One of the attractions of the visual foraging paradigm is the rich data that is produced by
participants. Each trial produces a sequence of item-selections, and these sequences are
influenced by features (i.e., colour, shape) and spatial factors such as how far away potential
targets are from the previously selected item. While this complexity allows us to study the
interaction of visual perception and decision-making, it can also complicate the analysis. Clarke
et al (2022, PloS Comp Bio) introduced a four parameter generative model for visual foraging
(FoMo) that aims to predict which item a participant will select next. In this talk, I will present a
new foraging experiment that explores how the layout of the foraging items influences the
spatial strategies that participants use while foraging. We find the when items are placed on a
cardinal grid, the foraging sequences generally follows the grid orientation. Rotating the grid to
an oblique orientation leads to a preference for the oblique directions. However, the overall
direction bias is stronger when the grid has a cardinal direction, suggesting that participants have
an underlying preference for cardinal directions above and beyond what is present in the
stimulus. We develop a von Mises mixture model to account for these patterns of inter-item
target selections. Finally, we incorperate this as a new component in FoMo and discuss the effect
this has on the accuracy of our predictions.