects’ (N = 8) responding during
the 20 sessions that each condition was in effect. Error bars represent ±1 standard
error of the mean.
578 G. Lupfer-Johnson et al. / Behavioural Processes 84 (2010) 573–580
Table 4
Parameter estimates of Eq. (2) for the constant and variety conditions.
Condition a b R2
M SEM M SEM M SEM
Constant 0.44 0.14 0.65 0.20 0.75
Variety 0.37 0.07 0.48 0.07 0.72
Note: Eq. (2) was fit to the mean proportion of total-session responses during succes-
sive 3-min intervals in the session. The variance accounted for by Eq. (2) is indicated
by the R2 values.
reinforcer flavor remained constant than when it varied within a
session, indicating more habituation in the constant condition than
in the variety condition. Eq. (2) explained 0.74 of the variance in the
data averaged across conditions.
5.2. Obtained rates of reinforcement
Subjects hoarded an average of 67.5% (SEM = 4.53%) of all rein-
forcers earned and consumed an average of 9.78 (SEM = 1.28) pellets
per session. Proportions of chocolate versus bacon reinforcers
hoarded were similar: 67.1% (SEM = 8.10%) of chocolate versus
62.5% (SEM = 7.40%) of bacon on average measured during constant
days, t(7) = 0.71, p = 0.51, or 70.8% (SEM = 4.73%) of chocolate versus
62.8% (SEM = 6.58%) of bacon measured during days when both fla-
vors were earned, t(7) = 0.97, p = 0.36. More importantly, absolute
quantities of reinforcers consumed were not significantly differ-
ent between days in which reinforcer flavor remained constant
(M = 9.44, SEM = 1.82) and days in which reinforcer flavor varied
(M = 10.13, SEM = 1.92), t(7) = 0.51, p = 0.62. Because subjects con-
sumed similar amounts of food during constant and variable flavor
days (see Table 2), post-ingestive factors cannot account for the
slower within-session decreases observed when reinforcer flavor
varied.
6. General discussion
The present studies showed that operant responding in P.
campbelli decreased within experimental sessions even when the
programmed conditions of food reinforcement were held constant
across the session. Previous experiments using food reinforcers
documented similar late-session decreases in responding in rats
(e.g., Melville et al., 1997) pigeons (e.g., Murphy et al., 2003), gerbils
(Roll and McSweeney, 1997), and humans (Epstein et al., 2003). Past
experiments with food reinforcers showed that operant respond-
ing for food primarily decreases within sessions when reinforcers
are provided at a high rate (e.g., 240 reinforcers/h), but may increase
and then decrease when reinforcers are provided at an intermediate
rate (e.g., 60 reinforcers/h; McSweeney, 1992).
The goal of the present experiment was to determine if the
observed within-session decreases in operant responding were
produced by habituation to food as it was repeatedly presented
over the course of the session. If P. campbelli did habituate to food
reinforcers, then the within-session changes in responding should
show the empirical properties of behavior undergoing habituation.
The present results are consistent with this hypothesis. In Experi-
ment 1, a stimulus rate effect occurred. Within-session decreases
in responding were steeper for the higher (i.e., 7.5 s), than the lower
(e.g., 60 s), rates of food reinforcement, and parameter estimates of
habituation obtained using Eq. (2) (McSweeney et al., 1996) were
largest for the 7.5-s schedule of reinforcement. In Experiment 2, a
variety effect (Broster and Rankin, 1994) occurred. Within-session
decreases in responding were steeper, and Eq. (2) parameter esti-
mates governing habituation were larger, when the flavor was held
constant than when it varied within the session.
The results of Experiment 1 are generally consistent with the
stimulus rate property of habituation (McSweeney and Murphy,
2000; Thompson and Spencer, 1966), although it is somewhat sur-
prising that robust within-session decreases were observed with
only the richest schedule of reinforcement. Perhaps the use of a
0.10
range of shorter interreinforcer intervals (e.g., VI 2 s, VI 4 s, VI 8 s,
0.12
and VI 16 s) would have resulted in a clearer relationship between
rate of reinforcement, rate of responding, and within-session
decreases in responding in this species. Nevertheless, hamsters’
within-session decreases were steepest on the richest schedule,
as would be expected if they were habituating to the repeatedly
presented reinforcers. However, other explanations may account
for these results. For example, as more reinforcers were consumed
within the session, satiety factors, such as calories and stomach
distension, should have increased, resulting in progressively lower
response rates across the session. The majority of food reinforcers
were stored in cheek pouches rather than consumed (see Table 1).
However, a small but statistically significant tendency for subjects
to consume (rather than hoard in cheek pouches) more food on the
richest schedule of reinforcement was also observed. Therefore,
either post-ingestive factors or the faster rate of stimulus (rein-
forcer) presentation could theoretically account for the steeper
decreases in response rates observed on the VI 7.5-s schedule of
reinforcement.
The fact that the within-session decreases in responding were
steeper when the reinforcer flavor was held constant, than when
it varied, in Experiment 2 question these explanations, however.
The amount of food consumed was similar for the constant and
variety conditions (see Table 2). Therefore, any effect of post-
ingestive satiety variables should have been similar. Nevertheless,
the within-session decreases in responding were steeper for the
constant than for the variable flavor condition. The variety effect
property of habituation provides a logical explanation for these
results.
Our findings resemble those from previous investigations of
variety effects on operant responding. For example, rats exhib-
ited steeper within-session decreases when reinforcers alternated
between grape liquid, sucrose pellets, and grain-based pellets
than when the type of reinforcer remained constant (Melville et
al., 1997). In similar studies using human participants, decreased
responding was restored by changing the reinforcer from one type
of food to another (e.g., from potato chips to M&MsTM ; Temple
et al., 2008). The results of these earlier studies, coupled with the
findings of Experiment 2, indicate that the effectiveness of a par-
ticular reinforcer is enhanced when it is presented in a variable
manner.
In the current experiments, subjects responded for palatable
reinforcers under ad libitum feeding conditions. We believe that
the elimination of food deprivation is desirable for both practical
and ethical reasons. However, it may be argued that this aspect of
our procedure affected our results. Pinkston et al. (2007) reported
minimal within-session changes in response rates when pigeons
were strictly maintained at 80% of their ad libitum weights, but
larger within-session decreases in subjects weighing more than
94% of their ad libitum weights. It is unknown whether P. camp-
belli hamsters’ within-session patterns of responding are affected
by deprivation level. Previous investigations with both Syrian (M.
auratus) and Turkish (Mesocricetus brandti) hamsters indicate that
hamsters’ food intake is surprisingly unaffected by food deprivation
periods of up to 72 h (Rowland, 1982), although the presence of a
high-fat dietary option may enhance post-deprivation hyperpha-
gia (DiBattista, 1987). In any case, differences in deprivation levels
cannot account for the results of Experiment 2, as our subjects expe-
rienced constant ad libitum feeding conditions but still decreased
their within-session responding more slowly when reinforcer fla-
vor varied compared to when it remained constant.
 G. Lupfer-Johnson et al. / Behavioural Processes 84 (2010) 573–580
To our knowledge, the current experiments represent the first
published account of P. campbelli subjects responding in a tra-
ditional operant conditioning paradigm (but see Wertz, 2007).
These results join those of previous studies documenting within-
session decreases in traditional laboratory animals. Although in the
current experiments dwarf hamsters’ within-session response pat-
terns resembled those of other species, the topography of their
lever-pressing was unusual. Wertz (2007, p. 27) observed that
dwarf hamsters “. . .often jumped on the lever using their back
feet, looking like they were trying to climb the chamber wall in
order to press the lever.” We also observed this behavior, which
strongly resembled the “scrabbling” described by Shettleworth
(1975) in Syrian hamsters during open-field behavioral observa-
tion and by Anderson and Shettleworth (1977) in Syrian hamsters
lever-pressing for food reinforcers. In addition to scrabbling, we
noted subjects digging on or around the lever in order to press
it; this behavior was also reported by Anderson and Shettleworth
(1977). In the current experiment, subjects were not videotaped,
so our behavioral observations were limited to occasional glances
through the operant chamber eyeholes. However, these informal
observations suggested to us that scrabbling and digging were more
common early in training and were replaced by more conventional
lever-pressing as subjects gained experience in the operant cham-
bers.
P. campbelli subjects seem well-suited for further studies on
habituation of operant responding. Their tendency not to consume
reinforcers until after an operant conditioning session ends enables
investigation of within-session decreases in operant response rates
accompanied by only minimal (and quantifiable) post-ingestive
changes. In the present experiments, both a stimulus rate effect
and a variety effect were observed. Experimenters in the future
may wish to investigate other empirical properties of habituation
(e.g., dishabituation) using P. campbelli dwarf hamsters. Such exper-
iments would not only provide information about the behavior of
P. campbelli hamsters, but would also extend our knowledge con-
cerning the generality of habituation (e.g., Thorpe, 1966) to include
a new species.
Acknowledgements
Partial funding of this project was provided by the University
of Alaska Anchorage Honors College. Preliminary results were pre-
sented at the 34th (Experiment 1) and 35th (Experiment 2) annual
meetings of the Association for Behavior Analysis International. We
thank Michelle Scott-Weber, Monique Harvey, and Lesly Krome for
their assistance in data collection.
References
Anderson, M.C., Shettleworth, S.J., 1977. Behavioral adaptation to fixed-interval and
fixed-time food delivery in golden hamsters. Journal of the Experimental Anal-
ysis of Behavior 25, 33–49.
Aoyama, K., McSweeney, F.K., 2001. Habituation contributes to within-session
changes in free wheel running. Journal of the Experimental Analysis of Behavior
76 (3), 289–302.
Bizo, L.A., Bogdanov, S.V., Killeen, P.R., 1998. Satiation causes within-session
decreases in instrumental responding. Journal of Experimental Psychology: Ani-
mal Behavior Processes 24, 439–452.
Broster, B.S., Rankin, C.H., 1994. Effects of changing interstimulus interval during
habituation in Caenorhabditis elegans. Behavioral Neuroscience 108, 1019–1029.
Campbell, D.T., Stanley, J.C., 1963. Experimental and Quasi-experimental Designs for
Research. Houghton Mifflin, Boston.
Cohen, J., 1977. Statistical Power Analysis for the Behavioral Sciences, rev. ed. Aca-
demic press, New York.
Cohen, J., Cohen, P., 1983. Applied Multiple Regression/Correlation Analysis for the
Behavioral Sciences, 2nd ed. Lawrence Erlbaum Associates, Hillsdale, NJ.
DiBattista, D., 1999. Operant responding for dietary protein in the golden hamster.
Physiology and Behavior 67, 95–98.
DiBattista, D., 1987. Dietary self-selection of golden hamsters in response to acute
food deprivation and chronic food restriction. Behavioral Neuroscience 101,
568–575.
579
Epstein, L.H., Saad, F.G., Handley, E.A., Roemmich, J.N., Hawk, L.W., McSweeney, F.K.,
2003. Habituation of salivation and motivated responding for food in children.
Appetite 41, 283–289.
Fleshler, M., Hoffman, H.S., 1962. A progression for generating variable-interval
schedules. Journal of the Experimental Analysis of Behavior 5, 529–530.
Herrnstein, R.J., 1970. On the law of effect. Journal of the Experimental Analysis of
Behavior 13 (2), 267–272.
Ishii, T., Watanabe, S., in press. Within-session changes in licking as a
function of sucrose concentration in marmosets. Behavioural Processes,
doi:10.1016/j.beproc.2009.12.014.
Jones, J.S., Wynne-Edwards, K.E., 2001. Paternal behavior in biparental hamsters,
Phodopus campbelli, does not require contact with the pregnant female. Animal
Behaviour 62 (3), 453–464.
Keppel, G., 1991. Design and Analysis: A Researcher’s Handbook, 3rd ed. Pearson
Prentice Hall, Upper Saddle River, NJ.
Lupfer, G., Frieman, J., Coonfield, D.L., 2003. Social transmission of flavor pref-
erences in two species of hamsters. Journal of Comparative Psychology 117,
449–454.
Lupfer-Johnson, G., 2008. Dwarf and golden hamsters respond differently to con-
specific signals for food. Journal of General Psychology 135 (1), 54–64.
Lupfer-Johnson, G., Hanson, K.L., Edwards, L.E., Elder, R.L., Evans, S.L., 2009. Social
cues influence foraging in dwarf hamsters. Journal of Comparative Psychology
123 (2), 126–129.
Mackintosh, N.J., 1987. Neurobiology, psychology and habituation. Behaviour
Research and Therapy 25, 81–97.
McSweeney, F.K., 1992. Rate of reinforcement and session duration as determinants
of within-session patterns of responding. Animal Learning and Behavior 20,
160–169.
McSweeney, F.K., 2004. Dynamic changes in reinforcer effectiveness: satiation and
habituation have different implications for theory and practice. Behavior Analyst
27 (2), 177–188.
McSweeney, F.K., Hinson, J.M., Cannon, C.B., 1996. Sensitization–habituation
may occur during operant conditioning. Psychological Bulletin 120 (2),
256–271.
McSweeney, F.K., Murphy, E.S., 2000. Criticisms of the satiety hypothesis as an expla-
nation for within-session decreases in responding. Journal of the Experimental
Analysis of Behavior 74 (3), 347–361.
McSweeney, F.K., Murphy, E.S., 2009. Sensitization and habituation regulate
reinforcer effectiveness. Neurobiology of Learning and Memory 92 (2),
189–198.
McSweeney, F.K., Swindell, S., 1999. General process theories of motivation revis-
ited: The role of habituation. Psychological Bulletin 125, 437–457.
Melville, C.L., Rue, H.C., Rybiski, L.R., Weatherly, J.N., 1997. Altering reinforcer variety
or intensity changes the within-session decrease in responding. Learning and
Motivation 28, 609–621.
Murphy, E.S., McSweeney, F.K., Kowal, B.P., 2003. Within-session decreases in oper-
ant responding as a function of pre-session feedings. The Psychological Record
53, 313–326.
Murphy, E.S., McSweeney, F.K., Kowal, B.P., 2007. Motivation to consume alcohol in
rats: the role of habituation. In: O’Neal, P.W. (Ed.), Motivation of Health Behavior.
Nova Science, Hauppauge, NY, pp. 111–126.
National Research Council, 1996. Guide for the Care and Use of Laboratory Animals.
National Academy Press, Washington, DC.
Olejnik, S., Algina, J., 2003. Generalized eta and omega squared statistics: measures
of effect size for some common research designs. Psychological Methods 8 (4),
434–447.
Pinkston, J.W., Saulsgiver, K., Branch, M.N., 2007. Within-session patterns in
variable-interval schedule performance: variation with deprivation level.
Behavioural Processes 75, 297–306.
Roll, J.M., McSweeney, F.K., 1997. Within-session changes in operant responding
when gerbils (Meriones unguiculatus) serve as subjects. Current Psychology 15,
340–345.
Rowland, N., 1982. Failure by deprived hamsters to increase food intake: some
behavioral and physiological determinants. Journal of Comparative and Phys-
iological Psychology 96, 591–603.
Shettleworth, S.J., 1975. Reinforcement and the organization of behavior in golden
hamsters: hunger, environment, and food reinforcement. Journal of Experimen-
tal Psychology: Animal Behavior Processes 104, 56–87.
Sokolov, Y.N., 1963. Perception and the Conditioned Reflex (S.W. Waydenfeld,
Trans.). Oxford, England, Pergamon Press.
Staddon, J.E.R., Higa, J.J., 1996. Multiple time scales in simple habituation. Psycho-
logical Review 103, 720–733.
Temple, J.L., Giacomelli, A.M., Roemmich, J.N., Epstein, L.H., 2008. Habituation and
within-session changes in motivated responding for food in children. Appetite
50, 390–396.
Thompson, R.F., 2009. Habituation: a history. Neurobiology of Learning and Memory
92 (2), 127–134.
Thompson, R.F., Spencer, W.A., 1966. Habituation: a model phenomenon for
the study of neuronal substrates of behavior. Psychological Review 73,
16–43.
Thorpe, W.H., 1966. Learning and Instinct in Animals. Harvard University Press,
Cambridge, MA.
Wagner, A.R., 1976. Priming in STM: an information processing mechanism for
self-generated or retrieval generated depression in performance. In: Tighe, J.J.,
Leaton, R.N. (Eds.), Habituation: Perspectives from Child Development, Animal
Behavior and Neurophysiology. Wiley, New York.
580 G. Lupfer-Johnson et al. / Behavioural Processes 84 (2010) 573–580

Wertz, K.A., 2007. Operant conditioning and hoarding behavior in djungarian
hamsters (Phodopus campbelli). Unpublished master’s Thesis. Kansas State Uni-
versity, Manhattan, Kansas.
Wertz, K., Frieman, J., 2005. Operant conditioning and hoarding in dwarf hamsters
(Phodopus campbelli). In: Paper Presented at the Meeting of the Psychonomic
Society, Toronto, November.
Wynne-Edwards, K.E., 2003. From dwarf hamster to daddy: the intersection of ecol-
ogy, evolution, and physiology that produces paternal behavior. Advances in the
Study of Behavior 32, 207–261.
Wynne-Edwards, K.E., Lisk, R.D., 1987. Behavioral interactions differentiate Djungar-
ian (Phodopus campbelli) and Siberian (Phodopus sungorus) hamsters. Canadian
Journal of Zoology 65, 2229–2235.