Psychoneuroendocrinology,Vol. 18, No. 7, pp. 521-531, 1993 03116-4530/93$6.00 + .

00
Printed in the U.S.A. © 1993PergamonPress Ltd.

NO DIFFERENCE IN COGNITIVE PERFORMANCE

BETWEEN PHASES OF THE MENSTRUAL CYCLE

H A R O L D W . GORDON 1 a n d PETER A . L E E 2
Departments of IPsychiatry and 2pediatrics, University of Pittsburgh School of Medicine,
Pittsburgh, Pennsylvania, U.S.A.

(Received 11 September 1991; in final form 5 May 1993)

SUMMARY
No differences were seen in performance on either verbosequentiai or visuospatial neuropsychologi-
cal tests among three groups of women. There were also no differences among the phases of the
menstrual cycle on any of the tests. The design of this study was to sample performance of women
on verbosequential and visuospatial neuropsychological tests during phases of the normal menstrual
cycle in regularly menstruating women, and at comparable time points among women taking oral
contraceptives and among those who had amenorrhea. Serum leutinizing hormone (LH), follicle-
stimulating hormone (FSH), estradiol, and progesterone were assayed to document these hormone
levels at each time point for each group. These data are consistent with previous results in our
laboratory but conflict with other reports in which shifts in cognitive function were related to
fluctuations in estrogen. The only indications of positive results were weak correlations with
gonadotropins and cognitive function across subjects. While some of these correlations replicated
previous results, they were weak and should be considered with caution.

THERE ARE MULTIPLE reports that males and females perform differently on a number
of tests of cognitive processing. For some verbal skills, such as producing as many words
as possible in a minute that begin with a given letter, females are consistently superior.
Conversely, for mental rotation and other visuospatial skills demanding mental manipula-
tion of geometric forms, males are consistently superior. However, there are also several
verbosequential and visuospatial tests for which males and females do not differ, making
it difficult to understand the basis of sex differences in cognitive function. Neither social
nor biological theories have provided totally adequate explanations for these persistent
inconsistencies (Halpern, 1986).
The strongest evidence in support of a biological model is a large study in which there
was a Sex × Hand x Performance Level interaction, wherein high-performing right-
handed males and nonright-handed females performed better on a composite of many
visuospatial skills, while nonright-handed males and right-handed females performed
better on a composite of verbal skills (Harshman et al., 1983). This interaction was

A d d r e s s c o r r e s p o n d e n c e a n d r e p r i n t r e q u e s t s to: H a r o l d W. G o r d o n , P h . D . , N a t i o n a l I n s t i t u t e o f
D r u g A b u s e , R o c k w a l l I I - - S u i t e 615, 5600 F i s h e r s L a n e , RockviUe, M D 20857, U . S . A .

521
522 H.W. GORDONet al.

replicated in a sample of several hundred subjects using a test battery consisting of only
four visuospatial and four verbosequential tests (Gordon & Kravetz, 1991). The Sex ×
Hand interaction was especially strong for tests in which there was also a main effect
for sex.
The Sex × Hand interaction, alone, still does not provide a clue as to the source
of sex differences in cognitive function. The compelling hormone differences between
males and females suggest one possible source. Accordingly, numerous studies have
tried to relate hormonal variations to differences in cognitive performance in patients
with elevated or reduced hormone levels such as in Turner's syndrome (Rovet &
Netley, 1980), precocious puberty (Money & Neill, 1967), or idiopathic hypogonadism
(Hier & Crowley, 1982), or in normal subjects who differ in levels of testosterone
(Klaiber et al., 1967).
Natural fluctuations of hormones in regularly menstruating women provide a different
model for comparison of task performance in which there is variation of actual and
relative levels of LH, FSH, estrogen, and progesterone within the same subject. Few
such studies have been reported. During menstruation, hormones (especially estrogen and
progesterone) are at their lowest levels, followed by steady increases of the gonadotropins
(FSH and LH) and estrogen over the next 2-3 weeks into the preovulatory period
(or follicular stage). Following ovulation (luteal phase), these hormones drop off while
progestrone increases 10- to 100-fold. Oral contraceptives reduce the overall level of
hormones and prevent the monthly fluctuations. In one study in which simple response
times and speeds of arithmetic calculations were measured every 2 days, there was
enhancement of both tasks during the luteal phase (Wuttke et al., 1975). No enhancement
was seen for a group of women taking oral contraceptives. However, only group data
were reported; there was no indication of whether all women had the same pattern of
cognitive functioning, nor whether the cognitive functioning was correlated with hormone
concentrations. Even though assays verified hormone levels, it was not reported whether
hormone fluctuations were consistent with ovulation.
Basal body temperature was used in another study to infer the maximum estrogen
peak (preovulatory), and the postovulatory progesterone peak (Broverman et al., 1981).
The hypothesis was that "automatic" tasks such as rapid naming would be facilitated
at maximum estrogen, while a "perceptual restructuring" task would be better performed
at maximal progesterone. Of the 87 subjects who participated, 21 were believed not to
have ovulated as determined by failure of a "postovulatory" temperature rise. Stringent
temperature requirements left only 19 subjects for analysis. The predicted effect was
observed for two of the four tests: relatively faster (color name) reading during the
preovulatory period, and faster performance on an embedded figures task during the
postovulatory (luteal) period.
These results were contradicted by a small 12-subject study in which a task of pattern
discrimination (matrices of paired dots vs. evenly spaced dots)wperceptual restructur-
ingwwas performed w o r s e in the luteal phase (Ward et al., 1978). The results were
also contradicted by b e t t e r flicker detection, faster response times, and better mental
arithmetic--automatic tasks--during the luteal phase (Wuttke et al., 1975).
A more recent series of studies has investigated the role of hormonal fluctuations for
cognitive tasks for which there is a sex bias in performance (Hampson, 1990a, 1990b;
Hampson & Kimura, 1988). These tasks are similar to those used in the earlier automatism
and perceptual restructuring tasks. Subjects were tested in two sessions--the menstrual
and mid-luteal p h a s e s w t o compare performance on cognitive tasks when there were
 COGNITIVE PERFORMANCE AND MENSTRUAL CYCLE 523

low levels of steroid hormones (estrogen and progesterone) to performance when there
were high levels of both. There was a significant Group x Phase interaction for the
composite ofaU tests, even though some tests did not show the expected effect (Hampson
& Kimura, 1988). In a spatial task for which males are usually superior--the rod and
frame task--there was better performance during the menstrual period. By contrast,
during the mid-luteal phase there was better performance for manipulative tasks such
as finger tapping, peg assembly, and some of the dependent measures of the Manual
Sequence Box (Kimura, 1977).
In an attempted replication of their initial report (Hampson, 1990a), the stronger effect
for spatial ability was, once again, found in Session 1 between the Menstrual vs. Luteal
group. However, there was no within-subject change for these tests between the phases.
For some of the manipulative tasks, the expected better performance was demonstrated
for the luteal phase both between groups and within subjects. Female-superior verbal
fluency failed to show group differences in Session l, although the expected interaction
did manifest between the menstrual and mid-luteal phases. Unfortunately, hormones
themselves were not measured, causing a potential source of error. Nevertheless, the
results gave preliminary support to the idea that female-superior tests are performed
better when both estrogen and progesterone are high; male-superior tests are performed
better when these hormones are low.
To test for the possibility that estrogen, and not estrogen plus progesterone, was
responsible for increasing female-superior skills, a third study was done (Hampson,
1990b). Subjects were tested during the menstrual period and the preovulatory period.
For most subjects, blood samples were taken and hormone levels were determined. Over
40% of the subjects had to be eliminated for nonspecific reasons or because they were
tested in the wrong cycle phase and estrogen was not at its proper elevation. The
results compared favorably to the menstrual/luteal comparison. Male-superior tests were
performed better in the menstrual period and female-superior tests were performed better
in the preovulatory period. It was concluded that estrogen alone was responsible for the
shift in cognitive performance.
However, performance on some individual tests did not support the composite results.
For example there was no difference across the cycle phases for male-superior deductive
reasoning while there was a significant difference in the predicted direction (menstrual
> preovulatory) for spatial skills. Only articulation and manual speed of the four female-
superior tests had increased in the preovulatory phase. Perceptual speed and, more
importantly, verbal fluency did not differ.
These contradictory findings and those of earlier studies are enigmatic and suggest
that the true picture is still elusive. Our own previously reported results have contributed
to this emerging puzzle. On the positive side, we had demonstrated both a within- and
between-subject relationship between g o n a d o t r o p h i c hormones (and not testosterone)
and specialized cognitive functions in m e n (Gordon et al., 1986; Gordon & Lee, 1986).
The strongest effects were a relationship whereby low levels of gonadotropins were
associated with high performance on spatial tasks. To a lesser extent a similar result
was found in women, but with the complementary effect that better verbal fluency was
associated with higher levels of gonadotropins. But we did n o t see a change within
subjects across the menstrual cycle, nor were there any trends. The present study was
intended to re-investigate cognitive changes at three cycle phases within subjects and
compare task performance across groups of women with and without oral contraceptives,
and to a group of amenorrheal women.
524 H.W. GOt~DONet al.

METHOD
Subjects
Three groups of female subjects were recruited from advertisements in the student
newspaper and on bulletin boards. Two groups consisted of 34 subjects each who reported
regular menstrual cycles; one group was taking oral contraceptives, one group was not.
The third group included 14 subjects who were amenorrheal, not having menstruated in
the 180 days prior to testing. Medical history did not suggest underlying pathophysiology
for this group other than amenorrhea. Hormone levels excluded either a hypergonado-
tropic or hypogonadotropic state. It is our impression that this group represents the
dysfunction of which results in the lack of synchrony, a relatively common form of
amenorrhea in this age group. The subjects ranged in age from 18 to 36 years.

Procedure
Volunteers were briefed on the purpose of the study by phone interviews or preliminary
group meetings. Part of the purpose of the interviews was to discourage heavy smokers,
drinkers, and substance abusers from volunteering. However, no laboratory assays were
conducted to prove this. Also the women were asked to estimate the date of their next
menses. For women on oral contraceptives, these dates could be stated with more
assurance. From these estimated dates, subjects were tentatively assigned to convenient
dates for their first testing that would correspond to one of the following: (1) Menstruation,
the second or third day after onset of menses; (2) Follicular, 10-14 days after onset
menses; or (3) Luteal, 20-24 days after onset of menses. Onset of menses was corrobo-
rated by interview before the subject was actually tested. Because hormones would be
measured by assay, more stringent verification at the time of testing was not deemed
necessary. Subjects were on combination estrogen/progesterone contraceptives and
therefore received both hormones each of 21 days of the cycle. This resulted in the
expected suppression of L H and FSH as verified by assay. Subjects on contraceptives
containing high dosages of progesterone, or on sequential estrogen and then progesterone,
were excluded.
Once subjects were assigned to one of the three first-test sessions, they were tested
at the estimated time of each subsequent cycle: Subjects who were tested at menstruation
were tested about 10 days later and then 10 days after that, and so on, with the constraint
that testing at the time of menstruation was confirmed by interview. Amenorrheal women
were scheduled as convenient, but with about 10 days between testing sessions.
Prior to the first testing, informed consent was obtained together with a medical
screening which included weight, temperature, and arm fat measurements. The medical
screening was also performed to assure that the volunteers were healthy and understood
the medical procedure for blood draws. Amenorrheal women were all healthy and experi-
enced normal pubertal development and onset of menses. Subjects were screened by
history and physical appearance. Those with any evidence suggestive of excessive exer-
cise, any major illness, bulimia, or anorexia were excluded. The review of history,
including past menstrual history and menarche plus gonadotropin measurements, were
used to verify that hypogonadotropic hypogonadism or hypergonadotropic hypogonadism
(ovarian failure, including Turner syndrome) were not present.
Subjects reported to the laboratory in the early morning at which time an indwelling
catheter was placed in a vein in the left arm. This was followed by a half-hour adaptation
period that included test instructions and relaxation.
 COGNITIVE PERFORMANCE AND MENSTRUALCYCLE 525

Neuropsychological Tests
Neuropsychological testing started with administration of an anxiety questionnaire
(Spielberger et al., 1970) followed by the Symbol Digit Modalities Test (Smith, 1973),
which is a sensitive test of performance level. There were eight tests of specialized
cognitive function collectively called the Cognitive Laterality Battery, which has been
used in several previous studies and normed on a large sample of adults (Gordon, 1986).
While it is true that these tests were originally chosen as measures of functions associated
with the left and right cerebral hemispheres, the important point for this study is that
the tests loaded on one of two orthogonal factors that can be labelled "verbosequential"
and "visuospatial," which have demonstrated sex differences in performance (Gordon
& Kravetz, 1991). Similar considerations have dictated test selection in other studies of
hormone/behavior relationships (e.g., Hampson, 1990a, 1990b).
The verbosequential tests included two tests of perception and recall of auditorily
presented sequences. One test used familiar sounds as stimuli; the second test used
digits. The subjects had to write the names of the sounds or the digits in the correct
sequence. There were two tests of written word production. In one test, the subject
wrote as many words as possible that began with a given letter of the alphabet (three
letters were used), or in a given category (two categories were used). The visuospatial
tests included a test of locating points in two dimensional space, rotating geometric
figures (constructed from figures of Shepard & Metzler, 1971), imagining blocks in three
dimensions (adapted from MacQuarrie, 1953), and perceptual closure (using stimuli from
Thurstone & Jeffrey, 1966, and from French et al., 1962). The tests were given in the
same order; tests of verbosequential ability were interspersed with tests of visuospatial
ability.
The tests were presented on 35 mm slides and audio tapes of a sound/sync projector.
The entire test session lasted about 2 h, including the preparation and adaptation period.

Hormone Assays
Four blood specimens of 8 ml each were taken at regular intervals throughout the 80-
min test session for later assay. For assay, the blood samples were pooled and all
hormones were determined in triplicate. Estradiol and progesterone were measured by
radioimmunoassay after prior separation by L H 20 Sephadex Column Chromatography
(Gutai et al., 1977; Radfar et al., 1976). FSH and L H were measured by a double antibody
radioimmunoassay (Odell et al., 1966).

RESULTS
Group Comparisons
There were no differences among groups for either the verbosequential composite or
the visuospatial composite, nor were there any interactions with the test session. There
were also no group differences, or interactions with the "cognitive profile" defined as
the difference between visuospatial and verbosequential performance (cognitive profile
= visuospatial - verbosequential). The data were analyzed by a two-way analysis of
variance with the three sessions as a repeated measure. Two subjects were eliminated
from all analyses because their progesterone levels during the menstrual phase were
greater than 150 ng/dl. There were no trends in the data; F-values for all between-group
measures were below 1.0. There was a strong effect for Sessions, of course, because all
groups improved with successive testing. Neither the Symbol-Digit Modalities Test nor
526 H . W . Goaoor~ et al.

TABLE I. COMPARISON OF COGNITIVE PERFORMANCE AMONG CYCLE PHASES

Cycle phase

Menstrual Follicular Luteai

Visuospatial composite
Regular menses 0.622 -+ 0.612 0.694 -+ 0.546 0.610 ± 0.725
Oral contraceptives 0.787 _+ 0.526 0.807 -+ 0.586 0.775 _ 0.500
Amenorrheal 0.627 -+ 0.728 0.547 ± 0.556 0.655 ± 0.672
All subjects 0.689 -+ 0.599 0.713 --- 0.565 0.685 --- 0.630
Verbosequential composite
Regular menses 0.826 -+ 0.794 0.848 - 0.699 0.864 --+ 0.748
Oral contraceptives 0.919 ± 0.574 0.868 -+ 0.598 0.799 - 0.718
Amenorrheal 0.592 -+ 0.953 0.684 - 0.832 0.593 --- 0.912
All subjects 0.821 ± 0.746 0.827 ± 0.681 0.789 ± 0.764
Cognitive profile
Regular menses -0.204 --- 0.796 -0.154 --- 0.693 -0.254 ± 0.808
Oral contraceptives -0.132 -+ 0.661 -0.061 ± 0.661 -0.024 --+ 0.718
Amenorrheal 0.035 --- 0.776 -0.137 ± 0.743 0.062 - 0.709
All subjects -0.131 -- 0.736 -0.114 ± 0.682 -0.104 -+ 0.757

All data are in standard units, with mean = 0, SD = 1.

the State-Trait Anxiety Index was related to the hormone levels nor did they affect the
results when used as co-variates.
Not unexpectedly, the groups did differ in levels o f all hormones, despite considerable
variation within groups. The regularly menstruating women who were not taking oral
contraceptives had 50% higher concentrations of both the gonadotropins and estradiol
(measurement of both endogenous and exogenous) than the other regularly menstruating
group taking oral contraceptives. The noncontraceptive group had lower gonadotropin
and higher estrogen and progesterone levels than the amenorrheal group. Results indicate
that the groups did not perform differently on any of the tests that have been purported
to be related to hormone levels.
A second analysis was performed among groups by rearranging the subjects to compare
across cycle phase as the repeated measure. The experimental design was such that for
each cycle phase (menstrual, follicular, luteal) one-third of the subjects would be taking
the tests for the first time, one-third for the second time, and one-third for the third time.
The repeated measures failed to demonstrate differences in cognitive function across
cycle phases (Table I). Again, there were no interactions or trends; all F-values were
less than 1.0. The hormone levels, however, are consistent with levels expected for the
phase o f the cycle, effect of oral contraceptives, or amenorrhea. Data for the hormone
levels are summarized in Table II.
Additionally, the one group of regularly menstruating women who were not taking
oral contraceptives were divided into subgroups according to cycle phase of their first
test session only, and compared across groups for that session. Once again, there were
no subgroup differences, indicating that performance on neither the visuospatial nor the
verbosequential neuropsychological tests differed among the menstrual, follicular, or
luteal phases of the menstrual cycle. The analysis was repeated once again with these
subjects plus the group of regularly cycling women who were taking oral contraceptives.
Again, there were no differences in cognitive function.
 COGNITIVE PERFORMANCEAND MENSTRUALCYCLE 527

TABLE II. C O M P A R I S O N OF H O R M O N E S A M O N G CYCLE PHASES

Cycle phase

Menstrual Follicular Luteai

Follicle stimulating hormone (mlU/ml)

Regular menses 7.5 +-- 3.4 8.6 -+ 3.7 4.8 -+ 3.3
Oral contraceptives 5.5 +- 4.0 4.7 +- 2.2 1.8 +- 1.2
Amenorrheal 8.4 --- 3.8 7.1 --- 2.1 7.7 --- 1.9
All subjects 6.9 --- 3.8 6.8 --- 3.4 4.2 --- 3.2
Luteinizing hormone (mlU/ml)
Regular menses 6.2 -+ 6.1 13.8 --- 15.0 5.0 -+ 5.3
Oral contraceptives 3.9 -+ 3.1 4.8 -+ 4.0 1.9 +-- 1.3
Amenorrheal 16.9 -+ 32.1 9.4 - 8.2 9.7 - 8.8
All subjects 7.2 --- 14.7 9.4 +- 11.2 4.6 - 5.7
Estradiol (pg/nl)
Regular menses 28.5 --- 18.50 33.0 --+ 18.6 35.5 --- 18.9
Oral contraceptives 20.3 - 11.2 23.0 - 12.3 21.6 - 12.8
Amenorrheal 25.6 - 12.8 27.4 -+ 10.7 27.9 --+ 13.2
All subjects 24.7 - 15.2 28.0 - 15.6 28.5 --+ 16.8
Progesterone (ng/dl)
Regular menses 39.2 --+ 26.2 42.5 --- 66.8 1368.9--- 1058.0
Oral contraceptives 16.0 --+ 10.9 14.5 -+ 9.0 36.3 -+ 95.5
Amenorrheal 23.5 --- 30.8 15.6 --- 9.6 13.2 - 7.8
All subjects 27.0 -+ 24.6 26.3 --- 45.3 585.9 --- 949.7

To ensure that the failure to find cognitive differences was not due to a washout by
a few subjects, a reanalysis was performed on the most hormone-variant subjects to
c o m p a r e the menstrual vs. luteal groups and the menstrual vs. follicular groups. Subjects
whose estrogen at menstruation was less than 75% of the value at either the follicular
or the luteal phase were included. Seventeen and 22 subjects fit these criteria for the
menstrual/follicular and menstrual/luteal comparisons, respectively. Separate two-group
comparisons were made for the first session, only. These analyses allowed direct compari-
son to previously reported results (Hampson, 1990a, 1990b). Neuropsychological perfor-
mance at the menstrual phase could be compared to increases of both estrogen and
progesterone (luteal phase), or only to a rise in estrogen alone (follicular phase). Never-
theless, the results for these select subgroups were the same as those for the group
as a whole: There were no differences or trends in performance on cognitive tests
between the menstrual and the follicular phases, or between the menstrual and luteal
phases.
Finally, 12 of the " b e s t " subjects were selected whose h o r m o n e levels were judged
to best fit expected fluctuations throughout the three observed phases of the cycle with
one-third having had their first test session in each of the three cycle phases. Once again
there was failure to find changes in cognitive function that related to cyclical h o r m o n e
fluctuation.

Regression Analyses
Multiple step-wise regressions were performed for the first test session on the
entire group in an attempt to look for hormone predictors for the composite visuospatial
and verbosequential scores, as well as for the cognitive profile. If significant predictors
528 H.W. GORDONet al.

were found for the group as a whole, the validity of the resultant model was checked
for each of the three subgroups: regular, with oral contraception, and amenorrheal.
Significant models were found, but only with the gonadotropins, FSH and LH, which
entered significantly into the regression models. FSH was always positively
related; L H was negatively related. However, even when the predictor models were
significant, the explained variance never exceeded 10%. In predicting the scores of
the visuospatial composite, the model included both FSH and L H and was significant
for the whole sample (F = 3.97, p < .05), but for only one of the three sub-
groups (amenorrheal). In predicting the verbosequential composite, only FSH entered
the model significantly for the whole sample (F = 8.55, p < .01) and for each of
the two regularly menstruating subgroups, but not the amenorrheal subgroup. For
the cognitive profile (i.e., the difference between the visuospatial and verbosequential
composite scores), L H was a significant predictor (F = 10.157, p < .01) for the
whole sample, but only for the one subgroup of regularly menstruating, noncontracep-
tive women.
It was decided to repeat the regression analysis with a stronger methodology. The
total sample was split into two random subsamples to determine if predictors from one
sample could be replicated in the second. While some of the models did affirm that the
gonadotropins, LH and FSH, were significant predictors, these were never replicated
across both random subsamples.
At best, these analyses suggest that gonadotropins may play a larger role than the
steroid hormones in predicting cognitive performance. The data are not strong, however,
so that future studies will have to validate these observations.

DISCUSSION
There was no support for a relationship between hormone levels, or changes in
hormone levels, and sex-biased visuospatial or verbosequential ability. This lack of
support for a hormone-behavior relationship actually replicates our previous failure
to find this relationship within female subjects (Gordon et al., 1986). We previously
reported a weak between-subject correlation in women (and a stronger relationship
in men) between the gonadotropins and cognitive function. FSH had been negatively
correlated to visuospatial skill and also, in women, positively correlated with word
production tests of verbosequential skill. The positive correlation with F S H was
replicated in this study but for all tests, both verbosequential and visuospatial ability.
LH had not previously correlated with cognitive function for either males or females;
it correlated negatively in this study. There was no apparent connection either within
or across subjects for the steroid hormones, a finding which is also at odds with
conclusions in other studies. We have pointed out previously that a correlation with
gonadotropins was reasonable based on animal work of lateralized receptors in the
brain to gonadotropins (Glick et al., 1977) and to a relationship between gonadotropins
and dopamine (Huseman et al., 1980).
The failure to find changes within subjects among the different phases of the menstrual
cycle in this and in our previous study is in direct contrast to the results recently reported
by Hampson (Hampson, 1990a, 1990b; Hampson & Kimura, 1988) and at odds with
several studies mentioned in the Introduction. We are unable to account for the discrep-
ancy but shall discuss each of the most logical confounding factors: the neuropsychologi-
cal tests, the subjects, and the hormone assays.
 COGNITIVE PERFORMANCE AND MENSTRUAL CYCLE 529

Neuropsychological Tests
The neuropsychological tests were first chosen because they are related to functions
of the left or right cerebral hemisphere. They were found to have a sex bias and were,
therefore, appropriate for this study. Although similar to tests used in hormone/behavior
studies of other investigators, these specific tests have been employed only in studies
in our laboratory. Qualitative differences among tests of various studies are not obvious.
Even in the "successful" studies there were always some tests that failed to show the
predicted changes with hormone fluctuations. This leaves one with an uncomfortable
feeling that if specific cognitive functions do relate to hormone levels, the relationship
is not well understood (at best) or unstable (at worst).

Subjects
Our subjects were selected from the university community, although not all were
university students. The level of performance in the present sample was above average,
although a ceiling effect did not occur, at least in the first test session. Multiple testing
may have confounded the results, however. It is possible that there are yet other intra-
subject differences, such as hormonal variations, whose adverse influence cannot be
completely ruled out.

Hormone Assays
The assays used were highly specific assays, double antibody for gonadotropins and
chromatographic separation before radioimmunoassay for steroid hormones. We feel,
therefore, that the hormone levels as reported are an accurate represent of actual and
comparative levels. Because there is normally extensive variation between individuals
during the same phase of the menstrual cycle or while receiving oral contraceptives,
findings such those reported here are inevitable in such studies. While variation in
measurements or specificity of the assays may have confounded the results, we believe
the hormone data represent the actual variation of hormone levels which exist in the
situation studied. Failure to show differences is more likely due to lack of acute hormone
effects, rather than lack of assay discrimination since, for example, the between-group
differencesmespecially between the regular women compared to the contraceptive
groupmfor cognitive performance, were absent. The group of women whose hormones
were significantly lower due to contraceptives did not perform differently from the
regularly menstruating, noncontraceptive women. Furthermore, subjects chosen for max-
imal or "ideal" hormone fluctuations, should still be the best subjects to demonstrate
cognitive fluctuation across the cycle phases, regardless of the absolute hormone levels
as measured by assay. However, the subgroups showed no differences.

CONCLUSION
We are left with a puzzle. Why do some studies show a relationship between hormones
and cognitive test performance and others not? As outlined in the Introduction, the so-
called positive studies, themselves, do not give a consistent picture. What is worse, it
is likely that many investigators have failed to find cognitive variation and either did not
seek to publish the negative results, or the results were not accepted for publication. In
either case, a preponderance of reports of positive results would bias the true picture.
A similar controversy has been raging with the conflicting evidence for performance
differences between sexes, and between handedness groups. One of the resolutions for
530 H . W . GORDONet al.

that argument was that level-of-performance was a mediating factor (Harshman et al.,
1983). The effect on cognitive performance of other hormones (e.g., cortisol, A C T H )
and neurobiological substances is poorly understood, not to mention the influence o f
environmental variables (e.g., stress, lack of sleep, or diet) on these substances. It is likely
that the confounding variables responsible for the current inconclusiveness regarding
hormone/behavior relationships are complicated.

Acknowledgment: This work was supported by NIH Grant HD 16264.

REFERENCES
Broverman DM, Vogel W, Klaiber EL, Majcher D, Shea D, Paul V (1981) Changes in cognitive
task performance across the menstrual cycle. J Comp Physiol Psychol 95:646-654.
French JW, Ekstrom RB, Price LA (1962) Kit of Reference Tests for Cognitive Factors. Educational
Testing Service, Princeton, NJ.
Glick SD, Jerussi RP, Zimmerberg B (1977) In: Harnad S, Doty RW, Goldstein L (Eds) Lateraliza-
tion in the Nervous System. Academic Press, New York, pp 213-249.
Gordon HW (1986) The Cognitive Laterality Battery: Tests of specialized cognitive function. Int
J Neurosci 29:223-244.
Gordon HW, Corbin ED, Lee PA (1986) Changes in specialized cognitive function following
changes in hormone levels. Cortex 22:399-415.
Gordon HW, Kravetz S (1991) The influence of sex, handedness, and performance level on
specialized cognitive functioning. Brain Cogn 15:37-61.
Gordon HW, Lee P (1986) A relationship between gonadotropins and visuospatial function.
Neuropsychologia 24:563-576.
Gutai JP, Kowarski AA, Migeon CJ (1977) Twenty-four hour integrated concentrations of proges-
terone 17-hydroxyprogesterone and cortisol in normal male subjects. J Clin Endocrinol Metab
44:116-120.
Halpern DF (1986) Sex Differences in Cognitive Abilities. Lawrence Erlbaum Associates, Hillsdale,
NJ.
Hampson E (1990a) Variations in sex-related cognitive abilities across the menstrual cycle. Brain
Cogn 14:26-43.
Hampson E (1990b) Estrogen-related variations in human spatial and articulatory-motor skills
Psychoneuroendocrinology 15:97-111.
Hampson E, Kimura D (1988) Reciprocal effects of hormonal fluctuations on human motor and
perceptual-spatial skills. Behav Neurosci 102:456-459.
Harshman RA, Hampson E, Berenbaum SA (1983) Individual differences in cognitive abilities and
brain organization: Part I Sex and handedness differences in ability. Can J Psychol 37:144-192.
Hier DB, Crowley WF (1982) Spatial ability in androgen-deficient men. N Engl J Med
306:1202-1205.
Huseman CA, Kugler JA, Schneider IG (1980) Mechanism of dopaminergic suppression of gonado-
tropin secretion in men. J Clin Endocrinol Metab 51:209-214.
Kimura D (1977) Acquisition of a motor skill after left hemisphere damage. Brain 100:527-542.
Klaiber EL, Broverman DM, Kobayashi Y (1967) The automatization cognitive style, androgens,
and monoamine oxidase. Psychopharmacology 11:320-336.
MacQuarrie TW 0953) MacQuarrie Test for Mechanical Ability. California Test Bureau, Monterey,
CA.
Money J, Neill J (1967) Precocious puberty, IQ and school acceleration. Clin Pediatr 6:227-282.
Odell WD, Ross GT, Rayford PL (1966) Radioimmunoassay for human luteinizing hormone.
Metabolism 15:287-289.
Radfar N, Ansusingha K, Kenny F (1976) Circulating bound and free estradiol and estrone during
normal growth and development. J Pediatr 89:719.
Rovet J, Netley C (1980) The mental rotation task performance of Turner syndrome subjects.
Behav Genet 10:437-443.
Shepard RN, Metzler J (1971) Mental rotation of three-dimensional objects. Science 171:701-703.
 COGNITIVE PERFORMANCE AND MENSTRUAL CYCLE 531

Smith A (1973) Symbol-Digit Modalities Test. Western Psychological Services, Los Angeles, CA.
Spielberger CD, Corsush RL, Lushene RE (1970) State-Trait Anxiety Inventory. Consulting
Psychologists Press, Palo Alto, CA.
Thurstone LL, Jeffrey TE (1966) Closure Speed. Industrial Relations Center, Chicago, IL.
Ward MC, Stone SC, Sandman CA (1978) Visual perception in women during the menstrual cycle.
Physiol Behav 20:239-243.
Wuttke W, Arnold P, Becker D, Creutzfeldt O, Langenstein S, Tirsch W (1975) Circulating
hormones, EEG, and performance in psychological tests of women with and without oral
contraceptives. Psychoneuroendocrinology 1:141-151.