Numerical cognition: Unitary or diversified system(s)?

Many researchers, including Clarke and Beck, describe the human numerical system as unitary. We offer an alternative view - the coexistence of several systems; namely, multiple systems (general magnitude, parallel individuation, and symbolic) existing in parallel, ready to be activated depending on the task/need. Based on this alternative view, we present an account for the representation of rational numbers.

Endogenous attention modulates automaticity of number processing.

Numerals (i.e., symbolic representations of numerical magnitude) are widespread in our environment and are fundamental to many decisions we make. It has been suggested that the processing of numerical magnitude is automatic. Various robust psychological effects, such as the distance effect (Moyer & Landauer, Nature, 215 (5109), 1519-1520, 1967) and the physical size-congruity effect (SiCE; Henik & Tzelgov, Memory & Cognition, 10 (4), 389-395, 1982), support this claim. Importantly, these effects demonstrate that the processing of numerical magnitude occurs unintentionally and while irrelevant to the task. These aspects often serve as criteria to assess the automatic nature of mental processes. However, evidence for the involvement of attention in the processing of magnitude of numerals somewhat subverts the automaticity account that was originally put forward. To reconcile between evidence in support of the automaticity account and evidence that compromises this account, we drew on another account of automaticity. This account distinguishes between strongly automatic and partly automatic mental processes based on their susceptibility to attentional influences. In the current study, we manipulated endogenous attention while participants completed numerical and physical comparisons of numerals, separately. We observed modulations of the SiCE for physical comparisons but not for numerical comparisons of numerals. That is, the processing of numerical magnitude when irrelevant was subjected to attentional influences, but the processing of their physical magnitude (i.e., size) was not. Therefore, we concluded that processing the numerical magnitude is partly automatic, whereas processing their physical magnitude is strongly automatic.

Numerical values modulate size perception.

The link between various codes of magnitude and their interactions has been studied extensively for many years. In the current study, we examined how the physical and numerical magnitudes of digits are mapped into a combined mental representation. In two psychophysical experiments, participants reported the physically larger digit among two digits. In the identical condition, participants compared digits of an identical value (e.g., "2" and "2"); in the different condition, participants compared digits of distinct numerical values (i.e., "2" and "5"). As anticipated, participants overestimated the physical size of a numerically larger digit and underestimated the physical size of a numerically smaller digit. Our results extend the shared-representation account of physical and numerical magnitudes.