Abstract
Neuroscientists frequently use a certain reasoning to establish the existence of distinct conscious and unconscious processes in the brain. In a typical paradigm, conscious judgements are shown to be at chance-level when categorizing stimuli, while those same stimuli have clear and significant indirect ("unconscious") effects on reaction times or neuronal activity. Based on such a pattern of results it is typically concluded that unconscious processes can classify the stimuli better than conscious processes (this reasoning is sometimes called the dissociation paradigm). We show that this reasoning is flawed for theoretical reasons: Indirect effects do not allow one to infer that they were based on a good underlying classification performance. In fact, indirect effects would be compatible with any above-chance classification performance. That is, the indirect effect is a necessary but not sufficient condition for good underlying classification performance. We illustrate the fallacy with a recent study (ten Brinke, Stimson, & Carney, 2014, Psychological Science, 25(5), 1098-1105) 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 classification performance was close to chance - just as the conscious performance. This illustrates the flaws of a widely used statistical reasoning, the benefits of open-data practices, and the need for careful reconsideration of other studies on the visual system (e.g., on unconscious priming) using the same rationale.
Meeting abstract presented at VSS 2015