Idiosyncratic Relation Between Human Brain Activity and Behavior

Human behavior is known to be idiosyncratic, yet research in neuroscience typically assumes a universal brain-behavior relationship. Here we test this assumption by estimating the level of idiosyncrasy by compare two different aspects of brain function: task- and behavior-based activation. Human subjects (N=50) perform a perceptual task with confidence inside an MRI scanner. For each subject, task-based activation strength is estimated from blocks of 8 trials. We first show that task-based activation maps are both stable within an individual (r = 0.99) and similar across people (r = 0.59). Behavior-based activation was estimated by calculating the differences between blocks with performance above and below the median reaction time (RT) and confidence (Conf), respectively. Critically, although behavior-based activation maps are also stable within an individual for reaction time (r = 0.68) and confidence (r = 0.49), they strongly diverge across people (RT: r = 0.11; Conf: r = 0.08). A computational model that jointly generates brain activity and behavior reveals that within-person factors have a 6x larger effect than group factors in determining behavior-based activations. These findings demonstrate that unlike task-based activity that is mostly similar among people, the relation between brain activity and behavioral outcomes is largely idiosyncratic.