Enhancing collision prediction in older adults via perceptual training in virtual reality emphasizing object expansion

Introduction: The ability to predict collisions with moving objects declines with age, partly due to reduced sensitivity to object expansion cues. This study examined whether perceptual training specifically targeting object expansion improves collision prediction more effectively than repeated practice on an identical collision prediction task. Additionally, the study verified whether such training could be employed to improve prediction accuracy in a more realistic context, using a virtual road-crossing scenario. Methods: Twenty older adults (71.35 ± 6.04 years; 11 females) participated. All tasks were constructed in virtual reality (VR) from a first-person perspective. Pre- and post-evaluation sessions comprised three tasks: a) an interception task assessing collision prediction ability, b) a target-approach detection task assessing the sensitivity of object expansion, and c) a road-crossing task. Participants were randomly assigned to one of two training groups: (a) a time-to-contact (TTC) estimation group (TE-group) or (b) an interception task group (IC-group). For the TE-group, participants repeatedly performed a TTC estimation task within a VR environment setting to isolate object expansion cues. This was achieved by restricting other visual cues and limiting the target's motion to a head-on collision approach. In the IC-group, participants repeatedly performed the same interception task used in the evaluation session. Results and Discussion: The TE-group showed significant improvement in collision prediction compared to the IC-group, indicating that training focused on the perception of object expansion was more effective than simple repetition of its evaluation task. However, neither sensitivity to object expansion nor the accuracy of road-crossing decisions improved significantly, suggesting that other factors may have contributed to the observed improvement.