Effects of Hormone Manipulations on Sex Differences In Learning, Visual Spatial Skills, Curiosity, and Exploration

Effects of Hormone Manipulations on Sex Differences In Learning, Visual Spatial Skills, Curiosity, and Exploration

Rhawn Gabriel Joseph, Ph.D., Steven Hess, MD & Elizabeth Birecree, MD


The maze learning ability and visual-spatial skills of male rats castrated at birth, males and females treated with cyproterone acetate or sesame oil, and females treated with testosterone within the first 10 days postpartum was compared to that of nonmanipulated males and females (Experiment 1). Neonatal castration and injections of cyproterone acetate and testosterone were found to significantly reverse normal sex differences in maze learning. In Experiment 2, differences in gonadal hormone secretions and activities during adulthood on maze learning ability were assessed. It was found that gonadectomy did not significantly influence the performance of males and females. In Experiment 3, females were found to exhibit a greater tendency to explore and make irrelevant section entries on a maze that was problem free. It is suggested that differences in maze learning and thus, visual spatial skills, are more greatly influenced by brain sex than by subsequent gonadal hormone secretions during adulthood and are also due to the tendency of females to make more exploration-linked errors.


In many species, differences in the neural regulation of sex behavior, pituitary activity, and other patterns of response associated with sexual differentiation have been clearly linked to the presence or absence of testosterone during a critical period before or after birth (Beach et al. 1969; Gorski, 1971; Whalen, 1968). In rats, the presence of endogenous or exogenously administered testosterone during the first 10 days postpartum causes a permanent alteration toward the normal male phenotype in the cellular structure and organization of the preoptic area of the hypothalamus (Raisman and Field, 1973), as well as in neuroendocrine function (Gorski, 1971).

In several recent studies it has been demonstrated that maze learning is a sexually dimorphic ability at which males excel (Davenport et al. 1970; Dawson et al. 1973, 1975; Joseph, 1979). These sex differences in visual spatial learning skills appear to be effected by absence or administration of testosterone shortly after birth. Dawson et al. (1975) have found that castration of male rats or injection of females with testosterone during the third and fifth days postpartum alters their adult maze learning ability toward that of their intact opposite-sex counterpart. However, although sex differences in maze learning appear to be dependent on the presence or absence of testosterone during specific critical periods, the role of gonadal secretions in the mediation of these differences during adulthood has not been directly evaluated.

In addition, several studies have indicated that females are more exploratory than males (Archer, 1975; Hughes, 1968; Russell, 1975, 1977); hence it is possible that the visual spatial superiority of male rats is a function of the females greater tendency to explore and consequently make more errors. Therefore, following the suggestions and procedures outlined by Joseph (1979) the competitive contribution of exploration to the accumulation of maze learning errors was also assessed.


In this experiment male and female rats were treated for the first 10 days of life postpartum with cyproterone acetate, a competitive inhibitor of androgens (Arai and Gorski, 1968; Berswordt-Wallrabe and Neumann, 1967; Bloch and Davidson, 1967; McEwen et al. 1970); females were injected with testosterone during the first 4 days of postpartum, and males were castrated 12 hr after birth.

Subjects and Treatment of Groups

Twenty-six male and twenty-five female Zivic-Miller rats born to seven females bred in this laboratory served as subjects. Four to twelve hours after birth subjects were removed from the breeding cage, counted, separated according to sex, group weighed, and assigned to treatment groups. Seven male subjects were anesthetized by cooling at ñ negative 10 Celsius for approximately 10 min. and then were bilaterally castrated while laid upon a towel-covered pan of ice. Eight males and seven females received 5 mg/0.05 ml of the antiandrogen steroid cyproterone acetate suspended in sesame oil during the first 12 hr immediately after birth (Day 1) and on Days, 4, 7, and 10 postpartum. Six females were injected with 5 mg/0.10 ml testosterone propionate dissolved in sesame oil on Days 1 and 4. The oil control groups consisted of six females and five males, which received 0.10 ml of sesame oil on Days 1, 4, 7, and 10 postpartum. All injections were given subcutaneously at the nape of the neck. Puncture holes and incisions were sealed with colloidion. In addition, six males and six females comprised an intact nonmanipulated control group.

At 21 days of age, all subjects were weaned and housed in same-sex/condition pairs in standard laboratory suspended cages. All subjects were maintained on Purina Lab Chow and water ad libitum except when training and testing procedures called for hunger motivation.


Maze. A symmetrical maze developed by Davenport et al. (1970) was used for behavioral testing (see Fig. 1). The maze field measured 76.2 cm^2 , the goal boxes measured 12.7 * 43.2 cm, and all were painted flat black. A flat black Plexiglas door was used to separate the goal box from the maze field when require by training or testing procedures. The floor of the maze consisted of 1.27-cm^2 wire mesh. The roof was of clear Plexiglas marked with grid lines and grooves into which flat black plastic barrier boards were affixed. The grid lines formed a series of 10.16-cm^2 areas which aided in marking error zones and barrier-board placements. At the end of each goal box was a small (2.4 cm in diameter) plastic Petri dish into which a food reward of two Noyes pellets was dispensed upon completion of a trial. Pellet dispensers (Davis Scientific Instruments, Model No. PD 104) were located outside and adjacent to each goal box.

Indirect illumination was provided in the testing room (4 * 5.5 m) by a fluorescent lamp with two 15-W bulbs. Throughout behavioral training and testing, an audio generator (BRS/LVE, Model No. AU-902/112-10) produced white noise at approximately 25 db to mask extraneous sound.


Throughout training and testing, the performance of the subjects was monitored via a video television camera positioned 2 m directly above the maze. To further minimize disturbances during testing, all observations of the televised sessions were made in a room separate from the maze.

Training. At approximately 180 days of age all subjects were weighed and immediately placed on a 23 hr/day food deprivation schedule. Water continued to be available ad libitum. When all subjects had lost approximately 15 to 20% of their initial weight they began magazine training.

Each subject was removed singly from the animal colony room, brought to the testing room, and placed in a holding cage for approximately 5 min. It was then removed and placed in one of two goal boxes which were closed off from the rest of the maze. During this initial training session each subject received a total of 24 Noyes food pellets, two at a time approximately every 20 sec. Each subject was then removed, weighed, and returned to the animal colony room.

On the following day this procedure was duplicated except that each subject was placed in the opposite goal box. Thus all subjects were trained once in each goal box.

On the third day of training, each subject was removed singly from the animal colony and placed in a testing room holding cage for approximately 5 to 15 min. The subject was then placed in the goal box in which it had been trained the first day. Following a 2-min habituation period, the door barrier was removed from the goal box, allowing the subject to enter the open field. When the subject crossed the open field and entered the opposing goal box it was rewarded with two Noyes food pellets. The subject was not rewarded again until it had recrossed the open field and entered the initial goal box. Each separate shuttle between goal boxes was rewarded and constituted a trial. Upon completion of the 12th training trial, the subject was removed, weighed, returned to its home cage, and fed an amount of food (depending on its weight) sufficient to maintain it at approximately 80 to 85% of its ad libitum weight.

Subjects were run every other day for the remainder of the training trials and tests for maze learning ability. In addition to open-field training trials, all subjects received 12 trials of training on each of Davenport et al.ís (1970) two practice mazes P-3 and P-4.

Testing. The regimen for handling, transporting, habituation periods, and weight maintenance remained as outlined for training. Seven of the maze problems of Davenport et al. (1970) were used as measures of maze learning ability (T-3 and T-7 through T-12). The following procedures were utilized for scoring errors and trials on these mazes. On each maze problem all subjects were run for a minimum of 12 trials, with a maximum of 45 trials and/or 99 errors. Achievement of four out of five errorless trials constituted the criterion for having learned each maze problem. Errors were scored according to the rules employed by Rabinovitch and Rosvold (1951, p. 125), except that crossing an error line was judged by the crossing of a subjectís head and shoulders due to the experimenterís overhead view. Entrances into error zones marked '1' and '2' (see Davenport et al. 1970, p. 113) were scored as individual errors for each zone entry. Entry or reentry of a '1' zone from a '2' zone was not scored as an additional error.

Two independent raters scored the errors and trials on the different maze problems. One hundred percent agreement was found between raters for judgments of total trials and errors accrued on maze problems jointly scored.

At the end of the experiment, all subjects were returned to ad libitum feeding conditions. Approximately 60 days later the genitalia of the females was inspected, and the presence of ovarian cyclic activity was determined by microscopic analysis of vaginal fluid over a 12-day period. In addition, males were anesthetized and their penis and/or testes were removed, measured, and weighed.


Since significant effects were found for both trials and errors (P < 0.01) across each of the seven maze problems (one-way analyses of variance), and the behavioral differences were comparable on all maze problems, the results were pooled, and separate one-way analyses of variance were performed for total trials and total errors accumulated by each group across the seven maze problems. Means are presented in Table 1. Significant effects were found for trials, F (7, 50) = 20.22, P < 0.0001, and errors, F(7, 50) = 19.50, P < 1.0001. A Newman---Keuls test for all ordered pairs of means of cumulative trials and errors is summarized in Table 2.

Trials and errors. Nonmanipulated males took significantly fewer trials (P < 0.01) and made significantly fewer errors (P < 0.01) in learning the seven maze problems as compared to males and females injected with cyproterone acetate, castrates, oil-injected females, and normal females. Likewise, testosterone-injected females required significantly fewer trials and made significantly fewer errors than did cyproterone males (P < 0.05), castrates, oil-injected females, and normal females ( P < 0.01). Similarly, significantly fewer errors and trials were taken to learn the maze problems by oil-injected males as compared to castrates, oil-injected females, normal females (P < 0.01), cyproterone-injected females (trials = P < 0.05), and cyproterone-injected males (trials = P < 0.01). It was also established that cyproterone acetate injections to females improved their maze learning performance, as they took significantly fewer trials (P < 0.01) and errors (P < 0.05) than castrates, oil-injected females, and normal females to learn the maze problems. In addition, it was found that males injected with cyproterone acetate made fewer errors and required fewer trials than castrates and normal females (P < 0.01).

Anatomical Measurements

Table 3 summarizes the effects of differential neonatal hormone manipulation on adult penis length and/or testes weight, as well as gross body weight for all groups. Compared to nonmanipulated males, neonatal castration exerted the greatest retardation of penis development (P < 0.01, NewmanóKeuls test), whereas neonatal treatment with cyproterone significantly reduced both penis (P < 0.01) and testicular weight (P < 0.05). Females injected with testosterone or cyproterone acetate were found to have severely modified vaginal openings, such that their genitalia resembled a combination penis-vagina. Females injected with testosterone demonstrated a total lack of ovarian cyclic activity, whereas the cyproterone acetate-injected females were found to experience absence or irregularity of ovarian cycling. Furthermore, females injected with testosterone were significantly heavier than normal and oil-injected females (P < 0.05, Newman-Keuls test).


Analysis of the trial and error to completion scores for all subjects indicated no significant differences between testosterone-injected females, nonmanipulated males, and oil-injected males. As expected, males neonatally injected with cyproterone acetate or neontally castrated were found to have a significantly depressed maze learning ability.

It is interesting to note that cyproterone acetate injections did not retard development of the male penis or maze learning performance as greatly as did castration of the males. Possibly some masculinization occurred after the cyproterone injections were stopped or before they were started. Also, the dose level of the cyproterone acetate may have been too small to completely saturate the receptor site. Another possibility is that these effects are due to high specificity of the cyproterone acetate to bind with, and therefore mimic, the masculinizing action of testosterone at the receptor site. As noted, females injected with cyproterone acetate performed the visual spatial maze problems with significantly fewer errors and trials than castrates, oil-injected females, and normal females. Furthermore, there was some masculinization of the genitalia as well as a disruption of normal ovarian cycling. In addition, an individual check of the data for each maze indicates that females injected with cyproterone acetate took fewer trials and made fewer errors than did males which received identical injections. This may due to differences in levels of steroid activity at the receptor site and the lack of competitive inhibition and antagonism between cyproterone acetate and testosterone in the female neonate.


This study was intended to discover if the sex differences discovered in Experiment 1, as well as the normal sex differences in maze learning, may be due to gonadal hormone activities.

Subjects and Treatment of Groups

Twenty-four experimentally naÔve adult male and female Zivic-Miller rats obtained from the Zivic-Miller company began the experiment at approximately 120 days of age. Upon arrival in our laboratory, subjects were randomly assigned to experimental and control groups, housed individually in standard laboratory suspended cages, and maintained on Purina lab chow and water ad libitum except when training and testing procedures called for hunger motivation. Subjects were anesthetized with 0.3 ml of Nembutal, sham gonadectomized (six males and females per group), or bilaterally gonadectomized (six males and females per group). In addition, all subjects were maintained on 25 mg of tetracycline per day for 7 days after surgery.

Twenty-eight days after surgery all subjects were weighed, immediately placed on a food deprivation schedule, and reduced to approximately 85% of their ad libitum weight. After 31 days all subjects began magazine training.


The maze and procedures for training and testing remained exactly as described in Experiment 1, with the exception that only four maze problems of Davenport et al. (1970) were used (T-8 through T-11).


One-way analyses of variance were performed separately on each maze problem for trials and errors accumulated. Significant differences were found for errors (P < 0.01) and trials (P <0.05) , however, Newman Keuls tests on all ordered pairs of means indicated that the normal sex differences in maze learning were maintained regardless of condition (errors and trials = P < 0.05) and that gonadectomy had no effect on acquisition between same-sex groups. Hence, the data were pooled and two separate one-way analyses of variance were run for total trials and errors. Significant differences were maintained between groups: errors F(3, 20) = 14.78, P < 0.0001; trials, F(3, 20) = 7.94, P < 0.002. As in each of maze problems, a Newman-Keuls test on all ordered pairs of cumulative means indicated that both sham-castrated and castrated males required significantly fewer trials and made fewer errors than did ovariectomized females (trials = P < 0.05; errors = P < 0.01) or sham-operated females (P < 0.01). Moreover, though castrated males performed more poorly than sham-operated males, and ovariectomized females performed slightly better than sham-operated females, these differences remained nonsignificant (see Table 5).


The results indicate that the presence or absence of testes or ovaries and, in effect, sex differences in the gonadal release of hormone during adulthood do not contribute significantly to sex differences in maze learning.


Several studies have shown that females are more exploratory than males (see Introduction). However, because the primary motivation of rats during the learning of a maze problem is assumed to be the desire to obtain food, then the contribution of exploration to errors accrued during maze learning must be determined and measured in a context which presents similar exploration-induced stimuli, as well as maintaining behavior directed toward the discovery of food (Joseph, 1979)


Sixteen adult Zivic-Miller rats obtained from the Zivic-Miller Company began the experiment at approximately 220 days of age. Subjects were maintained in same-sex pairs in standard laboratory cages and maintained on food and water ad libitum except when testing conditions called for hunger motivation.


All subjects were tested for maze learning ability prior to being tested for competitive exploratory behavior. The maze and procedures for training and testing remained as described in Experiment 1.

Two exploratory maze fields were used in this experiment (see Fig. 2). In each measure a clear path was maintained between the two goal boxes of the Davenport maze, providing all subjects with an unobstructed, problem-free path. On either side of the open path, barrier boards were set up symmetrically as in the maze problems, creating alleys and tunnels which could be freely explored. As in the maze problems goal-seeking behavior was maintained and rewarded.

Subjects were run for 4 consecutive days, 2 days on E-1 followed by 2 days of tests on E-4, for a total of 12 trials per day. Entering an alley or maze tunnel (exploratory zone) was scored as a ìsection entry.î The procedures outlined in Experiment 1 for testing maze learning ability, including transporting, habituation periods, and scoring remained the same for these measures.


Maze Learning. A one-way analysis of variance was performed independently for errors and trials accumulated by males and females across the seven maze problems. It was found that males took significantly fewer trials, F(1, 14) = 5.23, P < 0.03, and made fewer errors, F (1,14) = 9.12, p < 0.009, than females in learning the seven mazes.

Exploratory Behavior. A 2 * 2 * 2 analysis of variance was performed for sex with repeated measures on two factors, test days and exploratory mazes. Significant main effects were found for both exploratory mazes, F(1,14) = 28.307, p < 0.0002, such that females made significantly more section entries than did males. There were no differences between exploratory mazes E-1 and E-4 in the amount of exploratory behavior exhibited. However, there was a significant difference for test days, F (1, 14) entries on Day 1 as compared to Day 2 (E-1) and on Day 3 as compared to Day 4 (E-4). An interaction over days was also found, F(1,14) = 45.432, p < 0.0001, in that females differentially decreased their exploratory behavior at a greater rate over days than males. Means are presented in Table 6.


The data presented in this study supplement previous reports which have shown that females are more exploratory than males (Archer, 1975; Hughes, 1968; Russell, 1975, 1977) and strongly suggest that the poor maze learning performance of females is a function of their greater tendency to make ìirrelevantî section entries.

Although it is possible that the greater responsiveness of females is due to their greater activity levels (Archer, 1975), the finding that females made fewer section entries on the second day of testing, only to increase their section-entry responses on the third day of testing when the new exploratory field was introduced (see Table 6), is similar to effect described in other studies of exploration (Fowler, 1965; Joseph, 1979; Russell, 1977) and argues in favor of an exploration interpretation of their behavior.


The results of Experiments 1 to 3 indicate that nonmanipulated normal oil-injected, sham-operated, and ovariectomized females have inferior visual spatial skills and do not learn mazes as quickly as nonmanipulated, oil-injected, sham-operated, and castrated adult males. In particular, the finding that adult ovariectomy or castration of females or males does not significantly alter normal sex differences in maze learning, whereas neonatal castration of males and treatment with cyproterone acetate significantly reverses normal differences in maze learning due to sex, supports the hypothesis that these differences are determined by the presence or absence of a masculinizing steroid during a critical period extending 10 days of postpartum. That is, brain sex, rather than subsequent gonadal hormone activities during adulthood, greatly influences maze learning ability. However, alteration in peripheral hormone levels resulting from conditions described in Experiments 1 and 2 may significantly contribute to these differences in acquisition performance.

In addition, although no data are offered in regard to the influences of hormones on curiosity and exploratory behavior, it appears that sex differences in maze learning ability are due to the tendency of females to make more ìirrelevantî exploratory section entries than males. Although it is possible that differences in exploratory behavior as well as maze learning ability may be functions of differential influence of 15% weight reduction on the motivation levels of groups differing significantly in free-feeding weight, it is important to note that females injected with cyproterone or testosterone (Experiment 1), though they weighed less than cyproterone-injected males, still demonstrated a superior maze learning performance compared to this group overall.

Copyright: 1996, 2000, 2010, 2018 - Rhawn Joseph, Ph.D.