Abstract
Augmented reality (AR) is increasingly being used for spatial applications, such as navigation and training procedures. Accurate perception of action capabilities, or affordances, in AR is critical for the effectiveness of these applications. Initial research suggests that affordances in AR are perceived similarly to the real world, but the role of restricted field of view (FOV) inherent in AR devices and feedback about affordance judgment accuracy have not been addressed. In the present study, participants made judgments about passing through an AR aperture defined by two virtual walls presented via a Microsoft HoloLens. First, we investigated whether affordance judgments were affected by viewing distance. We hypothesized that judgments would be impaired when viewing at a close distance because the entirety of the aperture was not visible within the FOV of the HoloLens. To test this hypothesis, participants viewed the aperture from a near (0.85 m) or far (3.20 m) distance. Second, we explored whether verbal feedback on judgment accuracy would improve judgments over time. To assess this question, we asked participants to adjust the virtual walls until they resembled the smallest passable aperture. Following that, they were presented with an aperture and asked to judge whether the aperture was passable and then received verbal feedback. Then they performed the adjustment task again to estimate the smallest passable aperture followed by more feedback. Results indicate that passing through judgments were closer to actual shoulder width when viewed at a distance near the aperture compared to the far viewing distance, contrary to our prediction. Verbal feedback reduced error over trials at the farther distance but not the near distance, possibly because near distance judgments were close to ceiling performance. Our results provide a promising way to reduce error in affordance judgments in AR, but also open questions about generalizability to other tasks.