Abstract
Though it is well known that non-numerical magnitudes, such as cumulative area, element size, and density influence the perception of numerosity (Gebuis & Reynvoet, 2012; Leibovich et al., 2017), it is unclear whether these interactions reflect independent representations that interface vis-à-vis other systems (e.g., language) or, conversely, a holistic perception in which numerosity is not fully separable from other magnitudes. To dissociate these accounts, the present study drew from classic work on the perception of multidimensional stimuli in which perceived similarity for stimuli composed of integral dimensions is best explained by Euclidean distance and separable dimensions by city-block distance (Garner & Felfoldy, 1970). In Experiment 1, we utilized a restricted classification task (Ward et al., 1986) in which adult participants (n = 20) provided explicit similarity judgments about dot arrays that varied parametrically in number and cumulative area (CA). We found that similarity in this multidimensional space was best explained by Euclidean distance, suggesting that number and CA are integral dimensions akin to other classic integral dimensions (brightness/saturation) and in contrast to classic separable dimensions (shape/color), also tested here. In Experiment 2, we extended these findings by ruling out a possible effect of explicit categorization. Participants (n = 20) completed a match-to-sample task in which similarity was assessed implicitly as the degree of interference from distractor stimuli on reaction time. Consistent with Experiment 1, we found that similarity for dot arrays that varied in number and CA was best modeled by Euclidean distance, suggesting integrality. Again, number and CA were comparable to another example of integral dimensions (radial frequency contours) but different from separable dimensions (thickness/curvature) dimensions, also assessed here. Taken together, these findings provide support for integral representations of number and CA, suggesting a perception of magnitudes in non-symbolic sets that are processed holistically as unitary percepts.
Acknowledgement: This work was supported by a National Institutes of Health (NIH) institutional training grant (T32 HD071845) to LSA.