According to the Theory of Visual Attention (TVA; Bundesen, 1990), objects across the visual field compete in parallel for encoding into a capacity-limited visual short-term memory store. In the classical view, spatial location plays no special role and is treated as just another feature contributing additively to the attentional weight of an object. Nevertheless many TVA studies typically arrange stimuli on an imaginary circle mainly to avoid low-level perceptual confounds such as foveal–peripheral acuity differences.

However, recent work challenges this assumption. Nordfang and colleagues (2018) demonstrated robust, idiosyncratic spatial biases across individuals and showed that target and distractor weights vary across locations in a way incompatible with the classical additive TVA formulation. Instead, their results support a model in which attentional weights factor into separate spatial and nonspatial components that combine multiplicatively, implying that location exerts a structured and theoretically meaningful influence on attentional deployment. Similar spatial biases have been noted in earlier TVA datasets (e.g., Nordfang, Dyrholm, & Bundesen, 2013; Sørensen, 2012).

Here I will revisit some of our previous data (Sørensen, Vangkilde, & Bundesen, 2015) and show clear and systematic spatial asymmetries in encoding performance, even though spatial location was irrelevant to the task, and should not guide top-down prioritization. Moreover, these spatial biases appear to be systematically modulated by participants’ expectations, suggesting that spatial priority maps interact with anticipatory processes in a way not captured by the traditional TVA framework.