Upload
independent
View
0
Download
0
Embed Size (px)
Citation preview
Biological Psychology 49 (1998) 249–268
Sex differences in seasonal variations in P300
Mary H. Kosmidis 1,a, Connie C. Duncan a,b,*, Allan F. Mirsky a
a Section on Clinical and Experimental Neuropsychology, Laboratory of Brain and Cognition,National Institute of Mental Health, Building 15K, Room 101A, Bethesda, MD 20892-2668, USA
b Department of Psychiatry, Uniformed Ser6ices Uni6ersity of the Health Sciences,4301 Jones Bridge Road, Bethesda, MD 20814-4799, USA
Received 5 February 1996; received in revised form 16 March 1998; accepted 13 April 1998
Abstract
Previous reports of seasonal variations in P300 were based on cross-sectional observationsof subjects tested at different times of the year. In this study, we tested three groups ofsubjects in each of two seasons: winter and spring, spring and summer, and summer andwinter. We found winter or spring maxima in auditory and visual P300 and visual slow wave.This pattern of results, with the amplitude of P300 being inversely related to the amount ofsunlight in a season, supports the hypothesis that the allocation of processing resourcesvaries across the seasons. Our results also suggest a trend for an increased sensitivity ofwomen, as compared with men, to seasonal influences on P300. Although our findings do notprovide strong evidence that P300 varies systematically as a function of season, seasonalfactors appear to affect cognitive processing (as indexed by P300) differentially in men andwomen. © 1998 Elsevier Science B.V. All rights reserved.
Keywords: Event-related brain potentials; P300; Season; Sex differences
* Corresponding author. Tel.: +1 301 2952192; fax: +1 301 5649562; e-mail: connie–[email protected]
1 Present address: Experimental Therapeutics Branch, National Institute of Mental Health, Bethesda,MD 20892, USA.
0301-0511/98/$ - see front matter © 1998 Elsevier Science B.V. All rights reserved.
PII S0301-0511(98)00043-X
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268250
1. Introduction
Several studies have demonstrated seasonal variations in P300. These studieswere stimulated by an investigation of the effects of light therapy on depression, inwhich enhanced amplitude of visual, but not auditory, P300 was observed inpatients being treated for seasonal affective disorder (Duncan et al., 1990, 1991).The increase in P300 was directly proportional to the antidepressant effects oftreatment. Whereas brief light therapy did not affect the P300 of healthy controlsubjects in this study, the investigators speculated that it might be affected by thegradual and prolonged changes in sunlight over the seasons.
Subsequent studies investigated the relationship between season and P300 inhealthy subjects (Deldin et al., 1989a,b, 1994; Polich and Geisler, 1991). Deldin etal. used archival data from subjects, each of whom had been tested once, atdifferent times of the year, to examine seasonal variations in P300. They observedthat P300 was larger in summer and winter than in spring and larger in women thanin men. Their results also suggested that P300 in women was more sensitive to theeffects of season than it was in men. They found no interaction of modality ofstimulus with season or sex in their data. Given the unexpected finding ofaugmented P300 amplitudes in both summer and winter, the investigators specu-lated that some characteristic of season other than amount of sunlight must play amediating role.
Polich and Geisler (1991) also used archival data from a cross-sectional sample ofsubjects performing an auditory discrimination task. These data were collected in amore southern latitude than those reported previously. They found that theamplitude of P300 varied with season: P300 was larger in spring than in fall andwinter; moreover, P300 was larger in women than in men. Seasons in this studywere defined differently from those in the Deldin et al. (1989a,b, 1994) studies; thus,spring, fall, and winter in Polich and Geisler (1991) overlapped with summer,winter, and spring, respectively, in the Deldin et al. reports. They concluded thatP300 amplitude varied with season as a function of the amount of sunlight. Thisfinding is consistent with the hypothesis that sunlight may be one of the mediatingfactors in the seasonal variation in P300.
Previous research thus supports the notion that P300 varies as a function ofseason, and that this relationship may vary by sex, but apparently not by themodality of the stimuli eliciting the event-related potentials (ERPs). These findingswere all based on cross-sectional samples of subjects. We sought to extend previousfindings in a longitudinal investigation of healthy subjects, who were each tested inmore than one season. Within-subject comparisons would provide a more powerfulassessment of seasonal variations in P300 than a cross-sectional design, and wouldhelp to resolve the discrepancy between the patterns of variation reported previ-ously (i.e., increased amplitude in two seasons vs. one season). We also investigatedseasonal variations in the latency and scalp distribution of P300. Furthermore, weconsidered the possibility that any seasonal variation in ERPs might not beattributable directly to the amount of sunlight, but, perhaps, to associated changes,such as photoperiod. Finally, we predicted that women would show more substan-tial seasonal changes in P300 than men.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 251
2. Method
2.1. Subjects
A total of 17 men and 15 women were recruited through newspaper advertise-ments. Their mean age was 26.2 years (S.D.=5.1; range: 18–40), and their meanlevel of education was 15.1 years (S.D.=1.7; range: 11–19). A clinical psychologistinterviewed subjects using the lifetime version of the Schedule for Affective Disor-ders and Schizophrenia (Spitzer and Endicott, 1978): no subjects met the DSM-IV(American Psychiatric Association, 1994) criteria for any psychiatric disorder.Subjects were healthy, had been free of medications for at least four weeks prior toeach laboratory session, and had no history of head injury. They also reported nopsychiatric disorder among their first-degree relatives.
Only a small number of subjects could commit to being tested several timesthroughout the year. Because of this, and due to the transitory nature of the subjectpopulation available to us in the Washington, D.C. metropolitan area, we were ableto test most subjects in two seasons only. For this reason, we limited our analysesto comparisons of data collected from each subject in one pair of seasons: winterand spring (n=11; 5 women), spring and summer (n=12; 6 women), or summerand winter (n=9; 4 women). Fall testing sessions were excluded from the compari-sons because of the small sample sizes.
Seasons comprised months based on the average amount of sunlight per month.Thus, we defined winter as the months November, December, and January, springas the months February, March, and April, and summer as May, June, and July.Our definitions of seasons correspond to those used by Deldin et al. (1994).
The subjects tested in the three pairs of seasons did not differ significantly in sexdistribution, age, or level of education. Male and female subjects tested in each pairof seasons did not differ in terms of age or level of education. All subjects hadnormal or corrected-to-normal vision and reported no hearing impairment. Weobtained written informed consent from all subjects after explaining the nature ofthe experiment, potential risks, and their rights with respect to participation in aresearch project. Subjects were paid for their participation.
2.2. Stimuli and procedure
We used auditory and visual versions of a two-stimulus discrimination task toelicit ERPs. These tasks were identical to those used in the Deldin et al. (1989a,b,1994) studies. Auditory stimuli were tone bursts, either low (600 Hz) or high (1500Hz) in pitch. The two tones were equated for duration (100 ms), loudness (50-dBsensation level), and rise-fall time (10 ms) and were delivered binaurally via stereoheadphones over broad-band masking noise (60-dB sound pressure level). Visualstimuli were the letters S and H presented for 100 ms on a CRT display placed101.5 cm from the subject’s eyes. The letters subtended an angle of 1.0 degreevertically and 0.8 degrees horizontally. The low and high tones and the letters S andH were presented in separate series and in random sequences, with relativeprobabilities of 0.10 and 0.90, respectively.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268252
Subjects were instructed to press a button with one thumb in response to the 0.10stimulus and to press a second button with the other thumb in response to the 0.90stimulus. The buttons assigned to the two responses were counterbalanced acrosssubjects in each group. The stimuli were presented at a fixed interstimulus intervalof 1500 ms. We administered two blocks of 150 trials in each modality. Each trialblock was followed by a short rest period. The order of presentation of blocks ofstimuli in the two modalities was counterbalanced across subjects.
Testing took place in a sound-attenuated, dimly-lit room, with the subject seatedin a comfortable chair. The task followed standard instructions, practice trials toensure that subjects understood the instructions, and an opportunity to askquestions. We instructed subjects to respond as quickly and as accurately aspossible. We also requested that they minimize eye movements and blinking andremain still throughout the task.
2.3. Data collection
The electroencephalogram was recorded with Ag/AgCl electrodes, affixed withcollodion, from frontal (F3, Fz, F4), central (C3, Cz, C4), parietal (P3, Pz, P4), andoccipital (Oz) scalp sites, according to the International 10–20 system, all referredto linked earlobes. We recorded the electro-oculogram (EOG) between the supraor-bital and outer canthal positions of the left eye. A ground electrode was placed atFp2.
The EEG and EOG were amplified with a bandpass of 0.01–100 Hz (−3 dBpoints) (Duncan-Johnson and Donchin, 1979) and digitized at the rate of 200 Hzfor 1100 ms, beginning 150 ms prior to stimulus onset. Impedance of the EEGelectrodes was maintained below 3 kilohms.
Stimulus presentation and data acquisition were controlled by a PDP-11/34computer. Digitized single trial data and speed (to the nearest ms) and accuracy ofresponse on every trial were stored on digital magnetic tape for later analysis.
2.4. Data quantification
We quantified two measures of performance on each task: accuracy and reactiontime to the 0.10 stimuli. Accuracy was defined as the percentage of correct buttonpresses; reaction time was the time from stimulus onset to the button press. Abutton press within 175–950 ms of the onset of each stimulus was considered acorrect response, whereas a button press at any other time was considered an error.Prior to averaging, we discarded trials with incorrect responses.
We assessed eye movement artifact in the following manner: for each trial, wecalculated the absolute value of the difference between each point in the epoch andthe mean of the 150-ms prestimulus baseline. If a specified number of thesedifference scores exceeded a preselected criterion value, the trial was rejected. Thecriterion value, determined in previous work (Duncan-Johnson and Donchin,1977), led consistently to the rejection of trials with eye movement artifact. TheEOG of the accepted trials was averaged to verify the success of the algorithm foreach corresponding average ERP.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 253
Measures of ERP components used in the statistical analyses were derived fromthe individual subject ERPs elicited by the 0.10 stimuli, averaged separately bymodality, using standard baseline-to-peak measurements. Prior to quantification,each average ERP was smoothed using a boxcar digital filter with a low-pass cutofffrequency of 12.4 Hz (−3 dB) (Glaser and Ruchkin, 1976).
The ERP components elicited by the 0.10 stimuli were N100, P200, N200, P300,and slow wave. Each component was identified using a computerized detectionroutine, and the accuracy of the detections was verified visually. Amplitudes werecalculated by subtracting the average voltage in the 150-ms prestimulus baselinefrom the peak voltage in specified latency ranges. The latency of each componentwas defined as the time from stimulus onset to the peak amplitude. The latencyrange for each component was chosen following visual inspection of the individualsubject waveforms. The latency windows for the auditory N100, N200, and P300components were 75–125, 175–275, and 275–575 ms, respectively. The latencywindows for visual N100, N200, and P300 were 100–200, 200–300, and 300–600ms, respectively. P200 was defined as the maximum positive peak between N100and N200 in each modality. Because the amplitude of N200 can be reduced byoverlap with P200, we quantified N200 in the 0.10 minus 0.90 difference waveforms(Ritter et al., 1983; Novak et al., 1990).
Auditory N100, P200, N200, and visual P200 and N200 were quantified at Cz,where they were maximal in amplitude; visual N100 was quantified at Oz. P300 wasmeasured at nine of the electrode sites (F3, Fz, F4, C3, Cz, C4, P3, Pz, and P4).Mean slow wave activity at each of the midline electrode sites (Fz, Cz, Pz, and Oz)was calculated by summing the voltages from 505 to 700 ms for auditory ERPs andfrom 555 to 750 ms for visual ERPs, and dividing by the number of points sampled(i.e., 40).
2.5. Data analysis
We compared measures of ERPs (N100, P200, N200 and slow wave) andperformance between pairs of seasons (winter–spring, spring–summer, and sum-mer–winter) using repeated measures analyses, with one between-subjects variable(sex) and one within-subject variable (season). Because not all ERPs had N100,P200, and N200 components, we used an unbalanced repeated measures test(BMDP5V), which does not exclude from the analysis any observations withmissing data. Instead, this test replaces missing values with estimates derived froma linear interpolation of the available data, and uses a Wald-type x2 statistic to testthe null hypothesis for each term. We also investigated the effects of these variableson P300 using repeated measures analyses, with sex the between-subjects variable,and season, laterality (left, midline, right), and anterior/posterior electrode site(frontal, central, parietal) as the within-subject factors. Because the Wald statistic isliberal with sample sizes smaller than 20, we used PB0.01 as the criterion forsignificance in our comparisons across the seasons. Significant main effects wereinvestigated further with Bonferroni multiple comparison tests. Interactions of sexwith other factors were examined in Wald analyses for each sex separately.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268254
We conducted separate analyses of the auditory and visual data, because recentinvestigations have suggested modality-specific P300 activity (Barrett et al., 1987;Duncan et al., 1987, 1990; Johnson, 1989a,b; Johnson et al., 1991; Johnson, 1993).In view of data that suggest the functional independence of the frontal negative andparietal positive slow wave (Friedman, 1984; Friedman et al., 1984; Ruchkin et al.,1990), we analyzed slow wave amplitude at each of the midline electrode sitesseparately. Spearman correlations were used to assess the relationship of season,amount of sunlight, and photoperiod to P300 and slow wave. Data collected at Pzwere used in the correlational analyses because both components were maximal inamplitude at that electrode site.
3. Results
The most salient result is that P300 and slow wave varied according to season,and that, overall, more statistically significant changes in P300 were observed inwomen than in men. Auditory and visual P300 and visual slow wave tended to bemaximal in winter or spring. In contrast, task performance was stable acrossseasons. The scalp topography of P300 did not vary over seasons, but did differbetween men and women.
3.1. Amount of sunlight
We obtained estimates of the amount of sunlight per day from the localclimatological data monthly summaries (National Oceanic and Atmospheric Ad-ministration, 1984–1994); this measure is reported as the number of minutes ofsunlight between sunrise and sunset minus the number of minutes of cloud cover.We based our calculations of the amount of sunlight on the day preceding thetesting session because subjects were tested at different times of the day. An analysisof variance of sunlight on the day preceding each testing session yielded asignificant main effect of season (F(2,61)=7.5, PB0.005). Pairwise contrastsindicated that winter testing dates had significantly fewer minutes of sunlight thaneither spring or summer testing dates, while the latter two did not differ from oneanother.
3.2. ERP data
The effects of season on auditory and visual ERPs, for our female and malesubjects, are presented in Figs. 1–3. In these figures, the grand-mean ERPs for eachpair of seasons are superimposed at ten electrode sites. Analyses of N100, P200, andN200 amplitude and latency yielded minimal or no statistically significant differ-ences between seasons, despite some apparent differences in the grand-meanwaveforms. Given the large number of analyses and the lack of a coherent patternto these results, we have not presented them in this paper.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 255
Fig
.1.
Gra
nd-m
ean
ER
Pw
avef
orm
sel
icit
edby
the
0.10
audi
tory
(a)
and
visu
al(b
)st
imul
iav
erag
edov
erfe
mal
ean
dm
ale
subj
ects
test
edin
the
win
ter
and
spri
ng.
The
data
colle
cted
inth
ew
inte
r(s
olid
lines
)an
dth
esp
ring
(dot
ted
lines
)ar
esu
peri
mpo
sed
at10
elec
trod
esi
tes
and
EO
G.
Alt
houg
hF
ig.
1an
dsu
bseq
uent
figur
espr
esen
tth
eE
RP
sat
Oz,
this
elec
trod
esi
tew
asno
tin
clud
edin
the
stat
isti
cal
anal
yses
ofP
300.
An
1100
-ms
epoc
his
show
nfo
rea
chw
avef
orm
.St
imul
uson
set
isin
dica
ted
byan
‘S’
onth
eti
me
scal
e.P
osit
ivit
yof
the
scal
pel
ectr
ode
wit
hre
spec
tto
the
refe
renc
eel
ectr
odes
issh
own
asa
dow
nwar
dde
flect
ion
inth
isan
dsu
bseq
uent
figur
es.
(a)
Nei
ther
P30
0no
rsl
oww
ave
elic
ited
byth
eau
dito
ryst
imul
usdi
ffer
edbe
twee
nw
inte
ran
dsp
ring
.(b
)V
isua
lP
300
was
larg
erin
win
ter
than
insp
ring
for
the
fem
ale
subj
ects
but
did
not
diff
erbe
twee
nse
ason
sfo
rth
em
ale
subj
ects
.V
isua
lsl
oww
ave
was
larg
erin
win
ter
than
spri
ngfo
rth
een
tire
grou
pof
subj
ects
.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 257
Fig
.2.
Gra
nd-m
ean
ER
Pw
avef
orm
sel
icit
edby
the
0.10
audi
tory
(a)
and
visu
al(b
)st
imul
iav
erag
edov
erfe
mal
ean
dm
ale
subj
ects
test
edin
the
spri
ngan
dsu
mm
er.
The
data
colle
cted
inth
esp
ring
(sol
idlin
es)
and
the
sum
mer
(dot
ted
lines
)ar
esu
peri
mpo
sed
at10
elec
trod
esi
tes
and
EO
G.
(a)
Aud
itor
yP
300
ampl
itud
ew
asla
rger
and
its
late
ncy
shor
ter
insp
ring
than
insu
mm
erfo
rw
omen
;w
here
asfo
rm
en,
P30
0w
assm
alle
rin
spri
ngth
anin
sum
mer
.Sl
oww
ave
did
not
diff
erbe
twee
nth
ese
ason
s.(b
)V
isua
lP
300
was
larg
erin
spri
ngth
anin
sum
mer
acro
ssth
een
tire
grou
pof
subj
ects
,an
dit
sla
tenc
yw
aslo
nger
insp
ring
than
insu
mm
erfo
rm
enbu
tno
tw
omen
.T
here
wer
eno
seas
onal
vari
atio
nsin
slow
wav
e.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 259
Fig
.3.
Gra
nd-m
ean
ER
Pw
avef
orm
sel
icit
edby
the
0.10
audi
tory
(a)
and
visu
al(b
)st
imul
iav
erag
edov
erfe
mal
ean
dm
ale
subj
ects
test
edin
the
sum
mer
and
win
ter.
The
data
colle
cted
inth
esu
mm
er(s
olid
lines
)an
dth
ew
inte
r(d
otte
dlin
es)
are
supe
rim
pose
dat
10el
ectr
ode
site
san
dE
OG
.(a
)A
udit
ory
P30
0w
asla
rger
inw
inte
rth
anin
sum
mer
,w
here
assl
oww
ave
did
not
diff
erbe
twee
nth
ese
ason
s.(b
)T
hela
tenc
yof
visu
alP
300
was
long
erin
sum
mer
than
inw
inte
rfo
rw
omen
,bu
tdi
dno
tdi
ffer
betw
een
the
seas
ons
for
men
.T
here
wer
eno
seas
onal
vari
atio
nsin
the
ampl
itud
eof
visu
alP
300
orsl
oww
ave.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 261
3.2.1. Winter–spring comparisons
3.2.1.1. Effects of season and sex on P300. Fig. 1 presents the ERPs in winter andspring for women and men. We found no effect of season or sex on the amplitudeor latency of auditory P300 (Fig. 1a). For visual P300, however, there was asignificant Season×Sex interaction (x2(1)=10.6, PB0.001) (Fig. 1b). Post hoctests indicated that visual P300 amplitude was larger in winter than in spring forwomen, but did not differ between the seasons in men.
3.2.1.2. Scalp distribution of P300. Although the topographic distribution of audi-tory P300 did not vary across the seasons, it did differ between women and men(Sex×Anterior/posterior electrode site interaction: x2(2)=10.7, PB0.005). Posthoc comparisons revealed that the amplitude of auditory P300 in men was maximalat centro-parietal locations; whereas in women, it did not vary significantly alongthe anterior/posterior axis. Sex differences in the scalp distribution of auditory P300decreased from frontal to parietal electrode sites. In contrast, the scalp distributionof visual P300, which was maximal at parietal and minimal at frontal electrode sites(anterior/posterior electrode site main effect: x2(2)=103.1, PB0.0001), did notdiffer between men and women.
3.2.1.3. Effects of season and sex on slow wa6e. There were no seasonal or sexdifferences in auditory slow wave at any of the midline electrode sites. Visual slowwave, however, was larger in amplitude in winter than in spring at Cz (x2(1) =9.3,PB0.005), Pz (x2(1) =8.1, PB0.005), and Oz (x2(1) =7.8, PB0.01), but did notdiffer between the seasons at Fz. We found no effects involving sex.
3.2.2. Spring–summer comparisons
3.2.2.1. Effects of season and sex on P300. Fig. 2 presents the spring and summerERPs for women and men. Auditory P300 amplitude showed a significant Seasonx Sex interaction (x2(2) =10.3, PB0.001) (Fig. 2a). Post hoc tests indicated thatP300 was larger in the spring than in the summer for the female subjects, whereasthe reverse pattern occurred for the male subjects. The latency of auditory P300also showed a Season x Sex interaction (x2(2) =27.9, PB0.0001). Post hoc testsindicated that P300 was earlier in spring than in summer in women, but did notdiffer between the seasons in men. Season also had a significant effect on theamplitude of visual P300; it was larger in spring than in summer (x2(1) =6.9,PB0.01) (Fig. 2b). We found a Season×Sex interaction on visual P300 latency(x2(1) =21.2, PB0.0001). Post hoc tests confirmed that in men, but not women,visual P300 occurred later in the spring than in the summer.
3.2.2.2. Scalp distribution of P300. Analyses of auditory P300 revealed no interac-tions of season with laterality or anterior/posterior electrode site. The amplitude ofauditory P300 along the anterior/posterior axis differed between men and women(Sex×Anterior/posterior interaction: x2(2) =16.1, PB0.0001). Sex differences in
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268262
auditory P300 increased from frontal to parietal electrodes. As indicated by posthoc tests, in women, P300 was maximal in amplitude at parietal sites and minimalat frontal sites; whereas, in men, it did not differ significantly along the anterior/posterior axis.
3.2.2.3. Effects of season and sex on slow wa6e. No differences between seasons orbetween men and women were observed in auditory or visual slow wave in thisgroup of subjects.
3.2.3. Summer–winter comparisons
3.2.3.1. Effects of season and sex on P300. The ERPs recorded from women andmen in summer and in winter are displayed in Fig. 3. Auditory P300 was larger inwinter than in summer (x2(1) =8.0, PB0.005), but no differences were foundbetween women and men (Fig. 3a). Although there was no season effect on theamplitude of visual P300, there was a significant Season×Sex interaction on itslatency (x2(1) =7.3, PB0.01) (Fig. 3b). Post hoc tests indicated that P300 wasearlier in winter than in summer for women, with no seasonal difference for men.
3.2.3.2. Scalp distribution of P300. There were no seasonal variations in thetopography of either auditory or visual P300 and no sex differences in thetopography of auditory P300. Visual P300, however, showed an interaction of sexand anterior/posterior electrode site (x2(2) =20.5, PB0.0001). Post hoc testsconfirmed a parietal maximum in women but a central maximum in men.
3.2.3.3. Effects of season and sex on slow wa6e. In the summer–winter group,neither season nor sex affected slow wave in either modality, at any of the midlineelectrode sites.
3.3. Performance data
In the winter–spring group, women’s (mean=91% (S.D.=7)) performance onthe visual task was significantly more accurate than men’s (mean=74% (S.D.=14)) (x2(1)=9.9, PB0.005). There were no other seasonal or sex differences in thespeed or accuracy of performance.
3.4. Relationship of season, amount of sunlight, and photoperiod to ERPcomponents
We did not find any significant correlations between season, amount of sunlight,or photoperiod (amount of time between dawn and dusk) on the day prior totesting and measures of P300 or slow wave. We also correlated P300 with theaverage amount of sunlight during the 30 days prior to the testing session andfound no statistically significant relationships.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 263
4. Discussion
The present study provides additional evidence, albeit modest, that P300 and slowwave vary with season. Specifically, visual P300 was larger in winter than in springfor female subjects only, and it was larger in spring than in summer for both sexes.Auditory P300 was larger in spring than in summer for women, but the reverse wastrue for men. Auditory P300 was also larger in winter than in summer for men andwomen. Moreover, visual slow wave was larger in winter than in spring. Consistentwith the pattern reported by Deldin et al. (1989a,b, 1994), we found somewhatgreater seasonal variation in P300 in women than in men. Our data also indicatedsex differences in the scalp distribution of P300. These findings are generallyconsistent with previous studies reporting a larger P300 at Pz relative to Fz and Czin women than in men (Polich, 1986; Polich et al., 1988; Cahill and Polich, 1992).The absence of detectable seasonal variations in performance in the presence ofseasonal variations in P300 and slow wave suggests that some central mechanism,reflected in changes in P300 and slow wave, may compensate for seasonal fluctua-tions in sunlight, thus tending to ensure constant behavioral capacities. Alterna-tively, our tasks may not have been sufficiently challenging to detect subtledifferences in behavior over the seasons. Overall, however, the evidence that P300varies systematically as a function of season was rather weak. Therefore, our resultsdo not provide strong confirmation of previous findings.
Perhaps the most obvious variable related to season that may influence ERPs isthe amount of light exposure. In contrast to previous reports (Deldin et al., 1989a,b,1994; Polich and Geisler, 1991), however, we failed to find a correlation betweenamount of sunlight on the day (or month) preceding testing or photoperiod andmeasures of P300 or slow wave. This lack of a correlation between ERPs andsunlight and photoperiod, as well as the winter or spring maxima observed in bothauditory and visual P300 and visual slow wave, suggests that factors other thanphotoperiod may account for the observed seasonal variations in amplitude.
Also of interest is the finding that seasonal variations were more evident inwomen than in men. In order to explain the interaction of season and sex on P300,we considered the role of seasonal fluctuations in sex hormone levels. A number ofrecent studies have reported the effects of these hormones on cognitive capacities(Komnenich et al., 1978; Hampson and Kimura, 1988, 1992; Hampson, 1990a,b;Gouchie and Kimura, 1991; Kimura and Touissant, 1991; Kimura and Hampson,1994). Such studies have reported variations in women’s performance over themenstrual cycle on tests of spatial and verbal abilities (Hampson and Kimura, 1988,1992; Hampson, 1990a,b; Gouchie and Kimura, 1991; Kimura and Hampson,1994), as well as on tests requiring simple repetitive movement vs. inhibition of aresponse (Komnenich et al., 1978) [for a review, see Duncan and Gabbay, 1996].These variations in performance have been related to fluctuations in estrogen andprogesterone levels. In addition, seasonal variations in performance on cognitivetests in men have been found to vary with testosterone levels across seasons.Investigations of the circannual rhythm of testosterone in men have reported a
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268264
consistent peak of testosterone level in autumn (from July to December, peaking inOctober), with lowest levels recorded in spring (Reinberg and Lagoguey, 1978;Smals et al., 1976; Kimura and Hampson, 1994).
Because we did not measure hormone levels in our subjects, or gather informa-tion regarding the menstrual phase of our female subjects, we were unable to assesshormonal effects on our ERP measures. Whereas some investigators have reportedno significant effects of menstrual phase on ERPs elicited in an auditory oddballtask (Fleck and Polich, 1988), others have found increased P300 around the time ofovulation (high progesterone levels) when elicited by ‘emotional’ stimuli (i.e. nudemen, babies) (Johnston and Wang, 1991; Wang and Johnston, 1993). If we hadused ‘emotional’ stimuli or tasks in which male or female subjects tend to excel, thelikelihood of detecting significant effects of both season and sex in our group ofsubjects might have been greater. For example, men tend to score higher thanwomen on mental rotation of three-dimensional objects in space, whereas womenoutperform men on fine perceptual discriminations (Duncan and Gabbay, 1996).
Gonadal hormone fluctuations, however, may not account for our ERP findings.We found seasonal variations in P300 primarily in our female subjects, whereas thehormonal and cognitive variability reported previously in women was related tomonthly estrogen and progesterone levels. Seasonal fluctuations in cognitive perfor-mance in women have not been investigated.
Perhaps the seasonal variations in P300 observed in women are not mediated bygonadal hormones, but by other hormones or biological processes with a circadianand/or circannual rhythm. In a review of the literature on seasonal variations in abroad range of physiological functions, Lacoste and Wirz-Justice (1989) reported awinter maximum in most of these functions. They summarized studies that foundwinter peaks in endocrinological (thyroid function: basal thyroid stimulating hor-mone, thyroid releasing hormone response, T3 and T4), metabolic (glucose,glucagon, and basal metabolic rate), physiological (blood pressure, slow wave sleepand sleep time, and a maximum circadian phase delay of most functions), andepidemiological (onset of diabetes mellitus, and cardiovascular morbidity andmortality) variables. Moreover, some hormones exhibited a bimodal pattern, witha peak in winter and a smaller peak in summer (melatonin and cortisol) (Wirz-Jus-tice and Richter, 1979; Birau et al., 1981; Martikainen et al., 1985). The only seasonby sex interaction reported in their review of the literature was an inconsistentseasonal pattern of variation in prolactin levels in women but not in men. Theauthors suggested that these winter elevations constitute an adaptive strategy to theenvironmental challenges of harsh weather conditions (e.g., decreases in ambienttemperature and sunlight, changes in photoperiod).
The bimodal enhancement of P300 (summer and winter) reported by Deldin et al.(1994) could reflect the organism’s response to conditions of extreme photoperiodand temperature. Thus, whereas our results as well as those reported by Deldin etal. (1994) fail to support the hypothesis that sunlight is the determing factor in theseasonal variation of P300, they both suggest that other factors related to seasonmay be mediating these changes in P300. It appears that photoperiod and/orgeneral environmental conditions may affect ERPs. Our study and that of Deldin
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 265
et al. (1994) were conducted in the same laboratory and used the same methodol-ogy. In contrast, the Polich and Geisler (1991) study was conducted in a moresouthern latitude, and their methodology and definition of seasons differed fromours and that of Deldin et al. (1994). When we redefined their seasons inaccordance with our definitions, their data showed increased P300 in summer ascompared with other seasons, thus supporting the original hypothesis that seasonalvariation in P300 could be attributed to the amount of sunlight. The inconsistenciesin the findings of the three studies, however, suggest that there is not a simplerelationship between amount of sunlight and P300. Thus, it appears that in additionto sex, there are other variables affecting the relationship between season and P300that remain to be identified.
In the absence of a direct correlation of P300 and sunlight, photoperiod, oranother biological process, the observed seasonal fluctuations in P300 could berelated to changes in the level of physiological arousal associated with changes inseason. In a recent review of studies investigating cognitive and biological determi-nants of P300, Polich and Kok (1995) concluded that increased physiologicalarousal was associated with increased P300 amplitude and decreased P300 latency.Factors that affect physiological arousal include food intake (Geisler and Polich,1992), elevated body temperature (Geisler and Polich, 1990), ultradian cycles (Linand Polich, 1995), and sleep (Smulders, 1993). It is of interest, however, that theyreported no menstrual effects on P300 (Fleck and Polich, 1988). These studiessuggest that some biological variables may affect P300. Our finding of seasonalvariation in P300 suggests that the central mechanisms supporting informationprocessing are influenced by environmental factors. This result is consonant withthose seen in other physiological systems. Our data suggest, however, that men andwomen respond differently to environmental challenge.
Based on the Polich and Kok (1995) review, we also considered the possibilitythat ambient temperature may have influenced P300. We therefore examined theeffect of temperature (on the day of testing) across seasons. Outdoor temperaturesin winter and spring did not differ, and both were lower than temperatures insummer. As mentioned previously, we found increased P300 in spring as well aswinter. Temperature, however, was not correlated with the amplitude of eitherauditory or visual P300.
Several methodological factors may limit the scope of our findings. Perhaps themost obvious limitation was that we were able to collect enough data to compareeach subject in only two seasons. Due to a small sample of subjects tested in thefall, we had to exclude those data from the analyses. This is of concern becausetestosterone levels in men peak in the fall (Smals et al., 1976; Reinberg andLagoguey, 1978; Kimura and Hampson, 1994). Also, the number of subjects testedin each pair of seasons was modest. Finally, our suggestion that the pattern of ourresults may reflect a fundamental, autonomous built-in compensatory mechanismto ensure constancy of behavioral competence in non-hibernating species is clearlya post hoc explanation, which needs to be tested directly.
Despite these methodological limitations, the current study, which has the mostpowerful design, is the third to suggest seasonal effects on P300. Because the
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268266
previous studies used cross-sectional designs, their results could be attributed toindividual differences. Our study measured ERPs in two different seasons in eachindividual, and, thus, is more likely to reflect a genuine season effect. Thus, thetrends we found for sex differences in the seasonal variation of P300 warrantfurther investigation. Future studies could benefit from repeated testing of a largegroup of subjects in all four seasons. The use of tasks that have been shown toreflect sex differences in cognitive processing may maximize the possibility ofdetecting such differences in ERPs. Moreover, our finding of a sex difference in theseasonal variation of P300 may have implications for understanding the differencesin vulnerability to seasonal affective disorder between men and women, as well asthe physiological mechanisms involved in light therapy.
Acknowledgments
Preliminary reports of this study were presented at the 32nd annual meeting ofthe Society for Psychophysiological Research, San Diego (1992) and the 4thNational Conference of Psychological Research of the Hellenic PsychologicalSociety, Thessaloniki, Greece (1993). We wish to thank Lisa Slade, Ph.D., forconducting the psychiatric interviews, and John Ingeholm, Ph.D., and BryanFantie, Ph.D., for their technical assistance. We would also like to thank Sandra M.Wilkniss for her assistance in the data collection process, and Thalene T. Mallus,M.A., John Bartko, Ph.D., and Karen Pettigrew, Ph.D., for their consultation onthe statistical analysis. This research was supported by the Intramural ResearchProgram of the National Institute of Mental Health.
References
American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th ed.Washington, DC (Author).
Barrett, G., Neshige, R., Shibasaki, H., 1987. Human auditory and somatosensory event-relatedpotentials: Effects of response condition and age. Electroencephal. Clin. Neurophysio. 66, 409–419.
Birau, N., Birau, M., Schloot, W., 1981. Melatonin rhythms in human serum. In: Birau, N., Schloot, W.(Eds.), Melatonin: Current Status and Perspectives. Pergamon Press, Oxford, pp. 287-295.
Cahill, J.M., Polich, J., 1992. P300, probability, and introverted/extroverted personality types. Biol.Psychol. 33, 23–35.
Deldin, P.J., Duncan, C.C., Miller, G.A., 1989a. Evidence for seasonal effects on P300. Paper presentedat the Ninth International Conference on Event-Related Potentials of the Brain (EPIC IX),Noordwijk, The Netherlands.
Deldin, P.J., Duncan, C.C., Miller, G.A. 1989b. Light, gender, and P300. Psychophysiology (Suppl.) 26,20.
Deldin, P.J., Duncan, C.C., Miller, G.A., 1994. Season, gender, and P300. Biol. Psychol. 39, 15–28.Duncan, C.C., Gabbay, F.H., 1996. Sex differences in cognitive function: Implications for military
assignments. In: Gabbay, F.H., Ursano, R.J., Norwood, A.E., Fullerton, C.S., Sutton, L.K.,Duncan, C.C. (Eds.), Sex Differences, Stress and Military Readiness. Department of Psychiatry,Uniformed Services University of the Health Sciences, Bethesda, Maryland, pp. 61-86.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268 267
Duncan, C.C., Mirsky, A.F., Deldin, P.J., Skwerer, R.G., Jacobsen, F.M., Rosenthal, N.E., 1990. P300as an index of treatment response in seasonal affective disorder. In: Stefanis, C.N. et al. (Eds.),Psychiatry: A World Perspective, vol. 2. Elsevier Science Publishers B.V. (Biomedical Division),Amsterdam, pp. 398-401.
Duncan, C.C., Mirsky, A.F., Deldin, P.J., Skwerer, R.G., Jacobsen, F.M., Rosenthal, N.E., 1991. Brainpotentials index treatment response in seasonal affective disorder. In: Ansseau, M., von Frenckell,R., Franck, G. (Eds.), Biological Markers of Depression: State of the Art. Elsevier Science PublishersB.V, Amsterdam, pp. 117-120.
Duncan, C.C., Morihisa, J.M., Fawcett, R.W., Kirch, D.G., 1987. P300 in schizophrenia: State or traitmarker? Psychopharmacol. Bull. 23, 497–501.
Duncan-Johnson, C.C., Donchin, E., 1977. On quantifying surprise: The variation of event-relatedpotentials with subjective probability. Psychophysiology 14, 456–467.
Duncan-Johnson, C.C., Donchin, E., 1979. The time constant in P300 recording. Psychophysiology 16,53–55.
Fleck, K.M., Polich, J., 1988. P300 and the menstrual cycle. Electroencephal. Clin. Neurophysiol. 71,157–160.
Friedman, D., 1984. P300 and slow wave: The effects of reaction time quartile. Biol. Psychol. 18, 49–71.Friedman, D., Brown, C., Vaughan, H.G. Jr., Cornblatt, B., Erlenmeyer-Kimling, L., 1984. Cognitive
brain potential components in adolescents. Psychophysiology 21, 83–96.Geisler, M.W., Polich, J., 1990. P300 and time-of-day: Circadian rhythms, food intake, and body
temperature. Biol. Psychol. 31, 117–136.Geisler, M.W., Polich, J., 1992. P300, food consumption, and memory performance. Psychophysiology
29, 76–85.Glaser, E.M., Ruchkin, D.S., 1976. Principles of Neurobiological Signal Analysis. Academic, New York.Gouchie, C., Kimura, D., 1991. The relationship between testosterone levels and cognitive ability
patterns. Psychoneuroendocrinology 16, 323–334.Hampson, E., 1990a. Estrogen-related variations in human spatial and articulatory-motor skills.
Psychoneuroendocrinology 15, 97-111.Hampson, E., 1990b. Variations in sex-related cognitive abilities across the menstrual cycle. Brain
Cognit. 14, 26-43.Hampson, E., Kimura, D., 1988. Reciprocal effects of hormonal fluctuations on human motor and
perceptuo-spatial skills. Behav. Neurosci. 102, 456–459.Hampson, E., Kimura, D., 1992. Sex differences and hormonal influences on cognitive function in
humans. In: Becker, J.B., Breedlove, S.M., Crews, D. (Eds.), Behavioral Endocrinology. MIT Press,Cambridge, MA, pp. 357-398.
Johnson, R. Jr., 1989a. Auditory and visual P300s in temporal lobectomy patients: Evidence formodality-dependent generators. Psychophysiology 26, 633-650.
Johnson, R. Jr., 1989b. Developmental evidence for modality-dependent P300 generators: A normativestudy. Psychophysiol. 26, 651-667.
Johnson, R. Jr., Miltner, W., Braun, C., 1991. Auditory and somatosensory event-related potentials: I.Effects of attention. J. Psychophysiol. 5, 11–25.
Johnson, R. Jr., 1993. On neural generators of the P300 component of the event-related potential.Psychophysiology 30, 90–97.
Johnston, V.S., Wang, X-T., 1991. The relationship between menstrual phase and the P3 component ofERPs. Psychophysiology 22, 400–409.
Kimura, D., Hampson, E., 1994. Cognitive pattern in men and women is influenced by fluctuations insex hormones. Curr. Directions Psychol. Sci. 3, 57–61.
Kimura, D., Touissant, C., 1991. Sex differences in cognitive function vary with the season. Soc.Neurosci. Abs. 17, 868.
Komnenich, P., Lane, D.M., Dickey, R.P., Stone, S.C., 1978. Gonadal hormones and cognitiveperformance. Physiol. Psychol. 6, 115–120.
Lacoste, V., Wirz-Justice, A., 1989. Seasonal variation in normal subjects: An update of variablescurrent in depression research. In: Rosenthal, N.E., Blehar, M. (Eds.), Seasonal Affective Disordersand Phototherapy. The Guilford Press, New York, pp. 87-104.
M.H. Kosmidis et al. / Biological Psychology 49 (1998) 249–268268
Lin, E., Polich, J., 1995. P300 and ultradian rhythms. Unpublished manuscript.Martikainen, H., Tapanainen, J., Vakkuri, O., Leppaluoto, J., Huhtaniemi, I., 1985. Circannual
concentrations of melatonin, gonadotrophins, prolactin, and gonadal steroids in males in a geo-graphical area with a large annual variation in daylight. Acta Endocrinologica 109, 446–450.
National Oceanic and Atmospheric Administration, 1984–1994. Local climatological summary. Wash-ington, DC.
Novak, G.P., Ritter, W., Vaughan, H.G. Jr., Wiznitzer, M.L., 1990. Differentiation of negativeevent-related potentials in an auditory discrimination task. Electroencephal. Clin. Neurophysiol. 75,255–275.
Polich, J., Kok, A., 1995. Cognitive and biological determinants of P300: An integrative review. Biol.Psychol. 41, 103–146.
Polich, J., 1986. Normal variation of P300 from auditory stimuli. Electroencephal. Clin. Neurophysiol.65, 236–240.
Polich, J., Burns, T., Bloom, F.E., 1988. P300 and the risk for alcoholism: Family history, task difficulty,and gender. Alcoholism: Clin. Exper. Res. 12, 248–254.
Polich, J., Geisler, M.W., 1991. P300 seasonal variation. Biol. Psychol. 32, 173–179.Reinberg, A., Lagoguey, M., 1978. Circadian and circannual rhythms in sexual activity and plasma
hormones (FSH, LH, testosterone) of five human males. Archives of Sexual Behav. 7, 13–30.Ritter, W., Simson, R., Vaughan, H.G. Jr., 1983. Event-related potential correlates of two stages of
information processing in physical and semantic discrimination tasks. Psychophysiology 20, 168–179.
Ruchkin, D.S., Johnson, R. Jr., Canoune, H.L., Ritter, W., Hammer, M., 1990. Multiple sources of P3bassociated with different types of information. Psychophysiology 27, 157–176.
Smals, A.G.H., Kloppenburg, P.W.C., Benraad, Th.J., 1976. Circannual cycle in plasma testosteronelevels in man. J. Clin. Endocrinol. Metab. 42, 979–982.
Smulders, F., 1993. The Selectivity of Age Effects on Information Processing. Doctoral Thesis,University of Amsterdam.
Spitzer, R.L., Endicott, J., 1978. Schedule for Affective Disorders and Schizophrenia, 3rd ed. BiometricsResearch Division, New York State Psychiatric Institute, New York.
Wang, X-T., Johnston, V.S., 1993. Changes in cognitive and emotional processing with reproductivestatus. Brain Behav. Evolution 42, 39–47.
Wirz-Justice, A., Richter, R., 1979. Seasonality in biochemical determinations: A source of variance anda clue to the temporal incidence of affective illness. Psych. Res. 1, 53–60.
.