Neural correlates of static and dynamic social decision-making in real-time sibling interactions

Abstract

In traditional game theory tasks, social decision-making is centered on the prediction of the intentions (i.e., mentalizing) of strangers or manipulated responses. In contrast, real-life scenarios often involve familiar individuals in dynamic environments. Further research is needed to explore neural correlates of social decision-making with changes in the available information and environmental settings. This study collected fMRI hyperscanning data (N = 100, 46 same-sex pairs were analyzed) to investigate sibling pairs engaging in an iterated Chicken Game task within a competitive context, including two decision-making phases. In the static phase, participants chose between turning (cooperate) and continuing (defect) in a fixed time window. Participants could estimate the probability of different events based on their priors (previous outcomes and representation of other's intentions) and report their decision plan. The dynamic phase mirrored real-world interactions in which information is continuously changing (replicated within a virtual environment). Individuals had to simultaneously update their beliefs, monitor the actions of the other, and adjust their decisions. Our findings revealed substantial choice consistency between the two phases and evidence for shared neural correlates in mentalizing-related brain regions, including the prefrontal cortex, temporoparietal junction (TPJ), and precuneus. Specific neural correlates were associated with each phase; increased activation of areas associated with action planning and outcome evaluation were found in the static compared with the dynamic phase. Using the opposite contrast, dynamic decision-making showed higher activation in regions related to predicting and monitoring other's actions, including the anterior cingulate cortex and insula. Cooperation (turning), compared with defection (continuing), showed increased activation in mentalizing-related regions only in the static phase, while defection, relative to cooperation, exhibited higher activation in areas associated with conflict monitoring and risk processing in the dynamic phase. Men were less cooperative and had greater TPJ activation. Sibling competitive relationship did not predict competitive behavior but showed a tendency to predict brain activity during dynamic decision-making. Only individual brain activation results are included here, and no interbrain analyses are reported. These neural correlates emphasize the significance of considering varying levels of information available and environmental settings when delving into the intricacies of mentalizing during social decision-making among familiar individuals.

Keywords: fMRI; hyperscanning; mentalizing; siblings; social decision‐making; social interaction; theory of mind.

FIGURE 1

Hypothesized model of cognitive processes and regional brain activations related to each phase of decision‐making. The main cognitive processes in each phase are highlighted in bold and framed. (a) Static (strategic) phase includes the intention question “What is your plan?” and offers a binary choice (turn, and continue). During this phase, participants are expected to engage in more individualistic processes by considering what the other might do (mentalizing), evaluating the different possibilities, and planning their behavior accordingly. (b) Dynamic phase represents the car video phase in which participants can decide whether they want to turn (and when to turn within a 5‐s interval) or continue. This phase is expected to be associated with the predicting other's strategy, belief updating, monitoring the other's action, re‐evaluating one's own expectations, and real‐time behavior adjustment. This figure was created with BioRender.com. ACC, anterior cingulate cortex; aIns, anterior insula; Cun, cuneus; Ins, insula; PFC, prefrontal cortex; Prec, precuneus; Str, striatum; TPJ, temporoparietal junction.

FIGURE 2

Representation of the hyperscanning setup and the Interactive Chicken Game task. (a) Hyperscanning set‐up. Each scanner was connected to the server PC, which handled the communication to start the scanners simultaneously and managed the trial information from and to the task PCs. This figure was created with BioRender.com. (b) Interactive Chicken Game task payoff matrix. The numbers at the top and left side outside the table represent the seconds across the trial. The numbers inside the table represent the points each participant won or lost, and in the one‐turning condition depending on the waiting time to turn. For one participant, the outcomes are summarized in the following order from better to worse: One‐turning unilateral defection (participant defects/continuous, outcome: 1–5 points); (d) the other cooperates/turns; (c) >mutual cooperation (both turn, outcome: 0 points) (CC) > one‐turning unilateral cooperation (participant turns, outcome: −1 to −5 points); (c) the other continues; (d) >mutual defection (both continue, outcome −10 points) (DD). (c) Trial design. The waiting phase corresponds to a jittered fixation cross, followed by the intention question (static decision‐making process). Before the car video, a fixed brief fixation cross was presented to signal the participants to get ready. The car video included the dynamic decision‐making process and the feedback video with different endings depending on the outcome. Finally, participants received their outcomes in the feedback phase.

FIGURE 3

(a) Comparison of conditions (both crashing, both turning, and one turning) between and within decision‐making phases. The violin plots depict summary statistics and the kernel density estimation to show the frequency distribution of each condition. In the middle of each density curve, the thick dotted line represents the median, while the thinner solid lines represent the quartiles. (b) Distribution of outcome conditions based on the decision in the dynamic phase, and sex differences between them. (c) Distribution of consistency in the choices between the static and dynamic decision‐making phases, and sex differences. Cons., consistent; Incons, inconsistent. (d) Frequency of the feedback points across trials depending on the RT and choices from both participants. Y‐axis scale: 0 corresponds to the both turning condition. 1, 2, 3, 4, and 5 (−1, −2, −3, −4, −5 represent the same frequency mirrored for the other participant, so it was omitted) correspond to the one‐turning condition for each second of the turning RTs. −10 corresponds to the both crashing condition. *p < .05, **p < .01, ***p < .001.

FIGURE 4

(a) Conjunction analysis results showing the shared activation between static decision‐making and dynamic decision‐making. (b) T‐contrasts of static decision‐making versus dynamic decision‐making. The statistical threshold map is set to p < .05, family‐wise error‐corrected at voxel level. The figure was created with MRIcroGL1.2. software, and the coordinate values (sagittal plane) corresponding to the standard 2D slice coordinate system of the software. ACC, anterior cingulate cortex; aINS, anterior insula; DM, decision‐making; IFG, inferior frontal gyrus; MFG, middle frontal gyrus; MTG, middle temporal gyrus; PCC, posterior cingulate cortex; pINS, posterior insula; SFG, superior frontal gyrus; STS, superior temporal sulcus; TPJ, temporoparietal junction.

FIGURE 5

(a) Static decision‐making turning (cooperation) versus continuing (defection) contrast. (b) Dynamic decision‐making turning (cooperation) versus continuing (defection) contrast. The statistical threshold map is set to p < .05, family‐wise error corrected at voxel level. The figure was created with MRIcroGL1.2. software, and the coordinate values (sagittal plane) corresponding to the standard 2D slice coordinate system of the software. ACC, anterior cingulate cortex; aINS, anterior insula; DM, decision‐making; FP, frontal pole; MCC, midcingulate cortex; MFG, middle frontal gyrus; PCC, posterior cingulate cortex; pINS, posterior insula; SMA, supplementary motor area; SMG, supramarginal gyrus.