2.4. Experimental Design

The experiments were performed in a semi-anechoic chamber in Tianjin University. Participants were accompanied by their parents, who waited in a room outside the chamber during the experiment. Participants took part in the experiment individually, supervised by a researcher.

Before the formal experiment, participants completed the baseline measure of the cognitive test. When the experiment started, they were first given a 5-min oral calculation task, which was assessed to be an effective way to induce attentional fatigue [32]. The difficulty of oral calculation was adapted to children of different age groups to ensure the same difficulty level for all children as much as possible. Specifically, participants aged 7–8 were asked to perform continuous subtraction from a number with three digits with a step of 3 accurately as soon as possible. If he/she did a miscalculation, they would be asked to stop and start their calculation from the beginning. Participants aged 9–10 performed the subtraction from a number with four digits with a step of 7, while participants aged 11–12 performed the subtraction from a number with four digits with a step of 13. Especially, a few children’s oral calculation abilities were obviously better than those of their peers, so in those cases, a more difficult level of calculation was accordingly used. After that, participants were asked to perform the same cognitive test to assess attentional fatigue. The restoration period followed, when participants were exposed to an audiovisual soundscape for 3 minutes without any disturbance. Then, they were asked to perform the cognitive test again to examine whether their cognitive performance returned to normal level. Each experimental unit last around 13 minutes.

In order to control the whole experiment time within one hour, each participant could only experience four of seven soundscapes at most. Therefore, each participant performed the experimental unit four times, as shown in Figure 2. To ensure equal numbers of participants across soundscapes, the four soundscapes for each participant were randomly selected from seven soundscapes by computer. The four soundscapes were also counterbalanced in random order to minimize order effect. In total, each soundscape was experienced by 25 participants at least, which was a moderate sample size for statistical analysis. The sample size was set mostly based on the examples of other studies and our experience. In addition, data saturation was determined when the statistical values were rather steady with the addition of the last few participants. The cognitive baseline levels of participants in different soundscape groups would be examined to ensure no group differences across soundscapes.

Diagram of experimental procedure.

Finally, the participants were also asked to provide their demographic information, including age in years and gender.