Abstract

To evaluate unconscious processing of numbers, we used a classic priming paradigm: In a ‘direct’ task, participants classified a masked ‘prime’-stimulus (i.e., decided whether it is a number larger or smaller than 5). In this task, participants typically perform close to chance, suggesting no conscious awareness of the prime. In an ‘indirect’ task, the masked prime was also presented, but participants responded to another stimulus (decided whether that number was larger or smaller than 5). Typically, the prime has effects on reaction times in the indirect task, which is usually interpreted as preserved unconscious processing of the prime. However, this paradigm could be problematic, because participants are restricted to a binary response in the direct task. To assess whether participants have access to a richer, continuous representation of the prime in the direct task, we had 12 participants give a binary response and also judge their confidence on a continuous scale. We found that confidence modulated the accuracy of prime detection: Overall accuracy (mean+/-SEM in percent): 58.6+/-1.4; accuracy in high/low-confidence trials: 64.4+/-2.1 / 53.5+/-1.1, respectively. We will discuss to which degree this effect could be exploited to improve the response format in such paradigms.