The SNARC effect emerges only with symbolic numbers

Studying numerical interferences has become a widely used method for investigating the representations that underlie numerical cognition. Here, we contrast the classic Approximate Number System (ANS) framework and a more recently proposed hybrid ANS-Discrete Semantic System (DSS) framework with respect to their distinctive predictions for the symbolic and nonsymbolic SNARC effects (the most extensively studied interference between numbers and space). We compare the symbolic (Indo-Arabic numbers) to the nonsymbolic (arrays of dots) version of a SNARC paradigm (n=77). In contrast to previous studies, in the present experiment, (1) the magnitude is irrelevant for solving the task (a colorjudgment task) for the nonsymbolic task, too, and (2) the nonsymbolic stimuli contain arrays of dots outside the subitizing range, which would not activate the ANS. We found clear evidence for the SNARC effect in the symbolic color task. However, we found no indication of the SNARC effect in the nonsymbolic color task. This pattern of results supports the hybrid ANS-DSS framework, assuming that the SNARC interference is a symbolic effect while refuting the pure ANS view of the SNARC effect, which necessitates the presence of the SNARC interference using nonsymbolic notation, too.