Humans shift their attention to follow another person's gaze direction, a phenomenon called gaze cueing. This study examined whether a particular social factor, intergroup threat, modulates gaze cueing. As expected, stronger responses of a particular in-group to a threatening out-group were observed when the in-group, conditioned to perceive threat from one of two out-groups, was presented with facial stimuli from the threatening and non-threatening out-groups. These results suggest that intergroup threat plays an important role in shaping social attention. Furthermore, larger gaze-cueing effects were found for threatening out-group faces than for in-group faces only at the 200 ms but not the 800 ms stimulus onset asynchrony (SOA); the specificity of the gaze-cueing effects at the short SOA suggests that threat cues modulate the involuntary component of gaze cueing.
Gaze cueing, a tendency to follow the gaze of others and orient one's attention, is an important ability for non-language communication, and this ability is critical for social learning and discovering potentially relevant information . This ability, which is found in several animal species, appears at three to six months in human infancy . When we use this ability to observe others, the response time of congruent gazes is shorter than the response time of incongruent gazes, which is known as the gaze-cueing effect.
Previous studies provide evidence that gaze cueing is sensitive to social modulators. An animal study found high-status and low-status monkeys reflexively followed the gaze of high-status monkeys, but high-status monkeys did not follow the gaze of low-status monkeys . Similarly, there is evidence that people only follow the gaze of high-status individuals . Tellingly, greater gaze-cueing effects were found when humans observed masculine faces than feminine faces .
Previous research suggests that intergroup threat may affect gaze cueing, because males and social status are related to intergroup threat. Intergroup threat refers to the perception of group members that another group may cause them harm . The evidence from intergroup-threat research indicates that high-status groups pose more threat than low-status groups , and that masculine faces are more threatening than feminine faces .
We conducted this study to examine whether the effect of intergroup threat between nations influenced social attention in humans, independently of the sense of intergroup threat that was perceived implicitly from observed faces. Specifically, Chinese face stimuli were classified into two groups with a single random allocation: Chinese and Japanese in study a; Chinese and North Korean in study b. Participants were asked to learn to which group each face belonged. The intergroup relations between the two groups were manipulated as threatening in study a, and as non-threatening in study b. Then a standard gaze-cueing paradigm was employed . Our expectation was that participants would follow the gaze of threatening out-group faces in study a, while participants would not follow the gaze of non-threatening out-group faces in study b. Larger gaze-cueing effects were also predicted for threatening out-group faces than for non-threatening in-group faces in study a.
2. Material and methods
In study a, the participants were 42 students (26 females). In study b, the participants were 30 students (21 females). All participants were Chinese, from Southwest University, 19–25 years old (a: mean = 21.95 years, s.d. = 1.55; b: mean = 21.83 years, s.d. = 1.51), right-handed, reported no history of neurological disorder and had normal or corrected-to-normal vision.
Eight black-and-white photos with neutral expressions from the Chinese Facial Affective Picture System  were used as stimuli (four males and four females). Those photos were divided into two groups, Chinese (two males and two females) and Japanese (two males and two females) in study a, Chinese (two males and two females) and North Korean (two males and two females) in study b.
Participants first completed a 5-point scale (from 1 = ‘a little’ to 5 = ‘quite a lot’) with five emotions (angry, afraid, irritable, worried and anxious) related to a sense of threat . Then, they were given 3 min to learn which country each photo represented. Every photo was presented three times to make sure the participants learned them; the correct response rate was over 95%.
Two articles and six pictures were used to create conditions for threat versus non-threat. In study a, participants read an article about the Sino-Japanese War and observed three pictures about the war. In study b, participants read an article about the Sino-Korean friendship and watched three pictures about the friendship (see the electronic supplementary material). Then, the emotion measure was completed again to check the effectiveness of the intergroup threatening or non-threatening manipulation. In addition, the sense of intergroup threat was measured by the question: ‘How much intergroup threat did you feel from Japan (North Korea) to China after reading the article and watching the pictures?’ The response options ranged from 1 to 7; 1 = not at all, 7 = very much. Following the manipulation check, participants completed the gaze-cueing task.
The gaze-cueing task is based on those used in previous studies [4,11]: first, a white cross appeared for 500 ms. Second, a face with direct gaze appeared for 900 ms, after which the pupils moved to the left or right for 200 or 800 ms, and then a white target (‘T’ or ‘L’) appeared with a time-out limit of 2000 ms. After the target, a black screen appeared for 2, 4, 6 or 8 s (figure 1).
Response speed and accuracy were emphasized in the instructions. Participants made a response by pressing the ‘1’ or ‘2’ keys with their right index finger. Half the participants pressed the ‘1’ key if the target was ‘T’, and pressed the ‘2’ key if the target was ‘L’. The other half made the opposite responses. The trials were split into four blocks of 64 trials, which were counterbalanced in cue-target spatial congruency (congruent and incongruent), intergroup threat (a: threatening out-group and non-threatening in-group) (b: non-threatening out-group and non-threatening in-group) and stimulus onset asynchrony (SOA) (200 ms, 800 ms). Twenty-four practice trials were completed before the study trials.
Paired t-tests found significantly higher post-test scores for angry (t42 = 10.957, p < 0.001, d = 2.025), afraid (t42 = 2.995, p = 0.005, d = 0.496), irritable (t42 = 7.815, p < 0.001, d = 1.434), worried (t42 = 3.825, p < 0.001, d = 0.806) and anxious (t42 = 4.436, p < 0.001, d = 0.909) in study a. No significant difference in post-test scores was found for any emotion in study b (all p's > 0.223). Moreover, the mean score of the sense of intergroup threat was 4.76 (s.d. = 1.23) and greater than the median (4) in study a (t42 = 4.028, p < 0.001, d = 0.877), whereas it was 2.70 (s.d. = 1.39) and lower than the median in study b (t30 = −5.110, p < 0.001, d = 1.323). These results indicate that participants felt an intergroup threat from the threatening out-group (Japan) but not from the non-threatening out-group (North Korea). Errors and outlying latencies greater than 1500 ms or less than 100 ms were removed (a: 1.23% of all responses; b: 1.62% of all responses).
A 2 (cue-target spatial congruency: congruent versus incongruent) × 2 (intergroup threat: threatening out-group versus non-threatening in-group in study a; non-threatening out-group versus non-threatening in-group in study b) × 2 (SOA: 200 ms versus 800 ms) repeated-measures ANOVA was performed on the mean reaction times (RTs). In study a, the analysis found significant main effects of cue-target spatial congruency (F1,41 = 40.254, p < 0.001, ) and SOA (F1,41 = 81.479, p < 0.001, ), and a 3-way interaction (F1,41 = 4.733, p = 0.035, ). In study b, the analysis found significant main effects of cue-target spatial congruency (F1,29 = 7.095, p = 0.012, ) and SOA (F1,29 = 113.942, p < 0.001, ). No other significant effect emerged (all p's > 0.280).
A 2 (cue-target spatial congruency) × 2 (intergroup threat) repeated-measures ANOVA was conducted on the mean RTs for the 200 ms SOA. In study a, the analysis found a main effect of cue-target spatial congruency (F1,41 = 15.887, p < 0.001, ). The 2-way interaction was significant (F1,41 = 5.929, p = 0.019, ), and paired t-tests found participants followed the gaze of threatening out-group (t42 = 4.581, p < 0.001, d = 0.170) but not of non-threatening in-group faces (t42 = 1.254, p = 0.217, d = 0.041) (figure 2). In study b, the same ANOVA found a main effect of cue-target spatial congruency (F1,29 = 7.006, p = 0.013, ). Paired t-tests showed participants did not follow the gaze of non-threatening out-group (t30 = 1.593, p = 0.122, d = 0.088) or non-threatening in-group faces (t30 = 1.388, p = 0.176, d = 0.068).
We conducted the same ANOVA for the 800 ms SOA. In study a, a main effect of cue-target spatial congruency was found (F1,41 = 22.964, p < 0.001, ). Paired t-tests found participants followed the gaze of both threatening out-group (t42 = 2.845, p = 0.007, d = 0.110) and non-threatening in-group faces (t42 = 3.301, p = 0.002, d = 0.141). In study b, the main effect of cue-target spatial congruency (F1,29 = 3.302, p = 0.080, ) and the two-way interaction (F1,29 = 3.206, p = 0.084, ) were marginally significant. Paired t-tests found participants followed the gaze of non-threatening in-group (t30 = 2.364, p = 0.025, d = 0.127) but not of non-threatening out-group faces (t30 = 0.793, p = 0.434, d = 0.040).
A second ANOVA conducted on the percentage of errors found a significant main effect of SOA in study a (F1,41 = 10.391, p = 0.002, ). No significant effect emerged in study b (all p's > 0.054). Thus, the present findings cannot be accounted for by a speed–accuracy trade-off.
This study focused on whether intergroup threat would modulate gaze cueing. Participants felt an intergroup threat from the threatening out-group but not from the non-threatening out-group, so that they followed the gaze of threatening out-group faces at both SOAs in study a, but they did not follow the gaze of non-threatening out-group faces at either SOA in study b. Taken together, these results suggest that intergroup threat plays an important role in shaping social attention.
Paying more attention to a threat is a self-defence mechanism, and humans probably inherited the ability to detect threat from our ancestors . Throughout the evolutionary history of humans, the concept of threat is considered to be a central factor that helps to shape social attention, such as the nature of intergroup perceptions and relations . Because intergroup interaction is a basic activity for an individual's survival, quick and valid response to threat may alter an individual's and/or a group's destiny. Hence, when confronted with both threatening and non-threatening information, people orient to threatening stimuli , and notice facial stimuli displaying a threatening emotion much faster than they do either neutral or happy facial expressions . Based on the motive of self-defence, people should process information from threatening out-groups earlier and more quickly, as in this study, in which participants followed the gaze direction of the threatening person.
The study is the first to our knowledge to explore the effect of intergroup threat on gaze cueing. Our findings suggest that gaze cueing under intergroup threat is an involuntary process, because the effect of intergroup threat on the magnitude of the gaze-cueing effect is evident only at the 200 ms SOA, which is consistent with previous findings. For example, previous research found larger gaze-cueing effects for masculinized than for feminized faces at the 200 ms but not the 800 ms SOA . The specificity of the gaze-cueing effects at the short SOA suggests that threat cues modulate the involuntary component of gaze cueing. Future studies would benefit from exploring the interaction between intergroup threat and group status, because previous research found that a high-status out-group was a source of intergroup threat , like the Japanese in study a. Future research also would profit from exploring how intergroup threat works on gaze cueing using an experimental group and a control group directly. In addition, participants tended to follow the gaze of in-group faces in both threatening and non-threatening conditions, which is consistent with previous findings. For instance, a related study showed that White participants followed the gaze of White but not of Black faces, whereas Black participants followed the gaze of both . However, the reason why the in-group can modulate the gaze-mediated orientation of attention remains an open question.
Our ethical standards in conducting the research are in line with the Declaration of Helsinki and were granted approval by the Researcher's Review Board of Southwest University. Information about the study was provided to all participants after the study was finished. Furthermore, informed consent sheets were completed by all participants prior to commencing the study.
The datasets supporting this article can be found in the electronic supplementary material.
Y.C. and Y.Z. designed the study, carried out the analysis and wrote the paper.
This study was supported by the National Natural Science Foundation of China (31371055).
Conflict of interests
We have no competing interests.
We thank Oscar Ybarra, Chongzeng Bi and two anonymous reviewers for constructive comments.
- Received December 16, 2014.
- Accepted January 27, 2015.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.