Emergency Broadcasting Systems (EBS) are life-safety installations that deliver urgent information during fire events, enabling occupants to recognize hazards and initiate evacuation. Because response time is the most variable component of evacuation ...
Emergency Broadcasting Systems (EBS) are life-safety installations that deliver urgent information during fire events, enabling occupants to recognize hazards and initiate evacuation. Because response time is the most variable component of evacuation time, delays arising from cognitive appraisal and decision-making can substantially influence the Required Safe Egress Time (RSET). Thus, clarifying how EBS acoustic characteristics shape early recognition and evacuation initiation is essential for performance-based evaluation. Prior research has mainly assessed physical and environmental factors (e.g., sound pressure level, intelligibility, and message structure) using observations, full-scale experiments, and video-based analyses. Although these approaches quantify response time and route choice, they provide limited access to internal cognitive and attentional processes underlying delayed movement, misinterpretation, or weak behavioral transitions. This study investigated how combinations of acoustic parameters modulate EEG-based cognitive activation and how neural responses correspond to evacuation-initiation behaviors. A 3×3×3 factorial experiment was conducted in a fully anechoic chamber with sound pressure level (SPL), fundamental frequency (F0), and speech rate (SPM) as independent variables. A total of 27 (33) acoustic conditions were constructed for both male and female voices. Adult participants(N=547;≥20 years) were assigned using a between-subjects design. EEG was recorded using 19 channels based on the international 10– 20 system, and stimulus-related changes were computed relative to baseline. Cognitive activation was quantified using the RSMT(ratio of SMR–mid-beta to theta). Three-way ANOVA tested main effects and two and three-way interactions. Evacuation-related behaviors were coded within a Notification– Cognition–Activity framework and integrated with post-experiment interviews. Evacuation initiation was defined as pre-movement actions preceding physical egress, including orientation and information-seeking behaviors. To minimize expectancy effects, alarm presentation was not announced in advance. EEG features were derived from baseline-normalized ΔEEG values, ensuring comparability across participants and conditions. The chamber setting isolated acoustic factors from environmental confounders effectively. SPL produced the most consistent effects across electrodes for both voice types, supporting SPL as the primary determinant of alarm perceptibility. Significant three-way interactions at selected electrodes indicated that cognitive activation increased only under specific SPL–F0–SPM combinations, implying synergistic effects rather than a single-factor mechanism. Representative high-activation combinations included, for the male voice, C4 at 60㏈(A)–100 ㎐–300,360 SPM, and for the female voice, P4 at 78㏈(A)–210㎐–420 SPM or 95㏈(A)–180㎐–420 SPM. Conditions with higher cognitive activation were accompanied by clearer evacuation-initiation behaviors (e.g., exploration, gaze shifts, and exit checking). Notably, some participants with minimal overt movement still showed increased RSMT, suggesting internal appraisal not captured by behavior-only assessments. Overall, this study provides an integrated EEG–behavior assessment of how SPL, F0, and speech rate combinations influence cognitive activation and evacuation initiation. The derived optimal combinations may serve as reference stimuli for validation under realistic acoustic environments and support human-centered, performance-based design criteria for EBS.