Abstract
To prevent hunting accidents, people in forests should be well-visible to others. However, there is debate which kind of protective clothing should be required. We created a database containing photographs of 22 individuals in forestial settings. The database is available online; for the present study a subset of images was used. The photographed persons ("models") wore four different clothing variants adding to typical forest garment: Two conditions included protective clothing, either a reflective vest or a hat with reflective hatband. We included two conditions without protective clothing, either with a hat (without hatband) or with no hat, and a condition without person. For the present study, we cropped and scaled the image of the database to generate 800 stimuli (160 per condition). These were presented to N = 16 observers, who were asked to respond as quickly as possible without sacrificing accuracy whether a human was present. Reaction times (RTs) and miss rates depended on clothing. The two conditions without protective clothing had indistinguishable error rates, while fewer errors occurred with hatband, and even fewer with the vest. RTs were faster with protective clothing than without. In the absence of protective clothing, responses were faster without hat. Interestingly, vest and hatband led to indistinguishable RTs. However, individuals in vests were looked at earlier than individuals without. This difference between RTs and fixation latencies is likely explained by a frequent additional saccade from vest to face before the response, presumably to verify the human presence. In conclusion, although RTs suggest no advantage for wearing reflective vests over hatbands, error rates and eye-tracking data reveal a potentially critical safety benefit of wearing reflective vests in forestial environments. More generally, our study demonstrates the benefit of combining eye-tracking data with behavioral measures using a real-world example.