No evidence for unconscious lie detection: A significant difference does not imply accurate classification.
Date created: | Last Updated:
: DOI | ARK
Creating DOI. Please wait...
Description: This is the open material related to our Psychological Science paper with the same title. Please note that there is also an arXiv preprint with slightly different focus: In the Psychological Science paper we focus exclusively on the paper by ten Brinke, Stimson, and Carney (2014), but were not allowed to mention any of the more general implications of our critique for other studies using the same logic as ten Brinke et al. This is unfortunate, because we demonstrate in our paper how a widely used logic can go astray, such that there are indeed general implications for other studies. Those more general implications are (shortly) discussed in the arXiv preprint. Also, due to tight space-constraints at Psychological Science, all the math is only in the arXiv preprint. On the other side, we discuss 'boosting' only in the Psychological Science paper, but not in the arXiv preprint (because this issue arose later during the review process at Psychological Science). Links: Psychological Science paper: - http://www.ecogsci.cs.uni-tuebingen.de/pub/publications.php - http://dx.doi.org/10.1177/0956797615597333 arXiv preprint: - http://arxiv.org/abs/1407.4240 Abstract of Psychological Science paper: ten Brinke, Stimson, and Carney (2014, Psychological Science, 25(5), 1098-1105) reported that humans detect deceit better if they use unconscious processes instead of conscious deliberations. We show that the study is based on a flawed statistical reasoning. The study was published under a new open-data policy that allowed us to reanalyze their data. Using more appropriate methods we found that unconscious performance was close to chance --- just as the conscious performance. Therefore, the study does not provide any evidence for superior unconscious lie detection. Abstract of arXiv preprint: Neuroscientists frequently use a certain statistical reasoning to establish the existence of distinct neuronal processes in the brain. We show that this reasoning is flawed and that the large corresponding literature needs reconsideration. We illustrate the fallacy with a recent study that received an enormous press coverage because it concluded that humans detect deceit better if they use unconscious processes instead of conscious deliberations. The study was published under a new open-data policy that enabled us to reanalyze the data with more appropriate methods. We found that unconscious performance was close to chance - just as the conscious performance. This illustrates the flaws of this widely used statistical reasoning, the benefits of open-data practices, and the need for careful reconsideration of studies using the same rationale.