Daxing Wu, Shujing Xu, Huifang Yin
Medical Psychological Research Center, Second Xiangya Hospital, Central South University, Changsha 410011, Hunan Province, China
Daxing Wu☆, M.D., Asso-ciate professor, Medical Psychological Research Center, Second Xiangya Hospital, Central South University, Changsha 410011, Hunan Province, China
Corresponding author: Daxing Wu, Medical Psychological Research Center, Second Xiangya Hospital, Central South University, Changsha   410011, Hunan Province, China
Supported by: the National Natural Science Foundation of China, No. 30570609*
Received: 2010-04-16     Accepted: 2010-07-04   (N20090927001/ZW)
Wu DX, Xu SJ, Yin HF. Chinese emotional words in patients with major depres-sive disorder during a subliminal Stroop task: an event-related potential study. Neural Regen Res. 2010;5(16):1274-1280.
www.crter.cn
www.nrronline.org
doi:10.3969/j.issn.1673-5374.2010.16.013

Abstract
Patients with major depressive disorder (MDD) develop a negative cognitive bias, but how they respond to information in Chinese emotional words is unclear. Here we used a Stroop paradigm with subliminal Chinese emotional words to explore the event-related potential components of abnormal emotional processing in patients with MDD. The correct rate was similar in MDD and normal control groups, but MDD reaction time was longer than the normal controls, especially to the negative and neutral stimuli. In N270, repeated-measure analysis of variance demonstrated a significant main effect of the relation electrode and valence on peak amplitude and interactions between valence and electrode site. The peak amplitudes of the three kinds of words were different in the two groups (positive > negative > neutral). The topography of the difference waves indicated that the difference distributed in the frontal and left parietal-temporal sites across the scalp. In N400, there was a significant main effect of the relation electrode and valence on peak amplitude, and the latency showed a main effect of the electrode and an interaction between electrode and group. The amplitudes induced by type of words were significantly different from each other in both groups (positive > negative > neutral). The topography of the difference waves indicated that the effect of relation type was primarily at left and right frontal and central and left parietal-temporal regions. Both MDD patients and normal controls exhibited significant emotional Stroop effects during the processing of positive/negative Chinese emotional words. MDD patients showed interference in emotional stimuli in early cognitive processing that induced psychological resource intervention during late emotional information processing.
Key Words: Stroop test; subliminal stimulations; event-related potentials; depression; Chinese emotional words
INTRODUCTION
Cognitive theories posit that depressive schemas and associated biases play an important etiological role in the onset and maintenance of depression[1]. Major depressive disorder (MDD) is characterized by a negativity bias: an enhanced responsiveness to, and memory for, affectively negative stimuli[2]. MDD from biased thinking/depressive beliefs includes both self-reports and processing measures of cognitive bias[3]. Our previous behavior and event-related potential (ERP) studies have confirmed that patients with MDD appear to respond more slowly to positive words relative to control participants, with no difference to negative emotional words, and found that the opposite late positive component effects were elicited by recognition of Chinese emotional content words[4-6].
The Stroop interference effect refers to an increase in response time observed when the word meaning and the stimulus hue do not match[7]. The interference effect of naming the colors of the emotional words is the emotional Stroop effect. While the emotional Stroop test and the classic Stroop effect elicit similar behavioral outcomes, these tests engage different mechanisms of interference[8].
The emotional Stroop test has been used broadly in clinical studies in psychiatry using emotional words related to a particular individual’s area of concern, such as depression or anxiety related words for someone that has MDD or anxiety disorders[9]. Depressed participants will be slower to say the color of depressing words than non-depressing words[10]. Non-clinical subjects also name the color of an emotional word (e.g. “war”, “cancer”, “kill”) slower than naming the color of a neutral word (e.g. “clock”, “lift”, “windy”). For example, it takes longer time to name the colors of emotional words 
than matched neutral words in emotionally disturbed patients[11]. Attention bias and reduced attention control are typical symptoms of MDD. 
ERP results from a study[12] of gibbon subliminal emotional words suggested that there is indeed affective/semantic processing of unconscious words. Interestingly, this emotional Stroop effect can also be observed when the negative words and pictures are presented subliminally[13-16]. The Stroop interference of English words was lower than that of Chinese characters and Pinyin[17]. Despite reports of an emotional Stroop effect among Chinese words[18], little is known about the role of Chinese emotional words under subliminal conditions in Chinese patients with MDD.
We previously assumed that the opposite effects might reflect the lack of positive (not negative) emotional experiences in MDD, consistent with the valence hypothesis[4, 19]. Therefore, we tested subliminal Chinese emotional words in the Stroop task to examine the ERP effect of cognitive biases on emotion perception in MDD. We predicted that the double dissociation effects of ERP in MDD would be apparent in frontal and parietal sites.
RESULTS
Baseline data
The two groups (26 MDD patients and 24 controls) were similar in gender, age, and educational level (P > 0.05). Total score of 17 items of the Chinese Hamilton Depression Rating Scale and 13-item Beck Depression Inventory were significantly higher in MDD patients than normal controls (Table 1). 

Behavioral results
Both normal controls and MDD groups averaged over 90% accuracy of responses (t = 1.274, P > 0.05). Reaction time was faster for controls in overall emotional conditions compared with the MDD group  (t = 2.395, P < 0.05), especially in negative word condition (t = 2.226, P < 0.05) and neutral word condition (t = 2.696, P < 0.05). The sequences of reaction times were similar for all word types (P > 0.05) (Table 2).
ERP results for the subliminal Stroop task 
A grand mean ERP is shown in Figure 1, revealing negative peaks in the 60–130 ms, 220–300 ms, and 380–450 ms post-stimulus time window (identified as N80, N270, and N400, respectively), positive peaks in the 130–220 ms, 300–380 ms, and 450–750 ms window (P150, P350, and P500, respectively), the slow positive wave in the 750–1 000 ms post-stimulus time window (slow positive wave), and the sustained slow positive wave in the 1 000–1 500 ms post-stimulus time window (sustained slow positive wave).

N270 results for the subliminal Stroop task 
Using valence and electrode as the intragroup variable, there was a significant main effect of electrode and valence on peak amplitude (F9, 432 = 3.132, P < 0.01; F2, 96 = 7.415, P < 0.01). There were also significant interactions between valence and electrode site (F18, 864 = 1.951, P < 0.05). In both groups, the peak amplitudes of the three kinds of words were significantly different (F2, 777 = 3.139, P < 0.05 for MDD group; F2, 717 = 3.50, P < 0.05 for control group). The sequence of peak amplitude was positive > negative > neutral. Positive words induced significantly larger amplitudes than negative or neutral words (P < 0.05).
The topography of the difference waves (MDD minus control) for the valence conditions showed that valence influenced left parietal-temporal sites (T7/TP7/C3/CP3/P3/P7) for positive, left parietal sites (CP3/P3) in negative, and left parietal-temporal sites (T7/TP7/CP3/P3) in neutral (Figure 2A). The topography of the difference waves (positive minus negative) within MDD patients and controls indicated that the difference effect of positive/negative valence was primarily at anterior (FP1/FP2), left frontal (F3), and left parietal sites (P3) in MDD, but anterior (FP1/FP2) and left parietal-temporal sites (T7/TP7/CP3/P3/P7) in the control group (Figure 2B).