TMS-induced inhibition of the left premotor cortex modulates illusory social perception

Abstract

Communicative actions from one person are used to predict another person's response. However, in some cases, these predictions can outweigh the processing of sensory information and lead to illusory social perception such as seeing two people interact, although only one is present (i.e., seeing a Bayesian ghost). We applied either inhibitory brain stimulation over the left premotor cortex (i.e., real TMS) or sham TMS. Then, participants indicated the presence or absence of a masked agent that followed a communicative or individual gesture of another agent. As expected, participants had more false alarms in the communicative (i.e., Bayesian ghosts) than individual condition in the sham TMS session and this difference between conditions vanished after real TMS. In contrast to our hypothesis, the number of false alarms increased (rather than decreased) after real TMS. These pre-registered findings confirm the significance of the premotor cortex for social action predictions and illusory social perception.

Keywords: Behavioral neuroscience; Cognitive neuroscience; Neuroscience.

Graphical abstract

Figure 1

Experimental design The experimental design was adapted from studies of Friedrich,^, Zillekens, Manera, and colleagues. (A) Task conditions: A well-recognizable agent A performed either a communicative (COM) or an individual (IND) gesture. Agent B was blended into a cluster of randomly moving noise dots and performed a gesture in response of agent A in half of the trials (i.e., signal trials). In the other half of the trials, agent B was entirely replaced by the noise dots (i.e., noise trials). The task of the participants was to report the presence or absence of agent B. The Bayesian ghost was defined as a false alarm (i.e., FA) in the communicative noise trial. Please note that the gray silhouettes only serve illustration purposes and were not visible for the participants. (B) Timing: The main experiment was divided into two sessions on two separate days. In each session, participants received either offline real or sham TMS. Following the TMS application, participants performed the task. Each trial was initiated with a jittered intertrial interval (i.e., ITI) between 1.5 and 3 s, followed by a 1 s fixation cross indicating the subsequent location of agent A. Then, the video displaying agent A and the cluster of dots (with or without agent B) started. Although, agent B was present from the beginning of the video, its response action was delayed between 1.3 and 1.6 s. Participants were required to make a response as long as the video was still ongoing. The video duration was dependent on the specific action and was between 3.6 and 4.3 s long. Please note that the gray silhouettes again only serve illustration purposes and were not visible for the participants.

Figure 2

Real TMS eliminates the condition difference (COM-IND) for false alarms seen in sham TMS Number of false alarms (FA; left column) and correct rejections (CR; right column) as a function of condition (communicative: COM vs. individual: IND) and TMS session (sham vs. real). Translucent dots represent the sampling distribution and opaque dots with error bars represent the sample mean and its 95% confidence interval. Asterisks highlight statistical significance of Wilcoxon signed-rank tests at p_FDR < 0.05, whereas non-significant differences with p_FDR ≥ 0.05 are abbreviated with n.s.