Fragile Mentalizing: Lack of Behavioral and Neural Markers of Social Cognition in an Established Social Perspective Taking Task when Combined with Stress Induction.

The growing field of social neuroscience is reliant on the development of robust, ecologically valid paradigms for simulating social interaction and measuring social cognition in highly controlled laboratory settings. Perspective taking is a key component of social cognition, and accordingly several paradigms aimed at measuring perspective taking exist. A relatively novel paradigm is the ball detection task, in which participants and a virtual agent form independent beliefs about the presence of a target stimulus behind an occluder. Previous studies have shown that incongruent trials (in which the participant's and the agent's beliefs differ) affect participant reaction times and elicit increased neural activity in the so-called mentalizing network. This paradigm has important advantages over previous ones, in that experimental conditions can be fully randomized, and ceiling effects are not found even for adult populations. Here, we combined this paradigm with a stress induction and a nonstressful control task. In an online study, we found no evidence of perspective taking at the behavioral level. Combining the task with functional magnetic resonance imaging, we found no evidence of perspective taking at the behavioral or neural level, even for the control condition. While this paradigm is reliable on its own, implementing it in the context of a task-switching paradigm appears to reduce participants' focus on task-irrelevant perspective taking elements. Our findings highlight the fragility of existing social cognition paradigms and the need for reliable, simple, and ecologically valid measures of perspective taking.