Higher attentional costs for numerosity estimation at high densities

Antonella Pomè,Giovanni Anobile,Guido Marco Cicchini,David Charles Burr,Aurora Scabia

doi:10.3758/s13414-019-01831-3

Abstract

Humans can estimate numerosity over a large range, but the precision with which they do so varies considerably over that range. For very small sets, within the subitizing range of up to about four items, estimation is rapid and errorless. For intermediate numerosities, errors vary directly with the numerosity, following Weber’s law, but for very high numerosities, with very dense patterns, thresholds continue to rise with the square root of numerosity. This suggests that three different mechanisms operate over the number range. In this study we provide further evidence for three distinct numerosity mechanisms, by studying their dependence on attentional resources. We measured discrimination thresholds over a wide range of numerosities, while manipulating attentional load with both visual and auditory dual tasks. The results show that attentional effects on thresholds vary over the number range. Both visual and auditory attentional loads strongly affect subitizing, much more than for larger numerosities. Attentional costs remain stable over the estimation range, then rise again for very dense patterns. These results reinforce the idea that numerosity is processed by three separates but probably overlapping systems.

Highlights

Humans can estimate numerosity over a large range, but the precision with which they do so varies considerably over that range
We tested the effect of attentional load on numerosity perception over a wide range of numerosities
We first examined whether the attentional manipulations affected point of subjective equality (PSE)

Summary

Introduction

Humans can estimate numerosity over a large range, but the precision with which they do so varies considerably over that range. More recent evidence points to the existence of a third mechanism coming into play when judging numerosity at high densities, which might be linked to the perception of texture density This third system is thought to Sydney, Australia be activated when visual items are highly packed and difficult to segregate spatially (Anobile, Cicchini, & Burr, 2014). Evidence for subitizing comes largely from a discontinuity in reaction times, response variability, and accuracy These parameters are consistently lower for numbers of 1 to 4, with performance sharply declining for larger numbers outside the subitizing range (Atkinson, Campbell, & Francis, 1976; Choo & Franconeri, 2014; Mandler & Shebo, 1982; Revkin, Piazza, Izard, Cohen, & Dehaene, 2008). Studies have shown clear differences in the psychophysical laws governing precision for relatively sparse as compared with packed dot patterns: For sparse patterns, the discrimination thresholds are higher and obey Weber’s law; at higher numerosities, they decrease with the square root of numerosity (for a review, see Anobile et al, 2014; Anobile et al, 2015)

Methods

Results

Conclusion