The influence of observers’ sex on attention-demanding performance depends on performers’ sex

Lijun Wang; Jinfeng Tan; Jiangtao Chen; Antao Chen

doi:10.3389/fpsyg.2015.01217

. 2015 Aug 31;6:1217. doi: 10.3389/fpsyg.2015.01217

The influence of observers’ sex on attention-demanding performance depends on performers’ sex

Lijun Wang ¹, Jinfeng Tan ¹, Jiangtao Chen ¹, Antao Chen ^1,^*

PMCID: PMC4553392 PMID: 26379574

Abstract

Post-error slowing (PES) indicates the slower responses after errors than after correct responses. Prior studies mainly focus on how the observation errors influence one own’s performance, there is no study investigating how other’s monitoring influence one own’s performance. Additionally, the issue that whether social context influences the PES effect differently for females and males is still unclear. To address aforementioned issues, we required the participants to interact with a same-sex or opposite-sex partner to complete a color flanker task together (they sat next to each other, Experiment 1). One was the performer (perform the flanker task), and the other was the observer (monitor the error responses of performer). They alternated their roles in two successive blocks. To further verify the role of the interaction context, a control experiment was conducted in the individual context (Experiment 2). The results revealed that (1) larger PES effect was observed in females than in males in the interaction context; (2) the sex difference of PES effect mainly benefited from the opposite-sex interaction; (3) larger PES effect was observed in the interaction context than in the individual context; (4) females’ performance was influenced after an interaction with a same-sex or opposite-sex partner, whereas males’ performance was merely influenced after an interaction with an opposite-sex partner. Taken together, these findings may suggest that (1) interaction context modulates the PES effect differently for females and males; (2) females are more susceptible to social information and hence more effective to adjust the post-error behaviors.

Keywords: post-error slowing (PES), the same-sex interaction, the opposite-sex interaction, sex difference, color flanker task

Introduction

Cognitive control as a hallmark of cognitive flexibility plays an important role in monitoring and adjusting to the diverse environments and situations. An important aspect of cognitive control is the ability to detect and adjust error behaviors. Following error responses, three types of behavioral adjustments have been reported in the post-error trials (Danielmeier and Ullsperger, 2011; Ullsperger et al., 2014): post-error slowing (PES), post-error improvement in accuracy (PIA), and post-error reduction of interference (PERI). However, among these adjustments, the slowing in reaction time (RT) is the most stable phenomenon in the post-error trials. PES describes the prolonged RT in correct trials following errors (eC trial) compared to RT in correct trials following correct trials (cC trial; see Danielmeier and Ullsperger, 2011, for a review; Rabbitt, 1966; Laming, 1979; Dutilh et al., 2012; Wang et al., 2015).

When we perform a task with another person together, monitoring and learning from other’s actions is essential to efficiently adapt our behavior and to optimize task performance. In recent years, considerable evidences have demonstrated that the individual can adjust one own’s performance according to the error information of the person one is interacting with, as reflected in slower RT (Castellar et al., 2011; de Bruijn et al., 2011; Wang et al., 2015). Moreover, several studies have extended above finding by showing the difference in the slowing following observation errors depending on the social context (i.e., cooperative and competitive contexts; Castellar et al., 2011; de Bruijn et al., 2011). Research by de Bruijn et al. (2011) has found that participants slow down following observation errors only in the cooperation context by employing a social speeded choice RT task. Therefore, these studies suggest that the PES effect can be modulated by the social factor. However, it remains unknown whether and how other’s observation influence one own’s performance.

Additionally, the individual difference in cognitive control has aroused the researchers’ great interest (e.g., Hester et al., 2004; Garcia-Garcia et al., 2008; Li et al., 2009; Stoet, 2010; Kret and De Gelder, 2012; Wang et al., 2014). In particular, sex differences in cognitive monitoring have been documented within numerous literatures. For example, females are more distracted by the task-irrelevant information in the conflict-related tasks such as the flanker task (e.g., Bayliss et al., 2005; Garcia-Garcia et al., 2008; Stoet, 2010; Clayson et al., 2011). In a related vein, females have stronger inhibition ability to withhold inappropriate behaviors in the cognitive tasks such as the stop-signal task and the odd-ball task (e.g., Yuan et al., 2008; Li et al., 2009). However, although the ability of error monitoring is stronger in females, the adjustment following errors is not different between sexes (e.g., Li et al., 2009; Larson et al., 2011). It is worth noticing that heterosexual males’ cognitive performance is impaired after an interaction with one of female participants (Karremans et al., 2009), even when they merely anticipate an interaction with one female (Nauts et al., 2012). However, this cognitive impairment is not observed in females. To our knowledge, there have been no studies exploring the sex difference of the PES effect in the social context. Thus, it is appealing to investigate whether social contexts modulate the PES effect in a different way or to a different degree for females and males.

In the present study, we utilized two experiments to examine aforementioned issues by employing a modified flanker task (Eriksen and Eriksen, 1974). For the flanker task, Participants are required to respond to the central target while ignoring the flankers that may have the same response with the target (congruent trial) or an different response with the target (incongruent trial). Experiment 1 examined whether behavior was adjusted differently for females and males during social interactions. In this Experiment, the participants were instructed to interact with the other participant with same-sex or opposite-sex to execute a color flanker task (they sat next to each other). Two participants worked as a team, one of them was responsible for performing the task (performer), and the other was responsible for monitoring the performer’ responses (observer; cf. Koban et al., 2010). When the performer committed errors, the observer needed to press the spacebar in another keyboard to warn the performer. After completing one block, two participants altered their roles. In addition, to ensure the task involvement of the participants acting the observer, before the experiment, participants were informed that they would be rewarded as a team (Koban et al., 2010; Castellar et al., 2011; Dutilh et al., 2012). If both of them had good performance (more accurate than the other pairs not only performing the task but also counting the other’s errors), they would be paid extra bonus (Castellar et al., 2011). To further explore whether the social interactions influenced differently the behavioral adjustment for females and males, the control Experiment 2 was performed by the individuals alone.

Experiment 1

In this experiment, two participants with same-sex or opposite-sex were instructed to sit next to each other and complete the task together. Fukushima and Hiraki (2006) instruct participants to play a competitive two-person gambling game and find that females treat the loss of opponent the similar as one own’s. This finding may suggest females are more sensitive to socio-negative events such as unpleasant events or monetary loss. In addition, prior studies have affirmed that the error monitoring of females is stronger (e.g., Li et al., 2009; Larson et al., 2011). Based on these studies, we predict that the PES effect of females should be larger than males in the interaction context. Otherwise, the interaction context modulates the PES effect equally for the females and males.

Materials and Methods

Participants

Ninety-four healthy participants (47 females, all right-handed with normal color perception, and normal or corrected-to-normal vision, aged 19–30 years) took part in the experiment for payment. 44 participants (22 females) were assigned to the same-sex interaction and 50 participants (25 females) to the opposite-sex interaction. They were tested in pairs. All participants provided written informed consent and all of them were naive to the purpose of the experiment. Data from six participants (three pairs) were discarded, one pair (two females) in the same-sex interaction misunderstood the instruction and two pairs in the opposite-sex interaction lacked enough error trials. As suggested by a number of researchers (e.g., Olvet and Hajcak, 2009; Larson et al., 2010; Picton et al., 2012), between 6 and 14 trials are required for a reliable error processing. Thus, only those who had at least 15 trials for each condition were included in the analyses. Finally, data from 88 (43 females) participants were used for analyzing. Participants were free from neurological diseases and reported no history of psychoactive medication use. Approval of the study was made by the Human Research Ethics Committee of the Southwest University of China.

Apparatus and Stimuli

The experiment was conducted using E-Prime software (Psychology Software Tools, Inc. Pittsburgh, PA, USA) and run on a 17-inch monitor of a Dell computer (with a refresh rate of 85 Hz and a resolution of 1024 × 768). Participants were seated in a comfortable chair in a sound-attenuated chamber at a distance of approximately 60 cm away from the screen.

Eight color filled rounds (red, green, yellow, blue, orange, cyan, purple, and tawny) were employed to constitute the color flanker task, which was displayed on the gray background. The RGB values for stimulus colors were 255, 0, 0 (red); 0, 255, 0 (green); 255, 255, 0 (yellow); 0, 0, 255 (blue); 230, 110, 10 (orange); 0, 150, 80 (cyan); 110, 50, 160 (purple), and 150, 140, 80 (tawny), respectively. In each trial, a line of five color rounds was presented in the center of the screen. The central color round was the target and the remaining color rounds were the flankers. The 2-1 mapping rules (Zhang and Kornblum, 1998; De Houwer, 2003; Chen et al., 2011) was adopted to achieve enough error trials for the reliable analysis of the PES effect. When the target and the flankers were mapped onto the same response key, the trials were defined as congruent trials; when the target and the flankers were mapped onto different response keys, the trials were defined as incongruent trials. Congruent and incongruent trials were pseudo-randomly sequenced with equal frequency. Participants were instructed to respond to the central color round as quickly and correctly as possible on horizontally-arranged number keys on a standard keyboard. The stimulus-response mappings corresponding to the color rounds were counterbalance across the participants. For half of the participants, red and tawny rounds were mapped to the one key (left middle finger), cyan and orange rounds were mapped to the two key (left index finger), yellow and purple rounds were mapped to the nine key (right index finger), and blue and green rounds were mapped to the 0 key (right middle figure). And for the other half of the participants, red and tawny rounds were mapped to the 0 key (right middle figure), cyan and orange rounds were mapped to the nine key (right index finger), yellow and purple rounds were mapped to the two key (left index finger), and blue and green rounds were mapped to the one key (left middle finger).

Experimental Design

The sex differences in the PES effect were examined via a 2 by 2 by 2 mixed design, in which response type (eC and cC) as the within-subjects factor, sex of performer (female and male) and sex of observer (female and male) as the between-subjects factors. Before the formal experiment, participants first performed two practice blocks of 30 trials to get familiarized with the task and to reduce learning effects. The formal experiment consisted of six successive blocks (120 trial each, 720 trials in all), with a 30 s break between blocks, lasting approximately 60 min.

Experiment Procedure

Figure 1A displays the timing of one trial in the interaction context. Each trial started with a 300 ms fixation (+/*). The cross (+) fixation cued participant A as a performer, while the asterisk (*) fixation cued participant B as a performer. After completing one block, their roles exchanged. When the fixation disappeared, a blank screen was presented in a random interval for 300–500 ms. Then, an array of five color rounds presented in the center of the screen for 110 ms, followed by a 1,000 ms blank screen. Participants needed to respond to the central target in this interval, with a maximal time limitation of 1,000 ms. After a response was made, the blank screen immediately disappeared and the response of the performer (one of the four response keys) presented in the screen for 1,000 ms, which would help the observers to realize the correctness of key-press of their partners. Next, the number sign (#) presented the screen for 1,000 ms. Here, the observers were instructed to press the spacebar when they observed their partners made wrong responses (terminated after the spacebar was pressed within this interval). When the performers’ responses were correct, the observers did not need to respond, and the number sign would last for 1,000 ms. Finally, a 800–1,000 ms interval was presented (interval varied randomly).

**Schematic illustration of the procedures. (A)** The experimental setup of the interaction context used in Experiment 1. **(B)** The experimental setup of the individual context used in Experiment 2. Notably, in the interaction context, two participants were arranged to sit next to someone of the same sex or the opposite sex and to complete the task together. One was responsible for performing the flanker task (performer), and the other was responsible for observing the partner’s performance (observer). They exchanged their roles after completing one block. The cross fixation (+) cued participant A as the performer, and the asterisk (*) cued participant B as the performer. In the individual context, the participant completed the task alone.

Results

For the RT analyses, the error trials, and trials that the RTs were shorter than 150 ms and longer than 1,000 ms were eliminated. In total, 11% of all data was excluded. The results for PES, post-error accuracy and feedback accuracy in the interaction context as a function of experimental factors are shown in Figures 2 and 3.

**The result of feedback accuracy in Experiment 1.** Error bars denote standard error.

Post-Error Slowing

For the PES effect, a three-way analysis of variance (ANOVA) was conducted, with response type (eC and cC) as the within-subjects factor, sex of performer (female and male) and sex of observer (female and male) as the between-subjects factors. The Greenhouse–Geisser correction was employed where appropriate. The results revealed that the main effect of response type was significant [F_(1,84) = 102.36, p < 0.001], with slower response in eC trials. However, neither the main effect of sex of performer nor the main effect of sex of observer was significant [all F_(1,84) < 1]. Nevertheless, the three-way interaction between response type, sex of performer and sex of observer was significant, F_(1,84) = 15.55, p < 0.001. Post hoc tests revealed that females worked with another female observer, the mean RT of eC trials (654 ± 19 ms) was significantly larger than that of cC trials (625 ± 17 ms), F_(1,84) = 8.81, p < 0.05. Females worked with another male observer, the mean RT of eC trials (779 ± 18 ms) was significantly larger than that of cC trials (693 ± 15.7 ms), F_(1,84) = 89.55, p < 0.001. Males worked with another male observer, the mean RT of eC trials (661 ± 18 ms) was significantly larger than that of cC trials (632 ± 16 ms), F_(1,84) = 9.43, p < 0.05. Males worked with another female observer, the mean RT of eC trials (735 ± 18 ms) was significantly larger than that of cC trials (690 ± 15.7 ms), F_(1,84) = 24.38, p < 0.001. Moreover, the two-way interaction between response type and sex of performer was significant [F_(1,84) = 4.98, p < 0.05], showing the magnitude of the PES effect was significantly larger for females than for males. Post hoc tests revealed that the mean RT of eC trials (717 ± 13 ms) was significantly larger than that of cC trials (659 ± 11 ms) for females [F_(1,84) = 74.38, p < 0.001], suggesting the PES effect (RT_eC–cC = 57.70 ms) occurred in females. Likewise, the mean RT of eC trials (698 ± 13 ms) was significantly larger than that of cC trials (661 ± 11 ms) for males [F_(1,84) = 31.89, p < 0.001], suggesting the PES effect (RT_eC–cC = 36.85 ms) occurred in males. The two-way interaction between gender of performer and gender of observer was significant [F_(1,84) = 23.63, p < 0.001]. Post hoc tests revealed that the mean RT was significantly slower for the opposite-sex interaction (736 ± 16 ms) than for the same-sex interaction (640 ± 17 ms) in females, F_(1,84) = 16.32, p < 0.001. Likewise, the mean RT was significantly slower for the opposite-sex interaction (712 ± 16 ms) than for the same-sex interaction (647 ± 17 ms) in males, F_(1,84) = 7.95, p < 0.01.

Further, to make clear whether the sex difference of PES effect was influenced by the sex of observer, we analyzed the same-sex interaction and the opposite-sex interaction respectively. Two two-way ANOVAs were conducted, with response type as the within-subjects factor and the sex of performer as the between-subjects factor. In the same-sex interaction (Figure 2A), the results revealed that the main effect of response type was significant, F_(1,40) = 45.29, p < 0.001. However, neither the main effect of sex of performer nor the two-way interaction was significant, all F_(1,40) < 1. In the opposite-sex interaction (Figure 2C), the results revealed that the main effect of response type was significant, F_(1,44) = 67.17, p < 0.001. However, the main effect of sex of performer was not significant, F_(1,44) < 1. Importantly, the two-way interaction reached a significant level, F_(1,44) = 6.63, p < 0.05, showing the magnitude of the PES effect was significantly larger for females than for males. Post hoc tests revealed that the mean RT of eC trials (779 ± 22.7 ms) was significantly larger than that of cC trials (693 ± 19 ms) for females [F_(1,44) = 58.01, p < 0.001], suggesting the PES effect (RT_eC–cC = 86.35 ms) occurred in females. Likewise, the mean RT of eC trials (735 ± 22.7 ms) was significantly larger than that of cC trials (690 ± 19 ms) for males [F_(1,44) = 15.79, p < 0.001], suggesting the PES effect (RT_eC–cC = 86.35 ms) occurred in males.

Post-Error Accuracy

Post-error accuracy analysis focused on the performance accuracy of trials following errors when participants acted as performers. For the post-error accuracy analysis, a three-way ANOVA was conducted, with response type (post-error accuracy and post-correct accuracy) as the within-subjects factor, and sex of performer (female and male) and sex of observer (female and male) as the between-subjects factors. The Greenhouse–Geisser correction was employed where appropriate. The results revealed that the main effect of response type was significant [F_(1,84) = 22.99, p < 0.001], with more errors after error responses. The main effect of sex of performer was significant [F_(1,84) = 3.86, p < 0.05], with higher performance accuracy in females. Moreover, the interaction between sex of performer and sex of observer was significant [F_(1,84) = 14.90, p < 0.001]. Post hoc tests revealed that the performance accuracy was higher for males working with the opposite-sex partner (92 ± 1.5%) than those working with the same-sex partner (83 ± 1.5%), F_(1,84) = 14.64, p < 0.001. However, no any other significant results was found, all F_(1,84) < 2.20.

As the analysis of PES effect, we analyzed respectively the same-sex interaction and the opposite-sex interaction for the post-error accuracy. Two two-way ANOVAs were conducted, with response type as the within-subjects factor and the sex of performer as the between-subjects factor. In the same-sex interaction (Figure 2B), the results revealed that the main effects of response type [F_(1,40) = 17.07, p < 0.001] and sex of performer [F_(1,40) = 4.50, p < 0.05] were significant. However, the two-way interaction was not significant, F_(1,40) < 1. In the opposite-sex interaction (Figure 2D), the results revealed that the main effect of response type, F_(1,44) = 6.58, p < 0.05. However, neither the main effect of sex of performer nor the two-way interaction was significant, all F_(1,44) < 1.

Feedback Accuracy

Feedback accuracy analysis indicated the correct feedback rates of participants when they acted as the observer. An individual-sample t-test was conducted (Figure 3) to investigate the sex difference in feedback accuracy. The results revealed that feedback accuracy was not significantly different between females (67 ± 28%) and males (64 ± 27%), t₍₈₆₎ = 0.40, p > 0.1.

Discussion

Pairs of participants performing the color flanker task together revealed that the PES effect was observed for both females and males. The result pattern of PES effect manifested that the mean RT of eC trials was significantly prolonged compared to that of cC trials. Moreover, the PES effect was observed significantly larger in females than in males, suggesting sex difference exists in the post-error behavioral adjustment in the interaction context. More importantly, the sex difference of the PES effect only appeared in the opposite-sex interaction, but not in the same-sex interaction. Additionally, the mean RT was significantly slower for the opposite-sex interaction than for the same-sex interaction in both females and males. These findings may suggest that an interaction with unacquainted opposite-sex others requires more cognitive control to complete current task.

Results for the accuracy of post-error response were revealed that higher performance accuracy in females than in males. It might be that females were more conscientious when they worked with a partner and therefore put more emphasis on accuracy than speed after an error. This may explain why magnitude of the PES effect was higher for females in the interaction context. Additionally, the post-error accuracy of male participants was higher in the opposite-sex interaction relative to the same-sex interaction. This result may suggest that males’ cognitive functioning is influenced by the sex of partner, and in a way of improvement. However, for the feedback accuracy, the performance between sexes was comparable, suggesting that the task involvement of females and males did not different when they acted as the observers.

Experiment 2

Since previous studies have affirmed that there is no sex difference in the PES effect under the individual context (Hester et al., 2004; Li et al., 2009; Larson et al., 2011). To compare with the findings of previous studies and further verify the role of interaction context in sex difference of the PES effect, we conducted Experiment 2 without the interaction context. For this purpose, we arranged the participants to execute the color flanker task alone. In addition, the instruction that two of participants were as a team to be rewarded according to their joint performances was eliminated.