Neuropsychological evidence for a dissociation in counting and subitizing

Abstract

There is a long and ongoing debate about whether subitizing and counting are separable processes. In the present paper we report a single case, MH, who presents with a dissociation in subitizing and counting. MH was spared in his ability to enumerate small numbers accurately along with a marked inability to count larger numbers. We show that non-visual counting was intact and visual counting improved when a motor record of counting could be maintained. Moreover, when larger numbers of items were spatially grouped into 2 subitizable units, performance dramatically improved. However, color grouping did not aid MH's performance, despite his being sensitive to color segmentation. In addition, MH made more re-visits of inspected locations than controls, and he was less aware of a re-visitation being made. The data cannot be explained in terms of general working memory problems (verbal working memory was relatively spared), or general number comprehension problems (e.g., simple sums and counting of auditory items was intact); but they can parsimoniously be accounted for in terms of impaired visuo-spatial memory. The findings support the argument that at least some processes are specific to counting and are not required for subitization – in particular spatial coding and memory for previously inspected locations.

Keywords:

• Neuropsychology
• Enumeration
• Subitizing
• Counting
• Visuo-spatial

This work was supported by grants from the MRC and the Stroke Association. We thank MH for his kind participation.

Notes

¹ The counting range was considered to involve the numbers 5–8 here, since we used a maximum 9-item display and responses to 9 items may be affected by guessing (following previous enumeration studies (e.g., Trick & Pylyshyn, 1993).

² This slope was calculated on numerosities 2 and 3 only due to this patient demonstrating unreliable RTs for numerosity 1 (SD = 400.29) compared to numerosities 2 (SD = 165.18) and 3 (SD = 248.52). This was corroborated by Levene's test of homogeneity of variance in a comparison of numerosities 1–3, F(2, 82) = 4.194, p = .018.

³ The slopes here are expressed in order of sec given that RTs were recorded by stopwatch.