Abstract
The rapid and accurate estimation of small sets of objects, a process known as subitizing, has been mainly studied in isolation. However, objects rarely occur in isolation but are surrounded by other objects or embedded in textured surfaces. Here we investigate contextual effects on numerosity estimation by examining how grouping and regularity affect behavioural and event-related potential (ERP) responses to numerosity estimation. Stimuli were briefly presented (120ms) Gaussian-dot patterns consisting of small number (3 to 6) of bright (or dark) randomly-distributed target items embedded within dark (or bright) context elements. In our ERP experiment we varied the type of context as follows: Mirror-Symmetric, Random, Grid, and No-context. In a separate behavioural experiment, we varied element grouping within the Grid by using small and large displacements to create loose or tight groups, i.e., groupitize the Grid elements. In both experiments, participant’s task was to indicate the number of target elements as quickly and as accurately as possible. We found: (a) an increase in N2 ERP amplitude (250-350ms) for Grid compared to symmetric and random context conditions, which was independent of target numerosity; (b) that Mirror-Symmetric context did not increase N2 amplitude, nor differ from the Random context, implying that the symmetry had little unique influence upon participant’s task performance; (c) a late mid-frontal ERP between 400-650ms, regardless of context, that was greater in amplitude for small (3 & 4) compared to larger (5 & 6) numbers of elements; and separately, (d) higher performance for a regular Grid, but loose grouping led to poorer accuracy compared to tight grouping, with no effects on RTs. Altogether, these findings show that groupitizing facilitates numerosity estimation and Grid, but not random or symmetric context, increases the N2 amplitude, implicated in the perception of regularity. We conclude that context processing occurs before numerosity estimation.