Effects of Spatial and Visual Openness in Office Environments on EEG-Based Cognitive Efficiency

Abstract

Office openness comprises two physically distinct dimensions—spatial openness and visual openness—yet studies quantifying their independent contributions to cognitive efficiency at the individual level remain scarce. Prior research has predominantly reported group-mean effects, leaving bidirectional individual responses insufficiently examined. This study independently manipulated both dimensions and measured individual-level EEG responses in 24 adults using a 3 × 3 within-subject factorial design. The beta/alpha ratio change rate was computed as an index of cognitive efficiency. Substantial neurophysiological variation across conditions was confirmed in every participant. The absence of significant group-level effects was interpreted not as a lack of environmental influence but as the result of bidirectional individual responses canceling each other out in group averages. Spatial and visual openness induced response ranges of equivalent magnitude at the individual level, and individually optimal conditions were widely distributed across the nine experimental conditions. The correspondence rate between subjective preferences and EEG-identified optimal conditions did not exceed chance, and this bidirectional cancellation mechanism is proposed as an explanation for the contradictory findings that have long characterized open-office research. These results support design strategies that offer diverse combinations of spatial and visual openness within activity-based working environments, paired with feedback systems grounded in objective cognitive performance data.

1. Introduction

Open-plan offices have become a dominant spatial strategy worldwide, valued for their capacity to promote collaboration and enable the flexible utilization of space [1]. Contemporary knowledge workers spend an average of more than eight hours per day in office environments, and the efficiency with which they execute cognitive tasks during that time directly affects work performance. Although a growing body of research has examined how openness affects occupant concentration and task performance [2], studies that quantify the magnitude of openness effects at the individual level and systematically characterize the diversity of responses across individuals remain limited.

Open-plan offices refer to layouts in which multiple occupants share a space without fixed walls or partitions. Within this context, openness comprises two physically distinct dimensions. Spatial openness denotes the degree of physical enclosure determined by partial partitions or furniture arrangement, whereas visual openness refers to the visual connection with the external environment afforded by windows and blinds [3]. Prior research has largely compared wholly open versus wholly enclosed environments, and studies that systematically examine the independent contribution of each dimension to cognitive outcomes are scarce. “Cognitive efficiency,” as used here, denotes the capacity to process information and execute tasks with available cognitive resources [4].

Research investigating the relationship between openness and cognitive performance has yielded conflicting findings. Some studies report that greater openness enhances productivity [5], whereas others report the opposite [6]. Analyzing post-occupancy evaluation data, enclosed private offices outperformed open-plan layouts on most indicators of satisfaction with the indoor environment [1]. This variability suggests that the effects of openness may depend on context, task type, and individual characteristics [2,7]. Yet most existing studies have reported only group-mean outcomes. Consequently, inter-individual differences in response may go undetected in group-mean analyses, underscoring the need for individual-level approaches. If optimal conditions differ across individuals, design strategies that offer a range of environmental options are preferable to uniform configurations.

Capturing such variation at the individual level requires methods beyond subjective ratings and behavioral performance measures. Objective measures of cognitive function have revealed environmental effects that subjective assessments alone may not capture [8], and behavioral measures fail to detect differences in cognitive load when performance outcomes appear comparable across conditions [9]. Use of electroencephalogram (EEG) addresses these limitations by directly measuring cognitive processing, and the beta/alpha ratio has been validated as a reliable neurophysiological index of cognitive efficiency [10,11]. Advances in mobile EEG technology have made measurement feasible in real working environments, including architectural research contexts [12]. Nevertheless, no study to date has independently manipulated spatial and visual openness to quantify the effect of each dimension at the individual level.

This study aimed to empirically ascertain the effects of spatial and visual openness in office environments on cognitive efficiency as measured by EEG. To this end, a 3 × 3 (three levels of spatial openness defined by partition configuration × three levels of visual openness defined by blind aperture) within-subject factorial design was employed with 24 participants, independently manipulating both dimensions while recording EEG responses. Four specific objectives were pursued: quantifying response ranges at the individual level, comparing the relative contributions of the two dimensions, characterizing the distribution of individually optimal conditions, and assessing the correspondence between subjective preferences and objective optimal conditions.

2. Literature Review

2.1. The Evolution of Office Design and the Role of Openness

Office spatial layouts have evolved continuously since the Industrial Revolution, reflecting successive shifts in organizational logic and work practice [1]. The early twentieth-century bullpen office, designed around Frederick Taylor’s scientific management principles, prioritized supervisory oversight and standardized task execution but generated persistent problems, including noise exposure, insufficient privacy, and psychological stress [13]. The Bürolandschaft movement of the 1960s and Robert Propst’s “action office” responded to these limitations with organic, reconfigurable layouts; under corporate cost-reduction pressures, however, these were reduced to the cubicle office, which partially restored visual privacy while suppressing lateral communication [14]. From the 2000s onward, growth in the technology sector and an organizational emphasis on collaboration drove the re-emergence of open-plan offices. However, empirical findings diverged from expectations: removing walls reduced face-to-face interaction by approximately 70% while increasing electronic communication by 22–50% [15], and enclosed private offices outperformed open-plan layouts on most indicators of occupant satisfaction [2].

In response, office design has shifted toward the activity-based working (ABW) model, which eliminates assigned seating and enables occupants to select spatial zones according to task demands [16]. The effectiveness of ABW environments derives largely from occupants’ perceived control over their spatial conditions [17]. This trajectory of office design evolution is illustrated in Figure 1.

Empirical studies consistently demonstrate that openness is a consequential design variable within ABW environments. Higher openness levels are associated with increased perceived distraction and reduced work performance [6,18], while access to enclosed individual spaces improves cognitive task performance by 17–22% [19]. Openness thus directly mediates occupants’ spatial choices and work outcomes, and work performance is ultimately determined by cognitive efficiency—defined as the capacity to process information and execute tasks with available cognitive resources [4]. It is therefore necessary to examine, at a neurophysiological level, how openness acts upon occupants’ cognitive efficiency.

2.2. Spatial and Visual Dimensions of Openness

Office openness comprises two physically distinct dimensions: spatial openness and visual openness. Despite their conceptual distinctiveness, prior studies have rarely examined these dimensions independently. Spatial openness—defined by partition height, walls, and spatial configuration—has been the primary focus of open-plan office research [2], while visual openness, referring to connectivity with the external environment through windows and blinds, has been examined mainly in the context of daylighting and view access [3]. The independent contribution of each dimension to cognitive outcomes remains insufficiently characterized, and their conflation in prior research may partly explain the contradictory findings documented in the literature. Among the various environmental factors influencing cognitive efficiency in office settings, spatial and visual openness are particularly consequential, as they directly and independently determine the degree of physical enclosure and visual connectivity that occupants experience, and can be systematically adjusted through architectural design elements such as partition configuration [20] and blind or glazing systems [3].

One of the most rigorously controlled experimental tests of the effects of spatial openness has been reported in the literature [20]. Across three between-subjects experiments conducted in a windowless room—two measuring perseverance directly (Experiment 1: n = 65; Experiment 2: n = 60) and one testing the effort-allocation mechanism underlying this effect—the raised-partition condition significantly increased perseverance on difficult tasks compared with the lowered-partition condition. This effect was interpreted as reflecting reduced cognitive effort allocated to processing peripheral visual information. In a virtual reality-based study, systematically examined how spatial elements, including partition height, ceiling height, and contour shape, influenced 713 participants’ cognitive and aesthetic appraisals of safety, distraction, interaction, beauty, and pleasure [21]. Higher partitions were associated with greater perceived safety and reduced distraction, and the Bürolandschaft configuration significantly enhanced both perceived safety and interaction.

Access to natural light and outdoor views through windows exerts positive effects on cognitive performance, emotional wellbeing, and ocular health. A controlled crossover experiment demonstrated that ten office workers who performed tasks under three different window configurations over 14 weeks showed measurable differences in daily cognitive function [3]. Conditions providing access to natural light and outdoor views yielded significant improvements in working memory and inhibitory control, along with reductions in eye strain and increases in environmental satisfaction. A study of 100 office workers found that access to natural light and outdoor views was associated with significantly higher job satisfaction, with daylight amount and view quality emerging as particularly strong predictors [22]. Furthermore, proximity to windows has been shown to attenuate the negative aspects of open-plan offices, with satisfaction maximized when window proximity was combined with appropriate partition height [23]. The majority of existing studies on office layout either manipulate spatial and visual openness simultaneously or address only one dimension in isolation. Studies comparing enclosed private offices with open-plan layouts cannot isolate the independent contribution of each dimension, because removing partitions simultaneously alters window accessibility, spatial density, and noise levels [2]. Conversely, studies manipulating only spatial openness [19] hold visual openness constant as a control variable, precluding any examination of how the two dimensions interact. Similarly, studies manipulating only visual openness [3] maintain a fixed spatial configuration, making it impossible to explore potential interactions with the spatial dimension.

Only a small number of studies have independently manipulated both dimensions. A 2 × 2 design crossing window proximity with partition height was employed to examine the independent effects and interaction of each dimension, but each dimension was restricted to two levels, limiting the range of openness examined [23]. A field study varying workstation size, partition height, and window access was also conducted, but participants were measured in spaces to which they had been naturally assigned, resulting in insufficient experimental control [24].

Table 1. Methodological characteristics of experimental studies on office openness.

Category	Authors	Setting	Design	N	Focus	Spatial Type	Visual Type	Outcome Measures
Spatial only	Oldham (1988) [14]	Field	Quasi- experiment	~24	Spatial	Partition height/ density	—	Subjective: privacy, satisfaction
	Roberts et al. (2019) [20]	Laboratory	Between-subjects	~180	Spatial	Partition: 0.50 m/0.13 m	—	Behavioral: persistence
	Gath-Morad et al. (2024) [21]	Online 3D model	Between-subjects	713	Spatial	Partition height × ceiling height × contour	—	Cognitive & aesthetic appraisal; isovist
	Gao et al. (2025) [25]	Virtual Reality (Cave Automatic Virtual Environment, VR-CAVE)	Within-subjects	52	Spatial	3-level interior openness	—	Physiological: EEG, eye-tracking
Visual only	Jamrozik et al. (2019) [3]	Laboratory	Crossover	10	Visual	—	3-level enclosure conditions	Cognitive; subjective; physiological
Both dimensions	Bernstein & Turban (2018) [15]	Field	Quasi- experiment	152	Mixed	Cubicle wall removal	Visual /acoustic simultaneous change	Behavioral: interaction tracking
	Yildirim et al. (2007) [23]	Field	Quasi-experiment (2 × 2)	NR	Mixed	Partition: 1.20 m/1.40 m	Window proximity: near/far	Subjective: satisfaction, privacy
	Fich et al. (2014) [27]	VR (CAVE)	Between-subjects RCT	49	Mixed	Room enclosure: closed/ open	Aperture view provided	Physiological: cortisol, heart rate variability (HRV)
	Vartanian et al. (2015) [28]	2D images (functional magnetic resonance imaging, fMRI)	2 × 2 within-subjects	18	Mixed	Ceiling height × enclosure	Enclosure	Neural: fMRI; behavioral
	Boubekri et al. (2020) [26]	Field	Crossover	30	Visual	—	Electrochromic (EC) glass/blinds	Cognitive; sleep

NR: not reported.

systematically compares the methodological characteristics of recently published experimental studies on office openness. Analysis of ten experimental studies revealed that four manipulated only spatial openness [19,21,22,24,25], two manipulated only visual openness [3,26], one examined both dimensions without independently manipulating them [23], and three involved a confounded combination of both dimensions.

In sum, questions regarding which of the two dimensions exerts a greater influence on cognitive efficiency, how they interact, and whether preferred combinations differ across individuals remain insufficiently resolved. The interaction effect observed by The interaction effect observed in a prior study suggests that the two dimensions may not operate in a simple additive fashion, pointing to the need for factorial designs that manipulate them independently [23]. Furthermore, given that existing studies have predominantly reported group-mean effects, approaches that measure the contribution of each dimension at the individual level are warranted.

2.3. Measurement Approaches and Individual Differences in Environmental Response

Methods for assessing the effects of office environments on cognitive efficiency have advanced through three complementary approaches—subjective ratings, behavioral performance measures, and neurophysiological measurement—each providing distinct information.

Subjective ratings assess occupant satisfaction, perceived productivity, and environmental comfort through post-occupancy surveys [2]. This approach enables large-scale data collection and provides direct insight into occupants’ lived experiences, making it practically useful. However, respondents may interpret rating scales differently [29], and contextual factors such as green building certification labels can bias assessments [30]. conditions with low volatile organic compounds (VOCs) and enhanced ventilation improved cognitive function scores by 61% and 101%, respectively, yet participants remained unaware of these changes—suggesting that subjective ratings may not adequately capture actual variation in cognitive efficiency [8].

Behavioral performance measures quantify task execution directly, providing objective indices. Using quiet individual spaces within an ABW environment improved cognitive task performance by 17–22% [19], consistent with evidence that access to quiet workspaces in open-plan offices yields measurable benefits for task performance and wellbeing [31]. Nevertheless, task selection is critical because different types of task are differentially affected by environmental conditions [32], and repeated exposure to the same task may introduce learning effects that confound performance outcomes [33]. Differences in mental workload can persist even when behavioral performance outcomes remain comparable across conditions, indicating that additional measurement methods are needed to detect cognitive burden that does not manifest in performance results [19].

To address these limitations, mobile EEG technology has been introduced into research on architectural environments. Because EEG measures brain activity directly with high temporal resolution, it provides information that neither subjective reports nor behavioral observation can capture [34]. Zhang et al. (2024) systematically reviewed evidence that EEG can support evidence-based design by measuring spatial perception, emotional responses, and cognitive load in research on indoor environments [11]. Studies combining subjective ratings with EEG have demonstrated that the two methods provide complementary information, enabling a more integrated understanding of environmental effects [34,35]. Each of the three approaches to measurement has distinct strengths and limitations. EEG complements subjective and behavioral measures by capturing neurophysiological responses that the other approaches cannot detect, thereby affording a more complete account of how office environments affect occupants.

Individual differences represent a critical factor in environmental response. Responses to open-plan offices have been shown to vary substantially across individuals depending on personal characteristics [7], and EEG evidence indicates that individuals with high noise sensitivity exhibit distinct auditory cortex processing patterns, suggesting that individual differences in environmental response have a neurophysiological basis [36]. Despite this, existing studies have predominantly reported group-mean outcomes without providing individual-level analyses [15]. If individuals respond to the same environment in opposing directions, their responses will cancel each other out in group averages, statistically obscuring genuine environmental effects. In such cases, a non-significant group-mean result may reflect a structural limitation of the analytical approach rather than an absence of environmental influence.

EEG is particularly well-suited to capturing individual-level response patterns of this kind. It enables direct measurement of each individual’s neurophysiological response and can be used to examine the relationship between subjective preferences and neurophysiological efficiency. Research that quantifies environmental effects at the individual level and characterizes how optimal conditions vary across individuals can elucidate response structures that group-mean analyses fail to detect, thereby providing a neurophysiological basis for the design of spatial openness in offices.

3. Methods

3.1. Experimental Design and Participants

A 3 × 3 within-subject factorial design was employed to examine the effects of spatial and visual openness on cognitive efficiency. This design was chosen because having each participant experience all nine conditions automatically controls for individual differences in baseline brain activity, personality, and cognitive ability, allowing environmental effects to be isolated [37], and because the research objectives—quantifying individual-level response ranges, characterizing optimal condition distributions, and assessing subjective–objective correspondence—required observation of each participant’s full response pattern across all conditions. Each condition was measured once per participant. Repeating all nine conditions would have extended the session to an impractical duration, introducing fatigue and learning effects that could have confounded the EEG signal; a single measurement per condition was therefore adopted as a pragmatic design choice.

The independent variables were spatial openness, operationalised as the degree of physical enclosure determined by partition configuration on three sides of the workstation, and visual openness, operationalised as window-to-exterior visual connectivity as manipulated through blind aperture. Findings should be interpreted within the scope of these operationalisations. Spatial openness was set at three levels: 0% (partitions on three sides: front, left, and right), 50% (front partition only), and 100% (no partitions). Visual openness was likewise set at three levels: 0%, 50%, and 100%, yielding nine experimental conditions in total. The two dimensions were manipulated independently because prior research has predominantly treated overall openness as a single variable, making it impossible to separate the relative contributions of physical enclosure and visual obstruction.

Table 2. Experimental conditions and condition codes for spatial and visual openness.

	100%	50%	0%
Spatial openness (Partition)
Visual openness (Blind)
	Blind 100%	Blind 50%	Blind 0%
Partition 100%	C01	C02	C03
Partition 50%	C04	C05	C06
Partition 0%	C07	C08	C09

presents the nine conditions and their codes.

The dependent variable was the percentage change in the EEG-derived beta/alpha ratio relative to baseline. The beta/alpha ratio is a validated neurophysiological index of cognitive arousal and information-processing efficiency [4,10]. Increases in beta-band power are associated with sustained attention and execution of cognitive tasks, whereas decreases in alpha-band power reflect elevated cognitive load [38]; accordingly, an increase in the beta/alpha ratio indicates enhanced cognitive efficiency. While alternative metrics such as the theta/alpha ratio have been applied in mental workload research to capture task-difficulty-driven cognitive load [39], the beta/alpha ratio was selected in the present study because it simultaneously reflects both increased cortical activation (beta band) and reduced idling state (alpha band), making it particularly suited to detecting cognitive arousal changes induced by environmental conditions.

Twenty-four adults participated in the study (6 male, 18 female; mean age 26.8 years, SD = 3.2). All had normal or corrected-to-normal vision and reported no history of neurological or psychiatric disorders. Sample size was determined a priori using G*Power 3.1.9.7 (repeated-measures ANOVA; f = 0.24; α =0.05; power =0.80), yielding a minimum of 19 participants; the target was set at 24 to account for typical EEG attrition rates of 15–20%. The study was approved by the Institutional Review Board of the authors’ institution.

The experiment was conducted in a controlled laboratory room measuring 10,430 mm × 7620 mm with south-facing windows. Partitions measured 1200 mm in height and 1400 mm in width. In the 0% spatial openness condition, partitions were installed on three sides of the participant’s workstation; in the 50% condition, only the front partition was present; and in the 100% condition, all partitions were removed. Blinds were fully closed in the 0% visual openness condition, half-open in the 50% condition, and fully open in the 100% condition (Figure 2).

The experiment was conducted under consistent environmental conditions to minimise the influence of confounding variables on the two manipulated dimensions. The workstation was positioned perpendicular to the south-facing window wall, minimising the risk of direct or reflected glare. The external view consisted of a typical urban landscape. Illuminance ranged approximately 400–600 lux depending on blind condition. All sessions were conducted between 13:00 and 18:00 on days with stable weather conditions to minimise variation in natural light levels.

A one-minute baseline EEG recording was obtained under each participant’s first assigned condition. Baseline normalization controlled for inter-individual differences in absolute signal levels that would otherwise obscure genuine environmental effects [37]. Participants then completed the nine conditions in a counterbalanced order. Each condition comprised a one-minute baseline recording followed by three minutes of the arithmetic verification task (AVT), in which sequentially presented arithmetic equations were judged as correct or incorrect; up to 45 items were presented with a maximum of five seconds per item (Figure 3).

Upon completing each condition, participants rated the environment on a 7-point Likert scale across four items: concentration, visual comfort, visual privacy, and perceived spatial openness (1 = strongly disagree, 7 = strongly agree). At the end of the session, participants identified their most preferred, most focused, and most uncomfortable conditions and provided written explanations. These responses were used to assess the subjective–objective correspondence and to examine whether individuals could accurately identify their neurophysiologically optimal environment.

3.2. EEG Data Acquisition and Processing

EEG signals were recorded using the Neuroelectrics Enobio 32 wireless system, which comprises 32 channels arranged according to the international 10–20 electrode placement system [40] and samples data at 500 Hz. An average reference was applied [41,42].

Raw EEG data were preprocessed in MNE-Python (version 1.5) following a standard pipeline [43]. A 0.5–45 Hz bandpass filter was first applied to remove low-frequency drift and high-frequency noise [44]. Independent component analysis (ICA) was then used to eliminate physiological artifacts, including ocular movements and electromyographic activity [45]. Ocular components were identified by their characteristic high-loading pattern on frontal channels (Fp1, Fp2, AF3, AF4); a maximum of two artifactual components were removed per participant.

Preprocessed signals were decomposed into frequency bands. Spectral power in the delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz), and gamma (30–45 Hz) bands was computed using Welch’s method [46] with a two-second window (1000 samples). Band power was calculated by integrating across the full four-minute task segment of each condition, treated as a continuous signal.

The beta/alpha ratio was computed at each channel by dividing beta-band power by alpha-band power [10], and then averaged across all 32 channels to yield a single index of overall cognitive arousal [4,47]. Global averaging provides a more noise-robust and stable index than region-specific measures [48].

Baseline normalization was performed using the percentage change formula:

$$\mathrm{C}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}(\mathrm{\%})={\displaystyle \frac{\mathrm{Condition} \mathrm{beta}/\mathrm{alpha} \mathrm{ratio}\mathrm{-}\mathrm{Baseline} \mathrm{beta}/\mathrm{alpha} \mathrm{ratio}}{\mathrm{Baseline} \mathrm{beta}/\mathrm{alpha} \mathrm{ratio}}}\times 100$$

This approach controls for inter-individual differences in absolute signal levels and quantifies each participant’s change relative to their own resting state [37].

For data quality control, channels with excessive electrode impedance were restored via spherical spline interpolation [49]. The final dataset comprised 24 participants × 9 conditions = 216 data points.

3.3. Statistical Analysis

A multi-level analytical approach was adopted to examine both group-mean effects and individual-level heterogeneity simultaneously [50,51]. In repeated-measures designs, multiple observations are collected from the same participant, making it difficult for simple group-mean analyses alone to adequately characterize the directionality and magnitude of individual responses [52]. All analyses were conducted in Python 3.10 using the Pingouin, statsmodels, and SciPy packages.

Group-level mean effects of spatial openness (three levels) and visual openness (three levels) were examined using a two-way repeated-measures ANOVA [51]. The dependent variable was the percentage change in the beta/alpha ratio relative to baseline. Sphericity was assessed with Mauchly’s test, and the Greenhouse–Geisser correction was applied where the assumption was violated. Effect size was reported as generalized eta-squared (η²G), which corrects for the overestimation inherent in partial η² by including between-participant variance in the denominator [53,54]. Benchmarks of 0.01, 0.06, and 0.14 indicated small, medium, and large effects, respectively [55].

Individual-level environmental sensitivity was quantified using the response range, defined as the difference between the maximum and minimum beta/alpha ratio change rates across a participant’s nine conditions. Applying Cohen’s d in a max-minus-min fashion within a within-subject design produces inflated effect size estimates; response range was therefore adopted as an exploratory descriptive statistic for heterogeneity in individual responses rather than for inferential purposes.

To decompose the relative contributions of group-mean environmental effects and individual differences, a linear mixed-effects model was fitted with spatial and visual openness as fixed effects and participant as a random effect. The random-effects structure was determined via model comparison using the Akaike information criterion [AIC]; Ref. [56]: a random-intercept-only model (AIC = 1942.99) was preferred over a model with additional random slopes (AIC = 2024.10), as the random-slope variance was estimated at zero. Variance components were estimated by restricted maximum likelihood (REML). intraclass correlation coefficient (ICC) was derived from the between-participant and residual variance components and interpreted as a variance decomposition index—not a reliability coefficient—to distinguish whether non-significant group-mean results reflected an absence of environmental effects or statistical cancellation due to inter-individual heterogeneity. Marginal R2 (fixed effects only) and conditional R2 (fixed and random effects) were computed following the method of Ref. [57]; the difference between the two values directly reflects the magnitude of the individual-differences contribution (see

Table 3. Statistical indices used in the present study.

Analysis Purpose	Index	Definition	Interpretive Benchmarks	Source
Group-mean effect	Two-way repeated-measures ANOVA	Tests main effects of spatial openness (3 levels) × visual openness (3 levels) and their interaction; Greenhouse–Geisser correction applied if sphericity is violated.	p < 0.05: significant group-mean effect; p ≥ 0.05: no significant effect.	[53]; standard group-mean test for repeated-measures designs.
	η2G	Proportion of total variance explained by a factor in repeated-measures ANOVA; corrects for overestimation by partial η2.	0.01 = small, 0.06 = medium, 0.14 = large [55].	[53]; recommended by Bakeman (2005) as standard index for repeated-measures research.
Individual-differences contribution	Linear mixed-effects model	Fixed effects: spatial and visual openness; random effect: participant. Random-intercept-only model selected via AIC comparison; variance components estimated by REML.	Optimal model selected by AIC; random-intercept-only model adopted when random-slope variance = 0.	[50,51]
	ICC	Proportion of total variance attributable to between- participant differences; derived from variance decomposition of the linear mixed-effects model.	<0.50 = low, 0.50–0.75 = moderate, 0.75–0.90 = high, ≥0.90 = very high [58].	[58,59]
	R2marginal/R2conditional	R2m: variance explained by fixed effects (environmental factors) alone. R2c: total variance explained by fixed and random effects (individual differences).	R2c − R2m reflects the magnitude of the individual-differences contribution.	[57]; standard explanatory power indices for mixed-effects models.
Individual response range	Response range	Maximum minus minimum beta/alpha ratio change rate across each participant’s nine conditions; reflects absolute within-person environmental sensitivity.	No established threshold; used to confirm individual-level effects when group-mean results are non-significant.	Descriptive statistic for within-person variance in repeated-measures designs.
Subjective–objective agreement	Binomial test	Tests whether the agreement rate exceeds the chance level (1/9 ≈ 11.1%).	p < 0.05: systematic agreement; p ≥ 0.05: agreement not above chance.	[60]
	Pearson’s r	Strength and direction of the linear relationship between two variables.	0.10 = small, 0.30 = medium, 0.50 = large [60,61].	Standard correlation index; Ref. [61] benchmarks used as conservative reference.

for all index definitions and benchmarks).

For the subjective–objective agreement analysis, each participant’s subjectively preferred condition was compared with their EEG-identified optimal condition, coded as a binary agreement variable. A binomial test evaluated whether the observed agreement rate exceeded the chance level of 1/9 ≈ 11.1% [58], with p < 0.05 indicating systematic agreement beyond chance. The individual-level linear relationship between subjective concentration ratings and objective beta/alpha ratio change was quantified using Pearson’s correlation coefficient.

To evaluate individual-level differences in response to each dimension, one dimension was held constant while the response range across the three levels of the other dimension was computed per participant; the two sets of ranges were compared using a paired-samples t-test. All tests were two-tailed with α = 0.05. Table 3 summarizes the purpose, definition, interpretive benchmarks, and supporting references for each statistical index.

An outlier analysis applying IQR-based criteria (Q1 − 1.5 × IQR, Q3 + 1.5 × IQR) to all 216 observations identified a single outlier: P04’s value in condition C05 (+89.3%), which exceeded the upper bound of +71.5%. Because this value reflects a genuine neurophysiological response rather than measurement error or protocol deviation, the full sample (N = 24) was retained for all analyses.

4. Results

4.1. Individual Level Effects of Openness on Cognitive Efficiency

Analysis of beta/alpha ratio change rates across the nine conditions revealed a clearly identifiable optimal condition and a lowest-efficiency condition for every participant. This indicates that, even after baseline normalization, neurophysiologically distinct environmental conditions exist at the individual level—that is, spatial and visual openness induced meaningful neurophysiological change in all participants.

Table 4. Beta/alpha ratio change rates (%) across nine conditions by participant.

	Partition 100% (Spatially Enclosed)			Partition 50%			Partition 0% (Spatially Open)			Optimal Condition	Max (%)	Min (%)	Response Range (pp)
Participant	C01 B100	C02 B50	C03 B0	C04 B100	C05 B50	C06 B0	C07 B100	C08 B50	C09 B0
P01	−52.1	−22.2	−24.9	27.5	1.4	6.9	−31.1	−26.1	4.3	C04	27.5	−52.1	79.6
P02	−6.3	2.6	−0.9	8	10.3	−25	30.2	1	21	C07	30.2	−25	55.2
P03	0.1	−3.1	−4.4	7.8	8.6	10.4	8.2	6.6	0.9	C06	10.4	−4.4	14.8
P04	−7.5	3.9	−19.6	64.3	89.3	−3.2	−70.3	−33.9	−6.2	C05	89.3	−70.3	159.6
P05	−53.1	−26.7	9.1	−1.2	0.2	−3.4	7.3	28.3	23.8	C08	28.3	−53.1	81.4
P06	17.3	−22.1	19.9	34.9	19.1	18.3	28.8	30.3	15.9	C04	34.9	−22.1	57
P07	−18	9.9	−45.7	−47.9	−42.5	−10.5	−27.6	−35.6	−30.1	C02	9.9	−47.9	57.7
P08	−32.2	−9.9	−5.4	−13.4	−32.2	−16.3	4.3	−9.5	−12.2	C07	4.3	−32.2	36.5
P09	39.9	−5.4	−9.8	17.6	22.1	45.6	29.7	−12.6	27.9	C06	45.6	−12.6	58.3
P10	−18.1	−0.7	−29.4	37.6	8.1	29.1	−7.4	−36.1	−0.3	C04	37.6	−36.1	73.7
P11	28.2	13.6	1.5	28.9	−7.4	14.5	56.6	24.2	49.5	C07	56.6	−7.4	64
P12	61.4	25.1	43.2	24.7	34	32.2	−20.4	28.7	37.7	C01	61.4	−20.4	81.9
P13	−8.7	−16.3	−30.6	−23	−23.7	1.3	−11.9	−33	−7.4	C06	1.3	−33	34.2
P14	25	27	7.1	26	28.3	27.5	32.1	40.8	22.4	C08	40.8	7.1	33.6
P15	−18.3	−20.9	−0.9	−12.1	−46.5	9.1	−1.7	−37.6	−5.4	C06	9.1	−46.5	55.6
P16	−9.2	7.6	−18.3	−3	6.2	−25.4	9.1	2.6	−4.3	C07	9.1	−25.4	34.5
P17	−23.9	−8	−18.9	−16.8	−14.8	−18.5	−18	−17.7	−13.6	C02	−8	−23.9	15.9
P18	21.5	−11.7	−20.3	−10.1	4.7	−56.3	−16.9	4.7	−1.8	C01	21.5	−56.3	77.8
P19	9	−0.2	−18.6	32.4	7.5	4.1	17.8	10.8	−16.7	C04	32.4	−18.6	51
P20	11.7	−2.2	0.5	−1.1	9	6	4	5.4	−12.4	C01	11.7	−12.4	24.1
P21	−39.6	−43.2	−41.7	−35.5	−23.3	−38.9	−37.1	−36	−37	C05	−23.3	−43.2	19.9
P22	0.1	−55	−24.4	3.6	−14.4	−21	−3.3	−8.8	−18.7	C04	3.6	−55	58.6
P23	−20.4	−20.8	−12.6	9.3	8.3	1.3	19.3	−2.5	−29.1	C07	19.3	−29.1	48.4
P24	27.3	45.9	39.8	26.8	25.2	28.8	50	34.8	36.6	C07	50	25.2	24.8
Group mean	−2.8	−5.5	−8.5	7.7	3.2	0.7	2.2	−3	1.9	―	―	―	54.1
SD	28.5	22	22.1	25.8	28.2	24.2	29.1	25.4	22.9	―	―	―	30.8

Partition 100% = fully enclosed (spatial openness 0%); Partition 50% = front partition only; Partition 0% = no partitions (spatial openness 100%). B100/B50/B0 = blind fully open/half-open/fully closed. pp: percentage points. Green cells indicate the maximum beta/alpha ratio change rate (optimal condition) for each participant; red cells indicate the minimum (lowest-efficiency condition).

presents the beta/alpha ratio change rates (%) across the nine conditions, individually optimal conditions, and response range values for all 24 participants.

The response range—the difference in beta/alpha ratio change rate between each participant’s optimal and lowest-efficiency conditions across nine conditions—averaged 54.1 percentage points (pp; SD = 30.8 pp; range: 14.8–159.6 pp). The median was 55.4 pp, with a first quartile (Q1) of 34.1 pp and a third quartile (Q3) of 66.4 pp. A Q1 of 34.1 pp indicates that 75% of participants experienced inter-condition variation of at least 34 pp. The participant with the largest response range (P04) showed +89.3% in the optimal condition and −70.3% in the lowest-efficiency condition, yielding a total range of 159.6 pp; even the participant with the smallest range (P03) showed 14.8 pp of variation. Neurophysiological fluctuation across environmental conditions was thus confirmed in every participant; the structural interpretation of this variation is addressed through subsequent variance decomposition results (ICC, R²).

Figure 4 displays each participant’s response range in descending order. The substantial heterogeneity across individuals—ranging from P04 (159.6 pp) to P03 (14.8 pp)—and the even distribution of participants on either side of the mean line indicate that environmental effects were widespread at the individual level rather than driven by a small number of extreme cases.

At the group level, however, the two-way repeated-measures ANOVA yielded no statistically significant effects: main effect of spatial openness, F(2, 46) = 2.82, p =0.070, η²G = 0.024; main effect of visual openness, F(2, 46) = 0.39, p = 0.678, η²G = 0.002; interaction, F(4, 92) = 0.45, p = 0.775, η²G = 0.003. As an index that includes between-participant variance in its denominator, η²G corrects for the overestimation inherent in partial η² [54]; all three η²G values fell below the small-effect benchmark. This indicates that the openness factors account for limited variance at the group-mean level.

To determine whether these non-significant results reflect an absence of environmental effects, variance decomposition was conducted via a linear mixed-effects model. The ICC was 0.455, indicating that approximately 46% of total variance originated from between-participant differences. ICC is defined as the ratio of between-participant variance to total variance [59]; ICC = 0.455 signifies that response levels differ substantially across participants even under identical environmental conditions. The marginal R² was only 0.025, whereas the conditional R² rose to 0.469. The difference between these two values (0.444) indicates that between-participant variance contributes far more to explaining total variance than the effects of environmental conditions do. The non-significant group-mean results are therefore interpreted not as an absence of environmental effects but as statistical cancellation resulting from heterogeneity in the direction of individual responses.

This bidirectional cancellation was confirmed in a condition-level subgroup analysis. Classifying participants in each condition as either an improvement group or a deterioration group relative to their individual mean revealed that both groups coexisted in all nine conditions. In condition C07 (no partition, blinds fully open), for example, the group mean was only +2.2 pp, yet the improvement group (n = 16) averaged +17.5 pp while the deterioration group (n = 8) averaged −28.6 pp. Opposite neurophysiological responses to the same physical environment were thus present across participants, and these opposing responses canceled each other out in group-mean analyses, attenuating the apparent average effect.

In sum, spatial and visual openness constitute environmental factors that induce a mean response range of 54.1 pp at the individual level. The variance decomposition results (ICC =0.455; conditional R² =0.469 vs. marginal R² =0.025) show that approximately 46% of total variation originates from between-participant differences, while environmental factors account for only 2.5% of variance on average. These findings indicate that group-mean approaches alone are insufficient for understanding the neurophysiological effects of office spatial openness, and that individual-level response structures must be considered in conjunction.

4.2. Comparative Contributions of Spatial and Visual Openness

The effects of spatial and visual openness were analyzed at the individual level by holding one dimension constant and computing the response range across the three levels of the other dimension for each participant. The mean individual response range for spatial openness was 29.1 pp (SD = 17.7, Mdn = 25.3 pp, range: 7.3–87.4 pp), and for visual openness was 26.2 pp (SD = 12.6, Mdn = 26.7 pp, range: 4.8–60.0 pp). A paired-samples t-test indicated that the difference between the two dimensions was not statistically significant (t(23) = 1.33, p = 0.197), demonstrating that spatial and visual openness exert equivalent influences on cognitive efficiency at the group level.

At the group level, the highest mean across the nine conditions was observed in C04 (partition 50%, blinds fully open) at +7.7%, and the lowest in C03 (partition 100%, blinds fully closed) at −8.5%. Across levels of spatial openness, mean change rates were +0.4% for the no-partition condition, +3.9% for the half-partition condition, and −5.6% for the fully enclosed condition. This pattern is non-linear—the half-partition condition yielded the highest group mean, with the no-partition condition somewhat lower—but given that the main effect of spatial openness was non-significant (p = 0.070), these numerical differences should be interpreted as a consequence of the bidirectional cancellation identified in Section 4.1. Across levels of visual openness, the fully open blind condition had the highest mean at +1.1%, and the half-open condition the lowest at −1.8%. Simple main effects at each level were non-significant for both dimensions (spatial openness: F = 0.48–1.73, p = 0.185–0.620; visual openness: F = 0.01–0.30, p = 0.744–0.989), confirming that bidirectional cancellation operated equally across individual-level comparisons.

Analysis of the dominant dimension at the individual level revealed that 15 of 24 participants (62.5%) showed a larger response range for spatial openness than for visual openness (spatially dominant), while 9 (37.5%) showed the reverse (visually dominant). Among spatially dominant participants, the maximum difference between dimensions was +27.3 pp (P04: spatial 87.4 pp, visual 60.0 pp); among visually dominant participants, the maximum was 18.2 pp (P15: spatial 18.9 pp, visual 37.1 pp). Some participants showed near-equivalent sensitivity to both dimensions (P14: +0.3 pp; P20: +0.2 pp). Overall environmental sensitivity (response range) was comparable across groups (spatially dominant: M = 55.4 pp; visually dominant: M = 52.0 pp; t(22) = 0.25, p = 0.801), indicating that total neurophysiological responsiveness did not differ between groups regardless of which dimension was dominant (Figure 5).

Figure 5 presents two complementary views of these results. Panel A displays the difference between each participant’s spatial and visual response ranges (Spatial − Visual); blue bars indicate spatially dominant participants (n = 15) and red bars visually dominant participants (n = 9), with values distributed in both directions from zero. Panel B is a raincloud plot showing the response range distributions for both dimensions across all 24 participants; medians are indicated within each distribution. The two panels together illustrate that group means for the two dimensions are nearly equivalent while substantial heterogeneity exists at the individual level (t(23) = 1.33, p = 0.197).

A strong positive correlation was observed between individual response ranges for spatial and visual openness (r = 0.80, p < 0.001). This indicates that participants who responded sensitively to changes in spatial openness also tended to respond sensitively to changes in visual openness, suggesting that neurophysiological reactivity to environmental change may constitute an individual-specific trait spanning both dimensions. By contrast, neither the correlation between subjective perceived spatial openness ratings and EEG spatial response range (r = 0.01, p = 0.971) nor that between subjective visual comfort ratings and EEG visual response range (r = 0.18, p = 0.395) reached significance. Which dimension an individual is neurophysiologically more sensitive to is therefore unrelated to their subjective perception of sensitivity to that dimension.

These results demonstrate that spatial and visual openness exert equivalent effects on cognitive efficiency at the group level, yet the dominant dimension varies across individuals. From a design standpoint, this implies that environments permitting independent adjustment of only one dimension may fail to support the dimension to which a substantial proportion of users primarily respond. Configurations that allow both dimensions to be controlled independently can accommodate a broader range of users’ cognitive efficiency needs, and within activity-based working environments, offering diverse combinations of spatial and visual openness enables individuals to select settings that match their own response patterns.

4.3. Diversity of Individual Optimal Conditions and Subjective–Objective Mismatch

Analysis of the distribution of EEG-identified optimal conditions revealed that 7 of the 9 conditions served as the optimal condition for at least one participant (

Table 5. Frequency of optimal and lowest-efficiency condition selections by condition.

Condition	Partition	Blind	Optimal Selections (n)	Lowest-Efficiency Selections (n)	Rank
C07 (P0-B100)	0%	100%	6 (25.0%)	2 (8.3%)	1
C04 (P50-B100)	50%	100%	5 (20.8%)	1 (4.2%)	2
C06 (P50-B0)	50%	0%	4 (16.7%)	3 (12.5%)	3
C01 (P100-B100)	100%	100%	3 (12.5%)	4 (16.7%)	4
C02 (P100-B50)	100%	50%	2 (8.3%)	3 (12.5%)	5
C05 (P50-B50)	50%	50%	2 (8.3%)	3 (12.5%)	5
C08 (P0-B50)	0%	50%	2 (8.3%)	3 (12.5%)	5
C03 (P100-B0)	100%	0%	0 (0%)	3 (12.5%)	8
C09 (P0-B0)	0%	0%	0 (0%)	2 (8.3%)	8

). No single condition was optimal for a majority of participants (12 or more), indicating that a uniform environmental design capable of simultaneously optimizing cognitive efficiency for all occupants is not achievable in principle.

The most frequently identified optimal condition was C07 (no partition, blinds fully open) with 6 participants (25.0%), followed by C04 (partition 50%, blinds fully open) with 5 (20.8%) and C06 (partition 50%, blinds fully closed) with 4 (16.7%). Neither C03 nor C09—both fully closed-blind conditions—was identified as optimal for any participant, though this pattern should be treated as directional given the sample size of 24. Only 2 participants (8.3%) identified the intermediate condition C05 as their optimal condition, contradicting the design intuition that middle-ground conditions are broadly acceptable.

The distribution of lowest-efficiency conditions was similarly heterogeneous. C01 (partition 100%, blinds fully open) was most frequent at 4 participants (16.7%), followed by C02, C03, C05, C06, and C08 at 3 each (12.5%). Notably, C07 was simultaneously the optimal condition for 6 participants and the lowest-efficiency condition for 2—demonstrating that the same physical environment can be neurophysiologically most advantageous for some individuals and most disadvantageous for others, a condition-level reconfirmation of the bidirectional cancellation established in Section 4.1.

To assess the correspondence between subjective preferences and EEG-identified optimal conditions, each participant’s “most preferred condition” from the post-session survey was compared with their EEG-identified optimal condition. Only 3 of 24 participants (12.5%; P08, P11, P20) showed agreement; the remaining 21 (87.5%) did not. The observed agreement rate marginally exceeds the chance level of 1/9 = 11.1%, but a binomial test indicated no statistically significant difference (p = 0.508), demonstrating that participants were unable to accurately identify their neurophysiologically optimal condition through subjective appraisal alone.

Figure 6 plots each participant’s EEG-identified optimal condition (blue) and subjectively preferred condition (red) across the nine conditions. Green cells indicate agreement (P08, P11, P20). Red cells cluster toward the left (C01–C03, high-partition conditions), while blue cells are distributed across all nine conditions.

Table 6. Comparison of EEG-identified optimal conditions and subjective preferences by participant.

No.	Participant	EEG Optimal Condition	β/α Change (%)	Subjective Preference	Match
1	P01	C04 (P50·B100)	+27.5	C03 (P100·B0)	X
2	P02	C07 (P0·B100)	+30.2	C03 (P100·B0)	X
3	P03	C06 (P50·B0)	+10.4	C01 (P100·B100)	X
4	P04	C05 (P50·B50)	+89.3	C03 (P100·B0)	X
5	P05	C08 (P0·B50)	+28.3	C03 (P100·B0)	X
6	P06	C04 (P50·B100)	+34.9	C01 (P100·B100)	X
7	P07	C02 (P100·B50)	+9.9	C03 (P100·B0)	X
8	P08	C07 (P0·B100)	+4.3	C07 (P0·B100)	O
9	P09	C06 (P50·B0)	+45.6	C04 (P50·B100)	X
10	P10	C04 (P50·B100)	+37.6	C02 (P100·B50)	X
11	P11	C07 (P0·B100)	+56.6	C07 (P0·B100)	O
12	P12	C01 (P100·B100)	+61.4	C01 (P100·B100)	X *
13	P13	C06 (P50·B0)	+1.3	C02 (P100·B50)	X
14	P14	C08 (P0·B50)	+40.8	C01 (P100·B100)	X
15	P15	C06 (P50·B0)	+9.1	C09 (P0·B0)	X
16	P16	C07 (P0·B100)	+9.1	C03 (P100·B0)	X
17	P17	C02 (P100·B50)	−8.0	C05 (P50·B50)	X
18	P18	C01 (P100·B100)	+21.5	C08 (P0·B50)	X
19	P19	C04 (P50·B100)	+32.4	C09 (P0·B0)	X
20	P20	C01 (P100·B100)	+11.7	C01 (P100·B100)	O
21	P21	C05 (P50·B50)	−23.3	C02 (P100·B50)	X
22	P22	C04 (P50·B100)	+3.6	C06 (P50·B0)	X
23	P23	C07 (P0·B100)	+19.3	C01 (P100·B100)	X
24	P24	C07 (P0·B100)	+50.0	C01 (P100·B100)	X

* P12 selected both C01 and C09 as subjectively preferred conditions; as the preference was not exclusive to a single condition, this case was coded as a mismatch.

presents a participant-by-participant comparison of EEG-identified optimal conditions and subjectively preferred conditions, with the corresponding beta/alpha ratio change rate. Agreement is indicated for P08, P11, and P20; all remaining participants showed disagreement.

Several patterns in Table 6 merit attention. Beta/alpha ratio change rates at the EEG-identified optimal condition ranged from −23.3% (P21) to +89.3% (P04). P17 and P21 recorded negative values across all nine conditions, meaning their beta/alpha ratio did not recover to baseline throughout the session; for these participants, the optimal condition is defined as the one with the smallest magnitude of decrease. Among subjectively preferred conditions, C03 and C01 were each selected by 6 participants, meaning 15 participants (62.5%) preferred high-partition enclosed conditions (C01–C03), while C07 (no partition) was the condition most frequently identified as optimal by EEG, at 6 participants (25.0%). Of the 21 mismatched participants, 8 (38.1%) showed reverse-direction mismatch, in which subjective preference and neurophysiological optimality were diametrically opposed in terms of spatial openness (see Table 6).

The two findings together—an agreement rate (12.5%) not exceeding chance (p = 0.508) and individually optimal conditions distributed across 7 of 9 conditions—demonstrate that subjective preferences alone are insufficient to identify an individual’s neurophysiologically optimal condition in terms of spatial openness in office design.

5. Discussion

The present study aimed to empirically ascertain the effects of spatial and visual openness in office environments on cognitive efficiency as measured by EEG at the individual level. Results across the four research objectives converge on a single consistent structure: individual-level responses are substantial and genuine, yet they are statistically canceled out in group-mean analyses because of heterogeneity in response direction.

Neither spatial openness nor visual openness reached significance in the group-level ANOVA, but this does not indicate an absence of environmental effects. Variance decomposition results (ICC = 0.455; marginal R² = 0.025 vs. conditional R² = 0.469) demonstrate that between-participant differences contribute far more to explaining total variance than environmental conditions do. This bidirectional cancellation structure offers a potential mechanistic explanation for the coexistence of contradictory findings that has characterized open-office research for decades—namely, reports that greater openness enhances productivity [5] alongside reports that it reduces productivity [6]. Studies reporting only group-mean results in the presence of opposing individual response directions may have arrived at null conclusions as a consequence of this structural limitation of group-mean analysis rather than a true absence of environmental effects.

The bidirectional cancellation observed here is consistent with prior evidence that individual differences in environmental response have a neurophysiological basis. Responses to open-plan offices have been shown to vary substantially across individuals depending on personal characteristics [7], and individuals with high noise sensitivity have been demonstrated to exhibit distinct auditory cortex processing patterns in response to the same acoustic environment [36]. The present findings extend this line of evidence by showing that opposing response directions are not idiosyncratic exceptions but a systematic structural feature of how office openness affects cognitive efficiency—one sufficiently prevalent to statistically cancel out genuine environmental effects at the group level.

A further notable finding is the strong positive correlation between individual response ranges for spatial and visual openness (r = 0.80, p < 0.001). This indicates that participants who responded sensitively to changes in one dimension also tended to respond sensitively to changes in the other, suggesting that neurophysiological reactivity to openness may constitute an individual-specific trait spanning both dimensions. This cross-dimensional consistency raises the question of what neurophysiological or psychological characteristics underlie individual environmental reactivity. Candidate variables may include sensory processing sensitivity, introversion, or trait anxiety, each of which has been associated with heightened physiological reactivity to environmental stimulation. If such predictors can be identified, data-driven design approaches that estimate each occupant’s optimal condition prospectively may become feasible.

The subjective–objective mismatch rate of 87.5% warrants particular attention. One plausible account of the mismatch is that occupants’ spatial preferences are shaped by aesthetic and social norms—such as the widespread association of enclosed spaces with professionalism or privacy—rather than by direct awareness of their cognitive efficiency under each condition. Because the beta/alpha ratio reflects neurophysiological processes that are not directly accessible to introspection, subjective appraisal alone cannot reliably identify the physiologically optimal environment. This interpretation is consistent with findings showing that substantial improvements in cognitive function produced by environmental changes went unnoticed by participants [8].

These findings carry two implications for the spatial design of offices. First, no single condition can simultaneously support the cognitive efficiency of all occupants, and the design convention that intermediate conditions are broadly acceptable lacks empirical support. Within activity-based working environments, offering diverse combinations of spatial and visual openness that can be independently adjusted represents a design strategy with genuine justification for accommodating the diversity of individual neurophysiological responses. Second, providing choice alone is insufficient given an 87.5% mismatch rate. If occupants select spaces based solely on subjective preference, they may repeatedly fail to access their neurophysiologically optimal conditions. A complementary feedback system linking spatial usage patterns with objective performance data is needed to help occupants become aware of changes in their own cognitive efficiency. Although no significant group-level effects were observed, the present findings demonstrate that spatial and visual openness are meaningfully associated with cognitive efficiency at the individual level. Future research should therefore investigate what neurophysiological or personal characteristics underlie these individual differences in environmental response.

The present study has several limitations. The laboratory setting did not replicate the complex contextual factors present in actual office environments, and the sample of 24 participants limits the ecological validity and generalizability of the findings. The four-minute measurement period per condition and the restriction to a single attentional task preclude conclusions about long-term adaptation or response patterns during creative and collaborative tasks. Each condition was measured only once per participant, limiting the reliability of individual-level responses and making it difficult to fully separate stable neurophysiological traits from single-measurement variability; future studies should address this through repeated measurements or more robust statistical models. Visual openness was operationalised solely through blind aperture, and luminance distribution, glare, and view quality were neither controlled nor measured, so their contributions to the observed individual differences cannot be fully excluded; findings related to visual openness should therefore be interpreted within the scope of this operationalisation and may have limited direct applicability to architectural design practice.

Future research should examine long-term tracking in real office environments, condition-by-task-type interactions in optimal conditions, predictors of individual-specific environmental reactivity, and whether data-driven feedback can improve occupants’ self-awareness of their cognitive efficiency.