Experimental procedure. (A) Group identification manipulation. For each High Group Identification condition, the participants in each triad were invited to chat with each other to introduce themselves and find three-person-shared features for minutes. For each Low Group Identification condition, the participants in each triad were asked to chat with each other about the main courses they had been taking during this semester without being explicitly asked to find shared features. (B) The rate of group identification. The participants in each triad rated the group identification in their group before and after the manipulation. (C) Group identification manipulation check. We examined how the level of group identification changed when we manipulated it, for both High and Low Group Identification conditions. Group identification_1, Group identification before manipulation; Group identification _2, Group identification after manipulation. (D) The procedure of task. First, participants completed a series of individual difference questionnaires before the task. Then, each group member received 18 common information and 2 private information. They read their information within 5 minutes. After that, each triad was required to complete verbal information exchange, comprising both group sharing and group discussion Each group member texted the private information to other members by Tencent Meeting during group sharing for 5 minutes, and they discussed the information currently disclosed with others orally during the group discussion for 20 minutes. Ultimately, the groups had a period of 5 minutes to answer the questions.

Group identification leads to differences in collective performance. (A) Manipulated group identification led to group differences in collective performance. (B) Group identification was positively correlated with better collective performance. The Pearson correlation and its associated analyses were based on the data from group identification_2. *p < 0.05.

The differences in collective performance are correlated with single-brain activation. (A) Significant differences in single-brain activations between the High and Low Group Identification groups were observed in the DLPFC (CH4). (B) Greater single-brain activation in the DLPFC (CH4) was associated with higher individual performance. (C) A serial mediation model suggested that single-brain activation mediated the relationship between group identification of each person and individual performance. *p < 0.05, **p < 0.01, ***p < 0.001.

The differences in collective performance are correlated with neural synchronization. (A) A significant difference between the High and Low Group Identification groups in the OFC (CH21) was observed (p-value, FDR corrected). (B) Greater GNS in the OFC (CH21) was associated with higher collective performance. (C) A mediation model suggested that GNS mediated the relationship between group identification of each group and collective performance. *p < 0.05, **p < 0.01.

Brain activation connectivity links the single-brain activation and the corresponding GNS. (A) A significant difference between the High and Low Group Identification groups in the DLPFC-OFC (CH4-CH21) correlation was observed (p-value, FDR corrected). (B) A significant difference between the High and Low Group Identification groups in the similarity in individual-collective performance was observed. (C) A stronger DLPFC-OFC (CH4-CH21) correlation was linked to a decrease in the distance between individual and collective performance. (D) The analyses suggested that the relationship between individual single-brain activation and the corresponding GNS was regulated by brain activation connectivity. *p < 0.05, ***p < 0.001.

The dynamic neural process. There was a time delay in the transition from individual to collective decisions, and brain activation connectivity was in the middle. The processing of information flow in the single-brain activation (there was a significant increase in single-brain activation at approximately 7 minutes into the task), DPPFC-OFC connectivity (there was a significant increase in connectivity at approximately 12 minutes into the task), and ultimately the GNS (there was a significant increase in GNS at approximately 17 minutes into the task).

The two-in-one neural model explains how group identification influences collective performance. In the first step, group identification influences individual performance, which is associated with significant single-brain activation in the DLPFC of each group member. In the second step, group identification influences collective performance, which is linked to significant within-group neural synchronization (GNS) in the OFC. These two steps are connected by the DLPFC-OFC connectivity, which modulates the relationship between individual DLPFC activation and GNS in the OFC.

The quality of information exchange is correlated to the effect of group identification. (A) A significant difference between the High and Low Group Identification groups in the quality of information exchange was observed. (C) Higher quality of information exchange was linked to better collective performance. *p < 0.05.