ROVING VERSUS FIXED STANDARD IN NUMERICAL COMPARISONS: THE DISTANCE EFFECT REVISITED
A classic marker of (automatic) numerical processing is the distance effect: The time to select the larger member of a pair of numerals decreases as the numerical distance between the numbers increases. The most popular procedure for testing the distance effect is a task in which the participant decides whether the presented number is smaller or larger than a predetermined fixed standard (usually 5). The problem with this arrangement is that each number always maps into the same fixed response. Consequently the observer can bypass the numerical comparison, using the number-response association instead. We show that indeed the distance effect is reduced or eliminated all together when testing comparison with a fixed standard. A novel "roving standard" procedure in which a momentary standard varies from trial to trial forces the observer to make the comparison and thus can remove the confound with the traditional method.