Probability cueing in large-scale environmental search: The role of landmark cues in statistical learning

Efficient environmental search is an important and adaptive everyday skill. A particular feature of theoretical interest is whether large-scale search is informed by the spatial statistics of the environment - probability cueing is a robust effect in two-dimensional visual search tasks, but studies of large-scale search have generated equivocal findings. Here, we examined whether sensitivity to a statistical cue specified within an allocentric reference frame is modulated by the presence and location of environmental landmarks. Participants explored fully immersive virtual environments, wherein they were presented with an array of locations (columns) and required to search them for a hidden target (i.e. the column that changed color upon activation). A target was present on each trial, appearing within the cued hemispace on 80% of trials. In Experiment 1, the array was surrounded by a featureless circular wall and participants exhibited no reliable cueing effects. Experiments 2 and 3 introduced a stable landmark into the environment and manipulated its location to be either orthogonal or adjacent to the cued hemispace. Participants reliably biased their search in response to the probability cue, although learning was only observed when the landmark was positioned along the axis orthogonal to the midline separating hemispaces. These findings suggest that adapting search behavior in response to a statistical cue is facilitated by the presence of a stable landmark when it is specified independently of the searcher’s viewpoint, although this is dependent upon the spatial relationship between the landmark and the distribution itself.