- Email: [email protected]
Operant conditioning of response variability in male and female Wistar rats
Physiology & Behavior, Vol. 45, pp. 551-555. © Pergamon Press plc, 1989. Printed in the U.S.A.
0031-9384/89 $3.00 + .00
Operant Conditioning of Response Variability in Male and Female Wistar Rats ANNEMIEKE
VAN HEST, FRANS VAN HAAREN
AND NANNE
E. VAN DE POLL
Netherlands Institute f o r Brain Research, Meibergdreef 33, 1105 A Z Amsterdam, The Netherlands R e c e i v e d 3 O c t o b e r 1988 VAN HEST, A., F. VAN HAAREN AND N. E. VAN DE POLL. Operant conditioning of response variability in male and female Wistar rats. PHYSIOL BEHAV 45(3) 551-555, 1989.-It has previously been suggested that some of the behavioral differences between the sexes in food motivated operant procedures may be a function of the fact that males are more likely than females to exhibit stereotyped behavior. If such is the case, then it might be expected that behavioral variability is more easily conditioned in females than in males. The present experiment was designed to investigate this notion. Male and female Wistar rats were trained to respond in a procedure in which response variability was explicitly reinforced. In this procedure subjects had continuous access to two response levers in the experimental chamber. In the first experimental condition, each sequence of four responses was followed by the presentation of a food pellet, if the sequence differed from the two sequences which preceded it (Lag 2). Time-out was presented when such was not the case. During time-out the levers were retracted from the chamber and all stimulus lights were extinguished for 4 sec. In subsequent experimental conditions, subjects had to produce four-response sequences which differed from the preceding four, eight and sixteen sequences respectively (Lag 4, 8, 16). Response sequences were classified by the number of switches between levers. Behavioral variability increased as the lag requirement was increased, showing that variability is a conditionable dimension of behavior. Differences between males and females were however not observed. These results thus contradict the previously reported finding that males exhibit more behavioral stereotypy than females. It is suggested that procedural variables may account for these seemingly contradictory findings. Operant conditioning
Response variability
Sex differences
B E H A V I O R A L differences between male and female rats in food motivated operant learning tasks have been observed in a number o f experimental procedures (van Haaren, van Hest and Heinsbroek, manuscript submitted for publication). In general, males have been shown to exhibit more lever-directed behavior, whereas females tend to engage more often in other activities. Evidence has been presented that sex differences in activity levels may underlie at least some of the observed behavioral differences between the sexes (18,19). Recently, however, it has been suggested that sex differences in food motivated operant learning procedures might also be a function of the fact that males are more likely than females to engage in perseverative, stereotyped responding (22). Response perseveration in males has consistently been observed in a number o f different experimental set-ups. Males, but not females, have been reported to continue to respond on a lever which was no longer correlated with the presentation of food when subjects were exposed to an autoshaping procedure (15). Males have also been observed to be more likely than females to produce response runs which exceeded the run requirement on the work lever when behavior was maintained by different fixed consecutive number schedules (van Haaren and van Hest, manuscript submitted for publication). Extinction o f leverpressing in a two-lever G o - N o G o discrimination procedure occurred more slowly in males than in females when food was no longer presented (van Hest, Heinsbroek, van Haaren
and van de Poll, manuscript submitted for publication). Males also required more trials than females to reverse response patterns in a discrimination reversal procedure (8). Response perseveration or stereotyped behavior in males has been attributed to high levels o f the circulating male gonadal hormone testosterone (1, 2, 9). Testosterone-treated males have been shown to be less likely to complete a discrimination shift than oil-treated males, whereas preexperimental treatment with the antiandrogen cyproterone acetate facilitated the completion of this task as compared to control-injected subjects (12). In an experiment by van Hest et al. (22), different groups o f intact, gonadectomized or gonadectomized plus testosterone-treated male and female rats were exposed to an experimental procedure which has previously been shown to be sensitive to drug-induced response perseveration and stereotypy. In this procedure, reinforcers were randomly assigned to one o f two levers in an operant conditioning chamber. One response on the lever to which the reinforcer was assigned produced a food pellet. Males were observed to make more responses on the lever which was not selected for food presentation prior to switching to the selected lever than females. Furthermore, it was shown that females made more responses to the lever which was previously associated with the presentation of food, whereas males exhibited a preference for one of the two response levers, regardless of the previous experienced experimental contingencies. Between-group comparisons suggested that
~This research was conducted while Annemieke van Hest was supported by a grant from the Dutch Organization for the Advancement of Basic Research (NWO 560-258-024) awarded to Frans van Haaren. F. van Haaren is presently at the Department of Psychology, University of Florida, Gainesville, FL 32611.
551
VAN HEST ET AL.
552 this behavioral difference between the sexes might at least partly be mediated by testosterone. Because of the above results, it may be hypothesized that behavioral variability is more easily conditioned in females than in males. The present experiment was designed to explicitly investigate this notion. Male and female Wistar rats were trained in an experimental procedure in which response variability was selectively reinforced (10). They were exposed to an operant conditioning chamber in which two response levers were continuously accessible. Subjects were required to emit response sequences consisting of four lever presses. Sequences were classified by the number of switches [0, 1, 2 or 3] between the left and the right response lever. In the first experimental condition, each four-response sequence was followed by the presentation of food if, and only if, the sequence differed from the two, immediately preceding four-response sequences (Lag 2 requirement). Once behavior had stabilized, the lag requirement was systematically increased over the next three experimental conditions, and response sequences had to differ from the preceding four, eight or sixteen sequences in order for food to be presented (Lag 4, 8 and 16 respectively). METHOD
Subjects Eight male and 8 females Wistar rats were obtained from Animal House, TNO (Zeist, the Netherlands) when they were 9 weeks old. Upon arrival in the laboratory, they were housed in group cages (4 same sex subjects to a cage) under a reversed light-dark cycle (lights on 6:30 p.m.-6:30 a.m.) and constant temperature conditions. Subjects were food-deprived when they were 10 weeks old and fed daily to maintain body weights at approximately 85°70 corrected for growth throughout the experiment (5). Water was always available in the homecages. Experimentation started when subjects were 12 weeks old.
Apparatus Experiments took place in eight locally constructed rat chambers (34 cm wide, 33 cm long and 37 cm high). The side walls and intelligence panel were made of black Perspex. The front door of the chamber was made of translucent Plexiglas. The floor consisted of 26 grids, spaced 1.3 cm apart. Two retractable rodent levers (2.5 cm long, 2.8 cm wide and 0.75 cm thick, when extended) were located symmetrically to the side of the pellet retrieval unit. The levers required a force in excess of 0.20 N to be operated. Each leverpress produced auditory feedback for 0.10 sec from a Sonalert located approximately 16 cm above each lever. A stimulus light (green on the left- and red on the right-hand side of the intelligence panel) was located 9 cm directly above each lever. The pellet retrieval unit, located in between the two levers, could be illuminated by a white light. A houselight was mounted in the middle of the intelligence panel 3 cm from the ceiling of the chamber. All experimental chambers were enclosed in a sound-attenuated, ventilated cabinet; the front door of this cabinet was also made of translucent Plexiglas. The chambers were connected to a PDP 11-73 microcomputer (Digital Equipment Corporation, Maynard, MA), located in an adjacent room. Experimental contingencies and data acquisition procedures were programmed using SKED-11 (11), obtained from State Systems Inc. (Kalamazoo, MI).
Procedure Preliminary training. Subjects were first adapted to the experimental chamber for four sessions. During the first ses-
sion, both levers were extended and food pellets were delivered on a response-independent variable-time (VT) 30 sec schedule, or whenever subjects pressed one of the levers. The light in the pellet retrieval unit was illuminated for 1 sec during pellet delivery. The session was terminated once 40 pellets had been delivered. During the next three sessions, pellets were delivered whenever the subjects pressed one of the levers, but responseindependent food pellets were delivered on increasingly leaner VT schedules (VT 60, 120, 120 sec). To insure consistent responding on both levers, subjects were subsequently exposed to a procedure in which both levers were randomly presented, one at a time. During the first two sessions, the completion of a fixed-ratio 2 (FR 2) resulted in the presentation of a food pellet and the illumination of the pellet retrieval unit for 1 sec. The completion of an FR 5 was required to obtain a food pellet in the next two sessions. Sessions were terminated once 40 pellets had been presented. All subjects responded reliably on both levers at the end of preliminary training. Experimental procedure. The two levers were inserted into the chamber and houselight and stimulus lights were illuminated at the start of the session. Subjects were required to complete four-response sequences in order for food to be delivered. Each leverpress was followed by auditory feedback for 0.1 sec produced by the Sonalert located directly above the lever. Both the number of responses to the left and/or the right lever, as well as the number of switches between response levers were recorded for all individual sequences. In the first experimental condition, the first two four-response sequences were always followed by food presentation upon completion of the fourth response in each sequence. Subsequent four-response sequences were reinforced if, and only if, they differed from the two immediately preceding four-response sequences (Lag 2). The light in the pellet retrieval unit was illuminated for 1 sec during the delivery of the food pellet. When the current four-response sequence did not differ from the immediately preceding two four-response sequences, a time-out (TO) of 4 sec was presented immediately contingent upon the fourth response in the sequence. During TO the levers were retracted from the chamber, while houselight and stimulus lights were extinguished. Sessions were terminated after the presentation of 42 food pellets or after 45 min, whichever came first. The lag value was systematically increased in subsequent experimental conditions. Subjects were thus exposed to the same experimental contingencies, except for the fact that the lag value was increased from 2 to 4, 8 and 16 four-response sequences. The introduction of these manipulations resulted in the delivery of the first 4, 8 and 16 pellets without taking into account the different response sequences which preceded the reinforced four-response sequence. Once the first 4, 8 or 16 four-response sequences had been completed, the fifth, ninth or seventeenth four-response sequence had to be different from the preceding four, eight or sixteen sequences (Lag 4, Lag 8 and Lag 16). Sessions were terminated when 44, 48 or 56 (Lag 4, 8 and 16 respectively) pellets had been delivered or after 45 min, whichever came first. Experimental contingencies were changed when the averaged group response accuracy (reinforced fourresponse sequences divided by the total number of fourresponse sequences emitted) during the last five sessions did not statistically differ from response accuracy during the immediately preceding five sessions. Sessions were run five days a week (Monday through Friday) during the subject's dark hours. The following measures were analyzed: 1) Response accuracy, defined as the number of reinforced four-response sequences divided by the total number of fourresponse sequences emitted. The first 2 (Lag 2), 4, 8 or 16 (Lag
VARIABILITY IN MALE A N D F E M A L E RATS
553
LAG 4
LRG 2
LAG 8
LRG 16
100.0 80.0
o
:o
60.0 40.0 20.0 Jk 0.0 I
2
3
4
S
1
2
3
4
S
I
2
3
4
S
l
2
3
4
S
BLOCKS OF S SESSIONS
FIG. 1. Response accuracy (the number of reinforced four-response sequences/total number of four-response sequences) for male (triangles, dotted lines) and female (circles, solid lines) rats during the different experimental conditions.
100.0
80.0 v
GO. 0
LAG 2
LRG 4
2 Z
~
2
2
LAG 8
LRG IS
¢,
¢~
7-
40.0
20.0
0.0
•
.
.
I
,
2
.
3
r"l
0
1
2
2
3
.
0
.
1
.
2
.
.
3
.
.
0
~
I
2
3
MUMBER OF SWITCHES
FIG. 2. Relative frequency of response sequences as classified by the number of switches between levers, corrected for probability of occurrence, for male (open bars) and female (shaded bars) rats during the different experimental conditions.
4, 8 and 16 respectively) four-response sequences were omitted from the analysis. 2) Relative frequency of the number of response sequences with 0, 1, 2 or 3 switches between levers, corrected for probability of occurrence. With two response levers available, there are 16 possible different left (L)-right (R) four-response sequences. Sequences were classified according to the number of switches between levers. Two out of the 16 possible sequences consisted of four presses on the same lever. Six sequences contained one switch from the right to the left lever, or vice versa. Six other sequences required two switches. The remaining two sequences consisted of three L-R or R-L switches. If it is assumed that all possible individual response sequences have an equal a priori probability of occurrence than it can be gathered from the foregoing that response sequences which contain one or two switches will be more frequently observed than sequences with zero or three switches between levers. Transformations were thus made to correct for the unequal probabilities of occurrence. The number of sequences with zero or three switches was thus multiplied by 16/2, while the number of sequences with one or two switches was multiplied by 16/6. The first 2
(Lag 2), 4, 8 or 16 (Lag 4, 8 and 16 respectively) four-response sequences were omitted from the analysis. R E S U L T S
Response accuracy of males and females during the different lag conditions is shown in Fig. 1. Response accuracy was averaged over five consecutive sessions, and was subjected to a 2 x arcsine square-root transformation to increase homogeneity of variance (24) prior to analysis. Data were analyzed by means of analysis of variance (ANOVA) with the factors SEX, LAG requirement (Lag 2, 4, 8 or 16) and sessions (SES, 5 blocks of 5 sessions each), the latter two being repeated measures. ANOVA of response accuracy revealed no overall sex differences, F(1,14)=0.09, n.s. Response accuracy decreased as the lag requirement increased, F(3,42)=267.61, p<0.01. Accuracy increased over sessions, F(4,56)= 13.31, p<0.01. The effects of prolonged training were more strongly observed at the larger lag requirements as indicated by a significant LAG by SES interaction effect, F(12,168)=3.94, p<0.01. Figure 2 shows the relative frequency of sequences with 0,
554
VAN HEST E T A L .
1, 2 or 3 switches, corrected for probability of occurrence. All four-response sequences during the last five sessions of each experimental condition were classified according to the number of switches between levers. The number of observations in each of the four response classes was then corrected for probability of occurrence, and expressed as a percentage of the total number of observations. Next, relative frequency percentage scores were 2 × arcsine square root transformed, and subjected to ANOVA with the factors SEX, response class (as defined by the number of switches between levers (0, 1, 2 or 3) in each response sequence: SWI) and LAG requirement (lag 2, 4, 8 or 16), the latter two being considered repeated measures, SWI nested within LAG. Sex differences were again not observed, F(I, 14)=2.96, n.s. The relative frequency of response sequences with 0, 1, 2 or 3 switches between levers was inversely related to the number of switches in the sequence, F(3,42)=2119.82, p<0.01. The lag requirement did affect the frequency distribution of the response sequences classified by the number of switches, F(3,42)=44.62, p<0.01. A significant SWI by LAG interaction effect indicated that the frequency of response sequences in which all responses were made to the same lever decreased as the lag requirement was increased, whereas the frequency of other response sequences increased, F(9,126)=119.61, p<0.01. Inspection of Fig. 3, which shows the relative frequency distribution of all possible individual four-response sequences, further confirms this notion.
LRG 2 50.0 40.0 V'A 30.0 3= 20.0 Q
VA V~ I,'A VA
10.0
0.0
~. LRG 4
g,
30.0
:~ 20.0 o
OZ~10"0 0.0 LRG 8
g,
30.0
:~ 20.0 o
~
ZO.0
0o0 DISCUSSION
The experimental study of behavioral differences between males and females in learning and memory has thus far been mainly restricted to the effects of gonadal hormones on behavior in aversively motivated procedures (3), whereas the study of behavioral differences between the sexes in food motivated operant learning and memory procedures has received little attention. Recently however, a number of experiments have been reported concerning this issue (13-23). The results of these experiments have shown that gonadal hormones may also differentially affect schedule-controlled behavior. Behavioral differences between the sexes were observed in some, but not all appetitively motivated experimental procedures studied. Sex differences were most obvious when subjects were exposed to free-operant schedules of reinforcement. In such procedures, males have consistently been observed to be more likely than females to contact a lever which is associated with the presentation of food, both in autoshaping procedures in which the delivery of food is response-independent, as well as in procedures in which food presentation is contingent upon responding (4, 15, 17). Males have also been observed to respond at higher rates than females when behavior is maintained by freeoperant reinforcement schedules (4, 8, 20). Females on the other hand tend to engage more often in activities other than leverpressing (18,19). The results of previous experiments have provided evidence for the notion that behavioral differences between the sexes as observed in free-operant procedures may be mediated by sex differences in general activity levels, which in turn may be mediated by the organizational and activational effects of gonadal hormones, females being more active than males (3,18). However, the results of other experiments have suggested that at least some of the behavioral differences between the sexes in food motivated operant procedures may also be accounted for by the fact that males exhibit a hormonally (testosterone) determined disposition towards behavioral stereotypy
LAG 1.6 N 20.0 10.0 I
o.o
LLLL LLRL RLLL L R L R R R L L R L L R RRRL RLRR LLLR LRLL L L R R L R R L R L R L LRRR RRLR RRRR
FIG. 3. Frequency distribution of all possibleindividual four-response sequences for male (open bars) and female (shaded bars) rats during the different experimental conditions.
and response perseveration (17,22). On account of these results, it was hypothesized that males might be expected to behave less efficiently than females in an experimental procedure in which behavioral variability is selectively reinforced. The results of the present experiment showed that the response accuracy of both males and females decreased as the experimental contingencies required an increase in behavioral variability in order for food to be presented. However, response accuracy increased over sessions. In addition, it was shown that behavioral variability of both males and females increased, as can be inferred from the observed changes in the frequency distributions of the response sequences as a function of the lag requirement, thus suggesting that variability is indeed a conditionable dimension of behavior in this experimental set-up. Behavioral differences between males and females were however not observed in the present experiment. Males and females attained similar levels of response accuracy. The frequency of all individual response sequences of males did not differ from those of females. These results thus seem to contradict the previously reported finding that males exhibit more behavioral stereotypy and show less response variability than females. These seemingly contradictory results may be accounted for by procedural differences between the two exper-
V A R I A B I L I T Y IN M A L E A N D F E M A L E R A T S
555
iments. Males were m o r e likely t h a n females to c o n t i n u e to r e s p o n d to the lever n o t selected for r e i n f o r c e m e n t p r i o r to switching to the selected lever w h e n the occurrence o f such stereotyped b e h a v i o r h a d n o scheduled consequences (22). H o w e v e r , b e h a v i o r a l differences between the sexes were n o t observed when the experimental contingencies were arranged in such a way t h a t stereotyped b e h a v i o r was followed by the pres e n t a t i o n o f time-out. U n d e r these circumstances, males a n d females were equally likely to switch their responding between the two levers. These results thus seem to indicate t h a t the males' t e n d e n c y to engage in perseverative, stereotyped responding may only become evident when the experimental contingencies allow the development of such behavior. Males may show as m u c h variability as females when such is not the case.
T h e precise effects o f gender, or h o r m o n a l state, o n learning a n d m e m o r y processes is not a s t r a i g h t f o r w a r d issue, b u t r a t h e r seems to result f r o m a complex interaction between the type o f behavior under investigation and the experimental envir o n m e n t to which the subject is exposed. It thus seems the case that specific parameters o f the experimental procedure, including the d e p e n d e n t variable o f interest, d e t e r m i n e w h e t h e r behavioral predispositions (level o f activity, behavioral stereotypy), which are i n d u c e d by the o r g a n i z a t i o n a l effects o f gon a d a l h o r m o n e s d u r i n g the pre- a n d / o r p e r i n a t a l period, will be reflected in b e h a v i o r a l differences between the sexes in a d u l t h o o d (6,7). O t h e r experiments will have to be carried o u t to f u r t h e r disentangle the interactions between h o r m o n e s a n d b e h a v i o r in learning a n d m e m o r y .
REFERENCES 1. Andrew, R. J. Attentional processes and animal behaviour. In: Bateson, P. P. G.; Hinde, R. A., eds. Growing points irr ethology. Cambridge: University Press; 1976. 2. Archer, J. Testosterone and persistence in mice. Anim. Behav. 25: 479-488; 1977. 3. Beatty, W. W. Gonadal hormones and sex-differences in nonreproductive behavior in rodents: organizational and activational influences. Horm. Behav. 12:112-163; 1979. 4. Heinsbroek, R. P. W.; van Haaren, F.; Zantvoord, F.; van de Poll, N. E. Acquisition of random-ratio behavior in male and female rats: Effects of gonadectomy. Physiol. Behav. 39:269-272; 1987. 5. Hurwitz, M. H. B.; Davis, H. Depriving rats of food: a reappraisal of two techniques. J. Exp. Anal. Behav. 40:211-213; 1983. 6. Leshner, A. I. An introduction to behavioral endocrinology. New York: Oxford University Press; 1978. 7. Leshner, A. I. Kinds of hormonal effects on behavior: a new view. Neurosci. Biobehav. Rev. 3:69-73; 1979. 8. Millar, R. D. Free-operant comparisons of wild and domestic Norway rats. J. Comp. Physiol. Psychol. 89:913-922; 1975. 9. Rogers, L. J. Male hormones and behavior. In: Lloyd, B. B.; Archer, J., eds. Exploring sex differences. New York: Academic Press; 1976. 10. Schwartz, B. Development of complex stereotyped behavior in pigeons. J. Exp. Anal. Behav. 33:153-166; 1980. 11. Snapper, A. G.; Inglis, G. B. SKED-11: Reference manual. Kalamazoo, MI: State Systems Inc.; 1985. 12. Thompson, W. R.; Wright, J. S. "Persistence" in rats: effects of testosterone. Physiol. Psychol. 7:291-294; 1979. 13. van Haaren, F.; van Hest, A. The acquisition of visual and auditory discrimination in male and female Wistar rats. Behav. Brain Res.; in press. 14. van Haaren, F.; van Hest, A. Spatial matching and nonmatching in male and female Wistar rats: effects of delay interval duration. Anim. Learn. Behav.; in press.
15. van Haaren, F.; van Hest, A.; van de Poll, N. E. Acquisition and reversal of a discriminated autoshaped response in male and female rats: effects of long or short and fixed or variable intertrial interval durations. Learn. Motiv. 18:220-233; 1987. 16. van Haaren, F.; van Hest, A.; van de Poll, N. E. Self control in male and female Wistar rats..1. Exp. Anal. Behav. 49:201-211; 1988. 17. van Hest, A.; van Haaren, F. The effects of gonadectomy and chronic testosterone suppletion on the autoshaped response of male and female Wistar rats. Bull. Psychon. Soc. 27:45-48; 1989. 18. van Hest, A.; van Haaren, F.; van de Poll, N. E. Behavioral differences between male and female Wistar rats on DRL schedules: Effect of stimuli promoting collateral activities. Physiol. Behav. 39: 255-261; 1987. 19, van Hest, A.; van Haaren, F.; van de Poll, N. E. Behavioral differences between male and female Wistar rats in food rewarded lever holding. Physiol. Behav. 39:263-267; 1987. 20. van Hest, A.; van Haaren, F.; and van de Poll, N. E. The behavior of male and female Wistar rats pressing a lever for food is not affected by sex differences in food motivation. Behav. Brain Res. 27:215-221; 1988. 21. van Hest, A.; van Haaren, F.; van de Poll, N. E. Delayed response alternation: effects of stimulus presentations during the delay interval on the response accuracy in male and female Wistar rats. Bull. Psychon. Soc. 26:141-144; 1988. 22. van Hest, A.; van Haaren, F.; van de Poll, N. E. Perseverative responding in male and female Wistar rats: effects of gonadal hormones. Horm. Behav.; in press. 23. van Hest, A.; van Kempen, M.; van Haaren, F.; van de Poll, N. E. Memory in male and female Wistar rats: effects of gonadectomy, and stimulus presentations during the delay interval. Behav. Brain Res. 29:103-110; 1988. 24. Winer, B. J. Statistical principles in experimental design. New York: McGraw-Hill Book Company; 1971.
0031-9384/89 $3.00 + .00
Operant Conditioning of Response Variability in Male and Female Wistar Rats ANNEMIEKE
VAN HEST, FRANS VAN HAAREN
AND NANNE
E. VAN DE POLL
Netherlands Institute f o r Brain Research, Meibergdreef 33, 1105 A Z Amsterdam, The Netherlands R e c e i v e d 3 O c t o b e r 1988 VAN HEST, A., F. VAN HAAREN AND N. E. VAN DE POLL. Operant conditioning of response variability in male and female Wistar rats. PHYSIOL BEHAV 45(3) 551-555, 1989.-It has previously been suggested that some of the behavioral differences between the sexes in food motivated operant procedures may be a function of the fact that males are more likely than females to exhibit stereotyped behavior. If such is the case, then it might be expected that behavioral variability is more easily conditioned in females than in males. The present experiment was designed to investigate this notion. Male and female Wistar rats were trained to respond in a procedure in which response variability was explicitly reinforced. In this procedure subjects had continuous access to two response levers in the experimental chamber. In the first experimental condition, each sequence of four responses was followed by the presentation of a food pellet, if the sequence differed from the two sequences which preceded it (Lag 2). Time-out was presented when such was not the case. During time-out the levers were retracted from the chamber and all stimulus lights were extinguished for 4 sec. In subsequent experimental conditions, subjects had to produce four-response sequences which differed from the preceding four, eight and sixteen sequences respectively (Lag 4, 8, 16). Response sequences were classified by the number of switches between levers. Behavioral variability increased as the lag requirement was increased, showing that variability is a conditionable dimension of behavior. Differences between males and females were however not observed. These results thus contradict the previously reported finding that males exhibit more behavioral stereotypy than females. It is suggested that procedural variables may account for these seemingly contradictory findings. Operant conditioning
Response variability
Sex differences
B E H A V I O R A L differences between male and female rats in food motivated operant learning tasks have been observed in a number o f experimental procedures (van Haaren, van Hest and Heinsbroek, manuscript submitted for publication). In general, males have been shown to exhibit more lever-directed behavior, whereas females tend to engage more often in other activities. Evidence has been presented that sex differences in activity levels may underlie at least some of the observed behavioral differences between the sexes (18,19). Recently, however, it has been suggested that sex differences in food motivated operant learning procedures might also be a function of the fact that males are more likely than females to engage in perseverative, stereotyped responding (22). Response perseveration in males has consistently been observed in a number o f different experimental set-ups. Males, but not females, have been reported to continue to respond on a lever which was no longer correlated with the presentation of food when subjects were exposed to an autoshaping procedure (15). Males have also been observed to be more likely than females to produce response runs which exceeded the run requirement on the work lever when behavior was maintained by different fixed consecutive number schedules (van Haaren and van Hest, manuscript submitted for publication). Extinction o f leverpressing in a two-lever G o - N o G o discrimination procedure occurred more slowly in males than in females when food was no longer presented (van Hest, Heinsbroek, van Haaren
and van de Poll, manuscript submitted for publication). Males also required more trials than females to reverse response patterns in a discrimination reversal procedure (8). Response perseveration or stereotyped behavior in males has been attributed to high levels o f the circulating male gonadal hormone testosterone (1, 2, 9). Testosterone-treated males have been shown to be less likely to complete a discrimination shift than oil-treated males, whereas preexperimental treatment with the antiandrogen cyproterone acetate facilitated the completion of this task as compared to control-injected subjects (12). In an experiment by van Hest et al. (22), different groups o f intact, gonadectomized or gonadectomized plus testosterone-treated male and female rats were exposed to an experimental procedure which has previously been shown to be sensitive to drug-induced response perseveration and stereotypy. In this procedure, reinforcers were randomly assigned to one o f two levers in an operant conditioning chamber. One response on the lever to which the reinforcer was assigned produced a food pellet. Males were observed to make more responses on the lever which was not selected for food presentation prior to switching to the selected lever than females. Furthermore, it was shown that females made more responses to the lever which was previously associated with the presentation of food, whereas males exhibited a preference for one of the two response levers, regardless of the previous experienced experimental contingencies. Between-group comparisons suggested that
~This research was conducted while Annemieke van Hest was supported by a grant from the Dutch Organization for the Advancement of Basic Research (NWO 560-258-024) awarded to Frans van Haaren. F. van Haaren is presently at the Department of Psychology, University of Florida, Gainesville, FL 32611.
551
VAN HEST ET AL.
552 this behavioral difference between the sexes might at least partly be mediated by testosterone. Because of the above results, it may be hypothesized that behavioral variability is more easily conditioned in females than in males. The present experiment was designed to explicitly investigate this notion. Male and female Wistar rats were trained in an experimental procedure in which response variability was selectively reinforced (10). They were exposed to an operant conditioning chamber in which two response levers were continuously accessible. Subjects were required to emit response sequences consisting of four lever presses. Sequences were classified by the number of switches [0, 1, 2 or 3] between the left and the right response lever. In the first experimental condition, each four-response sequence was followed by the presentation of food if, and only if, the sequence differed from the two, immediately preceding four-response sequences (Lag 2 requirement). Once behavior had stabilized, the lag requirement was systematically increased over the next three experimental conditions, and response sequences had to differ from the preceding four, eight or sixteen sequences in order for food to be presented (Lag 4, 8 and 16 respectively). METHOD
Subjects Eight male and 8 females Wistar rats were obtained from Animal House, TNO (Zeist, the Netherlands) when they were 9 weeks old. Upon arrival in the laboratory, they were housed in group cages (4 same sex subjects to a cage) under a reversed light-dark cycle (lights on 6:30 p.m.-6:30 a.m.) and constant temperature conditions. Subjects were food-deprived when they were 10 weeks old and fed daily to maintain body weights at approximately 85°70 corrected for growth throughout the experiment (5). Water was always available in the homecages. Experimentation started when subjects were 12 weeks old.
Apparatus Experiments took place in eight locally constructed rat chambers (34 cm wide, 33 cm long and 37 cm high). The side walls and intelligence panel were made of black Perspex. The front door of the chamber was made of translucent Plexiglas. The floor consisted of 26 grids, spaced 1.3 cm apart. Two retractable rodent levers (2.5 cm long, 2.8 cm wide and 0.75 cm thick, when extended) were located symmetrically to the side of the pellet retrieval unit. The levers required a force in excess of 0.20 N to be operated. Each leverpress produced auditory feedback for 0.10 sec from a Sonalert located approximately 16 cm above each lever. A stimulus light (green on the left- and red on the right-hand side of the intelligence panel) was located 9 cm directly above each lever. The pellet retrieval unit, located in between the two levers, could be illuminated by a white light. A houselight was mounted in the middle of the intelligence panel 3 cm from the ceiling of the chamber. All experimental chambers were enclosed in a sound-attenuated, ventilated cabinet; the front door of this cabinet was also made of translucent Plexiglas. The chambers were connected to a PDP 11-73 microcomputer (Digital Equipment Corporation, Maynard, MA), located in an adjacent room. Experimental contingencies and data acquisition procedures were programmed using SKED-11 (11), obtained from State Systems Inc. (Kalamazoo, MI).
Procedure Preliminary training. Subjects were first adapted to the experimental chamber for four sessions. During the first ses-
sion, both levers were extended and food pellets were delivered on a response-independent variable-time (VT) 30 sec schedule, or whenever subjects pressed one of the levers. The light in the pellet retrieval unit was illuminated for 1 sec during pellet delivery. The session was terminated once 40 pellets had been delivered. During the next three sessions, pellets were delivered whenever the subjects pressed one of the levers, but responseindependent food pellets were delivered on increasingly leaner VT schedules (VT 60, 120, 120 sec). To insure consistent responding on both levers, subjects were subsequently exposed to a procedure in which both levers were randomly presented, one at a time. During the first two sessions, the completion of a fixed-ratio 2 (FR 2) resulted in the presentation of a food pellet and the illumination of the pellet retrieval unit for 1 sec. The completion of an FR 5 was required to obtain a food pellet in the next two sessions. Sessions were terminated once 40 pellets had been presented. All subjects responded reliably on both levers at the end of preliminary training. Experimental procedure. The two levers were inserted into the chamber and houselight and stimulus lights were illuminated at the start of the session. Subjects were required to complete four-response sequences in order for food to be delivered. Each leverpress was followed by auditory feedback for 0.1 sec produced by the Sonalert located directly above the lever. Both the number of responses to the left and/or the right lever, as well as the number of switches between response levers were recorded for all individual sequences. In the first experimental condition, the first two four-response sequences were always followed by food presentation upon completion of the fourth response in each sequence. Subsequent four-response sequences were reinforced if, and only if, they differed from the two immediately preceding four-response sequences (Lag 2). The light in the pellet retrieval unit was illuminated for 1 sec during the delivery of the food pellet. When the current four-response sequence did not differ from the immediately preceding two four-response sequences, a time-out (TO) of 4 sec was presented immediately contingent upon the fourth response in the sequence. During TO the levers were retracted from the chamber, while houselight and stimulus lights were extinguished. Sessions were terminated after the presentation of 42 food pellets or after 45 min, whichever came first. The lag value was systematically increased in subsequent experimental conditions. Subjects were thus exposed to the same experimental contingencies, except for the fact that the lag value was increased from 2 to 4, 8 and 16 four-response sequences. The introduction of these manipulations resulted in the delivery of the first 4, 8 and 16 pellets without taking into account the different response sequences which preceded the reinforced four-response sequence. Once the first 4, 8 or 16 four-response sequences had been completed, the fifth, ninth or seventeenth four-response sequence had to be different from the preceding four, eight or sixteen sequences (Lag 4, Lag 8 and Lag 16). Sessions were terminated when 44, 48 or 56 (Lag 4, 8 and 16 respectively) pellets had been delivered or after 45 min, whichever came first. Experimental contingencies were changed when the averaged group response accuracy (reinforced fourresponse sequences divided by the total number of fourresponse sequences emitted) during the last five sessions did not statistically differ from response accuracy during the immediately preceding five sessions. Sessions were run five days a week (Monday through Friday) during the subject's dark hours. The following measures were analyzed: 1) Response accuracy, defined as the number of reinforced four-response sequences divided by the total number of fourresponse sequences emitted. The first 2 (Lag 2), 4, 8 or 16 (Lag
VARIABILITY IN MALE A N D F E M A L E RATS
553
LAG 4
LRG 2
LAG 8
LRG 16
100.0 80.0
o
:o
60.0 40.0 20.0 Jk 0.0 I
2
3
4
S
1
2
3
4
S
I
2
3
4
S
l
2
3
4
S
BLOCKS OF S SESSIONS
FIG. 1. Response accuracy (the number of reinforced four-response sequences/total number of four-response sequences) for male (triangles, dotted lines) and female (circles, solid lines) rats during the different experimental conditions.
100.0
80.0 v
GO. 0
LAG 2
LRG 4
2 Z
~
2
2
LAG 8
LRG IS
¢,
¢~
7-
40.0
20.0
0.0
•
.
.
I
,
2
.
3
r"l
0
1
2
2
3
.
0
.
1
.
2
.
.
3
.
.
0
~
I
2
3
MUMBER OF SWITCHES
FIG. 2. Relative frequency of response sequences as classified by the number of switches between levers, corrected for probability of occurrence, for male (open bars) and female (shaded bars) rats during the different experimental conditions.
4, 8 and 16 respectively) four-response sequences were omitted from the analysis. 2) Relative frequency of the number of response sequences with 0, 1, 2 or 3 switches between levers, corrected for probability of occurrence. With two response levers available, there are 16 possible different left (L)-right (R) four-response sequences. Sequences were classified according to the number of switches between levers. Two out of the 16 possible sequences consisted of four presses on the same lever. Six sequences contained one switch from the right to the left lever, or vice versa. Six other sequences required two switches. The remaining two sequences consisted of three L-R or R-L switches. If it is assumed that all possible individual response sequences have an equal a priori probability of occurrence than it can be gathered from the foregoing that response sequences which contain one or two switches will be more frequently observed than sequences with zero or three switches between levers. Transformations were thus made to correct for the unequal probabilities of occurrence. The number of sequences with zero or three switches was thus multiplied by 16/2, while the number of sequences with one or two switches was multiplied by 16/6. The first 2
(Lag 2), 4, 8 or 16 (Lag 4, 8 and 16 respectively) four-response sequences were omitted from the analysis. R E S U L T S
Response accuracy of males and females during the different lag conditions is shown in Fig. 1. Response accuracy was averaged over five consecutive sessions, and was subjected to a 2 x arcsine square-root transformation to increase homogeneity of variance (24) prior to analysis. Data were analyzed by means of analysis of variance (ANOVA) with the factors SEX, LAG requirement (Lag 2, 4, 8 or 16) and sessions (SES, 5 blocks of 5 sessions each), the latter two being repeated measures. ANOVA of response accuracy revealed no overall sex differences, F(1,14)=0.09, n.s. Response accuracy decreased as the lag requirement increased, F(3,42)=267.61, p<0.01. Accuracy increased over sessions, F(4,56)= 13.31, p<0.01. The effects of prolonged training were more strongly observed at the larger lag requirements as indicated by a significant LAG by SES interaction effect, F(12,168)=3.94, p<0.01. Figure 2 shows the relative frequency of sequences with 0,
554
VAN HEST E T A L .
1, 2 or 3 switches, corrected for probability of occurrence. All four-response sequences during the last five sessions of each experimental condition were classified according to the number of switches between levers. The number of observations in each of the four response classes was then corrected for probability of occurrence, and expressed as a percentage of the total number of observations. Next, relative frequency percentage scores were 2 × arcsine square root transformed, and subjected to ANOVA with the factors SEX, response class (as defined by the number of switches between levers (0, 1, 2 or 3) in each response sequence: SWI) and LAG requirement (lag 2, 4, 8 or 16), the latter two being considered repeated measures, SWI nested within LAG. Sex differences were again not observed, F(I, 14)=2.96, n.s. The relative frequency of response sequences with 0, 1, 2 or 3 switches between levers was inversely related to the number of switches in the sequence, F(3,42)=2119.82, p<0.01. The lag requirement did affect the frequency distribution of the response sequences classified by the number of switches, F(3,42)=44.62, p<0.01. A significant SWI by LAG interaction effect indicated that the frequency of response sequences in which all responses were made to the same lever decreased as the lag requirement was increased, whereas the frequency of other response sequences increased, F(9,126)=119.61, p<0.01. Inspection of Fig. 3, which shows the relative frequency distribution of all possible individual four-response sequences, further confirms this notion.
LRG 2 50.0 40.0 V'A 30.0 3= 20.0 Q
VA V~ I,'A VA
10.0
0.0
~. LRG 4
g,
30.0
:~ 20.0 o
OZ~10"0 0.0 LRG 8
g,
30.0
:~ 20.0 o
~
ZO.0
0o0 DISCUSSION
The experimental study of behavioral differences between males and females in learning and memory has thus far been mainly restricted to the effects of gonadal hormones on behavior in aversively motivated procedures (3), whereas the study of behavioral differences between the sexes in food motivated operant learning and memory procedures has received little attention. Recently however, a number of experiments have been reported concerning this issue (13-23). The results of these experiments have shown that gonadal hormones may also differentially affect schedule-controlled behavior. Behavioral differences between the sexes were observed in some, but not all appetitively motivated experimental procedures studied. Sex differences were most obvious when subjects were exposed to free-operant schedules of reinforcement. In such procedures, males have consistently been observed to be more likely than females to contact a lever which is associated with the presentation of food, both in autoshaping procedures in which the delivery of food is response-independent, as well as in procedures in which food presentation is contingent upon responding (4, 15, 17). Males have also been observed to respond at higher rates than females when behavior is maintained by freeoperant reinforcement schedules (4, 8, 20). Females on the other hand tend to engage more often in activities other than leverpressing (18,19). The results of previous experiments have provided evidence for the notion that behavioral differences between the sexes as observed in free-operant procedures may be mediated by sex differences in general activity levels, which in turn may be mediated by the organizational and activational effects of gonadal hormones, females being more active than males (3,18). However, the results of other experiments have suggested that at least some of the behavioral differences between the sexes in food motivated operant procedures may also be accounted for by the fact that males exhibit a hormonally (testosterone) determined disposition towards behavioral stereotypy
LAG 1.6 N 20.0 10.0 I
o.o
LLLL LLRL RLLL L R L R R R L L R L L R RRRL RLRR LLLR LRLL L L R R L R R L R L R L LRRR RRLR RRRR
FIG. 3. Frequency distribution of all possibleindividual four-response sequences for male (open bars) and female (shaded bars) rats during the different experimental conditions.
and response perseveration (17,22). On account of these results, it was hypothesized that males might be expected to behave less efficiently than females in an experimental procedure in which behavioral variability is selectively reinforced. The results of the present experiment showed that the response accuracy of both males and females decreased as the experimental contingencies required an increase in behavioral variability in order for food to be presented. However, response accuracy increased over sessions. In addition, it was shown that behavioral variability of both males and females increased, as can be inferred from the observed changes in the frequency distributions of the response sequences as a function of the lag requirement, thus suggesting that variability is indeed a conditionable dimension of behavior in this experimental set-up. Behavioral differences between males and females were however not observed in the present experiment. Males and females attained similar levels of response accuracy. The frequency of all individual response sequences of males did not differ from those of females. These results thus seem to contradict the previously reported finding that males exhibit more behavioral stereotypy and show less response variability than females. These seemingly contradictory results may be accounted for by procedural differences between the two exper-
V A R I A B I L I T Y IN M A L E A N D F E M A L E R A T S
555
iments. Males were m o r e likely t h a n females to c o n t i n u e to r e s p o n d to the lever n o t selected for r e i n f o r c e m e n t p r i o r to switching to the selected lever w h e n the occurrence o f such stereotyped b e h a v i o r h a d n o scheduled consequences (22). H o w e v e r , b e h a v i o r a l differences between the sexes were n o t observed when the experimental contingencies were arranged in such a way t h a t stereotyped b e h a v i o r was followed by the pres e n t a t i o n o f time-out. U n d e r these circumstances, males a n d females were equally likely to switch their responding between the two levers. These results thus seem to indicate t h a t the males' t e n d e n c y to engage in perseverative, stereotyped responding may only become evident when the experimental contingencies allow the development of such behavior. Males may show as m u c h variability as females when such is not the case.
T h e precise effects o f gender, or h o r m o n a l state, o n learning a n d m e m o r y processes is not a s t r a i g h t f o r w a r d issue, b u t r a t h e r seems to result f r o m a complex interaction between the type o f behavior under investigation and the experimental envir o n m e n t to which the subject is exposed. It thus seems the case that specific parameters o f the experimental procedure, including the d e p e n d e n t variable o f interest, d e t e r m i n e w h e t h e r behavioral predispositions (level o f activity, behavioral stereotypy), which are i n d u c e d by the o r g a n i z a t i o n a l effects o f gon a d a l h o r m o n e s d u r i n g the pre- a n d / o r p e r i n a t a l period, will be reflected in b e h a v i o r a l differences between the sexes in a d u l t h o o d (6,7). O t h e r experiments will have to be carried o u t to f u r t h e r disentangle the interactions between h o r m o n e s a n d b e h a v i o r in learning a n d m e m o r y .
REFERENCES 1. Andrew, R. J. Attentional processes and animal behaviour. In: Bateson, P. P. G.; Hinde, R. A., eds. Growing points irr ethology. Cambridge: University Press; 1976. 2. Archer, J. Testosterone and persistence in mice. Anim. Behav. 25: 479-488; 1977. 3. Beatty, W. W. Gonadal hormones and sex-differences in nonreproductive behavior in rodents: organizational and activational influences. Horm. Behav. 12:112-163; 1979. 4. Heinsbroek, R. P. W.; van Haaren, F.; Zantvoord, F.; van de Poll, N. E. Acquisition of random-ratio behavior in male and female rats: Effects of gonadectomy. Physiol. Behav. 39:269-272; 1987. 5. Hurwitz, M. H. B.; Davis, H. Depriving rats of food: a reappraisal of two techniques. J. Exp. Anal. Behav. 40:211-213; 1983. 6. Leshner, A. I. An introduction to behavioral endocrinology. New York: Oxford University Press; 1978. 7. Leshner, A. I. Kinds of hormonal effects on behavior: a new view. Neurosci. Biobehav. Rev. 3:69-73; 1979. 8. Millar, R. D. Free-operant comparisons of wild and domestic Norway rats. J. Comp. Physiol. Psychol. 89:913-922; 1975. 9. Rogers, L. J. Male hormones and behavior. In: Lloyd, B. B.; Archer, J., eds. Exploring sex differences. New York: Academic Press; 1976. 10. Schwartz, B. Development of complex stereotyped behavior in pigeons. J. Exp. Anal. Behav. 33:153-166; 1980. 11. Snapper, A. G.; Inglis, G. B. SKED-11: Reference manual. Kalamazoo, MI: State Systems Inc.; 1985. 12. Thompson, W. R.; Wright, J. S. "Persistence" in rats: effects of testosterone. Physiol. Psychol. 7:291-294; 1979. 13. van Haaren, F.; van Hest, A. The acquisition of visual and auditory discrimination in male and female Wistar rats. Behav. Brain Res.; in press. 14. van Haaren, F.; van Hest, A. Spatial matching and nonmatching in male and female Wistar rats: effects of delay interval duration. Anim. Learn. Behav.; in press.
15. van Haaren, F.; van Hest, A.; van de Poll, N. E. Acquisition and reversal of a discriminated autoshaped response in male and female rats: effects of long or short and fixed or variable intertrial interval durations. Learn. Motiv. 18:220-233; 1987. 16. van Haaren, F.; van Hest, A.; van de Poll, N. E. Self control in male and female Wistar rats..1. Exp. Anal. Behav. 49:201-211; 1988. 17. van Hest, A.; van Haaren, F. The effects of gonadectomy and chronic testosterone suppletion on the autoshaped response of male and female Wistar rats. Bull. Psychon. Soc. 27:45-48; 1989. 18. van Hest, A.; van Haaren, F.; van de Poll, N. E. Behavioral differences between male and female Wistar rats on DRL schedules: Effect of stimuli promoting collateral activities. Physiol. Behav. 39: 255-261; 1987. 19, van Hest, A.; van Haaren, F.; van de Poll, N. E. Behavioral differences between male and female Wistar rats in food rewarded lever holding. Physiol. Behav. 39:263-267; 1987. 20. van Hest, A.; van Haaren, F.; and van de Poll, N. E. The behavior of male and female Wistar rats pressing a lever for food is not affected by sex differences in food motivation. Behav. Brain Res. 27:215-221; 1988. 21. van Hest, A.; van Haaren, F.; van de Poll, N. E. Delayed response alternation: effects of stimulus presentations during the delay interval on the response accuracy in male and female Wistar rats. Bull. Psychon. Soc. 26:141-144; 1988. 22. van Hest, A.; van Haaren, F.; van de Poll, N. E. Perseverative responding in male and female Wistar rats: effects of gonadal hormones. Horm. Behav.; in press. 23. van Hest, A.; van Kempen, M.; van Haaren, F.; van de Poll, N. E. Memory in male and female Wistar rats: effects of gonadectomy, and stimulus presentations during the delay interval. Behav. Brain Res. 29:103-110; 1988. 24. Winer, B. J. Statistical principles in experimental design. New York: McGraw-Hill Book Company; 1971.