Approach-avoidance conflict recruits lateral frontoparietal and cinguloinsular networks in a predator-prey game setting.

Objects associated with both reward and threat produce approach-avoidance conflict (AAC). Although our day-to-day encounters with AAC objects are dynamic and interactive, the cognitive neuroscience literature on AAC is largely based on experiments that use static stimuli. Here, we used a dynamic, interactive, video-game environment to test neural substrates implicated in processing AAC in a more ecologically valid setting. While undergoing functional magnetic resonance imaging (fMRI), subjects (N = 31) played a predator-prey video game, guiding an avatar through a maze containing six types of aversive or appetitive agents. Of the six agent types, two were "non-AAC" and either always healed or always harmed the player's avatar on contact. The other four were "AAC," healing or harming the avatar probabilistically. Results revealed that imminence (inverse of distance) between a player's avatar and an environmental agent was a strong predictor of activation in three brain networks: the cinguloinsular (CI), dorsal frontoparietal (DFP), and occipitotemporal (OT). Additionally, two distinct temporal patterns of heightened activation with AAC agents emerged in two networks: the CI network responded with a transient spike of activation at trial onsets, followed by rapid decay, whereas the lateral frontoparietal (LFP) network showed sustained activation across the whole trial. We conclude that, in an interactive, dynamic setting, the roles of the CI and LFP networks appear to be complimentary, with the CI involved in distinguishing between AAC and non-AAC agents when they first appeared and the LFP involved in maintaining a behavioral mode related to the level of AAC.