sp
elm
in a
l Psy
iplinar
al En
response for which the lowermost value was found to be 25 window from mainly the frontocentral recording sites and is
time-locked to the stimulus.
The early, time-locked gamma response is mainly
rmation processing
International Journal of Psychophysiolo* Corresponding author. Hacettepe University, Department of Experi-1. Introduction
The study of oscillatory responses in brain’s neuro-
electric activity proved useful in understanding the way
brain processes information (Adrian, 1942; for review, see
Bas¸ar, 1976, 1980, 1998, 1999; Berger, 1929). Among the
various oscillatory responses, the gamma response of the
brain has received considerable attention. Existing literature
includes a fair amount of knowledge about the gamma
Hz and the uppermost value 70 Hz (for review, see Bas¸ar-
Erog˘lu et al., 1996; Galambos, 1992; Gurtubay et al., 2001;
Karakas¸ et al., 2001). There are basically two types of event-
related gamma responses and these are discriminated on the
basis of temporal localization and time-locking (Gurtubay et
al., 2001; Bas¸ar-Erog˘lu et al., 1996; Karakas¸ et al., 2001).
The ‘‘late’’ gamma occurs in the 130–400 ms poststimulus
time window and has an induced character. The ‘‘early’’
gamma response occurs within the 150 ms poststimulusThe aim was to investigate whether gender is a causative factor in the gamma status according to which some individuals respond with
time-locked, early gamma response, G+, while the others do not show this response, G. The sample consisted of 42 volunteer participants
(between 19 and 37 years of age with at least 9 years of education). There were 22 females and 20 males. Data were collected under the
oddball paradigm. Auditory stimulation (10 ms r/f time, 50 ms duration, 65 dB SPL) consisted of target (2000 Hz; p =.20) stimuli that
occurred randomly within a series of standard stimuli (1000 Hz; p =.80). Gamma responses were studied in the amplitude frequency
characteristics, in the digitally filtered event-related potentials (f-ERPs) and in the distributions which were obtained using the recently
developed time–frequency component analysis (TFCA) technique. Participants were classified into G+ and G groups with a criterion of
full agreement between the results of an automated gamma detection technique and expert opinion. The 222 ANOVA on f-ERPs and
222 multivariate ANOVA on TFCA distributions showed the main effect of gamma status and gender as significant, and the interaction
between gamma status and gender as nonsignificant. Accordingly, individual difference in gamma status is a reliable phenomenon, but this
does not depend on gender. There are conflicting findings in the literature concerning the effect of gender on ERP components (N100, P300).
The present study showed that if the gamma status is not included in research designs, it may produce a confounding effect on ERP
parameters.
D 2005 Elsevier B.V. All rights reserved.
Keywords: Gamma response; Gender; Individual differences; Event-related potentials; Event-related oscillations; Time–frequency analysisAbstractReceived 28 December 2004; received in revised form 1 May 2005; accepted 24 May 2005
Available online 16 August 2005Early time-locked gamma re
Sirel Karakas¸ a,b,*, I˙lhan Tu¨fekc¸i c, B
Elvin Dog˘utepe a,b, O¨. Utku Erzeng
aHacettepe University, Department of Experimenta
bTU¨BITAK Brain Dynamics Multidisc
cBilkent University, Department of Electric0167-8760/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijpsycho.2005.05.009
mental Psycholo
2978335; fax: +90 312 2992100.
E-mail address: skarakas@hacettepe.edu.tr (S. Karakas¸).onse and gender specificity
a Bekc¸i a,b, Emine D. C¸akmak a,b,
,b, Arzu O¨zkan a,b, Orhan ArNkan c
chology, Beytepe Campus 06532, Ankara, Turkey
y Research Network, Ankara, Turkey
gineering, 06533 Bilkent, Ankara, Turkey
gy 60 (2006) 225 – 239
www.elsevier.com/locate/ijpsychorelated to the earlier operations of infogy, Beytepe Campus 06532, Ankara, Turkey. Tel.: +90 312
that culminate in sensation and perception (Gurtubay et al.,
2001; Karakas¸ et al., 2001; Karakas¸ and Bas¸ar, 1998,
event-related activity of the brain, the event-related
potentials (ERPs). The N200 component (N2a/mismatch
al of Pnegativity and N2b) is related to varying degrees of
attention, from preattention/passive attention to focused
attention/active attention (Karakas¸ et al., 2000a,b; Naata-
nen, 1990, 1992; Naatanen and Picton, 1986). Some
studies reported that N200 is higher in amplitude in males
(Go¨lgeli et al., 1999; Nagy et al., 2003). However, other
studies failed to find such an effect, or conversely, found
that this ERP component was higher in amplitude in
females (Barrett and Fulfs, 1998; Hoffman and Polich,
1999; Kasai et al., 2002). In females, the amplitude of the
P300 (P3b) was found to be higher (Orozco and Ehlers,
1998; Osterhout et al., 1997; Hoffman and Polich, 1999).
Such findings suggested that there was a gender effect on
attention allocation and memory updating processes which
are related to P300 (Sutton et al., 1965; Polich and Kok,
1995).
The ERP studies suggest gender as a causative factor for
the presence of the gamma response. To our knowledge, this
issue has not been studied in the relevant literature. The aim
of the present study was to study the effect of gender on the
gamma response and, using techniques of time-domain,
frequency-domain and time–frequency domain analysis, to
test the hypothesis that individual differences in this
response are due to gender.
2. Methods
2.1. Participants
The sample was obtained from a total of 58 volunteers
who met the below specified inclusion and exclusion
criteria. These participants were classified into G+ and
G groups. The sample consisted of 42 volunteer partici-
pants with respect to whom the automated gamma detection
technique and expert opinion showed 100% agreement with
regards to gamma status (G+ or G).
There were 20 males and 22 females in the sample.
Participants were young adults between 19 and 37 years of2004). However, in spite of its early occurrence, individual
differences were reported in this time-locked gamma
response such that some individuals showed time-locked
gamma response, G+, and others did not, G. Further-
more, these differences were found to be related to the
higher cognitive processes such as learning, memory and
executive functions (Jokeit and Makeig, 1994; Karakas¸ et
al., 2003).
Cognitive processes are differentially affected with
gender such that there is generally a female superiority
in verbal functions and a male superiority in visuo-spatial
functions (Kolb and Whishaw, 1996; Maccoby and Jacklin,
1974). Gender effect has also been studied on one type of
S. Karakas¸ et al. / International Journ226age and they had at least 9 years of education. Participants
were naive to electrophysiological studies. All were right-handed. Volunteering participants gave informed consent to
participate in the study after the purpose and nature of the
experiment were fully explained.
The sample included only those individuals who
reported being free of neurological or psychiatric prob-
lems. Individuals who were, at the time of testing, under
medication that could affect cognitive processes or who
stopped taking such medication were excluded. The
hearing level of the potential subjects was assessed through
computerized audiometric testing prior to the experimental
procedures. Individuals with hearing deficits were not
included in the study.
2.2. Stimulation and response parameters
The auditory stimuli had 10 ms r/f time, 50 ms
duration and were presented over the headphones at 65
dB SPL. Two types of stimuli were used: the standard
and the deviant. The standard stimuli (n=120–130) were
1000 Hz. Deviant stimuli (n=30–33) were 2000 Hz. The
deviant stimuli occurred randomly with a probability of
about 0.20 within a series of standard stimuli that were
presented with a probability of about 0.80. According to
the procedures of the oddball paradigm (OB), participants
had to mentally count the occurrence of deviant (target)
stimuli and to report them after the session was
terminated (for details of the methodology, see Karakas¸
et al., 2000a).
2.3. Electrophysiological procedures
Electrical activity of the brain, the prestimulus
electroencephalogram (EEG) and the poststimulus ERP,
were recorded in an electrically shielded, sound-proof
chamber. Recordings were taken from 15 recording sites
(reference: linked earlobes; ground: forehead) of the 10–
20 system under eyes-open condition using a commercial
electrode cap (Electro-Cap) of appropriate size. The
present study reports findings from the midline recording
sites (Fz, Cz, Pz).
Bipolar recordings were made of electroocular and
electromyographic activity for online rejection of responses
whose amplitudes exceeded T50 AV and offline rejection of
artifacts by means of visual inspection. Electrooculogram
(EOG) was recorded between the outer canthus and supra-
orbital area of the left eye. Electromyogram (EMG) was
recorded between two electrodes that were placed at the
submental area. Rejection occurred for epochs that con-
tained gross muscular activity, eye movements or blinks.
Electrical activity was amplified and filtered with a
bandpass between 0.16 and 70 Hz (3 dB down, 12 dB/
octave). It was recorded with a sampling rate of 500 Hz and
a total recording time of 2048 ms, 1024 ms of which served
as the prestimulus baseline. EEG–ERP data acquisition,
sychophysiology 60 (2006) 225–239analysis, and storage were achieved by a commercial system
(Brain Data 2.92).
al of P2.4. Data analysis
2.4.1. Time-domain analysis
Selective averaging of ERPs was used for obtaining
representative waveforms for the experimental conditions
and also for separating the ERP (evoked activity) from the
EEG (spontaneous activity) (Dawson, 1954).
2.4.2. Frequency-domain analysis
The frequency responses of brain’s neuroelectricity
were calculated through two techniques: transient response
frequency characteristics (TRFC) and digital filtering
(DF).
The amplitude frequency characteristics (AFCs) were
computed using the TRFC method (Bas¸ar, 1980). This
method computes the AFCs, )G(jx)), by the application
of Laplace transform (i.e., one-sided Fourier transform) to
the unfiltered transient (evoked) response, c(t), of the
system (Bas¸ar, 1998; Brandt and Jansen, 1991; Jervis et
al., 1983; Kolev and Yordanova, 1997; Parvin et al.,
1980; Ro¨schke et al., 1995, 1996; Ro¨schke and Alden-
hoff, 1991; Solodovnikov, 1960). The AFC is expressed
in relative units and it reflects the amplification in the
studied frequency channels. The presence of peaks in the
AFC thus reveals the frequency selectivities of the system
and these are interpreted as the most preferred oscil-
lations when responding to stimuli (for details of
methodology, see Karakas¸ and Bas¸ar, 1998). The AFCs
were used in the present study for a global and
simultaneous description of selectivities in the studied
frequency range (1–100 Hz). The technique was also
used for determining the limits of the resonant selectiv-
ities that appeared as maxima in the AFCs. These values
were used for determining the frequency cutoff values of
the digital filters.
In DF, the experimentally obtained transient (evoked)
response, c(t), is theoretically filtered by means of the
convolution integral using the weighting function, gKF(t),
of adequately determined ideal filter (for details of
methodology, see Cook and Miller, 1992; Farwell et al.,
1993; Karakas¸ and Bas¸ar, 1998; Ungan and Bas¸ar, 1976).
Digital filtering produces visual displays of the time
courses of oscillatory components within the frequency
limits of the utilized filters. Digital filtering thus displays
the oscillatory activity of different frequency bands over
the studied time interval (1024 ms to +1024 ms). In the
present study, digital passband filtering was response-
adaptive; filter limits were determined from selectivity
channels that were displayed in the AFCs as distinct peaks.
In other words, the bandwidth values of the digital filters
that were chosen coincided with the limits of the resonant
selectivities that were displayed in the AFC of each
condition. It was found that 28–46 Hz frequency limits
described the gamma range in the different experimental
S. Karakas¸ et al. / International Journconditions of the present study (also see Karakas¸ et al.,
2001; Karakas¸ and Bas¸ar, 1998).2.4.3. Time–frequency domain analysis
As the above explanations show, ERPs basically repre-
sent the analysis of brain electrical responses in the time-
domain. Fourier transform and the AFCs represent the
frequency components in the signal in the frequency
domain. Digital filtering is based on the findings of AFC
and displays the progression of oscillations of those
frequency ranges that are determined from the AFCs. Both
techniques of frequency analysis assume that the studied
system is linear and stationary.
The localization of signals simultaneously in the time and
frequency planes can be achieved using Short-Time Fourier
Transform (STFT). However, when ERP components are
closely localized in the time–frequency plane, STFT may
not be adequate to resolve them (Cohen, 1989, 1995).
Wigner distribution significantly improves the resolution of
the individual ERP components. The Wigner distribution
Wx(t, f ) of a signal x(t) is defined as
Wx t; fð Þ ¼
Z V
V
x t þ s=2ð Þx* t  s=2ð Þej2pf sds:
As it is seen from the definition, Wigner distribution is a
bilinear representation. Therefore, Wigner distribution of
multi-component signals or mono-component signals with
curved time–frequency supports will be cluttered by
spurious terms called cross-terms. The existence of cross-
terms may decrease the interpretability of the time–
frequency distribution. ERPs, like other non-stationary
signals, require an analysis technique that is free from
cross-terms and that also render a high-resolution time–
frequency distribution.
The simultaneous localization of the gamma component
in the time and frequency domains was achieved in the
present study using the recently developed time–frequency
component analysis (TFCA) technique (ArNkan et al., 2003;
O¨zdemir et al., 2001, 2005; O¨zdemir and ArNkan, 2000,
2001). Fig. 1 presents a flow diagram of the algorithm that
was used in TFCA. The basic aim of this algorithm was as
follows: Given a multi-component sampled signal x(n/Dx),
N/2nN/21, to extract its components and to
compute its time–frequency distribution. It is assumed that
x(t) is scaled before its sampling so that its Wigner
distribution is inside a circle of a diameter DxV
ffiffiffiffi
N
p
(see
Ozaktas et al., 1996).
The steps of the algorithm for TFCA are given below.
1. Initialize the residual signal and the iteration number
as r0(t)2x(t), i21, respectively.
2. Identify the time–frequency support of the component
si(t) using the watershed segmentation algorithm (Vincent
and Soille, 1991). After manually determining the appro-
priate rotation angle /i and the fractional domain ai=2/i/p,
estimate the spine wi,ai(t) of the fractional Fourier transform
rai
i1(t) using an instantaneous frequency estimation algo-
sychophysiology 60 (2006) 225–239 227rithm. Then determine the amount of the required frequency
shift dfi on the spine wi,ai(t).
r) =
t of the
d segm
riate r
          
tional 
ii a,ψ
al of PInitialize: 
(x t
Identify the t-f suppor
using the watershe
Identify the approp
and compute the spine
in the frac
S. Karakas¸ et al. / International Journ2283. Compute the sampled FrFT rai
i1(kT), ai =2/i/p, from
ri1(kT) using the fast fractional Fourier transform algo-
rithm (see Ozaktas et al., 1996).
4. Define the warping function fi(t)=Ci
1( fw i(t t1)),
where Ci tð Þ ¼ X
t
t1
wai tVð Þ þ dfi

dtV and fw i =Ci(tN)/(tN t1).
Compute the sampled warping function fi(kT).
5. Compute the sampled warped signal rai,f i
i1 (kT) as
r
i1;dfi
ai kTð Þ ¼ e j2pdfi kT ri1ai kTð Þ
r
i1;dfi
ai;fi
kTð Þ ¼ ej2pdfi kT ri1;dfiai fi kTð Þð Þ:
Compute the FrFT 
warping fu
r
By using the        order
transformation and the spine
warped FrFT of the si
Extract the ith signal compo
warped FrFT 
By using adaptively smoothing the WD slices, 
compute the high resolution t-f slices of the warped 
signal component                which are parallel to the 
time axis
Compute the corresponding high resolution  t-f slices 
of the fractional Fourier transformed signal component
By removing the rotation effect induced by the 
Fractional Fourier transformation, compute the 
corresponding high resolution   t-f slices of the signal 
component 
Compute the TFD of compo
TFDs of individual s
NO
ia
 
( )ˆ fii i
i
a ts
δ
ζ
,
,
ˆ ( )
i
i
as t
ˆ ( )is t
Fig. 1. Detailed flow diagram of time-freq0( ) ( )t x t:=
1i :=
ERP Signal
 component            by 
entation algorithm.
( )is t
otation angle         .      
      of the ith component 
domain     .
2i ia π=
ia
( )t
φ
sychophysiology 60 (2006) 225–2396. Estimate the ith component by incision of the time–
frequency domain as
sˆ
i;dfi
a;fi
tð Þ ¼ h2 tð Þ h1 tð Þ4r
i1;dfi
ai;fi
tð Þ
ih
;
where h2(t) is a time-domain mask and h1(t) is the
inverse Fourier transform of a frequency-domain mask
H1( f).
7. For each TFD slice of si(t), compute yai;fi kTð Þ ¼
sˆ
i;dfi
ai;fi
kTð Þe j2pDwfi kTð Þ, after choosing the slice offset Dw.
8. Compute the sampled TFD TFyai ;fi mT¯ ; fwi
 
; t1=T¯Vm
VtN=T¯ of yai,f i(t) using the directional smoothing algorithm
               , and 
nction 
1( )
i
i
a kT
-
( )i tΓ
 fractional Fourier 
               compute the 
gnal                   .
nent as               in the 
domain.
Compute the time domain estimate           of the ith    
signal component by using the inverse warping 
transformation
Compute the residual signal
site signal by summing the 
ignal components
Is any signal component left 
in residual signal
YES
( )
ii a
t
,
 
1 ( )
i i
i
ar kTζ
-
,
( )ˆ fii i
i
a ts
δ
ζ
,
,
ˆ ( )is t
1( ) ( ) ( )ˆi i ir kT r kT kTs-= -
( )ir kT 1i i= +
ψ
uency component analyzer (TFCA).
components have non-overlapping supports in the time–
frequency plane.
2.4.4. Detection of early gamma response based on time–
frequency domain characteristics
Detection of the early gamma response (0–150 ms
poststimulus time window) was realized using an automated
detection technique (Fig. 2).
PRE-PROCESSING
[26 – 48] Hz Band-pass 
Digital Filtering
x(t): EEG Signal
Weight filtered signal
g(t) = w(t) * y(t)
Compute STFT
 Sg
y(t)
Determine freq. support 
[fli – fhi] of gamma candidates 
from Sg
Sg
For each gamma candidate, i
Filter x(t) with a band-pass filter with cut-offs 
[fli - fhi]
yi(t)
x(t)
[fli – fhi]
Mgi
Compute STFT Syi
Find pre-stimulus 
marker, Iyi
Syi
Weight filtered signal
gi(t) = w(t) * yi(t)
gi(t)
Find early-gamma 
marker using multi-level 
thresholding, Igi
Compute STFT Sgi
Sgi
Myi
Construct masked TF 
representation of early-
gamma response, Mgi
Igi
Igi
Construct masked TF 
representation of pre-
stimulus signal, Myi
Iyi
yi(t)
g(t)
x(t)
al of Psychophysiology 60 (2006) 225–239 229(cf. O¨zdemir and ArNkan, 2000), where T¯ is the sampling
interval of the TFD slice.
9. The TFD slice of si(t) is given by
TFsi tr mT¯ð Þ; fr mT¯ð Þð Þ ¼ TFyai ;fi mT¯ ; fwi
 
;
where (tr(mT¯), fr(mT¯)) define a curve in the time–frequency
plane parameterized by the variable mT¯:
tr mT¯ð Þ ¼ f mT¯ð Þcos
aip
2
 
 w f mT¯ð Þð Þ þ Dw
 
sin
aip
2
 
fr mT¯ð Þ ¼ f mT¯ð Þsin
aip
2
 
þ w f mT¯ð Þð Þ þ Dw
 
cos
aip
2
 
;
t1=T¯VmVtN=T¯ :
10. Estimate the sampled si(t) by taking the inverse of
the warping, frequency modulation and the fractional
Fourier transformation on the sampled sˆai,f i
d f i (t)
sˆ
i;dfi
ai kTð Þ ¼ e j2pdfi f
1
i kTð Þsˆ
i;dfi
ai;fi
f1i kTð Þ
 
;
sˆiai kTð Þ ¼ e
j2pdfi kT sˆ
i;dfi
ai kTð Þ;
sˆi kTð Þ ¼ Fai sˆiai
n o
kTð Þ:
11. Compute the residual signal ri(kT) = ri1(kT)
sˆi(kT).
if any signal component is left in residual signal ri(kT)
then
Set i = i+1, and GOTO step 2,
else
Compute the time–frequency distribution of the
composite signal as the sum of the time–frequency
distributions of individual signal components.
endif
As the algorithm shows, TFCA suppresses the cross-
terms (both inner and outer interference terms), which are
associated with the Wigner distribution. TFCA can identify
the auto-terms in the time-and-frequency plane, and can do
this for mono- and multi-component signals with linear or
curved time–frequency supports. As such, TFCA identifies
and extracts the maximal energies of the oscillatory
components from the composite (multi-component) ERP
signal (O¨zdemir et al., 2005). Accordingly, as a high-
resolution signal analysis technique, TFCA can produce
the global distribution of uncontaminated signal compo-
nents in the form of spatially and temporally integrated,
S. Karakas¸ et al. / International Journtime-varying oscillatory activity of various frequency
ranges. The only assumption in TFCA is that signalExtract features
Detect presence of early-gamma response
Peak 
Ratio
Energy 
Ratio
Normalized
Enhancement
EARLY-GAMMA 
EXISTS
EARLY-GAMMA
 DOES NOT EXISTFig. 2. Flow diagram of the detection technique for the early gamma
response.
3. The ratio of energies of the masked TF representations,
al of PThe major steps of the detection algorithm were: (1) Pre-
processing of the EEG data to enhance detection. (2)
Computing Short-Time Fourier Transform (STFT) of the
pre-processed signal. (3) Locating time–frequency domain
peaks. (4) Identifying the frequency interval of each
detected peak. (5) Filtering the EEG recording within the
identified frequency intervals in Step 4. (6) Computing two
STFTs. (a) STFT of the signal obtained in Step 5: Syi(t,f);
(b) STFT of the signal obtained in Step 5 after time
weighting: Sgi(t,f). Here i denotes the index of the potential
early gamma response. (7) Extracting the support marker for
Sgi(t,f):Igi(t,f). (8) Forming the support marker for Syi(t,f) by
using Igi(t,f). (9) Extracting quantitative time–frequency
domain features. (10) Detecting the early gamma response
based on the extracted features.
As Fig. 2 shows, the detection algorithm starts with
filtering the EEG signal, x(t), in (26–48) Hz band (Karakas¸
and Bas¸ar, 1998) to obtain the filtered response, y(t). The
filter used is an infinite impulse response (IIR) filter with 14
coefficients which have 46 dB attenuation in stopbands, and
0.1 dB ripple in the passband.
The pre-processed signal g(t) is obtained by multiplying
the filtered EEG signal with the time-domain weighting
function, w(t):
g tð Þ ¼ w tð Þ  y tð Þ:
The support of w(t) is chosen to emphasize features in
50 to 300 ms. The algorithm looks for early gamma
response in the following STFT magnitude:
Sg t; fð Þ ¼
				
Z þV
V
g sð Þ 1ffiffiffiffiffi
2p
p
r
e
 tsð Þ2
2r2 ej2pf sds
				:
The width parameter, r, of the sliding STFT window is
chosen as 57 ms.
By multi-level thresholding Sgi(t,f), a compact time–
frequency support of gi(t) is obtained. In this way, a marker
Igi(t,f) indicating the regions with high energy on the STFT
magnitude is obtained. By multiplying Sgi(t,f) with Igi(t,f),
the masked time–frequency (TF) representation is obtained:
Mgi t; fð Þ ¼ Sgi t; fð Þ  Igi t; fð Þ:
Similarly, masked TF representation of the prestimulus
response Myi(t,f) is computed as:
Myi t; fð Þ ¼ Syi t; fð Þ  Iyi t; fð Þ;
where the prestimulus marker, Iyi(t,f), is obtained by dilation
of the gamma marker Igi(t,f) within the prestimulus region
boundaries.
The following set of features is chosen for robust
differentiation of the pre-and poststimulus response in the
presence of early gamma signal.
1. The normalized enhancement, ce:
pg  lp
S. Karakas¸ et al. / International Journ230ce ¼ rp
;Er:
Er ¼
Z
Mgi t; fð Þ
2dtdf
Z
Igi t; fð Þdtdf

Z
Myi t; fð Þ
2dtdf
Z
Iyi t; fð Þdtdf
3
775
2
664
1
:
To illustrate the efficiency of the extracted features, a
neural network (NN) classifier with 3 inputs corresponding
to each extracted feature and a single output indicating the
presence of the early gamma response was designed. The
neural network had two hidden layers, with 30 and 15
neurons, respectively. The transfer functions in all of the
layers were hyperbolic tangent transfer functions.
The designed neural network was trained by using a
back-propagation algorithm on a set, which contained 21
signals that exhibited gamma response, G+, and 14 signals
that did not exhibit gamma response, G. Preliminary
classification of the signals into G+ and G was achieved
through expert opinion. The trained network had a 95%
success rate on the training data. Then the neural network
classifier was tested on 102 signals, which were obtained in
response to the target and standard stimuli of the oddball
paradigm from the different recording sites. The expert and
the classifications of this preliminary neural network
coincided in 76% of the cases: G+ concordance was
47.06% and G concordance was 29.41%. Among the
misclassified cases, the false-positives were 7.84% and the
false negatives were 15.69%. This neural network has been
designed for the present study and may be further refined.
The expert opinion was based on digitally filtered wave-
forms which were, basically, time-domain representations of
the oscillatory responses whose frequency limits were
determined from the AFCs. Taking these into consideration,
the present study analyzed those gamma responses where
decision concerning the gamma status of both the expert and
the neural network classifier was identical.
3. Results
3.1. ERPs and time-locked gamma responses in gender
subtypes
Figs. 3 and 4 display the ERPs (first block of figures) for
females (F) and males (M) which were obtained in responsewhere pg is the peak value of Mgi(t,f), lp and rp are the
mean value and standard deviation of Myi(t,f), respectively.
2. The peak ratio, cp:
cp ¼
pg
pp
where pp denotes the peak value of the masked TF
representation of the prestimulus.
sychophysiology 60 (2006) 225–239to the target and the standard stimuli, respectively, from the
Fz, Cz and Pz recording sites. Table 1 shows the descriptive
Fig. 3. Findings obtained in response to the target stimuli. First block: Event-related potential waveforms (ERPs). Second block: Amplitude frequency characteristics (AFCs). Third block: Digitally filtered
waveforms (f-ERPs) for females (F: first rows in each block) and males (M: second rows in each block). Recording sites: Fz, Cz and Pz. Note that the Y-axis for ERPs, AFCs and f-ERPs have scales proportional to
the largest amplitude values in each representation. Stimulation onset was marked with ‘‘0’’.
S.
K
arakas¸
et
al.
/
International
Journal
of
P
sychophysiology
60
(2006)
225
–
239
231
Fig. 4. Findings obtained in response to the standard stimuli. First block: The event-related potential waveforms (ERPs). Second block: Amplitude frequency characteristics (AFC). Third block: Digitally filtered
waveforms (f-ERPs) for females (F: first rows in each block) and males (M: second rows in each block). Recording sites: Fz, Cz and Pz. Note that the Y-axis for ERPs, AFCs and f-ERPs have scales proportional to
the largest amplitude values in each representation. Stimulation onset was marked with ‘‘0’’.
S.
K
arakas¸
et
al.
/
International
Journal
of
P
sychophysiology
60
(2006)
225
–
239
232
AFCs of both types of gender and of both stimulus types.
Considering all experimental conditions, the frequency of
the gamma response varied between 28 and 46 Hz.
The third block in Figs. 3 and 4 displays the ERPs that
were filtered in the gamma range (f-ERPs). The figures
show that the time-locked gamma response occurred within
a time window of 0–150 ms poststimulus in the f-ERPs of
Table 2
Summary tables for 223 analysis of variance (ANOVA) with gender
(Female, Male), stimulus type (Target, Standard) and recording site (Fz, Cz,
Pz) as predictive (Independent) variables and ERP peak amplitudes as
predicted (Dependent) variables
Sources of variance Sum of squares df Mean square F-values
N100 Amplitude (lV)
Gender (A) 47.33 1 47.33 0.84
Error 2245.81 40 56.14
Stimulus type (B) 15.58 1 15.58 0.69
Error 908.69 40 22.72
Recording site (C) 970.85 1 760.80 73.48**
Error 528.47 51 10.35
AB 141.14 1 141.14 6.21*
Error 908.69 40 22.72
AC 2.82 1 2.21 0.21
Error 528.47 51 10.35
BC 22.51 1 17.11 5.52*
Error 162.97 53 3.10
ABC 3.26 1 2.48 0.80
Error 162.97 53 3.10
N200 Amplitude (lV)
Gender (A) 1.92 1 1.92 0.04
Error 1693.15 40 42.33
Stimulus type (B) 313.79 1 313.79 5.91*
Error 2124.75 40 53.12
Recording site (C) 1133.49 2 748.90 57.50***
Error 788.52 61 13.02
AB 114.70 1 114.70 2.16
Error 2124.75 40 53.12
AC 20.88 2 13.79 1.06
al of Psychophysiology 60 (2006) 225–239 233statistics for the amplitude variations of the ERP peaks of
males and females obtained in response to the target and
standard stimuli from the three recording sites. Females had
higher P300 response amplitudes than males. Target-evoked
N200 and P300 peaks were higher in amplitude than
standard-evoked ones. While target-evoked N100 was
higher in males, target-evoked P300 was higher in females.
For standard-evoked N100 and P300, gender effect shifted;
standard evoked N100 was higher in females and standard
evoked P300 was higher in males. Peak amplitudes changed
according to recording site. P300 increased in amplitude
towards the posterior recording sites and the negative peaks,
(N100 and N200), increased in amplitude towards the
anterior recording sites.
The effects of gender (female, male), stimulus type
(target, standard) and recording site (Fz, Cz, Pz) on
Table 1
Means and standard deviations of the amplitudes of target-elicited and
standard-elicited ERP components (N100, N200, P300) according to gender
(Female, Male), stimulus type (Target, Standard) and recording site (Fz, Cz,
Pz)
Amplitude (AV)
Fz Cz Pz
N100 Female Target 9.64 (T4.02) 9.02 (T3.60) 5.01 (T3.13)
Standard 10.26 (T4.51) 9.90 (T4.31) 6.51 (T3.26)
Male Target 12.52 (T5.08) 11.43 (T4.24) 6.82 (T3.39)
Standard 9.67 (T4.60) 9.17 (T4.10) 5.95 (T3.36)
N200 Female Target 5.39 (T6.07) 1.69 (T6.17) 2.30 (T4.46)
Standard 2.98 (T4.37) 0.08 (T4.50) 0.93 (T3.71)
Male Target 6.33 (T5.18) 2.86 (T4.85) 0.17 (T4.99)
Standard 1.40 (T3.67) 1.35 (T2.49) 1.44 (T2.25)
P300 Female Target 10.27 (T6.63) 15.61 (T6.76) 19.18 (T7.71)
Standard 1.34 (T3.38) 3.63 (T3.78) 5.12 (T3.50)
Male Target 5.99 (T5.35) 10.00 (T4.09) 12.03 (T3.87)
Standard 2.66 (T3.67) 5.42 (T4.06) 5.98 (T3.89)
S. Karakas¸ et al. / International Journamplitudes of the ERP peaks (N100, N200, P300) were
tested using 223 factorial design with repeated meas-
ures at the second and third factors. According to Mau-
chley’s test, the data were not spherical. Taking the sample
size into consideration, correction was performed using
Huyn–Feldt epsilon in performing analysis of variance
(ANOVA) (Vasey and Thayer, 1987). Table 2 presents a
summary table of ANOVAs which were separately per-
formed for each ERP peak (N100, N200, P300). Gender had
a significant effect on P300 amplitude. Stimulus type had a
significant effect on the N200 and P300 peaks. Recording
site was found significant for all the studied ERP peaks. The
interaction of gender and stimulus type was significant for
N100 and P300. The interaction of stimulus type and
recording site was significant for all peaks.
Fig. 3 displays the AFCs (second block of figures) for
females (F) and males (M) which were obtained in response
to target stimuli from the Fz, Cz and Pz recording sites. Fig.
4 displays the findings obtained in response to the standard
stimuli. The global representations that the AFCs provided
suggested that the gamma selectivity was present in the
Error 788.52 61 13.02
BC 146.84 2 96.72 17.10***
Error 343.57 61 5.66
ABC 0.65 2 0.43 0.07
Error 343.57 61 5.66
P300 Amplitude (lV)
Gender (A) 297.48 1 297.48 4.66*
Error 2554.61 40 63.86
Stimulus type (B) 4181.13 1 4181.13 67.77***
Error 2467.67 40 61.69
Recording site (C) 1312.86 2 873.94 75.78***
Error 692.94 60 11.53
AB 770.34 1 770.34 12.49***
Error 2467.67 40 61.69
AC 31.23 2 20.79 1.80
Error 692.94 60 11.53
BC 162.65 2 91.72 29.75***
Error 218.72 71 3.08
ABC 16.45 2 9.28 3.01
Error 218.72 71 3.08
* p <.05.
** p <.01.
*** p <.001.
both females and males in response to both the target and
the standard stimuli at specifically the frontocentral record-
ing sites. Data were analyzed using a 223 ANOVA for
the effect of gender (female, male), stimulus type (target,
standard) and recording site (Fz, Cz, Pz) with repeated
measures at the second and third factors. According to
Mauchley’s test, the data were not spherical; accordingly,
Huyn–Feldt was used as the epsilon value. The peak-to-
peak (Vp–p) amplitude of the gamma response, which was
measured between the most negative and most positive
were
left c
vefor
S. Karakas¸ et al. / International Journal of Psychophysiology 60 (2006) 225–239234Fig. 5. The event-related potentials (ERPs) and filtered ERPs (f-ERPs) that
females (F) and males (M) with time-locked early gamma response, G+ (
column). First four figures in each recording site: Event-related potential wa
gamma range. Note that the Y-axis for ERPs and f-ERPs have scales proportiona
was marked with ‘‘0’’.obtained for the target stimuli from the Fz, Cz and Pz recording sites in
olumn), and those without time-locked early gamma response, G (right
ms (ERPs). Second four figures in each recording site: ERPs filtered at the
l to the largest amplitude values in each representation. Stimulation onset
of gender (F(1, 38)=5.64, p =.023). (The mean amplitude of
females were 1.49 AV and of males 1.37 AV.) ANOVA did
not find a significant effect of stimulus type. The interaction
effect between gender and gamma status was not found
significant either. (In the G+ group, mean amplitude of
females was 1.97 AV and of males 1.41 AV; in the G
group, the mean amplitude of females was 1.01 AV and of
al of Psychophysiology 60 (2006) 225–239 235points in the 0–150 ms time window, was not affected by
stimulus type. Mean gamma response amplitudes of females
were higher than males (for target stimuli means for females
and males were, 1.49 AV and 1.24 AV, respectively, at Fz;
1.65 AV and 1.40 AV, respectively, at Cz; and 1.27 AV and
1.02 AV, respectively, at Pz). However, this effect was not
found significant. Recording site was found significant
(Huyn–Feldt epsilon: F(2,80) =9.72, p = .001). Post hoc
analyses revealed that the significant effect originated from
the difference between Cz and Pz recording sites (mean
difference: .31, p =.01). The difference between Fz and Cz,
recording sites, on the other hand, was not significant. These
findings showed that the early gamma response had a
frontocentral distribution. None of the interaction effects
were significant.
3.2. Gender subtypes and individual differences in the
time-locked gamma response
3.2.1. Findings from digital filtering
The f-ERPs of each participant were studied for the
presence of the early, time-locked gamma response.
Classification of the data of 58 participants was performed
by expert opinion and automated gamma detection techni-
que. Two raters independently rated the data for the
presence of the gamma response. There was 97% agreement
between these raters. The data of only those participants for
whom the detection technique and expert opinion showed
100% agreement were chosen for further analysis. Agree-
ment was obtained on 22 participants in the G+ group
(37.93%) and 20 participants in the G group (34.48%).
Both detection approaches found the gamma response of 16
participants unclassifiable (27.59%). Analyses of the present
study were thus conducted on the data of a total of 42
participants; the group consisted of 22 females and 20 males
(also see Section 2.1).
Fig. 5 displays the ERPs and the gamma responses (f-
ERPs) of G+ and G females and males in response to the
target stimuli at the Fz, Cz and Pz recording sites. Since the
gamma response was obtained from the frontocentral
recording sites (Section 3.1, Figs. 3 and 4; also see Karakas¸
and Bas¸ar, 1998), data that were obtained from the Fz
recording site were analyzed using 222 ANOVA for
the effect of gender (female, male), gamma status (G+ and
G) and stimulus type (target, standard) with repeated
measures at the last factor. According to Mauchley’s test, the
data were not spherical; accordingly, Huyn–Feldt was used
as the epsilon value. The Vp–p amplitude of the gamma
response was affected with gamma status (F(1, 38)=16.23,
p =.0001); the response amplitude of the G+ group was
higher than of the G group (mean amplitudes 1.69 AVand
1.02 AV, respectively). When individual differences in the
gamma response were not taken into account (Figs. 3 and
4), gender effect had not been found significant. However,
S. Karakas¸ et al. / International Journwhen gamma status was analyzed using the G+ and G
levels of the variable, ANOVA revealed a significant effectmales, 1.04 AV.)
Chi-square test for independent groups showed that the
distribution of gender to gamma status was not significant
(v2= .75, p> .05) (Table 3). This finding showed that G+
and G groups were homogeneously distributed over
gender.
Fig. 5 shows that, in females, gamma status does not
affect the ERP components. Even for target-evoked P300
for which females had a significantly higher amplitude than
males (Tables 1 and 2, Fig. 3), t-test for independent groups
showed that the difference between the G+ and G females
(measured as Vp–p between N2 and P3 in response to target
stimuli at Fz) as nonsignificant. Males had higher ampli-
tudes for target-evoked N100 and N200 (Tables 1 and 2,
Fig. 3). In G males, however, N200 lost its distinctiveness
as a peak and appeared as a side peak on an extensive
negativity that also included the N100. The t-test for
independent groups showed that the difference between
the G+ and G males (measured as Vp–p between P2 and
N2 in response to target stimuli at Fz) as significant
(t =2.08, p= .05).
The effects of gender (female, male), stimulus type
(target, standard) and recording site (Fz, Cz, Pz) on
amplitudes of the ERP peaks (N100, N200 and P300) were
tested using only G+ participants. A 223 factorial
design was used with repeated measures at the second and
third factors. Data were not spherical, thus correction was
performed using Huyn–Feldt epsilon in performing
ANOVA. The results were similar to those obtained for
the total sample that included G+ and G participants
(Table 2). The only critical difference was obtained for
P300. Gender had a significant effect on P300 in pooled
data (Table 2); however, this effect was not significant when
an ANOVA was conducted on only G+ participants.
3.2.2. Findings from TFCA
In the present study, the gamma responses were also
analyzed in the time–frequency plane using TFCA. Fig. 6
displays the application of TFCA to the averaged ERPs of
the females (F) and males (M) in the G+ and G groups.
Table 3
22 Chi-square test for the distribution of gender to gamma status
( p >.05)
Gamma present, G (+) Gamma absent, G ()
Female n =11 n =11
26.2% 26.2%
Male n =11 n =926.2% 21.4%
F 
M 
F 
M 
al of P20
30
40
50
-0.02
0
0.02
-1 -0.5 0 0.5 1
2
0
-2
20
30
40
50
-0.05
0
0.05
2
0
-2
GAMMA (+)
S. Karakas¸ et al. / International Journ236The first block that consisted of four figures shows the
localization of the gamma component in the time–
frequency plane. The second block that also consisted of
four figures shows the gamma components which were
produced by TFCA in the time plane.
Both the time–frequency distribution of components and
the time-domain activity that TFCA produced showed that
the gamma activity occurred in both the prestimulus and
poststimulus epochs. The left column in Fig. 6 shows the
findings for the G+ females and males. In the G+ group,
gamma activity was brought into resonance by stimulation.
The stimulus-locked nature of the response is seen in TFCA
as a high-energy component in the early time window (first
two figures in the left column) and as a high amplitude
component in the time-domain layout of the component
(third and fourth figures in the left column). The gamma
response occurred in a frequency range of 27.54 Hz and
42.46 Hz.
The right column in Fig. 6 shows the findings for the G
females and males. In both the TFCA distribution (first two
figures in the right column) and the time-domain layout
(third and fourth figures in the right column), it is clearly
seen that the component is not stimulus-locked. Unlike the
G+ group where the gamma response resonated upon
stimulation, gamma in the G group existed throughout
the pre- and the poststimulus recording period with a
Fig. 6. TFCA analysis of the ERPs that were obtained for the target stimuli in fe
column), and those without time-locked early gamma response, G (right c
representation produced by TFCA. Second four figures: Respective time-domain g
Y-axes/right: amplitude; X-axis: time in seconds. Note that each TFCA representatio
domain gamma components: Y-axes/left: amplitude; X-axis: time in seconds. Stim-0.02
0
0.02
-0.02
0
0.02
-1 -0.5 0 0.5 1
GAMMA (-)
sychophysiology 60 (2006) 225–239comparable magnitude in a frequency range of 25.28 Hz and
37.19 Hz. This gamma activity was not a response to the
stimulus but a component that spontaneously occurred in the
electroencephalogram (for a review, see Gottesmann, 1999).
The results of TFCA were statistically analyzed using
two parameters of the gamma activity as dependent
variables: maximum Vp–p amplitude in the poststimulus
early time window, ratio of the maximum Vp–p amplitude in
the poststimulus early time window to that in the
prestimulus interval. Since there were two dependent
variables, data were analyzed using 222 multivariate
analysis of variance (MANOVA) for the effect of gender
(female, male), gamma status (G+ and G) and stimulus
type (target, standard) with repeated measures at the last
factor. Data were derived from the Fz recording site.
As with the data on f-ERPs (Fig. 5), significant effects
were obtained in the total model for gender (Wilks lambda:
F(2,75)=3.58, p =.033) and gamma status (Wilks lambda:
F(2,75)=11.55, p =.0001). Stimulus type and the interaction
effects were not found significant. In MANOVA, each
individual dependent variable was analyzed using a
Bonferroni adjusted level of 0.25. The two groups of gender
differed in the maximum Vp–p amplitude in the poststimulus
early time window (F(1,76)=6.47, p =.013). The two groups
of gamma status differed in both the maximum Vp–p
amplitude in the poststimulus early time window
males (F) and males (M) with time-locked early gamma response, G+ (left
olumn), from the Fz recording site. First four figures: Time-frequency
amma components. For TFCA representations: Y-axes/left: frequency in Hz;
n has a scale that is proportional to the strength of its components. For time-
ulation onset was marked with ‘‘0’’.
gender was found nonsignificant with both types of data.
Lack of an interaction effect showed that gender is not a
al of Pdecisive factor in the gamma status. This conclusion was
also reached through the chi-square test which showed that
the gamma subtypes were homogeneously distributed over
gender. Accordingly, the causal factor for the individual
differences in the early, time-locked gamma response should(F(1,76)=18.08, p =.0001) and in the ratio of the maximum
Vp–p amplitude in the poststimulus early time window to
that in the prestimulus interval (F(1,76)=9.06, p =.004).
Main effect of stimulus type and the interaction effects were
not found significant for any of the dependent variables.
4. Conclusions
The present literature acknowledges the early, time-
locked gamma response as a reliable neuroelectic phenom-
enon (Gurtubay et al., 2001; Herrmann and Mecklinger,
2001; Karakas¸ et al., 2001; Karakas¸ and Bas¸ar, 1998, 2004).
Using time-domain (ERPs), frequency-domain (AFCs and
f-ERPS) and time–frequency domain (TFCA) representa-
tions, the frontocentral gamma response was also demon-
strated in the present study in response to the conditions of
the auditory oddball paradigm (Figs. 3 and 4).
Individual differences in the gamma response have
already been suggested in the previous literature (Jokeit
and Makeig, 1994; Karakas¸ et al., 2001, 2003). There are a
group of traits/characteristics in information processing such
as field-dependence and handedness (Annett, 1970; Witkin
et al., 1954). In terms of these traits/characteristics, there are
two groups at either extreme: these would be field-depend-
ence and field-independence or left handedness and right
handedness. There is, however, a third group between the
two extremes; this group shows varying amounts of the trait/
characteristics. Gamma status with G+ and G groups at the
two extremes and the unclassifiable cases in between is one
such trait/characteristics.
The two extremes, G+ and G, were demonstrated in the
present study with digital filtering (Fig. 5) and with TFCA
(Fig. 6), a technique where neuroelectric components are
simultaneously displayed in the time-frequency plane and
where no assumption is made on system linearity and/or
stationarity. Statistical analyses of the data obtained through
digital filtering technique (maximum Vp–p amplitude in the
poststimulus early time window) and through TFCA
technique (maximum Vp–p amplitude in the poststimulus
early time window, ratio of the maximum Vp–p amplitude in
the poststimulus early time window to that in the
prestimulus interval) showed that gender as well as the
gamma status affected the amplitude/energy of the gamma
component, with higher values in females than in males and
higher values in the G+ group than in G group.
However, the interaction between the gamma status and
S. Karakas¸ et al. / International Journbe searched for not in gender differences but in other
characteristics and/or processes. A genetic basis has beenfound of the beta frequency in human EEG (Porjesz et al.,
2002). The investigation of a probable genetic basis of the
time-locked gamma response should be among the pro-
spective research areas in the field.
According to a large group of studies, the time-locked,
early gamma response represents the early stages of
information processing that include sensory and perceptual
operations (Gurtubay et al., 2001; for a review, Karakas¸ et
al., 2001; Karakas¸ and Bas¸ar, 1998, 2004). Karakas¸ et al.
(2003) further found evidence for top-down influences that
involve higher cognitive processes on the bottom-up,
sensory-perceptual, early gamma response. The difference
in the early gamma response was found to be correlated with
neuropsychological test scores on attention, learning, short-
term memory and executive functions; G+ and G groups
could be predicted from these scores with an overall success
rate of 93.33%. The present study found individual differ-
ences in gamma status which, however, did not depend on
gender. The findings of the present study thus suggest that
sensory-perceptual processing that the early time-locked
gamma response represents (Karakas¸ and Bas¸ar, 1998) and
the top-down influences on the early gamma response is not
gender-specific (Karakas¸ et al., 2003). Such a conclusion
and its implication should be extensively investigated,
however, using standardized neuropsychological tests,
complex research designs and advanced techniques of
signal analysis.
4.1. ERPs considered within a context of the gamma
response
The poststimulus 0–150 ms time window of the early
gamma response includes the N100 and N200 ERP peaks.
As a group of other studies also did (Go¨lgeli et al., 1999;
Nagy et al., 2003), the present study found that amplitudes
of N100 and, in part, N200 were higher in males (Table 1,
Figs. 3 and 4). However, this relationship was complex.
Male superiority was obtained in the target-evoked response
(Table 1). Meanwhile, in the standard-evoked response,
there was a female superiority. There was thus an interaction
between gender and stimulus type which was found
significant (Table 2).
When the sample was divided into G+ and G
participants, the amplitudes of the negative peaks were not
affected in females (Fig. 5). However, in G males, the
differentiation of the N200 peak became less evident and
this difference was found significant. In G males, N200
took place on an extended negativity that also included the
N100 peak. Thus, if a sample consists of G males, the
result will be lower N200 in males, or conversely, higher
N200 in females. The higher amplitude N200 in females is
what another group of studies in fact found (Barrett and
Fulfs, 1998; Hoffman and Polich, 1999; Kasai et al., 2002).
The foregoing results on N200 suggest that the conflicting
sychophysiology 60 (2006) 225–239 237results in the literature may be due to experimental designs
where variables (stimulus type, gamma status) that have the
suggesting that females have a special position in attention
and memory updating (Sutton et al., 1965; Polich and Kok,
1 and 2, Figs. 3 and 4). Thus, higher amplitudes would be
obtained in females only in response to the target stimuli.
al of PFurthermore, the effect on P300 was obtained only
when the sample included both G+ and G participants
(Table 2). When analysis was conducted on only the G+
females, this effect was not found significant. As in
N200, the foregoing results on P300 suggest that the
conflicting results in the literature may be due to
experimental designs where variables (stimulus type,
gamma status) that have the potential of interacting with
the independent variable (gender) are not included in the
experimental design. Thus, when studies do not include
these factors as variables but one study uses one level of
the variable and the other one uses the other level,
conflicting results would be found.
Shortly, also using techniques of time–frequency analy-
sis, the present study showed that gender is not a factor in
the individual differences concerning the status of the early
gamma response. The amplitude variations in the ERP
components (N100, N200, P300) can also not be explained
solely on the basis of gender differences. Accordingly, even
the earlier sensory-perceptual operations of information
processing which the early gamma response, the N100
and the N200 ERP components represent, require an
approach which would enable the discovery of the multi-
factorial pattern of influences.
Acknowledgement
This study was partially supported by Project Number
TUBITAK-TBAG/CG-3.
References
Adrian, E.D., 1942. Olfactory reactions in the brain of the hedgehog.
J. Physiol. 100, 459–473.
Annett, M., 1970. A classification of hand preference by association
analysis. Br. J. Psychol. 61, 303–321.
ArNkan, O., O¨zdemir, A.K., Tu¨fekc¸i, I˙., C¸akmak, E.D., Karakas¸, S., 2003. A
new technique for joint time– frequency analysis of event-related1995). However, as in N200, there was a significant
interaction effect on P300 such that there was a female
superiority in target-evoked response amplitudes but a male
superiority in standard-evoked response amplitudes (Tablespotential of interacting with the independent variable
(gender) are not included in the experimental design.
As reported in the literature (Hoffman and Polich, 1999;
Orozco and Ehlers, 1998; Osterhout et al., 1997), the present
study found a significantly higher P300 amplitude in
females (Tables 1 and 2). The posterior topography of the
P300 is an indication that this is the P3b component,
S. Karakas¸ et al. / International Journ238signals of the brain: time–frequency component analyzer (TFCA).
EEG Clin. Neurosci. Soc. (ECNS) 34 (3), 170.Barrett, K.A., Fulfs, J.M., 1998. Effect of gender on the mismatch
negativity auditory evoked potential. J. Am. Acad. Audiol. 9 (6),
444–451.
Bas¸ar, E., 1976. Biophysical and Physiological Systems Analysis. Addison-
Wesley, Amsterdam.
Bas¸ar, E., 1980. EEG-Brain Dynamics. Relation Between EEG and Brain
Evoked Potentials. Elsevier/North-Holland, Amsterdam.
Bas¸ar, E., 1998. Brain Function and Oscillations: I. Brain Oscillations.
Principles and Approaches, Springer-Verlag, Heidelberg, pp. 87–91,
293–327.
Bas¸ar, E., 1999. Brain Function and Oscillations: II. Integrative Brain
Function. Neurophysiology and Cognitive Processes, Springer-Verlag,
Heidelberg, pp. 155–174, 331–435, 407–426.
Bas¸ar-Erog˘lu, C., Stru¨ber, D., Schu¨rmann, M., Stadler, M., Bas¸ar, E., 1996.
Gamma-band response in the brain: a short review of psychophysio-
logical correlates and functional significance. Int. J. Psychophysiol. 24,
101–112.
Berger, H., 1929. Uber des elektroenkephalogram. Arch. Psychiatr.
Nervenkr. 87, 527–570.
Brandt, M.E., Jansen, B.H., 1991. The relationship between prestimulus
alpha amplitude and visual evoked potential amplitude. Int. J. Neurosci.
61, 261–268.
Cohen, L., 1989. Time– frequency distributions: a review. Proc. IEEE 77
(7), 941–981.
Cohen, L., 1995. Time–Frequency Analysis. Prentice Hall, Englewood
Cliffs, NJ.
Cook III, E.W., Miller, G.A., 1992. Digital filtering: background and
tutorial for psychophysiologists. Psychophysiology 29, 350–367.
Dawson, G.D., 1954. A summation technique for the detection of
small evoked potentials. Electroencephalogr. Clin. Neurophysiol. 6,
153–154.
Farwell, L.A., Martinerie, J.M., Bashore, T.R., Rapp, P.E., Goddard, P.H.,
1993. Optimal digital filters for long-latency components of the event-
related brain potential. Psychophysiology 30, 306–315.
Galambos, R., 1992. A comparison of certain gamma band (40 Hz) brain
rhythms in cat and man. In: Bas¸ar, E., Bullock, T.H. (Eds.), Induced
Rhythms in the Brain. Birkha¨user, Boston.
Go¨lgeli, A., Su¨er, C., O¨zesmi, C¸., Dolu, N., As¸c¸Nog˘lu, M., S¸ahin, O., 1999.
The effect of sex differences on event-related potentials in young adults.
Int. J. Neurosci. 99 (1–4), 69–77.
Gottesmann, C., 1999. Neurophysiological support of consciousness during
waking and sleep. Prog. Neurobiol. 59, 469–508.
Gurtubay, A.M., Labura, A., Malanda, A., Iriarte, J., Artieda, J., 2001.
Gamma band activity in an auditory oddball paradigm studied with the
wavelet transform. Clin. Neurophysiol. 112, 1219–1228.
Herrmann, C.S., Mecklinger, A., 2001. Gamma activity in human EEG is
related to high-speed memory comparisons during object selective
attention. Vis. Cogn. 8 (3–5), 593–608.
Hoffman, L.D., Polich, J., 1999. P300, handedness, and corpus callosal
size: gender, modality, and task. Int. J. Psychophysiol. 31 (2),
163–174.
Jervis, B.W., Nichols, M.J., Johnson, T.E., Allen, E., Hudson, N.R., 1983.
A fundamental investigation of the composition of auditory evoked
potentials. IEEE T. Bio-Med. Eng. BME 30, 43–49.
Jokeit, H., Makeig, S., 1994. Different event-related patterns of gamma-
band power in brain waves of fast- and slow-reacting subjects. Proc.
Natl. Acad. Sci. U. S. A. 91 (14), 6339–6343.
Karakas¸, S., Bas¸ar, E., 1998. Early gamma response is sensory in origin: a
conclusion based on cross-comparison of results from multiple
experimental paradigms. Int. J. Psychophysiol. 31 (1), 13–31.
Karakas¸, S., Bas¸ar, E., 2004. Oscillatory responses of the brain and their
cognitive correlates. In: Adelman, G., Smith, B.H. (Eds.), Encyclopedia
of Neuroscience, 3rd ednR Elsevier, San Diego, CA.
Karakas¸, S., Erzengin, O¨.U., Bas¸ar, E., 2000a. A new strategy involving
multiple cognitive paradigms demonstrates that ERP components are
sychophysiology 60 (2006) 225–239determined by the superposition of oscillatory responses. Clin. Neuro-
physiol. 111, 1719–1732.
Karakas¸, S., Erzengin, O¨.U., Bas¸ar, E., 2000b. The genesis of human event-
related responses explained through the theory of oscillatory neural
assemblies. Neurosci. Lett. 285, 45–48.
Karakas¸, S., Bas¸ar-Erog˘lu, C., O¨zesmi, C¸., Kafadar, H., Erzengin, O¨.U.,
2001. Gamma response of the brain: a multifunctional oscillation that
represents a bottom-up with top-down processing. Int. J. Psychophysiol.
39 (2–3), 137–150.
Karakas¸, S., Bekc¸i, B., Erzengin, O¨.U., 2003. Early gamma response in
human neuroelectric activity is correlated with neuropsychological test
scores. Neurosci. Lett. 340, 37–40.
Kasai, K., Nakagome, K., Iwanami, A., Fukuda, M., Itoh, K., Koshida, I.,
Kato, N., 2002. No effect of gender on tonal and phonetic mismatch
negativity in normal adults assessed by a high-resolution EEG record-
ing. Brain Res. Cogn. Brain Res. 13 (3), 305–312.
Kolb, B., Whishaw, I.Q., 1996. Human Neuropsychology, 4th ednR W.H.
Freeman, New York.
Kolev, V., Yordanova, J., 1997. Analysis of phase-locking is informative for
studying event-related EEG activity. Biol. Cybern. 96, 229–235.
Maccoby, E., Jacklin, C.N., 1974. The Psychology of Sex Differences.
Stanford Univ. Pr., Stanford.
O¨zdemir, A.K., Durak, L., ArNkan, O., 2001. High-resolution time–
frequency analysis based on fractional domain warping. Proc. IEEE
Int. Conf. Acoust. Speech Signal Proc. 6, 3553–3556.
O¨zdemir, A.K., Karakas¸, S., C¸akmak, E.D., Tu¨fekc¸i, D.I˙., Arikan, O., 2005.
Timefrequency component analyzer and its application to brain
oscillatory activity. J. Neurosci. Methods 145, 107–125.
Parvin, C., Torres, F., Johnson, E., 1980. Synchronization of single evoked
response components: estimation and interrelation of reproducibility
measures. In: Pfurtscheller, G., Busser, P., Lopes da Silva, F., Petsche,
H. (Eds.), Rhythmic EEG Activities and Cortical Functioning. Elsevier,
Amsterdam.
Polich, J., Kok, A., 1995. Cognitive and biological determinants of P300:
an integrative review. Biol. Psychol. 41 (2), 103–146.
Porjesz, B., Almasy, L., Edenberg, H., Wang, K., Chorlian, D.B., Foroud,
D., et al., 2002. Linkage disequilibrium between the beta frequency of
the human EEG and a GABA (A) receptor gene locus. Proc. Natl. Acad.
Sci. 99, 3729–3733.
Ro¨schke, J., Aldenhoff, J.B., 1991. Excitability and susceptibility of the
brain’s electrical activity during sleep: an analysis of late components of
AEP and VEPs. Int. J. Neurosci. 56, 255–272.
Ro¨schke, J., Mann, K., Riemann, D., Frank, C., Fell, J., 1995. Sequential
analysis of the brain’s transfer properties during consecutive REM
episodes. Electroencephalogr. Clin. Neurophysiol. 96, 390–397.
S. Karakas¸ et al. / International Journal of Psychophysiology 60 (2006) 225–239 239as revealed by event-related potentials and other brain measures of
cognitive function. Behav. Brain Sci. 13 (2), 201–288.
Naatanen, R., 1992. Attention and Brain Function. Lawrence Erlbaum
Assoc., London.
Naatanen, R., Picton, T., 1986. N2 and automatic versus controlled
processes. In: McCallum, W.C., Zappoli, R., Denoth, I. (Eds.), Cerebral
Psychophysiology. Studies at Event-Related Potentials (EEG Suppl.
38). Elsevier, North-Holland.
Nagy, E., Potts, G.F., Loveland, K.A., 2003. Sex-related ERP differences in
deviance detection. Int. J. Psychophysiol. 48 (3), 285–292.
Orozco, S., Ehlers, C.L., 1998. Gender differences in electrophysiological
responses to facial stimuli. Biol. Psychiatry 44 (4), 281–289.
Osterhout, L., Bersick, M., McLaughlin, J., 1997. Brain potentials reflect
violations of gender stereotypes. Mem. Cogn. 25 (3), 273–285.
Ozaktas, H.M., ArNkan, O., Kutay, M.A., Bozdagi, G., 1996. Digital
computation of the fractional Fourier transform. IEEE Trans. Signal
Process. 44, 2141–2150.
O¨zdemir, A.K., ArNkan, O., 2000. A high-resolution time frequency
representation with significantly reduced cross-terms. Proc. IEEE Int.
Conf. Acoust. Speech Signal Proc. 2, 693–696.
O¨zdemir, A.K., ArNkan, O., 2001. Efficient computation of the ambiguity
function and Wigner distribution on arbitrary line segments. IEEE
Trans. Signal Process. 49, 381–393.Ro¨schke, J., Prentice-Cuntz, T., Wagner, P., Mann, K., Frank, C., 1996.
Amplitude frequency characteristics of evoked potentials during sleep:
an analysis of brain’s transfer properties in depression. Biol. Psychiatry
40, 736–743.
Solodovnikov, V.V., 1960. Introduction to the Statistical Dynamics of
Automatic Control Systems. Dover, New York.
Sutton, S., Braren, M., Zubin, J., John, E.R., 1965. Evoked potential
correlates of stimulus uncertainty. Science 150, 1187–1188.
Ungan, P., Bas¸ar, E., 1976. Comparison of Wiener filtering and selective
averaging of evoked potentials. Electroencephalogr. Clin. Neurophysiol.
40, 516–520.
Vasey, M.W., Thayer, J.F., 1987. The continuing problem of false positives
in a repeated measures ANOVA in psychophysiology: a multivariate
solution. Psychophysiology 24 (4), 479–486.
Vincent, L., Soille, P., 1991. Watersheds in digital spaces: an efficient
algorithm based on immersion simulations. IEEE Trans. Pattern Anal.
Mach. Intell. 13 (6), 583–598.
Witkin, H.A., Lewis, H.B., Hertzman, M., Machover, K., Meissner, P.,
Wapner, S., 1954. Personality Through Perception. Harper, New York.Naatanen, R., 1990. The role of attention in auditory information processing