The lighting environment has transcended purely functional illumination and has evolved into a critical medium for orchestrating narrative rhythm and modulating audience emotional responses. However, existing studies often examine photometric properties and human emotional responses in isolation, failing to establish a quantitative coupling mechanism to elucidate the relationship between light distribution, visual attention, and emotional states. This study aims to quantify the coupling mechanisms between luminous environmental parameters (illuminance and CCT), visual attention distribution, and emotional states (PAD) in immersive narrative exhibition spaces for the optimization of visitor experience. Four screen-based simulated narrative scenes were constructed with different illumination levels (low/high) and four levels of correlated color temperature (2700 K, 3000 K, 4000 K, and 5000 K). Using the SIFT algorithm, the illuminance pseudo-color map and the eye-tracking heat map were spatially registered to quantify the spatial correlation between the physical light field and the visual attention field. The results demonstrate a significant nonlinear coupling effect: high-illuminance cold light (4000 K, 544 lx) establishes a strong guidance mechanism, with a high spatial correlation between visual attention and brightness ($r$ = 0.82), which significantly enhances physiological arousal and perceived dominance. Conversely, low-illuminance warm light (2700 K, 150 lx) leads to a weak coupling state ($r$ = 0.62), which promotes free visual exploration, thereby improving pleasure and perceived immersion. These results suggest that lighting design should not be treated as a fixed set of parameters, but rather as an adjustable strategy that responds to changes in visual attention and emotional experience. By modifying the strength of visual and optical interaction, lighting conditions can influence how visitors move from initial perception to emotional engagement. This provides practical support for applying evidence-based lighting strategies in the design of cultural heritage spaces.

Public-space entrances are often assessed through ratings, although correspondence between favorable appraisal and entry choice remains unclear. We examined this relationship in an immersive virtual-reality experiment involving 72 adults and 24 scenes (1728 trials). A within-participant design varied entrance type, transition depth, pausing support, and natural-surveillance legibility. Participants made controller-based entry choices during each scene and rated safety, legibility, control, and retreatability afterward; gaze and electrodermal activity were recorded concurrently. The primary mismatch index compared standardized appraisal with entry rate, while latency was analyzed separately. Medium pausing support with high surveillance legibility (MH) combined favorable safety and legibility ratings with the lowest reported entry rate (9.95%) and longest mean latency (4.39 s). MH had the largest mismatch under eight of ten appraisal-weight specifications. High pausing support with high legibility combined favorable appraisal and frequent entry. These findings support evaluating entrances through both subjective experience and behavioral uptake, and assessing readable interior information alongside opportunities to pause and orient. Causal mechanisms and transfer to physical entrances require further verification.

Whether a well-organized scene facilitates visual search may depend on when in the search process the benefit is expressed. We tested this possibility in a virtual-office task in which 48 adults searched for targets varying in salience. Salience was manipulated between participants, whereas layout (grid, clustered, or random) and clutter (low, medium, or high) varied within participants. Mouse responses and eye position (120 Hz) were collected across 1296 target-present trials. Accuracy was broadly comparable across conditions, but reaction time (RT), time to first fixation (TTFF), fixation duration in the Target AOI, and fixation duration in the Clutter AOI each showed a three-way interaction. The transition data were particularly informative under high clutter. For grid layouts, clutter-to-target transitions occurred on 47.3% of transitions, clutter-to-clutter self-transitions on 20.5%, and the target was first reached after 0.89 intervening transitions on average. The respective values for random layouts were 11.2%, 64.2%, and 6.39. Thus, the benefit of the grid was not consistently present at initial target capture. It became most evident in the way gaze was redirected among regions as the search unfolded. Spatial regularity therefore seems to limit later search costs by making attentional reallocation and gaze-path correction more efficient.