Behavioural
Processes,
0 1990 Elsevier
BEPROC
22 (1990)
177-l
Science Publishers
177
86
B.V. 0376.6357/90/$03.50
00321
Within-session
responding
of post-session Frances
K. McSweeney,
Department
Julie
feedings
Hatfield
of Psychology, Pullman,
as a function
and Tammy
Washington
WA
State
99164-4830,
(Accepted
M. Allen
University,
USA
28 June 1989)
Abstract Ten
rats pressed
interval
schedule.
post-session Bacotti were
feeding
(1976),
These
results
Therefore,
in the
Experimenters
from
when
context
must
were
do not
always
that
do not confound
the interpretation
also
rewards
temporal
of their
variable
delivery
and
feedings.
across of
locations.
post-session
within
in response
by
feedings
increased
location
responding
of a
reported
to post-session
integrate
that changes
the
post-session
rates
delay
determines
take care to ensure
when
Response
the
and
to the results
higher
always
by a multiple
session
Contrary
of the
include
of variables
also
delivered
of the
immediate. regardless
subjects not
end
not significantly
session,
that need
reinforcers
the
0 to 240 minutes.
they
the
suggest
for food
between
rates were
within
theories
feedings session
varied
than
decreased
or keys
delay
response
delayed
then
levers The
the
rates
session.
within
the
results,
Introduction In a potentially ple-schedule than
when
reported
they
that
immediate.
increased delayed
were
Delayed
several
given
approximately
57% his
neglected higher
First,
20 and
150%
mean
for
subjects have
post-session
feedings
should
experimental
data. Such
this
valid
effect size
of
pressing
control
will
the
most
the
a key,
carefully
from
calculated
and 48%
be particularly
Bacotti
when
then
subjects
for
They
controlled
multi-
delayed
feedings
also were
trend. reported
implications.
that
were
session.
session
Bacotti
change,
reported
feedings
procedures,
large.
the
(1976)
experimental
decreasing
individual
practical be
for
was
for
Bacotti post-session
throughout
elimated
The results
after
decreased
the
study, when
immediately
rates
are generally
reasons.
feedings.
but were
feedings
results
between
Second,
rates
response
If Bacotti’s for
important
response
to
important
they that
the
subjects avoid
rates
immediate
from
suggest
are important response
his
figures,
pressing that
to
the
adding
in experiments
the was
a lever. time
of
variance
to
in which
178
changes results also
in
responding
(e.g.,
be reported
Third,
within
McSweeney results
current
theories
Many
the
reinforcers
situation
(e.g.,
rewards.
integrate
rewards
Bacotti’s
results
lower
response
rates
post-session feedings
With
those
the
also
reinforcers
because
conclusions inspection The
experimental layed
feedings
tion
in response
The
present
effect
of
pressing. minutes
rates
Experiment
responding
of his
open
and
outside
closed
of
because
within
placement,
economies.
just
is
the
they
governed session
session
indicate
the session.
not
rates
(open
that events
However,
the existence,
key
in
Bacotti
Therefore,
of their
weaken of
and records of
increase
over
because
study
patterns
records
post-session
Bacotti’s
of extra-ses-
responding
for
pressing
pressing
only for
over
presented
time,
and
rather
feedings
pecking
and His
on
were
a visual
presented.
immediate
than
immediate
key
importance conclusions.
based
4 of 9 subjects
immediate for
of his were
delayed
but he recovered the results
potential
the strength
feedings
an effect
of
followed
by de-
feedings
only
may represent
the
for the
a fluctua-
time. examine feedings
experiments,
in an effort
cage,
enclosure.
responding.
experiments
In both
apply
the experimental
to all of his subjects, subjects.
conclusions
from
distinction
de-
and the home
reinforcers
this
for
enclosure
all of their
be reexamined
manipulation.
lever-pressing
rates
Timberlake’s
between
some
the
those
of integration.
responding
the temporal
between an
that
that
within-session
represent
response
not
Lucas,
integrated
suggested
obtarn
do
and
have
they
animals
it is assumed
subjects
the experimental
distinction
1984)
can influence
of cumulative
also
within
which animals
of the window
that
are
may be part of
Gawley,
Higher
the experimental
support
characteristics
differences
may
results
should
about
the
when
influences
suggest
only
in the
reinforcers
over
if
that
outside
behavnot
the session.
window
conclusion
reinforcers.
results
what feedings
Timberlake,
indicate
for
present
circumstances,
(e.g.,
fell
should
include
reinforcers
during time
some
this
feedings
obtain
that
results
some
on
than
suggest
Bacotti’s
the
subjects
Bacotti’s
with
of rewards
Hursh,
in
minutes
between
bear
of the session
results
those
the context
in determining
the
feedings
Bacotti’s
(e.g.,
economies)
economies).
sion
results
when
16
within-session
integration
authors
differently
than
that
about other
emitted
that,
of the
feedings
of a behavior
also
about
interpretation
results.
that post-session
responses
argued
of rewards
as to the
arise
suggest
immediate
assumptions,
Finally,
for
with
indicate
to the integration
Problems
are consistent for
feedings
layed
but
the
of post-session
questions
behavior,
results
Timberlake
1987).
outside
1970).
more
time
that the determinants
bear on questions
over
confound
The
to answer
that
Bacotti’s
results
integrate
(closed
for
Herrnstein,
the
Several
argue
could
1988).
of experimental
help
of reinforcement
Fourth,
as well
may
obtained
part of the situation. the context
session
in all descriptions
Bacotti’s
ior.
the
and Melville,
the for
the delay
to determine
generality lever
of Bacottils
(Experiment
1) and
to post-session
the time
course
feedings
of this
effect.
in the
Johnson
findings. key
They
study
(Experiment
varies
from
2)
0 to 240
1
Method Subjecrs.
The
Washington
State
subjects University.
were
5 naive
They
were
rats
bred
approximately
120
days
Tower old
Vivarium
at the
beginning
at
179 of the
experiment
body
and were
The
Apparatus. rats,
20 by 24.5
metal;
the
apparatus
by 24.5
door
tenuating
was
panel,
made
this
floor
and extended
opening,
cm from
one
of their
free-feeding
allowed
apparatus
access
the floor.
on each side
The
Two
was
enclosure were
enclosed
to food
for
made of
in a sound-at-
reinforcers,
4 by 1.5 cm levers
of it. The
levers
were
light
was another
was
were
was
light,
the
located
in the front
2 cm in diameter
on
2.5 cm
5 cm above
light
was centered
centered
located
located
A 2 cm in diameter
2 cm in diameter
houselight
microcomputer, The
subjects
decreased
they
were
until
lever
pressing
the
were
(1962)
series.
minutes). An
lever
When
2.5
panel,
located
4
in the
delay
enclosure, to the
conducted the
post-session
following
order
at the first
30 sessions.
All
laboratory
was
of these
by
a shaping
for
pressing
(VI
l-m)
the first
pellet
and
interval
components times
screen
five
schedule.
had
5 m,
for 40 sessions,
60
m,
consisted
the
scheduled and
been
indication
that
removed
for
Hoffman
presented the
(60
session
the subject
room,
1, 2 and 3, the
used:
VI
lights
per week.
for
delays
The
but not during were
Fleshler
cage in another
Rats
a multiple
minutes.
component,
Noyes
to six
For
by Bacotti:
10
m,
0
following
m,
delays
and
Rats
240
m.
and at all subsequent
of one,
had
from
the
and began timing
60 m, 0 m, 240 m, 10 m, and 5 m. For
feedings
by was
approximately
were
4 and 5, Subjects
delays
for
6 g, pellet
of
to the beginning
of
chow.
the average,
post-session
darkened
every
the experimenter
feeding.
post-session
used
alternated
it to its home
order:
events
of the apparatus.
lever
interval
the schedule
mg
a computer
carried
right
on a variable
twelve five
outside
reinforcement
a twenty-five
when
appeared,
in the following
responded
On
indicator
45
to
the
of
the experimental
from
procedure.
during
one
conducted
watched
this
experimental
ended
were
experimenter
ended.
of
according
Sessions
Sessions
press
responded
illuminated
presented
noises
rate
components
consisted
right
to
The
employed
in which
Reinforcers
taught
on the experimental
procedure
schedule
above the right second.
subjects
room,
fan masked
procedure.
placed
experimental
VI l-m
in another
were
approximations
gradually The
located
the data. A ventilating
Procedure. successive
the
The
experimental of the enclosure
of the ceiling.
and recorded
l-m
which
A third,
the ceiling.
A SYM
the
85%
two-lever
and the ceiling
1.5 cm into the enclosure.
cm above each lever.
Then
a standard
sides
of plexiglas.
0.5 cm above
from
the
at approximately
chamber.
front
center
was
cm. Three
A 5 by 5.5 cm opening, the
maintained
weights.
35 seconds
delay.
The
experimental
cage. Subjects
were
enclosure weighed
end of the session
elapsed
subject
from
spent
the end of the session approximately
and approximately
before
and the beginning
the
session
of the
5 of these
30 seconds to minimize
post-session
35 seconds
in transit the
time
in the
to the home between
the
delay.
Results Table five
dividing Rates
1 presents
sessions the
for
the mean rates of responding which
number
are reported
each
condition
of responses
in responses
emitted per minute.
was
emitted
by each subject
conducted.
during
the
Rates
session
were
by the
over
the last
calculated session
by
length.
180
TABLE1 Mean
rates
of
Experiment
responding
(responses
per
minute)
for
each
delay
to
post-session
feeding
Subject
Delay
0
5
10
60
240
1
9.8
5.1
6.6
7.8
4.6
2
14.7
6.3
7.3
18.1
11.7
3
7.2
8.9
6.2
9.5
9.7
4
18.3
32.0
14.2
33.9
18.9
5
22.6
31.9
24.1
28.8
15.1
mean
14.5
16.8
11.7
19.6
12.0
Figure
results
the
for which emitted
the
are also
results for fell.
functions
averaged
10-m
delay,
rose
A two-way
fairly
of the
flat across
main
effect
of post-session
main
effect
1.
Mean
Experiment rates
are
of
rates 1. The
the
of
rates
the component
reported
which
per
rates
subjects
minute)
functions
Response
during The
for 1 for
the
well.
Most
peak
was
analysis
mean
during
represent
were
the
calculated
by the
five
are the means
responded
of all
individual
unrelated Only
by
minutes
of the rates
on each schedule.
subjects
also
represent
functions
rose
to a peak and
to the delay
one function,
was
the
to reinforcement.
that for subject
significant
was
minute)
(F(4,
(F(ll,
not significant
per
A
5 at the
in delays
over
last
sessions
presented.
five
16) = 2.36, 176)
successive
different
to
p
0.05).
The
0.0001).
The
p > 0.051.
components
post-session for
of respond-
of the session.
44) = 6.31, (F(44,
represent the
to the rates
components
not significant
(responses
subjects
was applied
the twelve
was
highly
functions all
of variance
during
feedings
responding for
feedings.
the session.
by component
different
means
(responses different
all subjects.
subjects
component
of delay
The
the session.
within-subjects by individual
Fig.
across
subjects
ing emitted
interaction
sessions Figure
across
responding
emitted
in
placement were
of
session.
was available. five
presented
The
few
the
of responses last
individual
rates
to post-session
the component
rates
then
mean within
delays
number
during
The
the
components
for different
dividing
results
presents
1
successive
The
in
1.
during
feedings. each
delay
The was
181
Experiment
2
Experiment also presents
2 examines the effect of post-session, feedings when rats press a key. It the 0 and 60 m delays as the first two delays in order to make the
procedure more similar to Bacotti’s. Large fluctuations often reported when subjects respond on multiple
in response schedules
rates over time are (e.g., McSweeney,
Dougan, Higa and Farmer, 1986; Spealman and Gollub, 1974). Because Bacotti conducted only the 0 and 60 m points while Experiment 1 conducted five delays, fluctuations in responding over time may have obscured differences in response rates between the 0 and 60 m delays in Experiment 1. If this is so, then the effect of delay should appear when the 0 and 60 m delays are conducted successively as they were by Bacotti.
Method Subjects. maintained
The same subjects used in Experiment 1 were used here. They were at approximately 95% of their freefeeding body weights. Deprivations
were decreased in this experiment to avoid a ceiling effect. As will become apparent, subjects pressed the key at rapid rates even at this mild deprivation. Had deprivations been more severe, the effect of post-session feedings may have been obscured by a ceiling. Apparatus. The apparatus was a 21 by 21.5 by 24 cm enclosure equiped with a lever and a response key. Two sides and the top of the enclosure were made of Plexiglas. The back and front were aluminum panels and the floor strips. The apparatus was enclosed in a sound-attenuating chamber.
was aluminum
A 6 cm in diameter opening allowed access to a 0.5 ml dipper. The opening was centered in the panel, 4 cm above the floor. A 2 cm in diameter, 5 W, light appeared 3 cm from each side of the panel and 5.5 cm from the top. The left light was clear and the right was opaque. A 2.5 cm in diameter response key was located 2 cm below the left light. A 3.5 cm lever, which below the right light.
extended
2 cm into
the chamber,
appeared 3 cm
A SYM microcomputer, located in another room, presented the experimental events and recorded the data. A ventillating fan masked noises from outside the apparatus. Procedure. The subjects were taught to press the key by a shaping by successive approximations procedure. The rate of reinforcement for pressing gradually decreased until subjects responded on a VI l-m schedule. Reinforcers consisted of 5-s access to the dipper which contained sweetened condensed milk, mixed one to one with water. When responding stabilized on the VI l-m schedule, subjects were placed on a multiple VI l-m VI l-m schedule. The light above the key was on during the first component and off during the second. The light located over the response used as a houselight and was illuminated throughout the session.
lever was
For subjects 1 and 2, the delays to post-session feedings were presented in the following order: 60 m, 0 m, 10 m, 240 m and 5 m. For subjects 4 and 5, delays were conducted in this order: 0 m, 60 m, 240 m, 10 m, and 5 m. For subject 3, the following order was used: 60 m, 0 m, 10 m, 5 m,and 240 m. Each condition lasted for 30 sessions. All other procedural details were identical to those used in Experiment 1.
182 TABLE
2
Mean
rates
Experiment
of
responding
(responses
per
minute)
for
each
delay
of
post-sessron
feeding
in
2.
Subject
Delay
0
5
10
60
240
77.5
52.6
45.0
45.8
40.4
2
59.8
21.5
63.6
31.8
40.6
3
34.6
23.6
50.7
35.3
14.5
4
21.2
70.0
32.7
57.9
48.4
5
24.3
29.4
35.3
42.0
64.7
mean
43.5
39.4
45.5
42.6
41.7
1
Results Table 2 presents the mean rates of responding emitted by each subject, and by the mean of all subjects, over the last five sessions for which each condition was conducted.
Rates were calculated as they were for Table 1.
Figure 2 presents the mean rates of responding (responses per minute) during successive components within the session. The separate functions represent the results for different delays to post-session feedings. Response rates were calculated as they were for Figure 1 and are averaged across all subjects. The within-session patterns of responding resembled those for Experiment 1. Response rates usually increased up to the fourth component and then decreased throughout
the rest of the session.
the post-session those presented
feeding. Again, in Figure 2.
This most
pattern was found functions
for
regardless
individual
of the delay to
subjects
resembled
A two-way within-subject analysis of variance was applied to the rates of responding emitted by individual subjects during the twelve components of the session. The main effect of postsession feedings was not significant (F(4, 16) = 0.07, p > 0.05). The
Fig.
2.
Mean
Experiment rates
are
rates 2. The
the
of
responding
different
means
for
(responses
functions all
subjects
per
minute)
in
successrve
represent
different
delays
over
last
sessions
the
presented.
five
to
components
post-session for
which
during
feedings. each
delay
The was
183
main effect of component was highly significant (F(11,44) = 22.03, p < 0.000 1). The interaction of delay by component was not significant (F(44, 176) = 1.04, p > 0.05).
General
Discussion
Bacotti reported that multiple-schedule sion feedings were delayed than when
response rates were higher when post-sesthey immediately followed the session. He
also reported that response rates decreased throughout the session were immediate. Delayed feedings eliminated this decreasing trend. The
present
experiments
did not reproduce
these
results.
when
Overall
feedings
response
rates
failed to change significantly with delay to post-session feedings in either experiment. Within the session, response rates reached a peak at the fourth component (20 minutes) and then decreased. The delay to post-session feedings did not alter this pattern of responding. That is, the interaction of component by delay was not significant for either experiment. There are many possible explanations
for
the differences
between
the
present
results and those of Bacotti. Because the present experiments studied the generality of Bacotti’s findings, many but not all procedural factors were held constant across the studies. Procedural differences include the following. The present study employed Sprague-Dawley rats; Bacotti employed hooded and Wistar rats. The present experiments used a Fleshler and Hoffman (1962) series to schedule reinforcers; Bacotti used an arithmetic series. The present subjects
the availability of responded at 85 or
95% of their free-feeding body weights; Bacottils subjects responded at 75 or 80%. The present experiments provided all rats with one, 6-g, pellet of laboratory chow as a post-session
feeding; Bacotti determined
the size of the post-session
feeding individu-
ally for each subject. Experiment 2 used sweetened condensed milk as the reinforcer; Bacotti used Noyes pellets. The present experiments conducted 30 or 40 sessions per condition; Bacotti conducted a variable number of sessions that averaged 18 (key pressing) or 26 (lever pressing). A light above the response key was illuminated in Experiment 2; the key itself was illuminated in Bacotti’s study. The present experiments conducted 5 delays to post-session Several of these procedural differences
feedings; Bacotti conducted only two. are not likely to explain the differences
in
results. First, differences in the reinforcers used and in the signalling conditions for the components cannot provide a complete explanation. Experiment 2 differed from Bacotti’s study on these dimensions but Experiment 1 did not. Differences in the number of delays to post-session feedings also cannot provide a complete
account. Experiment
2 which
conducted
Bacotti’s 0 and 60 m delays as the
first two delays nevertheless produced results which differed from Bacotti’s. Responding during the first two delays could not have been influenced by the other delays to be presented
in the future.
Differences in the number of sessions conducted per condition did not produce the differences in results. Our results also failed to replicate Bacottils when the data from the present experiments were examined for the same sessions used by Bacotti (sessions 14 to 18 for key pressing and sessions 22 to 26 for lever pressing). Neither rates of lever (t(4) = 0.73) nor key (t(4) = 1.80) pressing were significantly different (p < 0.05) for the 0 and 60 m delays studied by Bacotti. Finally, differences in the sizes of the post-sessions feedings probably did not produce the differences. A single pellet served as the post-session feeding in the
184
present
study;
in Bacotti’s because
his
varied
from
The
two
weights of
the sizes
study.
subjects only
constant
the
two
By
of
intervals.
Among
maintained
may
The when
rates in the
with nor
for
does
suggested
that studies
a condition
in which
they
of open
of post-session that,
always
and closed
feedings
does
imply
in
which
and closed
circumstances,
not
influence
within-session
post-
differed
is so,
then
study,
and
session
to find
changes
has several
The
need
results
are
However,
do not
given
(open results
the temporal
In contrast, temporal
response
because
Bacotti’s
emphasize
the
rates.
be ignored
economies
feedings
not
feedings.
response
present
in
practical
post-session
and closed
should
present
If this
may sometimes
existence.
the
have
present
within-session
economy).
as their
may
of these and
experiments
of
post-session
economies
as well
mentioned:
In all
apparatus
that
economies. open
of
study.
feedings
between
been
cage.
feedings
influence
less
offered
the experimental
placement
are not (closed
feedings,
under
results
the food
the failure
to post-session
present been
investigation.
not
in the
during
The
by the fact that
in Bacotti’s.
further
of post-session
open
has
of
at 75 or 80%
deserves home
effect
have
somewhat
better
in results,
temporal
not
may
enclosures
than
deserves
the
interreinforcer
maintained
the
behavior
these the
location
compare
imply
First,
between
to one
post-session
were
consumed
These
locations
factor
of the distinction
economy)
results
across
techniques
critical.
in an experimental
between
control,
most
that
and
cages.
two
in the delay
feedings
the temporal
a variable
for the differences
tightly
distinguishing
placement
m delays
do not account
weights,
immediately
experiment
the
the
subjects,
apparatus
present
cage. This
changes
to these
not
60
grams
examine
the explanation
in home
implications.
report,
do
weights
four
both
body
body
0 and
of scheduling
is contradicted
measured
discriminated home
Bacotti’s
consider
were
more
seems
experiments also
delivered
bear on the validity they
than
experimental
of the reason
theoretical delay
study
subjects
in holding
than
should
free-feeding
Nevertheless,
were
in the rates
Second,
the
probably
and method
of their
may have occurred
Regardless
always
present
the mean
Because
experiments
explanation
should
have
interaction
This
the
response
each other
and
The
more
techniques
deprivation
feedings
feedings.
feedings
response
each subject
in Bacotti’s
successful
between by no
by Bacotti.
future
in the present
between
and events
equally
grams
in the feeding
variables,
experiments
experiments,
less
conducted
at 85 or 95%
in the post-session
subjects
grams.
For example,
varied
type of subjects,
weights.
all of the subjects
session
3.2
weights
elimination,
these
free-feeding
from
when
in post-session
Future
for
necessary
to 796
approximately
by only
differences
subjects,
differences
354
conditions.
2. Body
in deprivations,
interested
from
determined
were
in results.
a process
differences
their
varied
points
successful,
were
experimental
subjects
the differences
individually
feedings
in weight
techniques
in Experiment
same
similarly
ranged
across
present
these
were
352 to 404 grams.
feeding
conducted
of the feedings
Individually-determined
the
present
placement
rates
of
and therefore
can be ignored. Third, delivered that
the
the
results
in the time
course
delivered
within
locations,
then
session
feedings
imply
experimental the
for
either
that
enclosure
integrating
experimental
responding
within
became
sufficiently
animals with
these
rewards
enclosure. the
session
delayed
do
those
not
always
delivered is different
If animals should so that
did have
they
integrate
in the from integrate
changed were
rewards
home that
cage,
for
rewards when
no longer
or
rewards the
across post-
integrated
185
with the within-session reinforcers. But, responding changes in delay to post-session feedings.
did not change with
the present
This may imply that animals do not always integrate homecage and experimental rewards. Alternatively, it may imply that the present experiments did not conduct delays that were long enough to reveal this integration. This seems unlikely because the range of delays used in the present experiments exceeds the 16-minute time window of integration suggested by previous research (e.g., Timberlake, Gawley, and Lucas, 1987). But, it remains possible that the time window for integration was longer in the present experiments than in those past studies. Regardless, the present results have important implications.
If the results
show that
animals do not integrate rewards across locations, then they imply that theories need not always include post-session feedings in the context of reinforcement that determines responding within the session (e.g., Herrnstein, 1970). If the present experiments did not employ delays that were long enough to reveal integration, locations mental control
then the results is different
imply
that the time course for integrating
from the time course for integrating
enclosure. Therefore, location should the time window over which animals
events
within
events across the experi-
be added to the list of variables that integrate rewards (e.g., Lucas, Gawley,
and Timberlake, 1988; Timberlake, Cawley, and Lucas, 1988). The systematic changes in response rates that occurred over the course of the session also have several implications. These changes join a growing body of data that suggest that molar measures of response rate mask changes in responding at more molecular levels. The within-session changes also suggest that care must be taken in designing studies when such changes would confound the interpretation of the results. This should not be taken to rule out the use of within-session experimental designs. For example, McSweeney and Melville (1988) argued that within-session designs have some advantage over between-session designs in the study of behavioral contrast. The present results do not necessarily question their design which measures baseline in the first half of the session and contrast in the second half. For example, the average rates of responding
emitted
during
the two halves of the session
cantly different (p ~0.05) for any delay to post-session delay, t(4) = 2.35; 5-minute delay, t(4) = 0.87; IO-minute
were not signifi-
feeding in Experiment 1 (0 delay, t(4) = 0.00; 60-minute
delay t(4) = 1.45; 240-minute delay, t(4) = 0.001. But, because within-session changes in response rates do occur, care must be taken when using a within-session design to ensure that these changes do not confound the interpretation of the results, The factors that govern within-session changes in response rates should be determined in future studies. For example, a study should determine whether response rates peak one-third of the way through the session or twenty minutes into the session. If response rates peak after 20 minutes, then absolute response rates will vary with session length. Studies that allow session length to vary across experimental conditions will be seriously confounded. If, however, peak response rates occur at a point that is relative to session length, then varying session length may not alter overall response rates. Studies that allow session length to vary will not necessarily be confounded. Events occuring outside of the experimental session undoubtedly influence responding within the session under some conditions (e.g., Clark, 1958; Hackenberg and Hineline, 1987). The present experiments suggest that they do not always do so.
186 Future not
experiments
influence
should
determine
within-session
exactly
when
extra-session
events
do
and
do
responding
Acknowledgements Preparation authors
of
would
its preparation. part
of
manuscript
to thank
Drs.
would
also
They
one
Department
this
like
experiment. of
was
Whipple
thank
Reprints
Psychology,
supported
James
Elizabeth
may
be
Washington
by
NIMH
MH
Hinson
for
Krumpter
for
help
obtained
State
grant
and John from
University,
her
Frances Pullman,
42466.
The
help
with
their
in conducting
K.
McSweeney,
WA
99164-4830,
USA.
References Bacotti,
A.V.,
Anim. Clark,
1976.
Home
cage feeding
Learn. and Behavtor,
time
F.C., 1958. The effect of deprivation
respondtng. Fleshier,
controls
responding
under
multiple
schedules.
4: 41-44. and frequency
of reinforcement
on variable-interval
J. Exp. Anal. Behav., 1: 221-228.
M., and Hoffman,
H.S., 1962. A progression
for generating variable-interval
schedules.
J.
Exp. Anal. Behav., 5: 5299530. Hackenberg,
T.D.,
and Hineline,
tion of appetitive Herrnstein, Hursh,
P.N.,
1987.
Remote
effects
of aversive contingencies:
behavior by adjacent avoidance sessions.
Disrup-
J. Exp. Anal. Behav., 48: 161-173.
R.J., 1970. On the law of effect. J. Exp. Anal. Behav., 13: 243-266.
S.R. ( 1984).
Lucas, C.A.,
Behavioral
Cawley,
horizons
economics.
D., and Timberlake,
J. Exp. Anal. Behav., 42: 435-452. W.,
in the rat: Some methodological
McSweeney,
F.K.,
function
Dougan,
of component
I.D., duration
Higa,
1988. Anticipatory determinants.
J., and Farmer,
and baseline
contrast
Anim. V.A.,
as a measure
of time
Learn. Behav., 16: 377-382.
1986.
Behavioral
rate of reinforcement.
Anim.
contrast
as a
Learn. Behav., 14:
1733183. McSweeney,
F.K.,
duration Spealman,
R.D.,
schedules. Timberlake,
control
and Gollub,
L.R.,
C.L.,
1988.
procedure. 1974.
Positive
contrast
Behav. Process.,
Behavioral
as a function
of
component
16: 21-41.
interactions
in mulitple
variable-interval
J. Exp. Anal. Behav., 22: 471-481. W., Cawley,
temporally Timberlake,
and Melville,
using a within-session
D.J., and Lucas, G.A., 1987. Time
separated patches. J. Exp. Psychol.: W., Gawley,
Anim.
D.J., and Lucas, G.A., 1988. Time
of access to future
horizons
in rats foraging
Behav. Process., horizons
food. J. Exp. Anal. Behav., 50: 405-417.
for food in
13: 302-309.
in rats: The effect of operant