Tl

Ta

b

h

•••••

a

ARRAA

KSAFEE

1

eeriss

tI

UT

h0

Neuroscience Letters 583 (2014) 148–153

Contents lists available at ScienceDirect

Neuroscience Letters

jo ur nal ho me p age: www.elsev ier .com/ locate /neule t

he subliminal affective priming effects of faces displaying variousevels of arousal: An ERP study

ian-Tian Lia, Yong Lua,b,∗

Academy of Psychology and Behavior, Tianjin Normal University, Tianjin 300387, ChinaCenter of Cooperative Innovation for Assessment and Promotion of National Mental Health, Tianjin 300387, China

i g h l i g h t s

Subliminal affective priming (SAP) effect of aroused faces was studied using ERP.The affective priming effect occurred exclusively in the negative condition.Valence affected the subliminal affective priming effect of aroused faces.Low-arousing faces tended to elicit greater LPC and N400 potentials.SAP effect occurs when the prime affects the late stage processing of the probe.

r t i c l e i n f o

rticle history:eceived 3 June 2014eceived in revised form 18 August 2014ccepted 9 September 2014vailable online 26 September 2014

eywords:

a b s t r a c t

This study on the subliminal affective priming effects of faces displaying various levels of arousalemployed event-related potentials (ERPs). The participants were asked to rate the arousal of ambiguousmedium-arousing faces that were preceded by high- or low-arousing priming faces presented sublim-inally. The results revealed that the participants exhibited arousal-consistent variation in their arousallevel ratings of the probe faces exclusively in the negative prime condition. Compared with high-arousingfaces, the low-arousing faces tended to elicit greater late positive component (LPC, 450–660 ms) and

ubliminal affective primingrousalacexpressionvent-related potentials

greater N400 (330–450 ms) potentials. These findings support the following conclusions: (1) the effect ofsubliminal affective priming of faces can be detected in the affective arousal dimension; (2) valence mayinfluence the subliminal affective priming effect of the arousal dimension of emotional stimuli; and (3)the subliminal affective priming effect of face arousal occurs when the prime stimulus affects late-stageprocessing of the probe.

. Introduction

Affective priming occurs when the probe acquires the samemotional information as the prime while individuals judge themotion of the probe [16]. If the presentation time of the prime iseduced from tens of milliseconds to a few milliseconds, the effects still observed [7,15,16]. Because the presentation time is veryhort, the participants are not aware of it; thus, the effect is calledubliminal affective priming.

Two longstanding theories regarding the classification of emo-ion exist: basic emotions theory and emotional dimension theory.n past research, faces chosen to demonstrate the existence of

∗ Corresponding author at: Academy of Psychology and Behavior, Tianjin Normalniversity, No. 241 Weijin Road, Hexi District, Tianjin 300074, China.el.: +86 022 23540172x8500.

E-mail address: ly6312@163.com (Y. Lu).

ttp://dx.doi.org/10.1016/j.neulet.2014.09.027304-3940/© 2014 Elsevier Ireland Ltd. All rights reserved.

© 2014 Elsevier Ireland Ltd. All rights reserved.

subliminal affective priming were classified based on the basicemotions theory; these classifications included emotions such asanger [12] and fear [13]. Based on emotional dimension theories,the valence and arousal dimensions of emotional classification havebeen widely accepted and applied in numerous studies [1,10,20].Valence is a measure of pleasure that ranges from happy tounhappy, while arousal is a measure of the activation level ofbody energy associated with an emotional state, which ranges fromstimulating to calm. Some studies have proven the existence ofsubliminal affective priming of valence based on the emotionaldimension theory [5,14]. However, no study has used arousal tomeasure subliminal affective priming. Therefore, we sought toanswer the following questions: If the valence is controlled, canthe arousal dimension of the prime produce subliminal affective

priming? If the answer is yes, then what is the neural mechanismresponsible for this process?

Unconscious emotional information may variably affect severalsteps of cognitive processing. For instance, such information can

ce Let

efisattaao

esaiacsn

2a

2

nvpp

2

lmahtfwm

wt(nfal((

2

sofsFT

to

T.-T. Li, Y. Lu / Neuroscien

nhance perceptual processing or accompany the transfer of theocus of attention and influence responses, decision-making andmplementation. To detect and distinguish between a processingeries and understand the neural mechanisms of the subliminalffective priming of arousal, we used event-related potential (ERP)echnology, which has a high temporal resolution. Therefore, inhe present study, the valence was controlled, and high- and low-rousing faces were selected as the priming stimuli. Behavioralnalyses were combined with ERP recordings to study the effectf arousal on subliminal affective priming.

The Chinese Affective Picture System (CAPS) is the universalmotional stimuli image system in China [1]. The materials of thistudy were selected from the CAPS. We initially used valence androusal as independent variables. However, after we matched themages for valence, arousal, dominance, attractiveness and gendernd removed some of the faces with obvious features, an insuffi-ient number of pictures was available. Therefore, we divided thetudy into two experiments; the materials for Experiment 1 wereegative, and those for Experiment 2 were positive.

. Experiment 1: the subliminal affective priming ofrousal by negative faces

.1. Participants

Seventeen right-handed undergraduate students (eight males,ine females) ranging from 20 to 26 years old (mean = 22.6 years)olunteered for this study for partial course credit. The partici-ants had normal or corrected-to-normal vision and no history ofarticipation in neurological or psychiatric experiments.

.2. Stimuli

The priming stimuli consisted of 10 high-arousing and 10ow-arousing pictures with negative emotional expressions (half

ale, half female). The probe stimuli consisted of 80 medium-rousing pictures with negative emotional expressions (half male,alf female). To avoid interference from the individual features ofhe probe faces on the subliminal affective priming effect, all probeaces were transformed into mosaic pictures. The mask stimulusas a neutral face picture that was transformed into an intenseosaic picture.The prime and probe stimuli were compared with a one-

ay analysis of variance (ANOVA). These analyses revealedhe following results. (1) All stimuli were matched in valenceF(2,97) = 0.79, p = 0.46). (2) The main effect of arousal was sig-ificant (F(2,97) = 197.80, p = 0.001). High-arousing primes differed

rom low-arousing primes in the dimension of arousal (p = 0.001),nd medium-arousing probes also differed from the high- andow-arousing primes in the dimension of arousal (p = 0.001).3) All stimuli were matched in dominance and attractivenessF(2,97) = 0.53, p = 0.59; F(2,97) = 0.81, p = 0.45, respectively).

.3. Procedure

The stimuli were controlled with E-Prime software. A trialtarted with the presentation of a black fixation cross at the centerf a white screen for 1000 ms. Next, a high- or low-arousing primeace was presented for 12 ms followed by the mask, which was pre-ented for 200 ms. Then, a white screen was presented for 300 ms.inally, the probe was presented until the participant responded.

he intertrial interval was 1200 ms.

The 20 prime faces were randomly presented to the participant 8imes. Eighty probe faces were presented to the participant twice,nce with a high-arousing prime and once with a low-arousing

ters 583 (2014) 148–153 149

prime. The participant sat in a dimly lit, soundproof electromag-netic shielding chamber facing a CRT monitor located 1 m from theparticipant. A total of 160 trials were presented on the center of thescreen, which subtended a visual angle of 5.25 × 6.05◦. The partici-pants were instructed to rate the arousal of the probe on a 4-pointscale using the D, F, J, and K keys of the keyboard. After the experi-ment, the visibility of the prime was assessed using the subjectivemeasure method.

2.4. EEG recording

Electroencephalograms (EEGs) were recorded from 64 scalpsites using the 10–20 system with Ag/AgCl electrodes with thereference on the left mastoid. All electrode impedances were main-tained below 5 k�. EEGs and EOGs were sampled at a digitizationrate of 1000 Hz and filtered with a 0.05–100 Hz bandpass. Verti-cal electrooculogram (VEOG) recording electrodes were positionedabove and below the left eye, and horizontal electrooculogram(HEOG) recording electrodes were positioned 1 cm from the outercanthus of each eye. EEGs and EOGs were re-referenced offlineusing the average of the right and left mastoid recordings.

2.5. Subject exclusion and behavioral data analysis

Two participants reported having perceived the effect of theprime. The ERP data from another participant contained too manyartifacts. Thus, we included only 14 subjects in the behavioral andERP analysis. Trials with reaction times (RTs) shorter than 200 msor longer than the mean + 3 SD were not included in the analysis.The elimination rate was 1.22%. The remaining data were analyzedusing repeated-measures ANOVA with one within-subject factor,i.e., prime arousal type (two levels: high and low), using the statis-tical software SPSS 16.0.

2.6. ERP data analysis

Offline correction of eye movement artifacts was performed.To exclude trials contaminated by artifacts, trials with voltagesexceeding ± 80 �V at any electrode were discarded. A low-passoffline filter of 30 Hz (24 dB/oct) was applied. The signals were aver-aged offline over 1200 ms periods, and an additional 200 ms wasrecorded prior to the probe onset to allow for baseline correction.The ERPs were averaged separately for the low- and high-arousingtrials.

The topographical distributions of the overall-averaged ERPactivities (Fig. 1) indicated that the two experimental condi-tions provoked negative components that peaked at approximately120 ms, 240 ms and 400 ms after the stimulus onset (i.e., the ante-rior N100, N200 and N400, respectively) and a positive componentthat peaked at approximately 180 ms after the stimulus onset(P200) on the front of the scalp. On the back of the scalp, thetwo experimental conditions provoked a positive component thatpeaked approximately 110 ms after the stimulus onset (P100) anda negative component that peaked approximately 140 ms afterthe stimulus onset (posterior N100). Furthermore, a positive com-ponent with an onset approximately 300 ms after the stimuluspresentation (P300) was distributed over the medial anterior of thescalp, but this component was not obvious over the occipital scalp.An additional positive component that peaked approximately560 ms after stimulus onset (i.e., the late positive component, LPC)was distributed over approximately the entire scalp.

Based on the topographical distribution of the grand-averaged

ERP activities and previous studies, the ERP components and theirtime epochs were as follows: P100: 90–120 ms; anterior N100:90–140 ms; posterior N100: 120–160 ms; P200: 150–210 ms;N200: 210–270 ms; P300: 270–330 ms; N400: 330–450 ms; and

150 T.-T. Li, Y. Lu / Neuroscience Letters 583 (2014) 148–153

Fig. 1. Grand average ERP to high-arousing and low-arousing primed trials. Subliminal primes differentially influenced N400 and LPC potentials, with larger amplitudes inlow-arousing priming trials.

T.-T. Li, Y. Lu / Neuroscience Let

Table 1Ratings of arousal (M ± SD) and tests of differences in negative conditions.

Ratings of arousal F p �2

LwFwPoP1PwmnpF2

oSnorac

2

mtwofat

2

mpptaaa(tp(eaps�

e�w

Low-arousing prime High-arousing prime

2.34 ± 0.27 2.41 ± 0.24 5.37 0.04 0.29

PC: 450–660 ms. The anterior N100, N200 and N400 componentsere obtained from the following nine electrodes: F3, FZ, F4, FC3,

CZ, FC4, C3, CZ and C4. The P100 and posterior N100 componentsere obtained from the following nine electrodes: CP3, CPZ, CP4,

3, PZ, P4, O1, O2 and POZ. The P200 and P300 components werebtained from the following nine electrodes: F3, FZ, F4, C3, CZ, C4,3, PZ and P4. The LPC component was obtained from the following5 electrodes: F3, FZ, F4, FC3, FCZ, FC4, C3, CZ, C4, CP3, CPZ, CP4, P3,Z and P4. Among these components, the N400 and LPC potentialsere measured using their mean amplitudes, and the others wereeasured using their peak amplitudes and latencies. The EEG chan-

els were included in anterior 1, anterior 2, middle, posterior 1 andosterior 2 as follows, anterior 1: F3, FZ, F4; anterior 2: FC3, FCZ,C4; middle: C3, CZ, C4; posterior 1: CP3, CPZ, CP4; and posterior: P3, PZ, P4.

Three-way repeated-measures ANOVAs on the measured dataf each component were conducted using the statistical softwarePSS 16.0. Prime arousal type (two levels: low and high), longitudi-al electrode position (three levels: anterior, middle and posteriorr five levels: anterior 1, anterior 2, middle, posterior 1 and poste-ior 2) and horizontal electrode position (three levels: left, centernd right) were used as within-subject factors. The p-values wereorrected with the Greenhouse–Geisser correction.

.7. Behavioral data

A one-way repeated-measures ANOVA was conducted on theean arousal ratings of the probes with prime arousal type as

he factor (see Table 1). The subliminal affective priming effectas demonstrated by significant difference in the mean ratings

f the probes that were preceded by high- versus low-arousingaces. The ratings for high-arousing trials tended to indicate a high-rousing effect (higher arousal) compared with the low-arousingrials (F(1,13) = 5.37, p = 0.04, �2 = 0.29).

.8. Electrophysiological data

330–450 ms effects (N400). The results revealed that theain effect of prime arousal type was significant, F(1,13) = 5.32,

= 0.04, �2 = 0.29, the N400 potentials differed between the trialsrimed by high- versus low-arousing faces, and larger ampli-udes were observed in the low-arousing trials. The mean N400mplitude for the low- and high-arousing trials was 3.12 ± 2.70nd 2.44 ± 2.64 �V, respectively. The interaction between primerousal type and horizontal electrode position was significantF(2,26) = 4.22, p = 0.03, �2 = 0.25). To explore this interaction fur-her, a simple effect analysis was conducted. For the left electrodeosition, the main effect of prime arousal type was significantF(1,13) = 9.39, p = 0.01, �2 = 0.42), and the low-arousing primelicited larger N400 amplitudes (2.92 ± 2.73 �V) than the high-rousing prime (2.05 ± 2.65 �V). For the center and right electrodeositions, the main effects of the prime arousal type were notignificant (F(1,13) = 4.13, p = 0.06, �2 = 0.24; F(1,13) = 2.87, p = 0.01,2 = 0.18, respectively).

450–660 ms effects (LPC). The results revealed that the mainffect of prime arousal type was significant, F(1,13) = 7.90, p = 0.02,2 = 0.38, the LPC potentials differed between the trials thatere primed by high- versus low-arousing faces, and increased

ters 583 (2014) 148–153 151

amplitudes were observed on the low-arousing trials. The mean LPCamplitude for the low- and high-arousing trials was 6.58 ± 3.41 and5.87 ± 3.38 �V, respectively. The main effect of electrode horizon-tal position was significant (F(2,26) = 5.66, p = 0.01, �2 = 0.30). Posthoc analysis using the Bonferroni correction revealed that the meanLPC amplitude for the center position (6.80 ± 3.62 �V) was largerthan those for the left hemisphere (5.84 ± 3.40 �V, p = 0.01) andthe right hemisphere (6.04 ± 3.14 �V, p = 0.05). The main effect oflongitudinal electrode position was also significant (F(4,52) = 12.31,p = 0.01, �2 = 0.49), and post hoc analysis using the Bonferroni cor-rection revealed that the mean LPC amplitude for the posterior 1position (7.43 ± 3.49 �V) was significantly larger than those for theanterior 1 position (4.89 ± 2.93 �V, p = 0.01), the anterior 2 position(5.33 ± 2.80 �V, p = 0.01) and the center position (6.02 ± 3.26 �V,p = 0.02). In addition, the mean LPC amplitude for the posterior2 position (7.48 ± 3.70 �V) was significantly larger than those forthe anterior 1 position (4.89 ± 2.93 �V) and the anterior 2 position(5.33 ± 2.80 �V, p = 0.05), and the mean LPC amplitude for the cen-ter position (6.02 ± 3.26 �V) was significantly larger than that forthe anterior 2 position (5.33 ± 2.80 �V, p = 0.02).

3. Experiment 2: the subliminal affective priming ofarousal by positive faces

3.1. Participants

The participants were the same as in Experiment 1.

3.2. Stimuli

The prime and probe stimuli were compared with one-wayANOVA, which revealed the following conclusions. (1) All stimuliwere matched in valence (F(2,97) = 0.69, p = 0.50). (2) The maineffect of arousal was significant (F(2,97) = 100.35, p = 0.01), with thehigh-arousing prime differing from the low-arousing prime withregard to arousal (p = 0.01) and with the probe differing in arousalfrom the high- and low-arousing primes (p = 0.01). (3) All stimuliwere matched in dominance and attractiveness (F(2,97) = 0.96,p = 0.39; F(2,97) = 1.34, p = 0.27, respectively).

3.3. Procedure and EEG recording

The procedure and the EEG recording were the same as in Exper-iment 1.

3.4. Subject exclusion and behavioral data analysis

Two participants reported perceiving the effect of the prime.However, the ERP data from two additional participants containedtoo many artifacts; thus, we included 13 subjects in the behav-ioral ERP analysis. The trials with RTs shorter than 200 ms or longerthan the mean + 3 SD were not included in the analysis. The elim-ination rate was 1.82%. The other analyses were the same as inExperiment 1.

3.5. Behavioral data

A one-way repeated-measures ANOVA on the mean arousal rat-ings for the probes was conducted with prime arousal type as

the factor (see Table 2). The results revealed that the differencebetween the mean ratings of the probes preceded by the high- orthe low-arousing faces was not significant (F(1,12) = 2.74, p = 0.12,�2 = 0.19).

152 T.-T. Li, Y. Lu / Neuroscience Let

Table 2Ratings of arousal (M ± SD) and tests of differences in positive conditions.

Ratings of arousal F p �2

3

tw

4

atfplpopg

td(earrdtliie

alw[tstlelttd

pipttbrah“p

pleasure and arousal in memory, J. Exp. Psychol. Learn. Mem. Cogn. 18 (1992)

Low-arousing prime High-arousing prime

2.31 ± 0.21 2.38 ± 0.27 2.74 0.12 0.19

.6. ERP data analysis and electrophysiological data

The ERP data analysis was the same as in Experiment 1. None ofhe differences in the any of the ERP components in the experimentsere significant.

. Discussion

This study investigated the subliminal affective priming effect ofrousal with ERP and behavioral methods. The results revealed thathis effect exclusively occurred in negative prime group; the ratingsor the high-arousing trials tended toward greater arousal com-ared with the low-arousing trials. The ERP results revealed that

arger LPC and N400 potentials were elicited by the low-arousingrimes than the high-arousing primes. However, this effect was notbserved in the positive prime group. Thus, the subliminal affectiveriming effect of arousal was more robust in the negative primeroup.

The emotional dimension theory was first proposed by Wundt,he founder of psychology. In his opinion, emotion includes threeimensions: valence, arousal and intensity. Although later scholarse.g., Sternberg and Osgood) proposed various schemes for dividingmotion [10,19], the majority of scholars thought that valence androusal were two important emotional dimensions. Valence hasemained the primary focus of prior subliminal affective primingesearch, although very few studies have investigated the arousalimension in isolation. The behavioral results of this study revealedhat the arousal dimension can independently influence the sub-iminal affective priming effect under conditions in which valences strictly controlled. Additionally, this study also found changesn N400 and LPC that provide behavioral and electrophysiologicalvidence for the arousal emotional dimension.

The changes in N400 and LPC illustrate that the subliminalffective priming effect of arousal occurs when the prime affectsate-stage processing of the probe. These results are consistent

ith the view that the arousal level influences late waveforms17]. Studies have reported that early perceptual processing ofhe probe is affected by the prime’s valence, demonstrating theubliminal affective priming effect [14]. These results are consis-ent with those of studies of emotional pictures showing that theatencies of the ERPs evoked by valence and arousal are differ-nt; the former latencies are shorter (100–250 ms), whereas theatter type is longer (200–1000 ms) [3,18]. These findings indicatehat the brain processes arousal and valence differently. Therefore,his study provides further evidence that arousal is an emotionalimension that is different from valence.

In ERP studies, the N400 is commonly associated with semanticrocessing. However, the N400 is not limited to situations involv-

ng words. For example, if a picture does not match a previouslyresented word or a series of images do not match in terms of con-ext, the N400 (or a wave similar to the N400) will appear [4]. Inhis study, for the negative face primes, larger N400s were elicitedy low-arousing primes than by high-arousing primes. This findingeflects the fact that information integration is more difficult with

low-arousing prime than a high-arousing prime. Compared withigh-arousing expressions, low-arousing expressions were moreexpressionless”. Thus, the difference between the low-arousingrimes and probes (medium-arousing) was equal to the difference

ters 583 (2014) 148–153

between expressionless and expressive, whereas the differencebetween the high-arousing primes and probes was equal to thedifference between expressive and expressive (the only differencebeing arousal level). Thus, the difference between the low-arousingprimes and probes was greater.

The feelings-as-information theory posits that, in cases of insuf-ficient information or time, people tend to judge a probe’s effectbased on a preexisting feeling. When people judge the effect ofa probe, they mistakenly believe that the feeling of the prime isthe feeling of the probe, thereby exhibiting the subliminal affectivepriming effect [16]. In this study, the probes were medium-arousingfuzzy faces. Therefore, when the participants judged the arousalof the probe, the prime became the reference upon which theprobe was judged given insufficient information. Compared withlow-arousing stimuli, high-arousing stimuli had greater intrinsicmotivation attributes, were assigned more psychological resources,and facilitated the emotional events, coding and memory storage[2,11]. When participants judged the arousal of the probe basedon the arousal of the prime, the high-arousing prime required lesspsychological resources than the low-arousing prime because ofthe facilitation of memory storage. Thus, the high-arousing primeelicited LPCs with reduced amplitudes.

In this study, participants exhibited arousal-consistent variationwhile rating the arousal of probe exclusively in the negative primecondition. This finding indicates valence might affect the subliminalaffective priming effect of arousal. Previous studies have reportedthat arousal is an important factor that influences the subliminalaffective priming effect of valence [5,18]. Those results suggest thatthe valence and arousal dimensions are not completely indepen-dent and they are closely related. Individuals preferentially processvalence information about emotion, and the valence influencesthe subsequent processing of arousal information. Thus, valenceis more basic than arousal. It is important to identify valence infor-mation because this determines the basic strategy for coping withenvironmental changes. From an evolutionary perspective, neg-ative information is closely related to survival and functions asa warning signal. Compared with negative information, the war-ning role of positive information is not as strong. The nature of thepriming effect is preparation for addressing “danger”; therefore,negative information may have played a stronger role in evolu-tion [6], which may be the reason that the emotion priming effectexclusively occurred in the negative prime group in the currentstudy. These data agree with the results of Jiang’s study of the sub-liminal affective priming effect of valence. In that study, the taskwas to judge the expressions in photos of human subjects aftera prime consisting of a crying, smiling or neutral face. The cryingface prime, but not the smiling face prime, elicited a significantemotionally congruent effect. The authors stated that their resultswere consistent with evolutionary psychology theory. Thus, thesubliminal affective priming effect of negative emotion is an earlywarning that benefits survival by serving as a warning of danger[8,9].

To summarize, the late-stage processing of probes is affectedby the arousal dimension of prime affective facial stimuli and maydemonstrate the subliminal affective priming effect.

References

[1] L. Bai, H.Y.X. Ma, Y.J. Luo, The development of native Chinese affective pic-ture system – a pretest in 46 college students, Chin. Ment. Health J. 19 (2005)719–722 (in Chinese).

[2] M.M. Bradley, M.K. Greenwald, M.C. Petry, P.J. Lang, Remembering pictures:

379–390.[3] M. Codispoti, V. Ferrari, M.M. Bradley, Repetition and event-related potentials:

distinguishing early and late processes in affective picture perception, J. Cogn.Neurosci. 19 (2007) 577–586.

ce Let

[

[

[

[

[

[

[

[

T.-T. Li, Y. Lu / Neuroscien

[4] G. Ganis, M. Kutas, M.I. Sereno, The search for common sense: an electrophys-iological study of the comprehension of words and pictures in reading, J. Cogn.Neurosci. 8 (1996) 89–106.

[5] H. Gibbons, Evaluative priming from subliminal emotional words: insightsfrom event-related potentials and individual differences related to anxiety,Conscious Cogn. 18 (2009) 383–400.

[6] P. Gilbert, The evolved basis and adaptive functions of cognitive distortions, Br.J. Med. Psychol. 71 (1998) 447–463.

[7] D. Hermans, A. Spruyt, J. De Houwer, P. Eelen, Affective priming with sublimi-nally presented pictures, Can. J. Exp. Psychol. 57 (2003) 97–114.

[8] Z.Q. Jiang, L.Z. Yang, Y. Liu, Study on adults’ subliminal affective priming effect,J. Liaoning Normal Univ. 28 (2005) 48–50 (in Chinese).

[9] Z.Q. Jiang, L.Z. Yang, Y. Liu, T. Bel-Bahar, Subliminal negative affective stimuliarouse opposite bias in adults and children, Psychol. Dev. Educ. 1 (2010) 9–13(in Chinese).

10] P.J. Lang, M.M. Bradley, B.N. Cuthbert, International Affective Picture Sys-

tem (IAPS): Instruction Manual and Affective Ratings, University of Florida,Gainesville, FL, 2001.

11] P.J. Lang, M.K. Greenwald, M.M. Bradley, A.O. Hamm, Looking at pictures:affective, facial, visceral and behavioral reactions, Psychophysiology 30 (1993)261–273.

[

[[

ters 583 (2014) 148–153 153

12] A.M. Leventhal, A.J. Waters, B.G. Breitmeyer, E.K. Miller, E. Tapia, Y. Li, Sublim-inal processing of smoking-related and affective stimuli in tobacco addiction,Exp. Clin. Psychopharmacol. 16 (2008) 301–312.

13] H.C. Liu, Y. Xu, R.L. Zhou, The influence of subliminal priming of fear and disguston person judgment, Psychol. Explor. 30 (2010) 37–41 (in Chinese).

14] Y. Lu, W.N. Zhang, D.L. Shen, Subliminal affective priming effect byfaces with different valence: an ERP study, Acta Psychol. Sin. 42 (2010)929–938.

15] Q. Luo, Neurocognitive Mechanisms of the Subliminal Priming Effect Induced byWords with Different Affective Meaning (Ph.D. dissertation), Graduate Schoolof Beijing Normal University, Beijing, China, 2003.

16] S.T. Murphy, R.B. Zajonc, Affect cognition, and awareness: affective primingwith optimal and suboptimal stimulus exposures, J. Pers. Soc. Psychol. 64 (1993)723–739.

17] J.K. Olofsson, S. Nordin, H. Sequeira, J. Polich, Affective picture processing: anintegrative review of ERP findings, Biol. Psychol. 77 (2008) 247–265.

18] J.K. Olofsson, J. Polich, Affective visual event-related potentials: arousal, repe-tition, and time-on-task, Biol. Psychol. 75 (2007) 101–108.

19] H. Sternberg, Three dimensions of emotion, Psychol. Rev. 61 (1954) 81–88.20] M.S. Yik, J.A. Russell, Chinese affect circumplex: I. Structure of recalled momen-

tary affect, Asian J. Soc. Psychol. 6 (2003) 185–200.