Clinical Neurophysiology xxx (2013) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Clinical Neurophysiology

journal homepage: www.elsevier .com/locate /c l inph

Correlation of risk-taking propensity with cross-frequencyphase–amplitude coupling in the resting EEG

Jaewon Lee a,b,c, Jaeseung Jeong a,⇑a Department of Bio and Brain Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 305-701, South Koreab Neuropsychiatry Research Laboratory, Department of Psychiatry, Gongju National Hospital, Chungcheongnam-do 314-200, South Koreac Addiction Brain Center, Eulji Addiction Institute, Gangnam Eulji Hospital, Seoul 135-010, South Korea

a r t i c l e i n f o h i g h l i g h t s

Article history:Accepted 7 May 2013Available online xxxx

Keywords:Cross-frequency phase–amplitude coupling(CFPAC)Risk-taking propensity (RTP)Barrett impulsiveness scale (BIS)Domain-specific risk-taking (DOSPERT)scale

1388-2457/$36.00 Published by Elsevier Ltd. on behahttp://dx.doi.org/10.1016/j.clinph.2013.05.007

⇑ Corresponding author. Address: Brain Dynamics Land Brain Engineering, KAIST, 373-1 Kuseong-dong,Korea. Tel.: +82 42 350 4319; fax: +82 42 350 4310.

E-mail address: jsjeong@kaist.ac.kr (J. Jeong).

Please cite this article in press as: Lee J, Jeong J.Clin Neurophysiol (2013), http://dx.doi.org/10.1

� The cross-frequency phase–amplitude coupling may be successfully adapted to the scalp resting EEG.� There are significant correlations between cross-frequency phase–amplitude couplings and the mea-

sures of risk-taking propensity.� It is suggested that the cross-frequency phase–amplitude coupling could be a promising indicator for

diagnosing the tendency to take risks.

a b s t r a c t

Objective: Recent evidence has suggested that the weak inhibitory influence of the prefrontal cortex onthe subcortical structures may be responsible for risk-taking behaviour. The aim was to determine thepossibility that this weakness in top-down control is reflected in changes in the cross-frequencyphase–amplitude coupling (CFPAC) in the electroencephalography (EEG).Methods: Nineteen-channel EEGs were recorded from 50 healthy volunteers with their eyes closed beforerisk-taking propensity was assessed by behavioural measures, the domain-specific risk-taking (DOSPERT)scale and the Barrett impulsiveness scale (BIS). Correlation analyses between the CFPACs and the behav-ioural measures were performed.Results: The CFPACs were negatively correlated with the risk-taking DOSPERT and BIS scores in frontal(Fp2) and centro–parietal (C3, C4 and P4) regions. By contrast, the CFPACs were positively correlated withthe risk-taking DOSPERT and BIS scores in the right hemisphere (T8 and P8).Conclusions: We suggest that frequent risk-taking behaviour is closely associated with the reduced inter-ference of the cortical control network on the reward-oriented system. The CFPAC, which reflects thedegree of interactions among functional systems, provides information about an individual’s risk-takingpropensity.Significance: The CFPAC may be a useful neurophysiological indicator of an individual’s tendency towardsrisk-taking behaviours, which thus potentially contributes to evaluating the severity of the psychiatricdiseases exhibiting abnormal risk-taking behaviours.

Published by Elsevier Ltd. on behalf of International Federation of Clinical Neurophysiology.

1. Introduction

Risk-taking propensity (RTP) is generally defined as an individ-ual’s tendency to perceive or interpret potentially risky situationsand to take endurable risks but avoid excessive risks (Byrnes,1998; Garon and Moore, 2004; Halpern-Felsher and Cauffman,2001; Mann et al., 1989; Steinberg and Scott, 2003). A wide variety

lf of International Federation of Cli

aboratory, Department of BioYuseong-gu, Daejeon, South

Correlation of risk-taking prop016/j.clinph.2013.05.007

of behaviours qualify as risky. Among these are behaviours that in-volve highly undesirable real-world risks, such as alcohol abuse,tobacco use, unsafe sexual activity, dangerous driving, interper-sonal aggression and even more serious delinquent and criminalbehaviours; these behaviours have been the major focus ofresearchers and clinicians interested in the neurobiology of RTP(Boyer, 2006).

To date, the studies on impulsive risky behaviour have focussedon the neural correlates of either the cognitive control system(Eshel et al., 2007; Van Leijenhorst et al., 2006) or the reward-pro-cessing system (Ernst et al., 2005; Galvan et al., 2006; Polezzi et al.,2010; Van Leijenhorst et al., 2010). Both systems are closely related

nical Neurophysiology.

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

2 J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx

to RTP, but recently, the interaction between the two systems hasreceived more attention. For instance, it has been postulated thathigh RTP in adolescents originates from differences in the develop-ment patterns of reward- and control-related circuitry that mightlead to an imbalance in the adolescent brain (Casey et al., 2008;Ernst et al., 2006; Galvan et al., 2006; Steinberg et al., 2008). Re-cently, RTP has been investigated in the context of interactions be-tween reward-related and control-related brain regions (VanLeijenhorst et al., 2010).

Reward processing is associated with activation of the ventralmedial prefrontal cortex, orbitofrontal cortex, ventral striatumand amygdala (Bjork et al., 2004; Cohen et al., 2009; Ernst et al.,2005; Eshel et al., 2007; Galvan et al., 2006; May et al., 2004).Reward processing is essential for motivation, volition and goal-di-rected behaviour of humans. However, the dysfunctional process-ing of reward is also frequently associated with behaviouralproblems such as addiction, impulsivity and risk-taking (Paulus,2007; Steinberg, 2007). It is supposed that reward processing ismediated by more complex mechanisms associated with the func-tional interaction between various reward-related brain regions(Camara et al., 2008). For instance, the reward system recognisessome stimuli as ‘liking’ and produces pleasure and positive affect.On the contrary, in other cases, the same stimuli are accepted as‘wanting’ and produce incentive motivation. It is quite differentlyprocessed in the reward-related brain regions, and it gives us an in-sight into understanding the aetiology of addiction or eating disor-ders (Berridge and Robinson, 1995; Berridge et al., 2009; Camerer,2006; Davis et al., 2009; Finlayson et al., 2007; Litman, 2005; Rob-inson et al., 2005).

On the other hand, it has been suggested that cognitive controlprocessing is associated with the dorsal anterior cingulate cortex,the lateral (dorsal and ventral) prefrontal cortex (Eshel et al.,2007; Van Leijenhorst et al., 2006), the anterior insular cortex (Wa-ger and Barrett, 2004), the inferior frontal junction (Brass et al.,2005) and the posterior parietal cortex (Andersen et al., 1997; Con-stantinidis and Steinmetz, 1996). Some of these regions are parts ofthe frontostriatal system known as the cortico–subcortical circuit(CSC). It is generally accepted that the sub-regions of this circuitdo not function independently but rather work together towardsspecific functions such as executive control processing (Gazzanigaet al., 1998). Undoubtedly, risk-taking behaviour is based in theCSC, where interaction between the reward-related and control-re-lated brain regions begins. This idea is supported by various stud-ies that have reported increased risk-taking behaviours in subjectswith prefrontal cortex disruptions caused by, for example, trau-matic brain injury (Bechara et al., 1996), transcranial direct currentstimulation (Fecteau et al., 2007), repetitive transcranial magneticstimulation (Knoch et al., 2006) and the interference of alcohol(Lane et al., 2004).

Since Robinson (1999, 2000) originally suggested that differentEEG rhythms might reflect different cortico–subcortical projec-tions, several studies have focussed on the relationship betweenthe EEG spectra and the CSC (Honk and Schutter, 2005; Knyazevand Slobodskaya, 2003; Schutter and Honk, 2004; Schutter et al.,2006). It has been suggested that the crosstalk in the CSC manifestsin the coupling of slow and fast wave activities and that this islikely the result of descending inhibition (DI), by which the corticalsystem executes inhibitory control over the subcortical brain struc-ture (Jackson, 1958). Thus, cognitive inhibitory control in the cor-tical system possibly alters the degree of cross-frequencycoupling between slow and fast oscillations. Therefore, we hypoth-esised that an individual’s RTP would be revealed by differences inthe degree of cross-frequency coupling observed in resting scalpEEG.

The cross-frequency coupling between slow and fast oscilla-tions can allow for even more complex cortico–subcortical

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

interactions (Buzsaki and Draguhn, 2004; Steriade, 2001) and hasbeen examined by EEG (Canolty et al., 2006; Jensen and Colgin,2007; Palva et al., 2005; Schack et al., 2002; Shils et al., 1996).Especially, recent studies on the human neocortex have demon-strated that the power of fast oscillations (30–150 Hz) is modu-lated by the phase of slow oscillations (1–8 Hz) (Canolty et al.,2006; Jensen and Colgin, 2007). From a theoretical point of view,there are several ways in which the cross-frequency interactionsmight occur, such as amplitude–amplitude, phase–amplitude andphase–phase interactions. So far, the cross-frequency amplitude–amplitude coupling has been investigated by many researchersinterested in the cortico–subcortical interaction related to the hor-mone level (Miskovic and Schmidt, 2009; Schutter and van Honk,2005; van Peer et al., 2008), anxiety (Knyazev, 2011; Knyazevet al., 2005, 2006; Miskovic et al., 2010; Miskovic et al., 2011a,b),emotion (Schutter and Knyazev, 2012) and obsessive–compulsivedisorder (Velikova et al., 2010). To our knowledge, there is a lackof studies that investigate the cortico–subcortical interaction usingthe phase–amplitude interaction. Moreover, the cross-frequencyphase–amplitude coupling (CFPAC) reported by Canolty et al.(2006) is of interest among clinical psychiatrists because of its po-tential as a neurophysiological measure of cortico–subcorticalinteractions (Jensen and Colgin, 2007).

The aim of the present study was to determine whether or notthe CFPAC in resting EEGs is correlated with self-reported mea-sures of RTP. To this end, the Korean versions of the Domain-Spe-cific Risk-Taking (DOSPERT) Scale (Hong et al., 2010) and theBarrett impulsiveness scale (BIS) (Chung and Lee, 1997) were usedas behavioural measures of RTP. Studies that have investigated theintra-individual stability of resting EEGs have demonstrated thestability of resting EEG traces over a period of years (Kondacsand Szabo, 1999). Moreover, because RTPs are rather long-standingbehavioural tendencies and would also be the result of sustainedcortico–subcortical interactions, we used resting EEG when thesubject’s eyes were closed in order to capture individual disposi-tional differences in RTP.

2. Materials and methods

2.1. Subjects

A total of 50 healthy volunteers (21 females; aged33.0 ± 7.8 years; Edinburgh laterality quotient, 87.9 ± 30.8) wererecruited using an online advertisement in Gongju, South Korea.None of the participants were taking any medications and all hadcompleted at least 12 years of education (14.9 ± 1.9 years). In addi-tion, each participant signed an informed consent form before theexperiment. This study was approved by the Institutional ReviewBoard (IRB) of the Gongju National Hospital (Gongju, South Korea)and was performed in accordance with the Declaration of Helsinki(World Medical Association: Ethical Principles for Medical Re-search Involving Human Subjects, 1964).

2.2. Risk-taking and impulsivity scales

The DOSPERT Scale, a self-reported psychometric scale, allowedus to assess the conventional risk attitudes (i.e., the reported levelof risk-taking) and the perceived risk attitudes (i.e., the willingnessto engage in a risky activity as a function of its perceived riskiness).The scale comprises five commonly encountered content domains:ethical, financial, health/safety, social and recreational (Blais andWeber, 2006; Weber et al., 2002). The DOSPERT generates twoscores: one for the risk attitude, which is called the risk-takingscore, and the other for the perceived risk attitude, which is calledthe risk-perception score. These scores provide integrative

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx 3

information about the five domains described above. In general,the greatest risk takers exhibit high risk-taking scores and lowrisk-perception scores. We used the Korean version of the 30-itemversion of the DOSPERT (Hong et al., 2010). In addition, the Koreanversion of the BIS standardised by Chung and Lee (1997), a widelyused and well-validated self-report measure of RTP, was used. It in-cludes 23 items that can be scored to yield three factors such asattentional, motor and non-planning impulsiveness. High scoresindicate a greater tendency towards impulsive risk-taking in deci-sion making and various social situations.

2.3. Experimental procedures

After the subjects arrived at the laboratory at Gongju NationalHospital, they were screened for neurological or psychiatric condi-tions. The use of alcohol was carefully screened for using a digitalbreath alcohol concentration (BAC) calculator (Alcoscan AL9000,Sentech, Seoul, South Korea) and self-report questionnaires. Thosesubjects with 0% BAC and a history of consuming fewer than fivealcoholic beverages per week were allowed to participate in theexperiment. First, each subject was seated in a comfortable chairin a dimly lit room and instructed to relax and keep movementsto a minimum. The resting EEG was recorded over 3 min withthe subject’s eyes closed. Next, the subjects filled out the DOSPERTand BIS self-report questionnaires. The subjects were debriefed atthe end of the experiment.

2.4. EEG recording and pre-processing

The EEG recording was conducted from the scalp using a Syn-Amps2 direct current (DC) amplifier and a 10/20 layout 64-channelQuik-Cap electrode placement system (Neuroscan Inc., NC, USA).The EEGs were recorded from 19 electrode sites (Fp1, Fp2, F7, F3,Fz, F4, F8, T7, C3, Cz, C4, T8, P7, P3, Pz, P4, P8, O1 and O2) basedon a standard international 10/20 system at a sampling rate of1000 Hz. We used the linked mastoid for reference and two addi-tional bipolar electrodes to measure the horizontal and verticaleye movements. The impedance of each electrode was kept below10 kO throughout the EEG recording session.

We used Matlab 7.0.1 (MathWorks, Natick, MA, USA) and theEEGLAB toolbox (Delorme and Makeig, 2004) to pre-process andanalyse the EEG recordings. First, the EEG data were downsampledto 250 Hz. Next, the EEG data were detrended and mean-sub-tracted to remove the DC component. A 1-Hz high-pass filter anda 60-Hz notch filter were applied to remove the eye and electricitynoise. Independent component analysis (ICA) was also performedto eliminate eye-blink and muscle artefacts. For the analysis,>2 min of artefact-free EEG data were selected from the 3-minrecording of each subject based on visual inspection by clinicalpsychiatrists and EEG experts.

2.5. Power-spectrum analysis of the EEG

Seven frequency bands were defined for further analyses: delta(1–4 Hz), theta (4–8 Hz), slow alpha (8–10 Hz), fast alpha(10–13.5 Hz), beta (13.5–30 Hz), slow gamma (30–58 Hz) and fastgamma (62–80 Hz). For the cross-frequency coupling analyses, theslow gamma and fast gamma frequencies were divided into fivesub-bands: 35-Hz gamma (30–39 Hz), 45-Hz gamma (40–49 Hz),55-Hz gamma (50–58 Hz), 65-Hz gamma (62–69 Hz) and 75-Hzgamma (70–80 Hz). We investigated the power spectra of theEEG data for each subject using the short-time Fourier transform‘spectrogram.m’ function of the signal processing toolbox in Mat-lab. Time windows of 1000 ms with an 800-ms overlap and theHamming window were used for the spectral analysis. The outliersthat were far from the spectral value distribution of each frequency

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

band at the significance level 0.05 were removed. Finally, the abso-lute powers were averaged over all of the time windows and fre-quency bands for further analysis.

2.6. CFPAC analysis

We employed the synchronisation index (SI) proposed by Cohen(2008) to assess the cross-frequency interactions between the low-frequency (1–8 Hz) phase and the power of the gamma (30–80 Hz)oscillations. Briefly, the SI is a phase coherence measurement be-tween the phase of the upper (gamma) power time series andthe phase of the lower time series (synchronisation index:SI ¼ 1

n

Pnt¼0ei½ult �uut�; n, the number of time points; uut, the

phase of the upper frequency power time series at time point t;and ult, the phase of the lower frequency time series at time pointt) (Cohen, 2008). To avoid distortions of the phase value during fil-tering, we used the two-way, least-squared finite impulse responsefilter (eegfilt.m) included in the EEGLAB toolbox (Delorme andMakeig, 2004). In addition, 1000-ms time windows with an 800-ms overlap and 3-Hz windowing for each of the five gamma sub-frequencies were used for the subjects. The gamma power timeseries was extracted as the squared magnitude of fðtÞ, which isthe analytic signal obtained from the Hilbert transform (power

time series: pðtÞ ¼ real½fðtÞ�2 þ imag½fðtÞ�2). The phases of the twotime series were extracted from the Hilbert transform

(phase ¼ arctan imag½fðtÞ�real½fðtÞ�

� �). The SI value is a complex number, and

its magnitude (SIm) reflects the extent to which the phases are syn-chronised (0 = completely desynchronised; 1 = perfectly synchron-ised). The outliers were also removed in the same way as describedfor the spectral analysis. Finally, the SIm values were averaged overall of the time windows and each of the five gamma sub-frequencywindows for further analyses. We investigated the two differentCFPACs: delta-phase gamma-power coupling (DGC) and the the-ta-phase gamma-power coupling (TGC).

2.7. Statistical analysis

The Matlab 7.0.1 statistical toolbox was used for the statisticalanalyses. All values of the RTP measures (the DOSPERT and BIS)were expressed as the mean and the standard deviation (SD). Stu-dent’s t-tests were used to test for gender differences in the age,education and the RTP measures (the DOSPERT and BIS). Statisticalsignificance was defined as p < 0.05. We used a correlation analysisto compare the linear relationship between the RTP measures andthe EEG data. We adjusted for gender, age and education using thePearson’s partial correlation method. The partial correlation coeffi-cients and the associated p-values were determined. To improveclarity, the topographical plots of the partial correlation coeffi-cients and p-values were used to represent the correlation results.

3. Results

To assess the characteristics of impulsive, risk-taking behav-iours in the subjects, we measured the DOSPERT and BIS scores.The demographic features and the findings are presented in Table1. The Cronbach’s alphas for the DOSPERT and BIS are 0.7186 and0.7827, respectively. To determine whether the risk-taking behav-iours were gender-dependent or not, we compared the values ofthe DOSPERT and BIS between the male and female subjects. Thestatistical analyses of the gender differences in the DOSPERT andBIS values revealed no significant differences between the groupsexcept in the DOSPERT risk-taking scores. The risk-taking DOSPERTscores were significantly higher for the male subjects than the fe-male subjects (t(48) = 2.01, p < 0.01). To assess the possible rela-tionship with the absolute powers of resting EEGs, we estimated

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

Table 1The characteristics of the demographic data and the measures of risk-takingpropensity in the subjects.

Mean ± S.D. Total(n = 50)

Male(n = 29)

Female(n = 21)

Age (years) 33.0 ± 7.8 31.8 ± 7.4 34.8 ± 8.2Education (years) 14.9 ± 1.9 15.0 ± 1.8 14.9 ± 2.1Edinburgh laterality quotient 91.4 ± 17.6 94.9 ± 12.5 86.5 ± 22.2BIS score 26.0 ± 7.0 26.1 ± 8.2 25.9 ± 5.1The DOSPERT risk-perception

score102.1 ± 12.1 100.7 ± 11.0 104.0 ± 13.4

The DOSPERT risk-taking score 72.0 ± 15.9 78.1 ± 16.1* 63.7 ± 11.3*

S.D., Standard deviation; BIS, Barrett impulsivity scale; DOSPERT, domain-specificrisk-taking scale.* p-Value < 0.05.

4 J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx

the partial correlations, corrected for gender, age and education,between the DOSPERT and BIS values and the 12 power spectra(seven frequency bands + five gamma sub-bands) of the restingEEGs from the subjects. The topographical representations of each

Fig. 1. Topographical representations of the Pearson’s partial correlations, corrected forscores. On the left is the topography of the Pearson’s partial correlation coefficients,topography pair corresponds to the color bar in the two enlarged samples above. r, Pearsscale; BIS, Barrett impulsiveness scale.

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

of the Pearson’s partial correlation coefficients and the associatedp-value are presented in Fig. 1.

From the results of partial correlation analyses, we found that,as a whole, the power spectra of EEGs were positively correlatedwith BIS scores at central and occipital regions and with DOSPERTrisk-taking scores at parietal and occipital regions. However, in thecase of DOSPERT risk-perception scores, the negative correlationsdominated at central and occipital regions. The partial correlationanalyses indicated significant positive partial correlations betweenthe Cz electrode and the BIS scores (45 Hz) and negative partialcorrelations between the central (C4 and Fz) and parieto–occipital(P8) regions and the DOSPERT risk-perception scores. However, nosignificant partial correlations were observed between the DOS-PERT risk-taking scores and the power spectra. The significant sta-tistical results are presented in Table 2.

To investigate whether RTP correlates with the CFPAC of restingEEGs, we also examined the partial correlations corrected for gen-der, age and education between the DOSPERT and BIS values andthe two CFPACs (DGC and TGC) of the resting EEGs in the subjects.

gender, age, and education, between the absolute powers and the BIS and DOSPERTand on the right is the topography of the associated p-values. The scale for eachon’s partial correlation coefficient; p, p-value; DOSPERT, domain-specific risk-taking

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

Table 2The electrodes show the significant (p < 0.05) partial correlation (adjusted for gender,age and education) between the power spectrum and the measures of risk-takingpropensity.

Risk-takingmeasure

Electrode location(type of powerspectra)

Pearson’s partialcorrelationcoefficient

p-Value

BIS score Cz (45 Hz) 0.319 0.027

DOSPERTrisk-perception score

C4 (delta) �0.297 0.040P8 (delta) �0.332 0.021Fz (slow gamma) �0.302 0.037C4 (fast gamma) �0.316 0.028Fz (35 Hz) �0.292 0.044Fz (45 Hz) �0.288 0.047C4 (65 Hz) �0.302 0.037C4 (75 Hz) �0.326 0.024

BIS, Barrett impulsivity scale; DOSPERT, domain-specific risk-taking scale.

Table 3The electrodes show the significant (p < 0.05) partial correlation (adjusted for gender,age and education) between the CFPACs and the measures of risk-taking propensity.

Risk-takingmeasure

Electrode location (type ofCFPAC and gamma band)

Pearson’spartialcorrelationcoefficient

p-Value

BIS score F8 (TGC 35 Hz) �0.340 0.0183T7 (TGC 35 Hz) �0.312 0.0310Fp2 (DGC 75 Hz) �0.391 0.0059*

F3 (DGC 75 Hz) �0.312 0.0307T8 (DGC 75 Hz) 0.298 0.0394P4 (DGC 75 Hz) �0.461 0.0009*

C4 (TGC 75 Hz) �0.387 0.0065*

DOSPERT risk-perceptionscore

C4 (DGC 35 Hz) 0.304 0.0356F3 (TGC 45 Hz) 0.298 0.0396P8 (TGC 45 Hz) 0.312 0.0307P8 (TGC 55 Hz) 0.294 0.0422P4 (TGC 65 Hz) 0.346 0.0161F8 (TGC 75 Hz) 0.294 0.0428

DOSPERT risk-taking score

F7 (DGC 35 Hz) �0.289 0.0463C3 (DGC 45 Hz) �0.376 0.0084*

T8 (DGC 45 Hz) 0.315 0.0289P8 (DGC 45 Hz) 0.355 0.0132Fp1 (DGC 55 Hz) �0.297 0.0403O1 (DGC 55 Hz) �0.263 0.0184Pz (DGC 65 Hz) �0.307 0.0340T8 (DGC 75 Hz) 0.346 0.0159

CFPAC, cross-frequency phase amplitude coupling; BIS, Barrett impulsivity scale;DOSPERT, domain-specific risk-taking scale; TGC, theta–gamma coupling; DGC,delta–gamma coupling.* p-Value <0.01.

J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx 5

Fig. 2 presents the topographical representations of each of thePearson’s partial correlation coefficients and the associated p-val-ues between the measures of RTP and the CFPAC on resting EEGs.

We found that the BIS score was negatively correlated with DGCand TGC. The significant findings were revealed mainly in DGC(75 Hz) and TGC (35 and 75 Hz) in frontal (Fp2, F8 and F3), central(C4) and temporo–parietal (T7 and P4) regions. However, the posi-tive partial correlation between the BIS score and DGC (75 Hz) wasfound in the right temporal (T8) region. Regarding the DOSPERTscore, positive partial correlations were found in the frontal (F3and F8) and centro–parietal (C4, P4 and P8) regions between therisk-perception score and CFPACs. In addition, negative partial cor-relations were observed in the frontal (Fp1 and F7), central (C3),temporo–parietal (T8, P8 and Pz) and occipital (O1) regionsbetween DGC and the risk-taking score. Further, the significant po-sitive partial correlations were found in the right temporal (T8 andP8) regions between DGC (45 Hz) and the risk-taking score. The

Fig. 2. Topographical representations of the Pearson’s partial correlations, corrected for gBIS and DOSPERT scores. On the left is the topography of the Pearson’s partial correlationof the topography pairs corresponds to the color bar enlarged in the two center examplrisk-taking scale; BIS, Barrett impulsiveness scale.

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

significant statistical results are presented in Table 3. The scatterplots of the four significant (p < 0.01) partial correlations are pre-sented in Fig. 3.

ender, age, and education, between the cross-frequency couplings (CFPACs) and thecoefficients, and on the right is the topography of the associated p-values. The scale

es. r, Pearson’s partial correlation coefficient; p, p-value; DOSPERT, domain-specific

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

Fig. 3. Scatter plots of the Pearson’s partial correlations corrected for gender, age, and education between the mean of the SIm values with the BIS and DOSPERT measures. (a)The partial correlation of the SIm DGC values (45 Hz) and the risk-taking DOSPERT score at C3. (b) The partial correlation of the SIm DGC values (75 Hz) with the BIS scores atFp2. (c) The partial correlation of the SIm DGC values (75 Hz) with the BIS scores at P4. (d) The partial correlation of the SIm TGC values (75 Hz) with the BIS scores at C4. SIm,magnitude of the synchronization index; r, Pearson’s partial correlation coefficient; p, p-value; DOSPERT, domain-specific risk-taking scale; BIS, Barrett impulsiveness scale;TGC, theta-phase gamma-amplitude coupling; DGC, delta-phase gamma-amplitude coupling.

6 J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx

4. Discussion

In the current study, we have demonstrated that high RTP as-sessed by the DOSPERT Scale and BIS is closely associated withlow degrees of CFPACs in the EEG in the frontal (Fp2) and cen-tro–parietal (C4, P4 and C3) regions. By contrast, the high degreesof CFPACs in the right temporal and parietal lobe (T8 and P8) arerelated to high risk-taking scores. Otherwise, comparing the meth-ods for EEG analysis, the spectral analyses had a weakened powerto demonstrate significant differences between the absolute pow-ers and the measures of RTP. This finding implicates that frequentrisk-taking behaviour is reflected in the EEG potentially measuringthe reduced interference of the cortical control network in the sub-cortical regions. We suggest that CFPAC of the EEG, which likely re-flects the degree of interactions among functional systems, can be auseful tool for quantifying individual’s risk-taking propensity.

Clearly, unlike spectral analysis, the CFPAC mainly uses thephase information from EEG oscillations. The conceptual frame-work for the difference between the phase and amplitude of EEGoscillations has been outlined in detail by Klimesch et al. (2007).The idea is that the phase represents the timing of neuronalactivity, whereas the amplitude indicates the extent of task

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

involvement of the relevant neurons in the ongoing EEG (Klimeschet al., 2007). The discrepancy in our findings between the CFPACand spectral analysis might come from this dissimilar point of viewabout the method for neuronal activity. Another intriguing differ-ence between the two methods is that the correlation coefficientsof the CFPAC exhibit more local alterations than the spectral anal-ysis in our topographic findings. In the case of DGC (75 Hz) of theBIS as shown in Fig. 2, the topographic patterns of the partialcorrelation coefficient vary considerably. This is likely becausethe CFPAC might serve as a mechanism to transfer informationfrom large-scale brain networks to the local cortical processing re-gions (Jensen and Colgin, 2007). Taken together, these findingssuggest that the degree of CFPAC may differ across various localbrain areas. Furthermore, the CFPAC measures the amount of tim-ing information from the interacting functional systems acrossmultiple spatiotemporal scales, unlike the spectral results, whichdescribe the amount of excitation of the functional neuronal sys-tem. Recently, our studies using the CFPAC of scalp EEG haveshown the similar characteristics of CFPAC in different tasks (Parket al., 2013, 2011).

Our results indicate that the topographic features of the corre-lation vary across the RTP measures. This variation possibly results

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx 7

from dissimilar properties in each RTP measure that might focus ondifferent neuronal systems involved in RTP. The assessment of RTPis limited, and its degrees are varied depending on the type of mea-sures. All of our measures assess different aspects of RTP (Boyer,2006; Reynolds et al., 2006). The DOSPERT reflects the social as-pects of RTP (Weber et al., 2002). The scale directly measuresRTP across a number of everyday, real-life situations, such asinvesting in stocks, buying a lottery ticket, bungee jumping, engag-ing in unsafe sex and drunk driving (Harrison et al., 2005). Further-more, previous work has shown that the risk-taking DOSPERTscores are negatively correlated with the risk-perception DOSPERTscores. Unlike DOSPERT, the BIS reflects the emotional and affec-tive aspects of RTP, including aggression, depression and impulsiv-ity. However, a number of previous studies have demonstratedmeaningful correlations among these measures of RTP (Honget al., 2010; Lejuez et al., 2002).

As expected, the DOSPERT risk-taking and BIS measurementswere negatively correlated with CFPACs, mainly in the frontaland centro–parietal regions. These findings suggest that increasedinteractions between the functional neuronal systems in the fron-tal and centro–parietal regions would be associated with the ten-dency to avoid risk. This result may be related to the top-downinhibitory control mechanisms underlying decision-making pro-cesses that address risky environments and situations. This couldbe evidence for DI (i.e., higher systems inhibiting the lower), whichwas proposed by Jackson (1958). A number of studies have demon-strated the key role of the prefrontal and frontal cortex in inhibi-tory control of behaviour, reward processing and decisionmaking (Badre, 2008; Brass et al., 2005; Byrnes, 1998; Cole andSchneider, 2007; Logan, 1994; Reuter and Kathmann, 2004; VanLeijenhorst et al., 2006, 2010). Recently, Cole and Schneider(2007) suggested that a set of brain regions, including the dorsolat-eral prefrontal and posterior parietal cortex, forms a cognitive con-trol network of anatomically distinct component-processing brainregions that interact in a tightly coupled fashion to implement cog-nitive control of a variety of tasks (Cole and Schneider, 2007). Inaddition, the association between our findings and DI is supportedby the observation that patients with traumatic brain injuries orother pathologies affecting the PFC show a tendency towards riskyand impulsive decision making and an apparent disregard for thenegative consequences of their actions (Bechara et al., 1996; Rah-man et al., 2001). Moreover, this observation appears to be partic-ularly true for patients with right-sided lesions (Clark et al., 2003;Tranel et al., 2002). We found meaningful negative correlations be-tween the CFPAC and the BIS measurements in the right frontal(Fp2) region. The importance of the right frontal lobe in inhibitorycontrol has been well demonstrated experimentally in previousworks on repetitive transcranial magnetic stimulation and trans-cranial direct current stimulation-induced functional disinhibitionof the right prefrontal cortex (Fecteau et al., 2007; Knoch et al.,2006).

On the one hand, we found significant positive correlations be-tween the CFPAC and the measures of RTP in the right temporal(T8) and parietal (P8) lobes. From the perspectives of the brain–behaviour relationship for impulsivity and risk-taking, violentand impulsive behaviour could be categorised as ‘impulsive–emo-tional’ or ‘controlled–instrumental’ (Vitiello and Stoff, 1997).Impulsive–emotional behaviour occurs suddenly in reaction tothreat within the context of increased impulsivity and emotional-ity. Controlled–instrumental behaviour, on the other hand, mani-fests itself as a relatively non-emotional display of aggressiondirected at obtaining the goal. According to the result from theneuropsychological studies for impulsivity, it appears that impul-sive–emotional behaviour is closely associated with the greaterright-hemisphere activation, particularly in temporal regions andconnections to limbic and hypothalamic structures, which are

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

important in the expression and regulation of emotion (Knyazevet al., 2002; Raine, 2002; Scarpa and Raine, 2000). A previous studyfor the psychological differences between left and right temporallobe epilepsy has shown that the patients with right temporal lobeepilepsy were more impulsive and more externalised aggressiveresponses than those with left temporal lobe epilepsy (McIntyreet al., 1976). The patient with temporal lobe epilepsy involuntarilyfeels the unexpected phenomena, called ‘experiential phenomena’,of the nonspecific irritation from seizure discharges such as per-ceptual hallucinations or illusions, memory flashbacks, forced thin-kings or emotions (Gloor et al., 1982). It is suggested that theexaggeration of the emotional function of the right temporal loberesults in affective instability and emotional expressiveness (McIn-tyre et al., 1976). Therefore, it is supposed that the positive corre-lations between the CFPAC and the measures of RTP found in theright temporal (T8) and parietal (P8) regions are closely relatedto the overexpression of the impulsive–emotional function of theright hemisphere.

So far, many studies for RTP have investigated the neural sub-strate of risk-taking during the performance of a decision-makingtask that might have directly tapped into some of the neurocogni-tive operations presumed to be involved in mediating risk-takingbehaviours in real-world situations. However, the result of thepresent study comes from the analysis of EEG for the restingeyes-closed condition. How can the resting EEG effectively expressthe neural information for RTP? In the present study, there wereseveral reasons for using the resting EEG for examining RTP. First,as mentioned in the Introduction section, the RTP for a long-stand-ing trait is explained by the individual state of equilibrium or bal-ance between the reward-related and cognitive control-relatedfunctional networks of the brain (Casey et al., 2008; Steinberg,2007). From the result of a recent study using functional magneticresonance imaging (MRI), the highest global brain connectivity wasfound in both the cognitive control network (Cole and Schneider,2007) and the default mode brain network, the so-called restingstate brain network (Cole et al., 2010). Further, it is thought thatthe quantitative EEG (QEEG) for the resting state often providesvaluable information about the balance between large-scale func-tional brain networks such as the cognitive control network (Can-tor, 1999; Prichep and John, 1992). Consequently, it is supposedthat two different types of EEG (resting and during task) mightbe evaluating different aspects of RTP, so that the risk-taking deci-sion task mainly assesses reward processing, whereas the restingEEG mainly reflects the cognitive control ability. Second, the re-search findings for the default mode brain network have shownthat the resting state is temporally anti-correlated with the taskstate (Broyd et al., 2009; Cole et al., 2010). Moreover, a close asso-ciation between task performance and the strength of anti-correla-tion has been suggested (Fox et al., 2005). In essence, a healthyresting state promotes better task performance. It is suggested thatthe resting state might reflect the performance of the risk-takingdecision task to some degree. Finally, considering the clinicalimplication, it is more beneficial to use ‘resting’ than ‘during task’in terms of applicability, time and simplicity.

The most significant limitation of the present study is that wecould not control the false positives arising from multiple compar-isons. To our knowledge, there is no previous study regarding thedifferences in topographical pattern between spectral analysisand CFPAC. We thought that it is meaningful to compare the de-tailed topographical pattern and characteristic of various frequen-cies between two different analyses (spectral and cross-frequencycoupling) for the resting EEG. Hence, the design of the analysis ofthe present study was complex and involved many comparisons.However, this is also the study’s weak point. Therefore, the resultsshould be interpreted with caution as preliminary findings and re-quire further replication.

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

8 J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx

Additionally, we could not exclude the possible confoundingfactors of personality traits, intelligence and psychological symp-toms such as anxiety and depression. Many previous studies havefocussed on the relationships among these factors, and specific fre-quency oscillations have been frequently reported as having a closerelationship with various factors, such as personality traits, intelli-gence and symptoms of anxiety and depression (Chi et al., 2005;Lutzenberger et al., 1992; Thatcher et al., 2005; Thibodeau et al.,2006). In addition, various scales and tasks have been presentedto measure RTP and impulsivity recently. It is our weak point thatwe used the limited type of scales or tasks for measuring RTP.

Although the CFPAC is not specific to RTP, and thus should beinterpreted with caution, the present study suggests that theCFPAC in resting EEGs can be used as a potential novel neurophysi-ological indicator of RTP. We suggest that the CFPAC is likely to beuseful for assessing RTP in individuals who have social, criminal orhealth problems and patients with neurological and psychiatricdiseases who are driven to take excessive risks that lead to nega-tive consequences.

Acknowledgments

The authors thank Sujin Kim, Kyung Hee Kim, So Yul Kim,Yul-mai Song and Young Sung Kim (Neuropsychiatry Research Lab-oratory, Gongju National Hospital, Chungcheongnam-do, SouthKorea) for their valuable help with this study. This work was sup-ported by the Chung Moon Soul Research Center for Bio Informa-tion and Bio Electronics (CMSC) in KAIST and a Korea Science andEngineering Foundation (KOSEF) Grant provided by the Koreangovernment (MOST) (No. R01-2007-000-21094-0 and No.M10644000028-06N4400-02810; No. 20090093897 and No. 20090083561).

References

Andersen RA, Snyder LH, Bradley DC, Xing J. Multimodal representation of space inthe posterior parietal cortex and its use in planning movements. Annu RevNeurosci 1997;20:303–30.

Badre D. Cognitive control, hierarchy, and the rostro-caudal organization of thefrontal lobes. Trends Cogn Sci 2008;12:193–200.

Bechara A, Tranel D, Damasio H, Damasio A. Failure to respond autonomically toanticipated future outcomes following damage to prefrontal cortex. CerebCortex 1996;6:215.

Berridge KC, Robinson TE. The mind of an addicted brain: neural sensitization ofwanting versus liking. Curr Dir Psychol Sci 1995;4:71–6.

Berridge KC, Robinson TE, Aldridge JW. Dissecting components of reward:‘liking’,‘wanting’, and learning. Curr Opin Pharmacol 2009;9:65–73.

Bjork J, Knutson B, Fong G, Caggiano D, Bennett S, Hommer D. Incentive-elicitedbrain activation in adolescents: similarities and differences from young adults. JNeurosci 2004;24:1793.

Blais A, Weber E. A domain-specific risk-taking (DOSPERT) scale for adultpopulations. Judgement Decis Making 2006;1.

Boyer T. The development of risk-taking: a multi-perspective review. Dev Rev2006;26:291–345.

Brass M, Derrfuss J, Forstmann B, Cramon D. The role of the inferior frontal junctionarea in cognitive control. Trends Cogn Sci 2005;9:314–6.

Broyd SJ, Demanuele C, Debener S, Helps SK, James CJ, Sonuga-Barke EJ. Default-mode brain dysfunction in mental disorders: a systematic review. NeurosciBiobehav Rev 2009;33:279.

Buzsaki G, Draguhn A. Neuronal oscillations in cortical networks. Science2004;304:1926.

Byrnes J. The nature and development of decision making: a self-regulationmodel. Lawrence Erlbaum; 1998.

Camara E, Rodriguez-Fornells A, Münte TF. Functional connectivity of rewardprocessing in the brain. Front Hum Neurosci 2008;2.

Camerer CF. Wanting, liking, and learning: neuroscience and paternalism. Univ ChicLaw Rev 2006;73:87.

Canolty R, Edwards E, Dalal S, Soltani M, Nagarajan S, Kirsch H, et al. High gammapower is phase-locked to theta oscillations in human neocortex. Science2006;313:1626.

Cantor DS. An overview of quantitative EEG and its applications toneurofeedback. Academic Press; 1999.

Casey B, Jones R, Hare T. The adolescent brain. Ann NY Acad Sci 2008;1124:111–26.

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

Chi SE, Park CB, Lim SL, Park EH, Lee YH, Lee KH, et al. EEG and personalitydimensions: a consideration based on the brain oscillatory systems. Pers IndivDiff 2005;39:669–81.

Chung Y, Lee C. A study of factor structures of the Barratt impulsiveness scale inKorean university students. J Korean Assoc Clin Psychol 1997;16:117–29.

Clark L, Manes F, Antoun N, Sahakian BJ, Robbins TW. The contributions of lesionlaterality and lesion volume to decision-making impairment following frontallobe damage. Neuropsychologia 2003;41:1474–83.

Cohen M. Assessing transient cross-frequency coupling in EEG data. J NeurosciMethods 2008;168:494–9.

Cohen M, Axmacher N, Lenartz D, Elger C, Sturm V, Schlaepfer T. Good vibrations:cross-frequency coupling in the human nucleus accumbens during rewardprocessing. J Cogn Neurosci 2009;21:875–89.

Cole M, Schneider W. The cognitive control network: integrated cortical regionswith dissociable functions. Neuroimage 2007;37:343–60.

Cole MW, Pathak S, Schneider W. Identifying the brain’s most globally connectedregions. Neuroimage 2010;49:3132.

Constantinidis C, Steinmetz MA. Neuronal activity in posterior parietal area 7a duringthe delay periods of a spatial memory task. J Neurophysiol 1996;76:1352–5.

Davis CA, Levitan RD, Reid C, Carter JC, Kaplan AS, Patte KA, et al. Dopamine for‘‘wanting’’ and opioids for ‘‘liking’’: a comparison of obese adults with andwithout binge eating. Obes 2009;17:1220–5.

Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trialEEG dynamics including independent component analysis. J Neurosci Methods2004;134:9–21.

Ernst M, Nelson E, Jazbec S, McClure E, Monk C, Leibenluft E, et al. Amygdala andnucleus accumbens in responses to receipt and omission of gains in adults andadolescents. Neuroimage 2005;25:1279–91.

Ernst M, Pine D, Hardin M. Triadic model of the neurobiology of motivated behaviorin adolescence. Psychol Med 2006;36:299–312.

Eshel N, Nelson E, Blair R, Pine D, Ernst M. Neural substrates of choice selection inadults and adolescents: development of the ventrolateral prefrontal andanterior cingulate cortices. Neuropsychologia 2007;45:1270–9.

Fecteau S, Knoch D, Fregni F, Sultani N, Boggio P, Pascual-Leone A. Diminishing risk-taking behavior by modulating activity in the prefrontal cortex: a direct currentstimulation study. J Neurosci 2007;27:12500.

Finlayson G, King NA, Blundell JE. Liking vs. wanting food: importance for humanappetite control and weight regulation. Neurosci Biobehav Rev 2007;31:987–1002.

Fox MD, Snyder AZ, Vincent JL, Corbetta M, Van Essen DC, Raichle ME. The humanbrain is intrinsically organized into dynamic, anticorrelated functionalnetworks. Proc Natl Acad Sci USA 2005;102:9673–8.

Galvan A, Hare T, Parra C, Penn J, Voss H, Glover G, et al. Earlier development of theaccumbens relative to orbitofrontal cortex might underlie risk-taking behaviorin adolescents. J Neurosci 2006;26:6885.

Garon N, Moore C. Complex decision-making in early childhood. Brain Cogn2004;55:158–70.

Gazzaniga M, Ivry R, Mangun G. Fundamentals of cognitive neuroscience. NY,USA: WW Norton; 1998.

Gloor P, Olivier A, Quesney LF, Andermann F, Horowitz S. The role of the limbicsystem in experiential phenomena of temporal lobe epilepsy. Ann Neurol1982;12:129–44.

Halpern-Felsher B, Cauffman E. Costs and benefits of a decision: decision-makingcompetence in adolescents and adults. J Appl Dev Psychol 2001;22:257–73.

Harrison J, Young J, Butow P, Salkeld G, Solomon M. Is it worth the risk? Asystematic review of instruments that measure risk propensity for use in thehealth setting. Soc Sci Med 2005;60:1385–96.

Honk E, Schutter D. Salivary cortisol levels and the coupling of midfrontal delta–beta oscillations. Int J Psychophysiol 2005;55:127–9.

Jackson J. Evolution and dissolution of the nervous system. In: Taylor J, editor.Selected writings of John Hughlings Jackson. New York: Basic Books; 1958[Reprinted].

Jensen O, Colgin L. Cross-frequency coupling between neuronal oscillations. TrendsCogn Sci 2007;11:267–9.

Hong Jong In, Kim Kyoung Hee, Park Jin Young, Park Sung Hyuk, Lee Gunsuk, LeeJaewon. Risk-taking propensity of Korean people that were measured by BARTand K-DOSPERT. J Korean Assoc Soc Psychiatry 2010;15:3–14.

Klimesch W, Sauseng P, Hanslmayr S, Gruber W, Freunberger R. Event-related phasereorganization may explain evoked neural dynamics. Neurosci Biobehav Rev2007;31:1003–16.

Knoch D, Gianotti L, Pascual-Leone A, Treyer V, Regard M, Hohmann M, et al.Disruption of right prefrontal cortex by low-frequency repetitive transcranialmagnetic stimulation induces risk-taking behavior. J Neurosci 2006;26:6469.

Knyazev G, Slobodskaya H. Personality trait of behavioral inhibition is associatedwith oscillatory systems reciprocal relationships. Int J Psychophysiol2003;48:247–61.

Knyazev GG. Cross-frequency coupling of brain oscillations: an impact of stateanxiety. Int J Psychophysiol 2011;80:236–45.

Knyazev GG, Savostyanov AN, Levin EA. Uncertainty, anxiety, and brain oscillations.Neurosci Lett 2005;387:121–5.

Knyazev GG, Schutter DJ, van Honk J. Anxious apprehension increases coupling ofdelta and beta oscillations. Int J Psychophysiol 2006;61:283–7.

Knyazev GG, Slobodskaya HR, Wilson GD. Psychophysiological correlates ofbehavioural inhibition and activation. Pers Indiv Diff 2002;33:647–60.

Kondacs A, Szabo M. Long-term intra-individual variability of the background EEGin normals. Electroencephalogr Clin Neurophysiol 1999;110:1708–16.

ensity with cross-frequency phase–amplitude coupling in the resting EEG.

J. Lee, J. Jeong / Clinical Neurophysiology xxx (2013) xxx–xxx 9

Lane S, Cherek D, Pietras C, Tcheremissine O. Alcohol effects on human risk taking.Psychopharmacol 2004;172:68–77.

Lejuez C, Read JP, Kahler CW, Richards JB, Ramsey SE, Stuart GL, et al. Evaluation of abehavioral measure of risk taking: the balloon analogue risk task (BART). J ExpPsychol Appl 2002;8:75.

Litman J. Curiosity and the pleasures of learning: wanting and liking newinformation. Cogn Emotion 2005;19:793–814.

Logan GD. On the ability to inhibit thought and action: a users’ guide to the stopsignal paradigm. Academic Press; 1994.

Lutzenberger W, Birbaumer N, Flor H, Rockstroh B, Elbert T. Dimensional analysis ofthe human EEG and intelligence. Neurosci Lett 1992;143:10–4.

Mann L, Harmoni R, Power C. Adolescent decision-making: the development ofcompetence. J Adolesc 1989;12:265–78.

May J, Delgado M, Dahl R, Stenger V, Ryan N, Fiez J, et al. Event-related functionalmagnetic resonance imaging of reward-related brain circuitry in children andadolescents. Biol Psychiatry 2004;55:359–66.

McIntyre M, Pritchard III P, Lombroso C. Left and right temporal lobe epileptics: acontrolled investigation of some psychological differences. Epilepsia1976;17:377–86.

Miskovic V, Ashbaugh AR, Santesso DL, McCabe RE, Antony MM, Schmidt LA. Frontalbrain oscillations and social anxiety: a cross-frequency spectral analysis duringbaseline and speech anticipation. Biol Psychiatry 2010;83:125–32.

Miskovic V, Campbell MJ, Santesso DL, Van Ameringen M, Mancini CL, Schmidt LA.Frontal brain oscillatory coupling in children of parents with social phobia: apilot study. J Neuropsychiatry Clin Neurosci 2011a;23:111–4.

Miskovic V, Moscovitch DA, Santesso DL, McCabe RE, Antony MM, Schmidt LA.Changes in EEG cross-frequency coupling during cognitive behavioral therapyfor social anxiety disorder. Curr Dir Psychol Sci 2011b;22:507–16.

Miskovic V, Schmidt LA. Frontal brain oscillatory coupling among men who vary insalivary testosterone levels. Neurosci Lett 2009;464:239–42.

Palva J, Palva S, Kaila K. Phase synchrony among neuronal oscillations in the humancortex. J Neurosci 2005;25:3962.

Park JY, Jhung K, Lee J, An SK. Theta-gamma coupling during a working memory taskas compared to a simple vigilance task. Neurosci Lett 2013;532:39–43.

Park JY, Lee YR, Lee J. The relationship between theta–gamma coupling and spatialmemory ability in older adults. Neurosci Lett 2011;498:37–41.

Paulus MP. Decision-making dysfunctions in psychiatry – altered homeostaticprocessing? Science 2007;318:602–6.

Polezzi D, Sartori G, Rumiati R, Vidotto G, Daum I. Brain correlates of risky decision-making. Neuroimage 2010;49:1886–94.

Prichep LS, John E. QEEG profiles of psychiatric disorders. Brain Topogr1992;4:249–57.

Rahman S, Sahakian BJ, Cardinal RN, Rogers RD, Robbins TW. Decision making andneuropsychiatry. Trends Cogn Sci 2001;5:271–7.

Raine A. Biosocial studies of antisocial and violent behavior in children and adults: areview. J Abnorm Child Psychol 2002;30:311–26.

Reuter B, Kathmann N. Using saccade tasks as a tool to analyze executivedysfunctions in schizophrenia. Acta Psychol 2004;115:255–69.

Reynolds B, Ortengren A, Richards J, de Wit H. Dimensions of impulsive behavior:personality and behavioral measures. Pers Indiv Diff 2006;40:305–15.

Robinson D. The technical, neurological, and psychological significance of ‘alpha’,‘delta’ and ‘theta’ waves confounded in EEG evoked potentials: a study of peaklatencies. Electroencephalogr Clin Neurophysiol 1999;110:1427–34.

Robinson D. The technical, neurological, and psychological significance of ‘alpha’,‘delta’ and ‘theta’ waves confounded in EEG evoked potentials: a study of peakamplitudes. Pers Indiv Diff 2000;28:673–93.

Please cite this article in press as: Lee J, Jeong J. Correlation of risk-taking propClin Neurophysiol (2013), http://dx.doi.org/10.1016/j.clinph.2013.05.007

Robinson S, Sandstrom SM, Denenberg VH, Palmiter RD. Distinguishing whetherdopamine regulates liking, wanting, and/or learning about rewards. BehavNeurosci 2005;119:5.

Scarpa A, Raine A. Violence associated with anger and impulsivity. NeuropsycholEmotion 2000:320–39.

Schack B, Vath N, Petsche H, Geissler H, Moller E. Phase-coupling of theta-gammaEEG rhythms during short-term memory processing. Int J Psychophysiol2002;44:143–63.

Schutter D, Honk J. Decoupling of midfrontal delta–beta oscillations aftertestosterone administration. Int J Psychophysiol 2004;53:71–3.

Schutter D, Leitner C, Kenemans J, Honk J. Electrophysiological correlates of cortico–subcortical interaction: a cross-frequency spectral EEG analysis.Electroencephalogr Clin Neurophysiol 2006;117:381–7.

Schutter DJ, Knyazev GG. Cross-frequency coupling of brain oscillations in studyingmotivation and emotion. Motivation Emotion 2012;36:46–54.

Schutter DJ, van Honk E. Salivary cortisol levels and the coupling of midfrontaldelta–beta oscillations. Int J Psychophysiol 2005;55:127–9.

Shils J, Litt M, Skolnick B, Stecker M. Bispectral analysis of visual interactions inhumans. Electroencephalogr Clin Neurophysiol 1996;98:113–25.

Steinberg L. Risk taking in adolescence new perspectives from brain and behavioralscience. Curr Dir Psychol Sci 2007;16:55–9.

Steinberg L, Albert D, Cauffman E, Banich M, Graham S, Woolard J. Agedifferences in sensation seeking and impulsivity as indexed by behaviorand self-report: evidence for a dual systems model. J Appl Dev Psychol2008;44:1764–78.

Steinberg L, Scott E. Less guilty by reason of adolescence. Am Psychol 2003;58:1009–18.

Steriade M. Impact of network activities on neuronal properties in corticothalamicsystems. J Neurophysiol 2001;86:1.

Thatcher R, North D, Biver C. EEG and intelligence: relations between EEGcoherence, EEG phase delay and power. Electroencephalogr Clin Neurophysiol2005;116:2129–41.

Thibodeau R, Jorgensen RS, Kim S. Depression, anxiety, and resting frontal EEGasymmetry: a meta-analytic review. J Abnorm Psychol 2006;115:715.

Tranel D, Bechara A, Denburg NL. Asymmetric functional roles of right and leftventromedial prefrontal cortices in social conduct, decision-making, andemotional processing. Cortex 2002;38:589–612.

Van Leijenhorst L, Crone E, Bunge S. Neural correlates of developmental differencesin risk estimation and feedback processing. Neuropsychologia 2006;44:2158–70.

Van Leijenhorst L, Moor B, Op de Macks Z, Rombouts S, Westenberg P, Crone E.Adolescent risky decision-making: neurocognitive development of reward andcontrol regions. Neuroimage 2010;51:345–55.

van Peer JM, Roelofs K, Spinhoven P. Cortisol administration enhances the couplingof midfrontal delta and beta oscillations. Int J Psychophysiol 2008;67:144–50.

Velikova S, Locatelli M, Insacco C, Smeraldi E, Comi G, Leocani L. Dysfunctional braincircuitry in obsessive–compulsive disorder: source and coherence analysis ofEEG rhythms. Neuroimage 2010;49:977–83.

Vitiello B, Stoff DM. Subtypes of aggression and their relevance to child psychiatry. JAm Acad Child Adolesc Psychiatry 1997;36:307–15.

Wager TD, Barrett LF. From affect to control: functional specialization of the insulain motivation and regulation. 2004. Published online at PsycExtra. Available at:http://www.affective-science.org/pubs/2004/Wager_Edfest_submitted_copy.pdf.

Weber E, Blais A, Betz N. A domain-specific risk-attitude scale: measuring riskperceptions and risk behaviors. J Behav Decis Making 2002;15:263–90.

ensity with cross-frequency phase–amplitude coupling in the resting EEG.