Perceptual judgments are resistant to the advisor’s perceived level of trustworthiness: A deep fake approach

As we navigate our environment, we frequently make spontaneous judgments about other's characteristics. Trustworthiness is a particularly important trait, often judged instantly and used to guide decisions, especially in uncertain situations. Although the impact of trustworthiness on social behaviour is well-documented, its influence on more fundamental cognitive processes, such as perceptual decision-making, remains unclear. The present study aims to fill this gap. In the first experiment (N = 100), we validated a new trustworthiness manipulation by applying deep fake technology to create animated versions of perceptually trustworthy, untrustworthy, and neutral static computer-generated faces. In the second experiment (N = 199), the deep fake procedure was applied to a new set of trustworthy and untrustworthy faces that served as advisors during a perceptual decision-making task. Here participants had to indicate the direction of dots that were either moving coherently to the left or the right (i.e., random dot motion task). Contrary to our predictions, participants did not align more with the advice of trustworthy advisors than that of untrustworthy advisors. While participants made faster decisions and reported greater confidence when aligning with the advice, these effects were not influenced by the advisor's perceived trustworthiness. We integrate our findings within theoretical frameworks of advice taking, domain specificity of facial trustworthiness, and task requirements.