How many can you see at a glance? The role of attention in visual enumeration

There is considerable controversy as to how the brain extracts numerosity information from a visual scene and as to how much attention is needed for this process. Traditionally, it has been assumed that visual enumeration is subserved by two functionally distinct mechanisms: the fast and accurate apprehension of 1 to about 4 items, a process termed “subitizing”, and the slow and error-prone enumeration of larger numerosities referred to as “counting”. Further to a functional dichotomy between these two mechanisms, an attentional dichotomy has been proposed. Subitizing has been thought of as a pre-attentive and parallel process, whereas counting is supposed to require serial attention. In this work, the hypothesis of a parallel and pre-attentive subitizing mechanism was tested. To this aim, the amount of attention that could be allocated to an enumeration task was experimentally manipulated. In Experiment 1, attentional set was manipulated such that attention could either be drawn to the relevant of two subsets to enumerate or had to be distributed to both subsets. Furthermore, the relationship of enumeration to perceptual grouping and item discrimination was explored. In Experiment 2, a dual-task approach was employed in which the amount of attentional resources available to enumeration was systematically modulated by imposing an additional task and by varying its attentional load. Experiment 3 investigated the neural correlates of visual enumeration under attentional load using functional magnetic resonance imaging (fMRI). Results indicated that (1) enumeration, particularly subitizing, was clearly compromised under conditions of distributed or reduced attention. (2) Both the enumeration of small and large numerosities was affected by such attentional manipulations. (3) Subitizing selectively activated brain areas associated with stimulus-driven attention. (4) Enumeration is contingent on other potentially attention-demanding visual processes such as perceptual grouping. The evidence presented here seriously challenges the traditionally held claim of a parallel and preattentive subitizing mechanism and suggests instead that small numerosity judgement requires visual attention. This weakens the argument of an attentional as well as a functional dichotomy and strengthens the idea that enumeration may be subserved by a single, continuous mechanism.