Beach lifeguards: visual search patterns, detection rates and the influence of experience

Abstract

Lifeguard surveillance is critical to any water safety program. This study determined the rates of detection of a 'drowning' individual by beach lifeguards, and whether scanning patterns differed between groups of lifeguards (experienced/less experienced, male/females, surf/non-surf). It was hypothesized that (1) Experienced lifeguards would perform better and produce less fixations of longer duration than inexperienced; (2) A greater detection rate would be seen in a 'biased' compared to a 'non-biased' condition; (3) There would be no differences between the surf compared to non-surf lifeguards, and male compared to female lifeguards with regard to scanning patterns or detection rates. A mobile eye tracker was worn by each lifeguard (n = 69, 52 males, 17 females) as they watched 12 min of animated beach footage projected onto a screen in two conditions: a. 'Non-biased' (uniform scene). b. 'Biased' (uniform scene with presumed 'rip' on right side of screen). The lifeguards were informed that at any point during the 12 min a person may or may not disappear and to highlight if and where, a person disappeared. Unknown to the participants, a person always disappeared after 10 min at the same position within, but not between, conditions. Data were analysed using anova, t-tests and binary logistic regression. Experienced lifeguards were five times (p < 0.05) more likely to detect the drowning individual than inexperienced lifeguards. There were no significant differences between the visual search patterns of the groups between 2 and 10 min. The detection rates averaged 16% in the non-biased condition and 29% in biased conditions (p < 0.1), probably because lifeguards searched more on the right of the water. Furthermore, 40% (biased) and 42% (non-biased) did not detect the person disappearing, even though they fixated in the correct location in the 3.5 s before the person completely disappeared. This suggests that some lifeguards may have fixated on, but not processed, relevant visual data ('looked at but not seen'). 25% (biased) and 36% (non-biased) of the lifeguards did not fixate in the location of the person disappearing, but were able to identify their disappearance. Visual search patterns used by lifeguards can be altered by instruction and detection rates improve as a consequence. Peripheral vision is used effectively by some lifeguards, but cue extraction may be problematic for others.