Food deprivation enhances both autoshaping and autoshaping impairment by a latent inhibition procedure

Food deprivation enhances both autoshaping and autoshaping impairment by a latent inhibition procedure

Behavioural Processes, 23 (1991) 59-74 0 1991 Elsevier Science Publishers B.V. 0376-6357/91/$03.50 BEPROC 59 00331 Food deprivation enhances both ...

1MB Sizes 4 Downloads 47 Views

Behavioural Processes, 23 (1991) 59-74 0 1991 Elsevier Science Publishers B.V. 0376-6357/91/$03.50 BEPROC

59

00331

Food deprivation

enhances both autoshaping

and autoshaping

impairment

inhibition Sheldon Department

B. Sparber,

of Pharmacology,

George

by a latent

procedure L. Bollweg

and Rita B. Messing

Medical School, University Minnesota 55455, USA

of Minnesota,

Minneapolis,

(Accepted 28 August 1990)

Abstract The

influence

of food deprivation

on acquisition

of autoshaped

operant behavior

was measured. In one study separate groups of young, male rats that were deprived to 75%, 80%, 85%, 90%, and 95% of ad lib weight were subjected to an autoshaping procedure

in which

a 6 s delay was

interposed

between

lever

retraction

(which

occurred when rats made a lever touch, or automatically after 15 s) and food pellet delivery. In a second study, groups of rats were deprived to 80% or 90% of ad lib weight prior to testing in a latent inhibition variation of the same autoshaping procedure. This learning which,

was done to determine if greater food deprivation would enhance because of the latent inhibition manipulation, is manifest as less

lever-directed behavior. Greater food deprivation was associated both with fast acquisition of autoshaped lever responding and with more reliable failure to increase lever responding in the latent inhibition paradigm. Thus, increasing food deprivation was associated with enhanced acquisition regardless of whether the required performance was an increase or a failure to increase the same behavior, indicating effect on learning.

Key words:

Autoshaping;

Food deprivation;

a specific

Latent inhibition

Introduction Food intake by experimental subjects in a laboratory setting is usually restricted when studying behavior which is reinforced with food, and behavioral consequences

60 associated with changes in levels of deprivation are often thought of as the result of changes in motivation. For example, Hovancik et al. (1984) found that increasing deprivation positively influenced the speed of responding, but not the level of discrimination performance during acquisition of a straight-alley runway task. But much evidence exists which suggests that deprivation interpreted. Peterson and McHose (1980) used a runway

effects are not so easily to a goal to show that the

speed of running (interpreted as a measure of reinforcer incentive value) is positively related to deprivation state for the first exposure to the reinforcer, but does not change in the expected way if deprivation state is changed, replicating much earlier work of Butter and Campbell (1960). Other work has shown that behaviors which are maintained by reinforcers other than food can be dramatically

affected by the level of food deprivation,

questioning

the utility of a simple explanation relying upon motivation or incentive for the deprived stimulus, since there is clear generalization to other reinforcing stimuli. Thus, Carroll (1985) found that food deprivation nearly doubled the response rate of monkeys on a fixed interval (FI) schedule of reinforcement for oral delivery of the abused hallucinogen

phencyclidine;

Gaiardi et al. (1987) similarly

concluded

that the

reinforcing and discriminative stimulus properties of morphine are greater in food deprived than in food satiated rats, using a place conditioning task. In tests of unconditioned

locomotor

activity

(i.e.

behavioral

arousal),

Campbell

and

Fibiger

(1971) found that food deprivation markedly potentiated the stimulant effect of amphetamine, and vice versa, suggesting that food deprivation may increase the responsiveness of an organism to stimuli other than food, although the role of drug disposition between nontarget tissue (e.g., adipose) and target (e.g., brain), under various degrees of deprivation Coveney and Sparber, 1990).

has not been adequately studied

(Sparber et al., 1977;

Autoshaping (Brown and Jenkins, 1968) combines aspects of classical conditioning, where the event sequence in an experiment occurs no matter what the subject does, and operant conditioning, where emission of a specific behavior brings reward or punishment.

We have previously

observed that autoshaped

amenable to parametric manipulations

lever-directed

of the delay of reinforcement,

behavior is

a consequential

variable (Messing et al., 1986), and herein systematically manipulate the antecedent variable, degree of food deprivation. Further, we wished to determine if another treatment (administration of the vasopressin vasopressin; DGA VP) that enhances acquisition optimally,

because of age or a CNS lesion,

analog desglycinamide-8-arginine in rats that may not be learning

may do so by amplifying

or mimicking

the

positive influence of increased food deprivation upon learning and performance. We have reported that acquisition of forward autoshaped behavior, with a 6 second delay of reward following retraction of a lever can be enhanced in older rats by administration of DGA VP (Messing and Sparber, 1985). Furthermore, this peptide, which is devoid of significant pressor and endocrine properties (Kovacs and DeWied, 1983), reversed a deficit in autoshaped learning caused by treatment of young, mature rats with the neurotoxin trimethyltin (Sparber et al., 1988). In one study, the rate of learning (acquisition) was examined in separate groups of male rats deprived and maintained at 75%, 80%, 85%, 90%, and 95% of ad lib weight. In a second study, rats were deprived to 80% or 90% of their ad lib weight and injected with either DCA VP or saline and tested in a latent inhibition variation of the autoshaping

task (Lubow

and Moore,

1959).

In this

procedure,

subjects

repeatedly

61

exposed stimulus

to the to-be-conditioned stimulus show retardation of conditioning

not yet paired with an unconditioned upon subsequent introduction of the

unconditioned stimulus (reinforcer). This study was carried out to determine if greater food deprivation and/or peptide injection increased the likelihood that animals would fail to autoshape after reinforcement was instated. By using a latent inhibition manipulation, we hoped to further control for the possibility that enhanced learning by more food deprived rats was nothing more than a reflection of a generalized arousal effect, thereby increasing the probability that lever-directed behavior would be emitted

during

the conditioning

process.

Lest the reader think that a 75% deprivation level is unduly harsh or cruel, such regimens, or even more severe regimens, have been found to produce robust decreases

in

experiments

age related

changes

and increases

in

life

span

conducted over the past 50 years (cf. Goodrick

of

rats

in

numerous

et al., 1982).

Methods

Subjects Autoshaping. Thirty five male Long Evans rats, nine weeks of age (Blue Spruce, Altamont, NY) were individually housed and maintained in a temperature (22 + 1°C) and humidity (40-50%) controlled environment on a 12:12 hour light/dark schedule, with

lights on at 0700. Food (Purina

rat chow) and tap water were available ad lib for

one week. After this period of acclimation the rats weighed 302 f 10.6 g (mean k s.d.). They were then divided into 5 groups of 7 rats/group and food deprived over one week and maintained within +2% of 95%, 90%, 85%, 80% and 75% of their ad lib weights.

Before

behavioral

testing

began, a dish

containing

twenty

45

mg rein-

forcement pellets (#0021, Bioserv Inc., Frenchtown, NJ) was given to each animal in its home cage prior to its normal feeding. Each animal consumed all of the pellets within two

hours.

reinforcer

This

was

done

to assure

the

acceptability

and effectiveness

of

the

and to obviate neophobia.

Forty male Long Evans rats were obtained and Latent inhibition manipulation. treated as above. Weights after ad lib feeding (330 + 16.5 g), and maintenance at weights during behavioral testing (80 or 90% of ad lib weight, means for individual rats over

5 preexposure

sessions

and 10 autoshaping

sessions

+2%)

were

similar

to

Experiment 1. Subjects were maintained at the same deprivation level during the behavioral sessions in four treatment groups of 10 rats per group: 90% DGA VP, 90% saline, 80% DGA VP, and 80% saline. However, one rat in the 90% saline group had to be excluded from data analyses because of illness during reinforced autoshaping sessions. DGA VP (kindly supplied by Organon International, Oss, The Netherlands) was dissolved in 0.9% NaCl solution (saline). DGA VP (7.5 pg/kg body weight) or saline (1 ml/kg)

was injected S.C. 60 minutes

prior to the preexposure

sessions.

Apparatus Laurel, MD) Rats were tested in Skinner boxes (Model 143-22, BRS/LVE, isolation chambers with closed circuit television specially constructed

placed in cameras

62 (Sparber, 1980). Skinner boxes were 31 cm wide X 25 cm deep X 25 cm high, with grid floors consisting of 0.5 cm diameter stainless steel bars spaced 2 cm apart, center to center. Each box had standard house and cue lamps and a speaker that delivered a white masking noise. Boxes contained a metal strip 7.5 cm wide on two walls, the bottom edge of which was 15 cm above the grid floor. Contacts with the strip (strip touches)

indicated exploratory

rearing behavior, and were monitored

eter-type touch circuit standardized with a retractable lever (BRS/LVE

against a 2M D resistance. Model RRL-005,

with a drinkom-

Each box was equipped

Laurel, MD),

and contacts with the

lever, extended or retracted, were monitored with a touch circuit similarly calibrated. The boxes were controlled and data were collected and data were collected by TRS80 Color Computers

(Tandy Corp., Fort Worth,

each treatment group were distributed potential source of variation.

TX) and custom

equally

built interfaces,

Rats from

among the boxes to control

for that

Procedure Autoshaping. Nine autoshaping sessions consisting of 24 trials per session were conducted on Tuesdays, Wednesdays and Thursdays for three weeks. Sessions began with illumination of the house lamp and a white cue lamp above the lever; both remained

lit

throughout

each 24 trial

session.

A trial

consisted

of

a computer

generated random interval with an average of 45 s and a range of 22-68 s (the intertrial interval, or ITI), followed by a lever extension into the Skinner box. The lever retracted when the rat touched the extended lever (extended lever touch, ELT). If 15 s elapsed with no ELT, the lever also retracted. A food pellet was delivered 6 s after lever retraction; thus, 6 s constituted the reinforcement delay (RFD). Food pellet delivery ended the trial. While older rats can be autoshaped with a 6 s delay of reinforcement, their rate of acquisition with a longer delay is much slower (Messing et al., 1986). Moreover,

delays shorter

in acquisition

than 6 s may not be optimal

rates caused by low doses of a neurotoxin

for studying

subtle

differences

(Cohen et al., 1987). Since we

have used a 6 s delay of reinforcement for such studies, keeping deprivation levels constant, we chose to characterize the effects of various levels of deprivation, using a 6 s delay. Latent inhibition manipulation. Sessions (12 trials per session) were identical to those for autoshaping, but the pellet dispenser was disconnected for the first 5 sessions ml/kg)

and the animals were injected S.C. 60 min prior to testing.

with either DGA VP (7.5 pg/kg) or saline (1 These preexposure sessions, without rein-

forcement, were conducted Monday-Friday, and were followed by 10 autoshaping sessions (12 trials/session) with reinforcer delivery and no injections, which were conducted Monday-Friday over the following 2 weeks (no sessions were run over weekends). Deprivation levels were maintained all I5 behavioral sessions.

Data collection

at 80% or 90% of ad lib weight during

and statistics

Autoshaping. Data consisted of ELT, interim lever touches (paw and nose poking behavior toward the retracted lever during the ITI and RFD), latencies to respond to the extended

lever, and strip

touches

(measuring

unconditioned

exploratory

rearing

63

behavior).

Interim

lever touches

represent

adjunctive

and/or

superstitious

and rate increases in this variable provide a measure of learning

behavior,

not as constrained

by

ceiling effects as ELT data (Messing et al., 1986). Strip touches occur to a greater extent shortly after introducing the subject to the novel chamber and tend to diminish as habituation occurs and/or attention is focused on the lever and food delivery through. However, we have found strip touches to be a convenient measure of drug induced arousal or locomotor stimulatory effects (Sparber, 1980), which may contribute to, or be responsible 1987).

Failure

measures

for, altered acquisition

to respond

to the extended

of total latency. Additionally,

of autoshaped behavior (Huang et al.,

lever was treated

latencies

as a I5

for actual touches

s latency

for

of the extended

lever (ELT latency) were analyzed to determine if deprivation affected response speed after learning occurred. To control for the effect of “practice” in subjects that acquired the lever-food association in early, rather than in later sessions, speed of responding was thus analyzed on the day each rat achieved a learning criterion: the number of ELT exceeding the 95% confidence interval for basal ELT established for all rats on the first day of autoshaping, during 2 of 3 consecutive sessions. It is probable that some rats begin to form an association between the manipulandum and food pellet delivery during the first session. Therefore, using the first session statistic for baseline (unconditioned) behavior yields a conservative measure of learning. Square root transformations

were used for analyses of interim

lever touches and strip touches

to deal with heterogeneity of variance for these measures. Error terms from ANOVAs were used to compute Duncan’s t-statistics on overall means (mean of nine sessions) of the behavioral

measures

(Winer,

1971).

Evidence of learning

by individual

groups

was obtained by performing serial repeated measures ANOVAs across sessions (l-2, l-3, l-4, etc.) until a significant difference was found between the first and last session

analyzed (in the appropriate direction)

by Fisher’s

LSD test. ELT were analyzed

with a criterion for purposes of direct comparison with data from experiment 2 (see below). This criterion was defined as the number of ELT greater than the upper limit of a 95% confidence interval established on the first autoshaping session (computed as 18.54 of 24 possible ELT). For a 3 session (weekly) block, learning was therefore defined as > 56 ELT per 3 session block. Rats were divided into a low deprivation (90% and 95% of ad lib weight)

and a high deprivation

(75% and 80% of ad lib weight)

group and x’analysis was then carried out, comparing totals of learners and nonlearners in each 3 session block. Latent inhibition manipulation. Data were collected as in autoshaping. Due to the bimodal distribution of data in the reinforced sessions, x2-comparisons of ELT were performed, using a criterion value to divide rats into two categories; those achieving higher levels of performance than observed in preexposure sessions, and those failing to achieve such levels (i.e., showing latent inhibition). The criterion for failure to increase autoshaped lever responding was defined as the upper limit of a 95% confidence interval for ELT for all animals during the preexposure sessions, or 23 or fewer ELT (5 sessions

Xl2

possible

ELT/session

= 60 possible

ELT).

Results The results indicate autoshaping acquisition tion.

that, within the range of deprivation levels utilized, both and latent inhibition are enhanced by greater food depriva-

64

u 360 -

95%

-

90%

-

85%

-

80%

300 z z

240 -

z + 5

180 -

ti

120-

60 -

0 : 0

I

I

12

I

I

I

I

I

I

,

3

4

5

6

7

8

9

SESSION Fig. 1. Relationship between deprivation state and total latencies to touch the extended lever (24 trials X 15 s maximum/session). Each point represents the mean for a group of 7 rats.

Autoshaping. Figure 1 total latency to touch the extended lever. As is apparent, more food deprivation was associated with a shorter latency. No Group differences were seen in Session 1 (one factor ANOVA, F,,,o= 0.434, p = 0.783). Repeated measures ANOVA over the nine autoshaping sessions showed significant effects of Group (F4.30 = 3.229, p = 0.026), Session (F8,zW = 27.39, p < O.OOl), and a Group by Session interaction (F32,240= 1.831, p = 0.006). There were significant differences between 95% and 75% groups (t = 4.13, p < 0.05) and 90% and 75% groups (t = 3.88, p < 0.05) by Duncan’s test on the combined 9 session means. The stepwise nature of the group differences

over the last four sessions

evidence of the lever-pellet

is also apparent. The 95% group showed

little

associated even after nine sessions, and did not attain a (F,,, = 0.83, p = 0.580). The 75%, 80%, 85% and 90% groups

significant effect of Session showed the effect of Session

after 4 (Fj,,8 = 8.3, p = O.OOl), 4 (F3,18 = 4.20, p = 0.020), 5 = 3.85, p = 0.015), and 7 sessions (F,,,, = 4.161, p = 0.003), respectively. (F‘W As did latency analyses, similar analyses of ELT behavior (not shown) indicated that greater deprivation was generally associated with more extended lever touching. There

was

F4,3,,= 0.604,

no evidence p = 0.662).

of Group

differences

in

session

1 (one

factor

ANOVA,

Analysis

by repeated measures ANOVA over all 9 sessions showed significant effects of Group (F,,,, = 3.028, p = 0.033) and Session (F8,zm = 11.86, p c 0.001). As was the case with latency, significant differences by Duncan’s test on 9 session means emerged between 95% and 75% groups (t = 4.11, p c 0.05) and 90% and 75% groups (t = 3.79, p i 0.05). Because of the bimodal distribution of the data derived from the latent inhibition procedure (see below), ELT data from experiment 1 were also analyzed by x2 to facilitate comparison between experiments. Thus, ELT responding of individual animals in high and low deprivation groups was plotted in 3 session blocks (sessions l-3, 4-6, and 7-9)

in Figure 2, omitting

the intermediate

deprivation

group (85% ad lib weight)

65

Deprlvztion

low

low

high

Sess. 1-3 Fig. 2. Relationship shaping.

Sessions

deprivation: omitting

between were

“high”

combined

session)

95% confidence

defined

learning.

and

combining

groups (“low

represents interval

In sessions

p =0.105).

learned

into

(85%) group.

The dashed line at 56 ELT

subjects

state

Sess. 7-9 lever

blocks

of

extended

3 and groups

l-3, 4-6,

established

4-6

and 7-9,

(ELT)

touches were

x2 = 7.146, df =I, p = 0.008; p = 0.020).

to

autoreflect

(90% and 95% of ad lib weight), by a “+I’

in session

significantly

during

combined (high)

greater than the upper

high and low deprivation

In sessions

(sessions

or “low”

Each rat is represented ELT

high

Sess. 4-6

a value 3 times

for

low

and

(75% and 80% of ad lib weight)

the intermediate

2.625, df =I,

deprivation

hfgh

or a “0” (low).

limit

1 by all subjects;

did not differ

of a (single 2 56 ELT

significantly

(x2 =

more high than low deprivation sessions

7-9,

x2 = 5.39,

df =I,

the 75% and 80% groups (“high” deprivation) and the 90% and 95% deprivation). Analysis of block 1 indicated nonsignificant differences in

the number of subjects reaching the criterion, when comparing high (4 of 14) and low (0 of 14) deprivation groups. Significant differences did emerge in block 2, with the number of subjects reaching criterion in the high deprivation group (10 of 14) greater than the number in the low deprivation group (2 of 14). This difference continued into the third block; the number of high deprivation group (5-14). Analysis of ELT by both x2 and ANOVA thus reported the latency results, showing a significant positive influence of greater food deprivation on delayed reinforcement autoshaped learning.

66 g

35

B

1

Ill 30u iTi 0 25St!. Y

20Y

5 c

15”

E >

W A

loY

z = ii

z

5-

02 0123456789

SESSION Fig. 3. Relationship

between deprivation

interval plus reinforcement

state and square root of interim

delay) lever touching.

Each point represents

(the sum of intertrial

the mean for a group of

7 rats.

Interim There

lever

were

touching

no session

data

p = 0.138), but significant measure)

( Fa,240= 25.569,

p = 0.037 emerged

(square

1 responding effects

for Group

p -C 0.001)

over 9 sessions,

on the means for all 9 sessions

root

transformed)

differences (F4,,,

are

depicted

in

Figure

3.

ANOVA (F,,a, = 1.89, = 3.303, p = 0.023), Session (repeated by one

factor

by Session interaction (F32,240= 1.546, as well as significant differences by Duncan’s test

and a Group

between the 95% and 80% groups (t = 4.24, p < O.Ol),

the 95% and 75% groups (t = 3.49, p -C0.05), and the 90% and 80% groups (t = 3.64, p < 0.05). A significant effect of Session was reached by the 95% group on session 9 (F&W3= 2.45, p = 0.026). This effect occurred on session 8 in the 90% (F,,d2 = 4.66, p = 0.001) and 85% (F7,42 = 3.05, p =O.Oll) groups, and on session 4 in the 80% (Fa.18= 7.14, p = 0.002) and 75% (F3,,8 = 4.91, p = 0.002) groups. The 9 session mean of the 80% group was higher than (but not significantly different from) the 75% group for

interim level touches. To better characterize the learning enhancement associated with greater food deprivation, total ELT latency per session, counting only those trials in which a touch actually occurred, and deprivation were regressed and are displayed in Figures 4A and 4B. Figure 4A shows a significant

positive

correlation

between the number of sessions

to attain a learning criterion (19 ELT, rounded upward from 18.54 [the upper limit of the 95% confidence interval for all groups on session I], on 2 of 3 consecutive sessions), and the average session 9 ELT latency for individual animals (y = 0.439 X 0.178; r = 0.65, p = 0.001). Thus, rats which reached the criterion sooner when more “practiced” by session 9 and took less time to respond to the lever when it was extended. In contrast, Figure 4B shows no significant association between deprivation status and average ELT latency on the session this criterion was reached. Thus, when practice effects are eliminated, deprivation level was not related to speed of responding during the acquisition

phase.

67 3

' 1

y =

- 0.178 + 0.439x

R=0.65, P=.OOl

Sessions Fig. 4A. Correlation

between first session

ELT, 2 of 3 consecutive

sessions)

to

on which

Criterion

a learning criterion

and mean ELT latency (s) on session

animal that reached the criterion,

7

regardless

P y = 1

CV ._

1, 70

I

Fig. 46. Lack of correlation

q

I

1

I

Strip

between

on which a learning criterion

the

touching

data,

as the session

groups

did

not

I 95

deprivation

an

differ

p < 0.001) was (FWO = 13.01, p = 0.071) was not significant.

state and average ELT

latency (s) on the first sessions).

an animal that reached the criterion.

indication

progressed.

VVEght)

was reached (19 or more ELT, 2 of 3 consecutive

Each point represents

creased

an

status.

R = ,237, ~~300

Grbsup (% sFOree FeEling session

was reached (19 or more

9. Each point represents

of deprivation

7.3717 - .0434x

i;j6

Q

of

As with

(F4,so = 1.81, observed

exploratory the other

rearing

behavior,

measures,

session

p =0.153). over

9 sessions;

A

significant the

Group

generally 1 levels

effect effect

of

de-

among Session

(F4,s0 = 2.41,

68 The autoshaping data thus support the idea that within the deprivation range we tested, more deprivation is associated with more rapid acquisition of the lever-pellet association, manifest as more behaviour directed toward the extended and retracted lever. Furthermore, greater deprivation is not associated with a systematic, general increase in activity level in the operant chamber during acquisition sessions. This suggests

that enhanced learning

of the lever-pellet

animals is not secondary to greater probabilities

association

of touching

in the more deprived

the extended lever during

early sessions. Repeated measures ANOVA of ELT, interim lever Latent inhibition manipulation. touching, total latency, and strip touches for preexposure sessions l-5, during which

00 0

* -c +++ *

: 0

+

90

80

L

% Ad Ilb S.W.

90

90

80

between

deprivation

state and ELT

The dashed line at 23 ELT is the upper limit preexposure by a ‘I+”

sessions, (“hi”,

reinforcement) were

and defines

80%

no significant

p > 0.25); in sessions (x2 = 5.72, df =l,

with

differences 6-10,

p <0.02),

for the latent

of a 95% confidence

inhibition

interval

manipulation.

established

during

the

failure

ad lib weight)

is contrasted

Sess.6-lo

Sess. i-5

Pre-exposure Fig. 5. Relationship

a0

to increase lever responding. Each rat is represented Preexposure (no or a “0” (“lo”, 90% ad lib weight).

sessions during

1-5

and sessions

preexposure,

the 80% group responded demonstrating

6-10

or during significantly

greater likelihood sponding.

(reinforcer

session

l-5

delivered). (x2=0.27,

There df =I,

less than did the 90% group

of failure

to increase

lever

re-

69

time pellets were not delivered (data not shown), showed only an effect of Session (except for ELT, see below), as follows: ELT, F,,,,, = 2.959, p = 0.022 (ELT decreasing over sessions); interim lever touching: F,,,,, = 10.863, p -C0.001 (decreasing over sesover sessions); and strip sions); total latency: F,,,,, = 2.746, p = 0.031 (increasing touching: F,,,,, = 3.877, p = 0.005 (decreasing over sessions). After the preexposure sessions, rats were autoshaped for IO sessions with delivery of the reinforcer 6 s after each lever retraction, as in experiment 1. ELT responding during preexposure increase

and autoshaping

lever responding

below). The saline-DGAVP

was less

comparison

is depicted in Figure 5. The criterion than or equal to 23 responses/5 did

not

reach significance

for

for failure to sessions

either

(see

block

of

reinforced sessions (x2 = 0.03, df = 1, p > 0.75; x2 = 0.94, df = 1, p > 0.25, respectively). The 80%-90% comparison (collapsed across DCAVP or saline treatment) for the first 5 session

block of autoshaping

was not different

(x2 = 0.27, df = 1, p > 0.25); behavior

toward the lever appeared to be depressed in all groups as a result of the latent inhibition manipulation. In the second block, sessions 6-10, a difference emerged. The greater deprivation of the 80% group was thus associated with more failures to increase responding, with 9 of 20 subjects remaining below the criterion, against only 2 of 19 in the less deprived group. The possibility that exploratory strip touches were affected by peptide treatment or food deprivation was also investigated. Repeated measures ANOVA of the preexposure sessions (where reinforcement was not delivered) showed no effect of Treatment (F,,,, = 0.408, p = 0.563), Deprivation (F1,36 = 0.0003, p = 0.985), nor Deprivation by Treatment interaction (F,,,, = 0.037, p = 0.849). There was an effect of Session by Deprivation (F,,,, = 0.982, p = 0.419), nor Session by Deprivation by Treatment (F,,,,, = 0.928, p = 0.449). These results indicate a lack of difference among the groups in strip touching during the preexposure sessions. Repeated measures ANOVA of strip touches for the subsequent 10 autoshaping sessions showed no effect of peptide the 90% Treatment (F,,35 = 2.276, p = 0.140), but there was an effect of Deprivation: group touched the strip more (a ten session average of 5.5) than the 80% group (a ten session average of 4.7; both values, square root strip touches) (F,,35 = 7.081, p = 0.012). There was no overall Treatment by Deprivation interaction (F1,35 = 0.060, p = 0.809), but there was a Session

effect (F9,31s = 10.873,

p < 0.001). There

was no interaction

of

Session by Treatment (F9,3,5 = 1.055, p = 0.396), Session by Deprivation (F9,3,5 = 0.567, p = 0.824), nor Deprivation by Treatment by Session interaction (F9,3,5 = 0.829, p = 0.590).

Discussion We found significant enhancement of acquisition of autoshaped behavior with increasing deprivation, as well as greater likelihood that more deprived rats would fail to increase the same behavior after a latent inhibition manipulation. Thus, learning was enhanced by food deprivation

when food was used as a reinforcer

regardless

of

whether the same behavior was increased or decreased. There was no effect of food deprivation on session 1 measures for autoshaping, or on lever-directed behaviors during preexposure sessions for latent inhibition. This argues against the possibility that food deprivation generally increases reactivity or unconditioned behavior toward

70

the lever, which may have thereby enhanced autoshaped acquisition by virtue of a greater probability of the emitted behaviors being reinforced. When we analyzed the speed of responding conservative criterion

(latency to touch the extended lever) on the day when a for learning was achieved in the autoshaping experiment, we

found

between

no correlation

deprivation

lever

support the notion that learning, not response food deprivation. Moreover, since lever-directed

and response

latency.

Such

data

speed or reactivity, is enhanced by behaviors during the 5 preexposure

sessions of the latent inhibition manipulation were also not different when comparing the more and less food deprived groups, the greater likelihood that the subjects in the more deprived group would fail to increase responding cannot be accounted for by their experiencing a greater number of unreinforced lever touches. Further

evidence that food deprivation

performance

are data showing

effects on learning are not due to effects on

that rats which acquired the autoshaped lever response

under food deprivation (80% of ad lib. weight) continued to respond at high levels three weeks later, when they were reintroduced to the apparatus after being allowed to return to 100% of their former ad lib body weights (Kim and Sparber, unpublished observations). These results complement work showing that the establishment and maintenance of associations between events are enhanced by food deprivation even if food is not used as a reinforcing stimulus: reinforcing effects of drugs are increased as a result of greater food deprivation (reviewed by Carroll and Meisch, 1984). Thus, the enhancing consequences of food deprivation simply explained by generalized

up cognitive processes, broadly defined, cannot be arousal, specific incentive, or other motivational

constructs which have been used to interpret which the subject is more deprived. Exploratory

rearing

behavior

behavior

was not systematically

directed

toward

stimuli

affected by treatment

to

in the

autoshaping study, but was affected by deprivation level after the latent inhibition manipulation. Exploratory rearing behavior was significantly greater in the less deprived, compared to the more deprived group, which also showed more lever-directed behaviors during autoshaping after the latent inhibition manipulation. We interpret this as evidence for less latent inhibition in the less deprived group. It is most likely that strip touch behavior under these special conditions is a measure of early adventitiously reinforced behaviors, and correlates with other behaviors directed by the less deprived group toward its environment, upon which it is operating during the establishment of the association between retraction of the lever and food pellet delivery. Like the more deprived group, this group experienced 5 sessions during which lever extension and retraction were unrelated to any food pellets. When the food pellets were later introduced, the fact that the less deprived group was also less affected by the latent inhibition manipulation most likely was responsible for the emergence

of multiple

types of behaviors

until

they autoshaped

to the lever by a

process of successive approximations. Observation of several rats during acquisition of autoshaped behavior (through closed circuit television systems in the sound attenuating environment) also supports this interpretation. In contrast to the present data, prior work in this laboratory showed that DCAVP enhances acquisition of autoshaped behavior (Messing and Sparber, 1983; 1985); however, another laboratory did not replicate these results (Mundy and Iwamoto, 1987). These conflicting results are an indication of the general controversy surrounding the study of vasopressin and its analogs in learning and memory. Strupp and

71

Levitsky “ceiling

(1985) argue that vasopressin and its analogs do affect cognition, but that effects” may militate against observations of positive effects. Mundy and

lwamoto used 265-345 g male Sprague Dawley rats, which learned the autoshaping association very quickly. Our earlier studies (in contrast to the present one) utilized much older, heavier rats (up to 625 g), which learned more slowly, perhaps because of an aging process, or because maintenance of rats at similar percentages of ad lib body weights

might produce very different

levels of effective deprivation

when ad lib body

weights vary by a factor of two. Thus, it may be that the subjects used by Mundy and lwamoto were already learning and performing optimally, and therefore were not susceptible

to the peptide’s enhancing property.

In fact, we too have failed to observe

enhancement of autoshaped behavior acquisition when younger, with fast acquisition rates have been given DGAVP (unpublished the present

study,

when

DGAVP

was given during

preexposure

lighter, healthy rats observations), or in sessions

for latent

inhibition, which also utilized younger, lighter, healthy (i.e., nonlesioned) rats, presumably also capable of learning optimally. In contrast, we did find that DGAVP reversed

the learning

deficit

of rats compromised

by administration

of the

limbic

forebrain neurotoxin trimethyltin (TMT) (Sparber et al., 1988). The consequences of varying degrees of deprivation of food and/or water under laboratory conditions upon learning or performance of learned, stabilized behavior probably involve multiple systems which can also be activated by other experimental manipulations (e.g., drug administration) which induces physiological changes associated with inefficient

life sustaining natural needs. From an evolutionary viewpoint, it would be to activate, via separate and distinct chemical signals, diverse organ sys-

tems, each of which

produces separate and distinct

receptor mediated responses

for

altering function in an uncoordinated manner. It is more parsimonius to posit the evolution of hierarchical systems, with each level activated by a neurohumoral factor functioning for maintenance of homeostasis in diverse organ systems. By having relatively few biochemical signals (e.g., steroids, peptides, biogenic amines), each with diverse physical chemical properties and time courses of action, and diverse receptor types and subtypes for these signals, great economy can be achieved. This systems analysis perspective appears to be useful for describing many biological phenomena, including what may be the biological bases for our observations. For example, the ability to conserve water, focus attention, and otherwise cognitive function are of obvious utility under conditions of life threatening or food shortage,

which

necessitate

exploration

of new territory

sharpen drought

and coping

unknown predators. It is possible that there is a critical level of deprivation, laboratory conditions, for activation of physiological and behavioral processes

with under which

are comparable to those activated by such naturally occurring emergencies. The finding by Berry and Swain (1989) that nictitating membrane conditioning improved in rabbits that were water deprived for 22 hrs, compared to controls with ad lib access to water, supports this idea. Dorsa and Bottemiller (1983) found that vasopressin-like immunoreactivity (VP-LI) is differentially affected by water deprivation: simultaneously reduced in the neurointermediate pituitary and increased in plasma. They also found reduced VP-LI in septum and concurrent increases in amygdala, demonstrating responsiveness to water deprivation in brain regions involved with functions other than fluid regulation. Thus, it is tempting to speculate that some of the effects of food deprivation upon learning and performance may be produced via similar endocrinological and humoral changes. Recent work supports the notion that physiological

72

modulation of hormones and neurotransmitters may be involved in the influence of food deprivation on learning. For example, food deprivation or fasting engenders increased plasma corticosterone in rats (Suemaru et al., 1986) and increased plasma cortisol in humans (Fichter and Pirke, 1984). The hippocampus, a structure very much involved

in learning

and memory,

shows

the highest

concentration

of corticosterone

and its receptors in rat brain (McEwen, 1982), and there is evidence that autoshaping and operant behavioral adaptation in general are dependent upon normal functioning of this brain region and its adrenal steroid receptors. We have found that deficits in acquisition

of autoshaped

lever responding

and in operant response

progressive fixed ratio challenge (in which the ratio requirement in rats given TMT are correlated with decreases in corticosterone campus (Gerbec et al., 1988; Messing et al., 1988). Further, depriving rats to 75% of ad lib weight attenuated deficits

levels

during

a

is double each day) receptors in hippo-

like DGAVP administration, in autoshaping seen in rats

treated with TMT (Bollweg and Sparber, unpublished observations). Other work indicates interactions among endogenous glucocorticoids,

vasopressin,

and catecholamines that may be related to the food deprivation associated learning enhancement that we observed. For example, Stone et al. (1987) reported interactions between

catecholaminergic

receptor

stimulation

and the

presence

or absence of

corticosterone. When fi-adrenergic receptors in rat cortex slices were stimulated, there was an increase in second messenger cyclic AMP. This was further augmented by concurrent stimulation of a-adrenergic receptors. The augmentation was suppressed by chronic ACTH or corticosterone treatment or enhanced by adrenalectomy. Moreover, in rat hippocampal slices, AVP potentiated norepinephrine (NE) induced padrenergic-mediated cyclic AMP accumulation (Brinton and McEwen, 1989). Food deprivation is associated with a decrease in fi-adrenergic receptors (Stone, 1983), while TMT induces increased P-receptor density in forebrain (Messing and Sparber, 1986). Since food deprivation can affect corticosterone levels (e.g., Suemaru et al., 1986) and hypothalamic and/or other brain region vasopressin concentrations (Jhanwar-Uniyal et al., 1988), it may influence transduction processes of monoaminergic the positive

influence

the modulation of the signal reception and receptor types and subtypes. Thus, some of

of food deprivation

observed

in the present

experiments

may

have been mediated by influence upon glucocorticoids, vasopressin, catecholamines and their respective receptors. Experiments that directly determine relationships between food deprivation,

these endogenous

substances,

and learning

are necessary

to test this hypothesis.

Acknowledgements This

research was partially

supported

by USPHS

grants DA 01880

and HD 20111.

References Berry,

SD.

facilitates Brinton,

R.E.

and Swain, nictitating

R.A.

(1989).

membrane

and McEwen,

Journal of Neuroscience,

B.S.

Water

deprivation

conditioning. Vasopressin

9 (3), 752-759.

optimizes

Behavioral

hippocampal

Neuroscience,

neuromodulation

in the

activity

and

103 (I), 71-76. hippocampus.

(1989).

73 Brown,

P.L. and Jenkins,

Experimental Butter,

H.M.

Analysis

C.M. and Campbell,

drive level. (1960). Campbell,

B.A.

starvation. Carroll,

(1985). Carroll,

B.A. Running

and Fibiger,

H.C.

Performance

speed as a function

(1971).

maintained

and fixed-interval

(1968).

of successive

and Physiological

Potentiation

M.E. and Meisch,

Vol. 4 (pp. 47-88).

by orally

schedules

Journal of Pharmacology

Journal

of

reversals

Psychology,

53 (I),

of the

in hunger

52-54.

amphetamine-induced

delivered

and Experimental

R.A. (1984).

phencyclidine

in food-satiated

Orlando,

under

and food-deprived

Therapeutics,

arousal

by

Increased drug-reinforced

second-order,

rhesus

monkeys.

232 (2), 351-359. behavior

P.B. Dews and J.E. Barrett (Eds), Advances

C.A., Messing,

reinforcement

key-peck.

233, 424-4125.

tion. In T. Thompson, Cohen,

of the pigeon’s

11, l-8.

Journal of Comparative

Nature,

M.E.

tandem

Auto-shaping

of Behavior,

due to food depriva-

in Behavioral

Pharmacology,

FL: Academic Press.

R.B. and Sparber,

autoshaped

S.B. (1987).

Selective

learning

impairment

behavior caused by low doses of trimethyltin.

of delayed

Psychopharmacol-

ogy, 93, 301-307. Coveney,

J.R. and Sparber,

Interaction

S.B.

(1990).

of food deprivation,

Delayed

stress

effects

and dose.

of amphetamine

Pharmacology

or

phencyclidine:

Biochemistry

and Behavior,

36, 443-449. Dorsa,

D.M.

and Bottemiller,

L.A. (1983).

logic and chromatographic Peptides, Fichter,

and their

peptides

response

in the rat brain:

to water

deprivation.

immunoRegulatory

6, 393-403.

M.M.

and Pirke,

subjects.

In K.M.

124-135).

Berlin:

the stimulus Behavior,

K.M.

Pirke

(1984).

Hypothalamic

and D. Ploog

(Eds.),

The

pituitary

function

Psychobiology

in starving

of Anorexia

healthy

Nervosa

(pp.

Springer-Verlag.

Gaiardi, M., Bartoletti,

Cerbec,

Vasopressin-like

behavior

M., Bacci, A., Cubellini,

properties

of morphine

C. and Babbini,

in food deprived

M. (1987).

Increased sensitivity

rats. Pharmacology

Biochemistry

to and

26, 719-723.

E.N.,

Messing,

R.B.

and Sparber,

adaptation and hippocampal

S.B.

(1988).

corticosterone

Parallel

binding

changes

in operant

in rats treated with

behavioral

trimethyltin.

Brain

Research, 460, 346-351. Goodrick,

C.L.,

Ingram,

intermittent Hovancik, of

D.K.,

J.R., Gruning,

positive

97 (I),

in

refeeding.

Life Sciences,

Chapleur,

content

in

Society

fragment, Lubow,

M.L.,

J.R. and Cider,

N.L.

(1982).

Effects

of

28, 233-241.

Effect of deprivation

straight-alley

level on the magnitude

performance.

American

Journal

of

D. (1983).

A.J. and Leibowitz,

nuclei:

for Neuroscience

Learning enhancement

relation

Abstracts, Hormonally

to

Neuroendocrinology, stimulus.

S.F.

(1988).

deprivation

Vasopressin

and

and subsequent

active arginine-vasopressin and a behaviorally

suppresses

endo-

active vasopressin

37, 258-261.

A.U. (1959).

to the conditioned

food

59.11.

fever in rats: lack of effect of oxytocin

R.E. and Moore,

and behavior arousal

41, 1083-1088.

Burlet,

hypothalamic

Kovacs, G.L. and DeWied, toxin-induced

A. (1984).

discriminative

R.B. and Sparber, S.B. (1987).

induced by yohimbine. M.,

Freeman,

37-46.

Huang, M.S., Messing, Jhanwar-Uniyal,

M.A.,

and life span in rats. Gerontology,

R. and Intorre,

induction

Psychology,

oxytocin

Reynolds,

feeding upon growth

Latent inhibition:

Journal

the effect of non-reinforced

of Comparative

and

Physiological

preexposure

Psychology,

52,

415-419. McEwen,

B.S. (1982).

Glucocorticoids

Ganten and D. Pfaff (Eds.), (pp. l-22). Messing,

New York:

R.B., Bollweg,

tin poisoning 491-502.

Current

and hippocampus: Topics

receptors

in search of a function.

in Neuroendocrinology.

Adrenal

In D.

actions on brain

Springer-Verlag.

G., Chen, Q. and Sparber, S.B. (1988).

on learning

and hippocampal

corticosterone

Dose specific effects of trimethylbinding.

Neurotoxicology,

9 (3),

74 Messing,

R.B.,

Kleven,

MS.

and Sparber, S.B. (1986).

task generates adjunctive Messing,

R.B.

extinction Messing,

and superstitious

and Sparber,

S.B.

of autoshaped

behavior.

R.B. and Sparber,

desglycinamide

(1983).

reinforcement

Behavioral

Des-Cly-vasopressin

improves

Greater task difficulty

vasopressin

on appetitively

amplifies

motivated

in an autoshaping

Processes,

13, 327-338.

acquisition

European Journal of Pharmacology,

S.B. (1985).

arginine

Delaying

behaviors.

and slows

89, 43-51.

the facilitatory

learning.

effect of

Behavioral

Neuro-

science, 99,1114-1119. Messing,

R.B. and Sparber, S.B. (1986).

by trimethyltin. Mundy,

W.R.

Toxicology

and Iwamoto,

scopolamine

Biochemistry

D. and McHose,

as a function

Frankos

methods

P-adrenergic

ligand binding

induced

Studies

on desglycinamide

autoshaping

and Behavior,

model

arginine

vasopressin

of learning/memory

and

in rats.

27, 307-315. deprivation

effect in instrumental

level at the time of exposure

to the reinforcer.

conditioning

Animal

Learning

8 (4), 642-646.

S.B. (1980).

V.H.

(1987).

J. (1980). The residual

of deprivation

and Behavior, Sparber,

E.T.

Increased forebrain 32, 107-112.

in a modified/lever-touch

Pharmacology Peterson,

Letters,

Use of learned behavior

(Eds.),

Effects of foods

for predicting

toxicity.

in testing

for neurotoxicity.

and drugs on the development

Proceedings

of the Fifth

F.D.A.

In M.R.

Gryder

of the nervous Science

and

system:

Symposium

(pp.

49-61). Sparber, S.B., in adipose thology

Nagasawa, S. and Burklund, after daily

K.E.A.

administration

and Pharmacology,

(1977).

to rats. R.B. (1988).

deficit by desglycinamide-8-arginine

vasopressin.

E.A. (1983).

Reduction

tent food deprivation. Stone,

E.A., McEwen,

nents

B.S.,

Reversal

in Chemical

of a trimethyltin-induced

Life Sciences,

Pa-

Herrera,

Letters,

AS.

response

after chronic

intermit-

40, 33-37.

and Carr, K.D.

cyclic AMP

learning

42, 171-177.

in cortical beta adrenergic receptor density

Neuroscience

of noradrenergic

pool of d-amphetamine

Communications

18 (3), 423-431.

Sparber, S.B., Cohen, C.A. and Messing, Stone,

A mobilizable

Research

(1987).

in cortical

Regulation slices.

of a- and P-compo-

European

Journal

of Phar-

macology, 141, 347-356. Strupp,

B.J. and Levitsky,

D.A.

against. Neuroscience Suemaru,

S., Hashimoto,

(1985).

A mnemonic

and Biobehavioral K., Hattori,

role for vasopressin:

Reviews,

T., Inoue, H., Kageyama, J. and Ota, Z. (1986).

duced changes in rat brain corticotropin-releasing response. Winer,

Life Sciences,

B.). (1971).

Statistical

the evidence

factor (CRF)

Starvation-in-

and pituitary-adrenocortical

39, 1161-1166. principles

in experimental

for and

9, 399-411.

design.

New York:

McGraw-Hill.