5.3.1.2. Results

For each participant, incorrect response times (RTs) and correct RTs more than the value calculated by the following equation: third quartile + 2 x (third quartile – first quartile) or less than 300 ms (Tukey, 1977) were discarded. Occasionally, some hesitation or mixed pronunciation (e.g., /bl....red/) occurred, but errors were very low (less than 3%), and we eliminated those RTs. A 2 (Stroop congruency) X 4 (emotional face) X 2 (block) within-subjects ANOVA was performed using mean RTs. The main effect for emotional face was significant, F (3, 87) = 7.12, p < .001. RTs with the negative face were faster than those with the positive or neutral face, in addition to those in the absence of a face (see Table 1). There was a significant Stroop congruency level effect, F (1, 30) = 30.25, p < .0001. Participants responded more rapidly in the congruent condition between colour word and colour patch than in the incongruent condition. The mean RTs were 554 ms for the congruent and 580 ms for the incongruent condition. The two-way interaction between face type and block was significant, F (3, 90) = 10.98, p < .0001. As shown in Table 1, participants colour-named more rapidly with negative information in Block 2. The three-way interaction between face type, Stroop condition, and block was also significant, F (3, 90) = 3.02, p < .05. Finally, there was no significant interaction between Stroop congruency and emotional condition, F (3, 90) = .46, p < .71. It seemed that emotional processing did not influence the level of semantic activation of a colour word. A post hoc PLSD comparison between emotional faces and the absence of face (test of single bias variables against zero) revealed a significant effect only with the positive face (p < .05).

TABLE 1 - Mean reaction times in Emotional face x Block conditions
Emotional Face
  Absence Negative Neutral Positive
Block 1 563 msec 567 msec 577 msec 565 msec
Block 2 566 msec 551 msec 564 msec 584 msec

Bias scores. Facilitation and interference scores were calculated for each emotional face type by subtracting the mean RT when the emotional face was absent from the mean RT with each emotional face condition. A 2 x 3 x 2 mixed design ANOVA was performed on the RT bias scores with two Stroop congruency levels, three face types and two blocks. There was a main effect of face, F (2, 60) = 10.86, p < .0001. In general, participants showed rapid colour-naming latency after priming with a negative face and longer latency after priming with a positive face (Figure 9). There was an interaction between face and block, F (2, 60) = 14.38, p < .0001. Participants demonstrated more rapid colour-naming latency with negative and positive face stimuli than with the neutral face in Block 1, but in Block 2, RTs with a positive face were longer than those with the neutral face. A three-way complex interaction between face, Stroop condition and block was also significant, F (2, 60) = 4.48, p < .05, but this tendency was difficult to be interpreted. Finally, in post hoc Neman-Keuls tests, there were significant differences between negative and neutral faces (p < .01), and between negative and positive faces (p < .001).

Figure 9. Attentional bias for faces in emotional priming Stroop task.
Figure 9. Attentional bias for faces in emotional priming Stroop task.