Introduction and Purpose:
This study investigated whether the sound of a running river produces greater self-reported restorative outcomes (Restoration Outcome Scale, ROS) when described as "live-streamed from a national park" versus "artificially synthesized." This comparison was motivated by the assumption in restorative-environment research that natural stimuli are restorative, whereas artificial connotations are typically associated with non-restorative features.
Method:
A total of 168 participants completed an online experiment and submitted valid responses. After a stress-induction task, each participant was randomly assigned to one of three groups, in which they listened to the same 10-minute river sound but received different descriptions of its source: (1) no-information (control), (2) live-streamed from a national park, or (3) artificially synthesized. ROS was measured three times: T0 (baseline), T1 (after the stress-induction task and before the listening), and T2 (after the listening). The source information was presented immediately before the listening phase, i.e., after the completion of the T1 ROS assessment. We used a linear mixed-effects model, specifying Group (reference = no-information group), Time (reference = T1), and their interaction as fixed effects, and subject ID as a random effect.
Results:
Likelihood ratio tests showed that, among the fixed effects, only the main effect of Time significantly improved model fit, χ²(2) = 123.81, p < .001. An additional planned contrast showed that T2 (vs. T1) significantly increased the ROS score, b = 0.91, SE = 0.08, 95% CI [0.76, 1.06], t = 12.15, p < .001. Also, 93% of the participants in the no-information group correctly identified that the sound was at least related to water in natural surroundings.
Conclusions:
These results suggest that natural sound may be intrinsically restorative regardless of its semantic source attributions, at least when the source is easily identifiable (as opposed to abstract sounds, such as pink or white noise).
