Atmosphere perception of dynamic LED lighting over different hue ranges

Abstract

This paper investigates the atmosphere perception of dynamic coloured lighting over warm and cool hue ranges in a living room. Experiments were carried out with eight fully functional multi-channel tunable LED luminaires. Thirty Chinese observers (14 males and 16 females) evaluated 36 light conditions (2 luminance ×2 chroma ×3 hue range types ×3 speeds) using a questionnaire of 25 response scales. Four atmosphere factors were extracted: coziness, spaciousness, liveliness and warmth. Both dynamics and hue range significantly affected the atmosphere perception.

1. Introduction

In recent years, LEDs have become a popular choice for illumination, not only because of their long service life and low energy consumption, but also because of their good performance with rich colours and dynamics. The LED provides more freedom to create more complex and attractive effects in luminaire design.

Many investigations have confirmed that lighting affects human emotion, performance and cognition.^1–6 Models have been developed from investigations of the emotional effects of lighting. In 1974, Mehrabian and Russell⁷ proposed the PAD (Pleasure, Arousal and Dominance) model using the emotional scales: pleasure–displeasure, arousal–nonarousal and dominance–submissiveness. In 1988, Watson et al.,⁸ proposed the PANAS (Positive and Negative Affect Schedule) model, which consists scales of positive affect and negative affect. Research based on these two models confirmed that lighting does influence human emotion. However, there were some disagreements among the previous work.^3,5,9 This is due to the complexity of emotions: They are not only affected by environmental factors, but also by non-environmental factors, such as the observer’s internal state.

Flynn^10–12 was one of the pioneers to measure the subjective impression of environments for various lighting systems using semantic differential rating scales. Through factor analysis, three dominant categories were discriminated: evaluative impression (e.g. pleasant–unpleasant), visual clarity (e.g. clear–hazy) and spaciousness (e.g. spacious–cramped). The last two categories describe how the environment is visually perceived. Only the evaluative category contains a few atmosphere descriptors. In 2008, Vogels¹³ introduced the concept of atmosphere. Different from the affective state such as emotion and mood, atmosphere is the experience of ambient surroundings in relation to observers. They also developed a metric to measure the atmosphere perception, and extracted two factors as coziness and liveliness. Employing Vogels’s method, many scholars have investigated lighting atmosphere from various angles: Erp¹⁴ investigated perceived atmosphere of static neutral light and coloured light; Custers et al.¹⁵ studied the atmosphere perception in retail environments; Liu et al.¹⁶ developed a word-pair questionnaire for Chinese observers and investigated the atmosphere perceptions in a living room with different luminares and lighting sources. Kuijsters et al.¹⁷ focused on effects of ageing on atmosphere perception.

However, all these atmosphere perception studies were carried out with static light. As one of the major features of LED-based lighting, the affective effects of dynamic lighting need more attention. For the dynamic property of colour fluctuation speed, Sekulovski et al.¹⁸ and Wan et al.¹⁹ found that orange light conditions created a more relaxed atmosphere than white light, and adding slow pulsations to some coloured lighting might support stress-reduction while creating a relaxed atmosphere. Wang et al.⁹ studied the atmosphere of full hue-range colour-varied dynamic lighting and extracted three atmosphere factors: tenseness, coziness and liveliness. She confirmed the effect of dynamic lighting on liveliness, but few investigations have been carried out on the effects of dynamic lighting over different hue ranges. According to colour psychology, the basic colour wheel can be split into two halves, warm and cool. This study explores the atmosphere perceptions of dynamic lighting with colour varying over warm and cool hues. The impact of lighting parameters on visual scaling is also investigated. Besides, gender and academic background effects on atmosphere perception are also reported.

2. Method

2.1. Experiment environment

Our experiments were conducted in a lighting laboratory (length 5.4 m, width 4.7 m, height 2.7 m) decorated as a living room. Figure 1 shows the test environment. An elliptical tea table (1.7m × 0.6 m × 0.45 m high) was placed directly under the ceiling panel, covered with a piece of grey paper. Two dark brown leather sofas were adjacent to the table. One hundred wooden cubic blocks (3 cm on each side), some colourful fruits (including oranges, apples, Indian jujubes and straw berries), flowers and plastic balls were put on the table to help simulate a living room environment. The mean CIELAB²⁰ value (L*a*b*) of the grey paper surface (L*= 76.6, a*= −2.0, b*= 8.0), brown leather sofas (L*= 26.6, a*= 5.7, b*= 4.0) and wooden blocks (L*= 71.8, a*= 6.1, b*= 32.2) were measured by an Xrite i1Por Spectrophotometer under D65/10 degrees condition. Room temperature was kept at 23 ± 1℃ during experiment.

Figure 1

The experiment environment.

2.2. Lighting conditions

The room was illuminated from above by eight LEDCubes© (Thouslite Ltd.). Each LEDCube is a multi-channel LED lighting system. Six different types of LED are used in the system giving a good coverage of the visible spectrum. Two types of red LEDs were together controlled as an R channel, similarly two for green (G) and two for blue (B). Figure 2 shows the spectral power distributions.

Figure 2

Normalised spectrum power of LED cubes used in the experiment.

The PLCC model (Piecewise Linear interpolation assuming Constant Chromaticity coordinates)²¹ was used to characterise the signal of each channel (RGB) to the CIE XYZ tristimulus values. In this way, we can precisely specify the desired colour of light. Then, 20 colours were chosen from the experimental conditions to test the model’s accuracy between the target and measured colours. The XYZ values of the target colours were measured, pointing at a white reference at the centre of the table at a distance of 1.85 m away from the ceiling LED lighting. A JETI Specbos 1211UV tele-spectroradiometer was used for all the colour measurement work. The results suggested a good model accuracy with a mean of 0.56 CIELAB colour difference. Finally, all the 36 light stimuli were transformed to CIELAB in terms of L*, a*, b*, C_ab* and h_ab values. Table 1 lists lightness (L*), chroma (C*_ab), h_ab range (HR) and speed (S). An equal energy white (illuminant E) was set as reference white. Luminance values were measured at different lightness values. For lightnesses of 40 and 70, luminance values correspond to 23 and 69 cd/m², respectively. For the experimental lighting choosing and modeling, we used the CIELAB lightness. For colour specification and later technical discussion, luminance was used.

Table 1

Physical parameters.

Lightness (L*)	Chroma (C* _ab )	Hue range (HR in h_ab angle)	Speed (S in ΔE_ab/s)
40 70	18 36		0.75 1.5 3

Here, chroma is the radial component of the cylindrical coordinate CIE L*C_ab*h_ab representation of the CIELAB colour space. The definitions of C*_ab and h_ab are (1)C_{ab}^ \ast = \sqrt {a^{ \ast 2} + b^{ \ast 2} } (2)h_{ab} = \tan ^{ - 1} \left( {{{b^ \ast } / {a^ \ast }}} \right) where a* and b* are respectively the CIELAB redness–greenness and yellowness–blueness values. Colour difference (ΔE*_ab) is the Euclidean distance between two colours in CIELAB colour space. This was introduced to define and control dynamic lighting speed, as the quantity of ΔE*_ab displayed per second.

To produce smooth dynamic colour-varied lighting, perceived flicker caused by an inappropriate minimum distance between sampling stimulus should be avoided.²² Note that since CIE LAB is not a uniform space, the perceived colour differences in the 4th quadrant of the a*b* plane are larger than those in the other quadrants. With the constant ΔE*_ab interval set in the experiment, it causes flicker in the 4th quadrant hue range, especially around pink and purple regions. Hence, the 4th quadrant hue range was excluded in the experiment.

The remaining hue ranges were set as three types: (0°–135°) which covers warm hues, (135°–270°) which covers cool hues and the full hue-range (0°–270°). They were assigned as 1, −1 and 0 respectively, according to the proportion of warm hue in each range type. Each range was displayed back and forth at different speeds.

2.3. Questionnaire

Three questionnaires from previous studies by Vogels,¹³ Wang et al.⁹ and Liu et al.¹⁶ were taken as references in the present study. Vogels¹³ applied 38 words, originally Dutch and translated into English in her paper. Liu et al.’s¹⁶ experiment contained 71 word-pairs, and was conducted in Chinese and also translated into English. However, they were both concerned with static lighting environments. Wang et al.⁹ applied 21 words to study the atmosphere produced by dynamic coloured light. This experiment was also carried out in the Chinese language. Taking Liu et al.’s questionnaire¹⁶ as a basic reference, word pairs were used to avoid linguistic misunderstanding. Each word pair is also considered as one scale.

All scales were classified into three categories as light appearance, atmosphere and preference. Every scale was looked up through the online Merriam Webster and Oxford Advanced Learner’s English-Chinese Dictionary. According to the words’ meaning, some words were eliminated because they are not commonly used to describe room atmosphere in Chinese, such as ‘musty’ (‘bedompt’ in Dutch). When two words have similar meanings, such as ‘exciting’ and ‘inspiring,’ only one of them was retained (in this example, ‘inspiring’ was retained). The scale ‘natural–artificial’ was added to test the effect of light from the aesthetic experience perspective on the interior space. And ‘ugly–beautiful’ was added to assess observers’ preference.

Finally, 25 scales were used in the present study. As shown in Table 2, both Chinese and the corresponding English translations were listed. Thirty-eight words can be found in Liu et al.’s paper,¹⁶ including 29 words translated from Vogel¹³ and 10 words from Wang et al.⁹ The scale ‘unsmooth–smooth’ was included to monitor any visually observed flicker from the experimental lighting. Because of the language usage difference, translations from Chinese to English may be not precise.

Table 2

Questionnaire composition.

Constructive dimensions	Scale index	Chinese scales		English scales
Constructive dimensions	Scale index	Negative term	Positive term	Negative term	Positive term
Appearance of the room	Q1	冷的	暖的	cool ^a,b	warm ^a,b
	Q2	昏暗的	明亮的	dim^b	bright^b
	Q3	闪烁的	不闪烁的	unsmooth	smooth
	Q4	慢的	快的	slow	fast
Atmosphere of the room	Q5	非正式的	正式的	informal^b	formal ^a,b,c
	Q6	狭小的	宽敞的	crowded	spacious ^a,b
	Q7	难受的	舒适的	uncomfortable ^a,b,c	cozy^a,b,c
	Q8	乏味的	有趣的	boring ^a,b	interesting ^a,b
	Q9	昏昏欲睡的	令人清醒的	drowsy ^b	alert^b
	Q10	令人焦躁的	令人平静的	agitated	tranquil ^a,b,c
	Q11	令人压抑的	鼓舞人心的	depressing ^a,b	inspiring ^a,b,c
	Q12	令人紧张的	令人放松的	tense^a,b,c	relaxed^a,b,c
	Q13	死气沉沉的	充满活力的	lifeless^b	lively^a,b,c
	Q14	古板的	浪漫的	unromantic	romantic ^a,b
	Q15	有敌意的	友好的	hostile^a,b,c	friendly^a
	Q16	冷清的	热闹的	detached ^a,b	attached
	Q17	神秘的	熟悉寻常的	mysterious ^a,b	familiar^b
	Q18	私密个人的	开放公开的	personal ^a,b	open^b
	Q19	简朴的	华丽的	plain^b	luxurious ^a,b
	Q20	冰冷的	温馨的	distant	intimate ^a,b
	Q21	令人不快的	令人愉快的	unpleasant^b	pleasant^a,b,c
	Q22	危险的	安全的	dangerous ^a,b	safe^a,b,c
	Q23	令人恐惧的	引人亲近的	terrifying ^a,b	delightful
Association	Q24	乡村田园的	都市气息的	natural^b	artificial^b
Preference	Q25	丑的	美的	ugly	beautiful^b,^c

Words from Vogels’ questionnaire.

Words from Liu et al.’s questionnaire.

Words from Wang et al.’s questionnaire.

Observers were asked to apply a 6-point rating to each scale in the questionnaire. For each light condition, observers were forced to choose one term from the scale to describe it. Then they were asked to report how applicable the term is from 1 the least to 3 the most. If the observer chose the negative term, the scale score would be negative. Otherwise, it would be positive. The questionnaire was presented in Chinese and the responses were oral. All scores were used directly in the following calculations.

2.4. Observers

Previous studies^23,24 revealed perception differences existing between designers and users in many areas. Since designers play a great part in products such as lighting, observers with and without design background were both involved. Thirty Chinese observers (14 males, 16 females) having a mean of 25.4 years and a standard deviation of 2.9 years participated.²⁵ Twelve came with design-related background. All observers passed the Ishihara test.¹⁴ None had experience of assessing lighting atmosphere.

2.5. Experimental procedure

Observers performed the experiment individually. When an observer entered the Lighting Lab, it was illuminated with a neutral white light (luminance: 68 cd/m², 6500 K). He/she was seated in the middle of the sofa facing the front white wall. The experimental procedure and the questionnaire were then explained (Table 3).

Table 3

The experiment procedure.

	Phase 1	Phase 2	Phase 3
Light condition	Neutral light	Experimental light	Experimental light
Time duration	10 seconds	45 seconds	Normally 100–120 seconds
Task	Accommodate	Building blocks	Answer questionnaire

When an experiment started, the light conditions were displayed one by one in random sequence. Every light condition included three phases. Phase 1 was 10 seconds neutral light (luminance: 68 cd/m², 6500 K) for the observers to prepare. Phase 2 was 45 seconds²⁶ of experimental light. To help observers accommodate, they were asked to build certain forms with 100 wooden blocks placed on the table. A sound signal rung at the end of this phase. Phase 3 was to answer the questionnaire with the experimental lighting displaying.

In order to avoid order bias, all 36 light conditions were randomly arranged through all observers. The order of the questionnaire scales was also randomised. Each observer finished the experiment in two sessions. Each session contained 18 light conditions and took 60 minutes. The interval between two sessions was one week.

3. Results and discussions

3.1. Light quality and reliability of the questionnaire

The score of unsmooth-smooth for all observers across all light conditions (Mean ±Standard error = 0.09 ± 0.057) indicates that all experimental lighting was smooth without perceptible flicker. Reliability describes the consistency of measuring tool or method. Cronbach’s Alpha (α) is usually used as estimation of reliability. A commonly accepted rule²⁷ is: α greater than 0.6 indicates acceptable reliability, beyond 0.9 shows excellent reliability. In this study, an α of 0.85 demonstrated a high reliability for the experiment.

3.2. Inter-observer agreement

Inter-observer agreement describes the data consistency between different observers scoring the same conditions. This can be measured by calculating the standard deviations associated with each scale for each condition. The nearer the mean standard deviation (MSD) gets to zero, the more the observers agreed with each other. An MSD value of 1.0 indicates an inter-observer variability of 1.0 on the −3 to 3 category scale used.¹⁶

The MSD for all 36 conditions were calculated. Table 4 shows the overall mean MSD value was 1.52 ranging from 1.31 to 1.71. Comparing with MSD from similar experiments,^9,16 this MSD result was acceptable. The top four inconsistent scales were: unsmooth–smooth, lifeless–lively, plain–luxurious and slow–fast. MSD for males and females were also calculated. MSD values of the males were similar to females for most scales, except for informal–formal and lifeless–lively, where the difference of MSD was more than 0.18.

Table 4

The MSD values for male, female and all observers

Scales	Male	Female	Total
Cool–warm	1.66*	1.56	1.60
Dim–bright	1.43	1.40	1.44
Unsmooth–smooth	1.65	1.63	1.65*
Slow–fast	1.64	1.57	1.62
Informal–formal	1.59	1.40	1.51
Crowded–spacious	1.46	1.42	1.46
Uncomfortable–cozy	1.44	1.42	1.45
Boring–interesting	1.57	1.52	1.55
Drowsy–alert	1.58	1.56	1.58
Agitated–tranquil	1.44	1.48	1.46
Depressing–inspiring	1.48	1.46	1.48
Tense–relaxed	1.54	1.47	1.51
Lifeless–lively	1.54	1.71*	1.64
Unromantic–romantic	1.48	1.48	1.49
Hostile–friendly	1.41	1.38	1.41
Detached–attached	1.57	1.55	1.56
Mysterious–familiar	1.60	1.56	1.60
Personal–open	1.50	1.53	1.54
Plain–luxurious	1.65	1.62	1.64
Distant–intimate	1.56	1.46	1.52
Unpleasant–pleasant	1.42	1.36	1.39
Dangerous–safe	1.51	1.43	1.48
Terrifying–delightful	1.37**	1.31**	1.35**
Natural–artificial	1.54	1.66	1.61
Ugly–beautiful	1.42	1.31**	1.38
Mean	1.52	1.49	1.52

Marks the maximum values.

Marks the minimum values.

3.3. Gender differences in atmosphere scale perception

To avoid the effects of light condition differences and reveal gender effects, we applied the partial correlation analysis with Spearman rho correlation method: Spearman’s rho test is a non-parametric measure to deal with the interval data and partial correlation is the correlation between two variables after removing the effect of one or more additional variables. Across all physical parameters, eight scales showed significant differences with p values lower than 0.05. They are ‘slow–fast’ (p = 0.011), ‘uncomfortable–cozy’ (p = 0.004), “agitated–tranquil’ (p = 0.010), ‘tense–relaxed’ (p =0.011), ‘hostile–friendly’ (p = 0.001), ‘dangerous–safe’ (p = 0.022), ‘unpleasant–pleasant’ (p = 0.044) and ‘ugly–beautiful’ (p = 0.000). Mean values of these scales for males and females showed that males were more sensitive to speed (‘slow–fast’) while females were more inclined to dynamic light and felt comfortable, relaxed and safe. This result is in line with the findings of Wang et al.⁹

3.4. Academic background differences in atmosphere scale perception

The same partial correlation analysis method was applied to investigate the perception difference related to academic background. Across all physical parameters, significant differences were found in ‘dim–bright’ (p = 0.01), ‘boring–interesting’ (p = 0.012), ‘depressing–inspiring’ (p = 0.023), ‘lifeless–lively’ (p = 0.031), ‘detached–attached’ (p = 0.000), ‘personal–open’ (p = 0.001), ‘plain–luxurious’ (p = 0.022), ‘tense–relaxed’ (p = 0.002) and ‘agitated–tranquil’ (p = 0.021). Mean values of these scales show that, people from a design background tended to feel the light conditions as brighter and more inspiring, while people without a design related background were more inclined to feel relaxed and tranquil.

3.5. Effects of the physical lighting parameters

The effects of physical lighting parameters on the perception were revealed through Spearman rho correlation analysis. Scales significantly affected by the parameters are listed in Table 5. Mean values of these scales against lighting parameters are listed in Figure 3. It further revealed their relationships with physical lighting parameters:

Luminance had a significant positive impact on the perception of most scales. Observers were more inclined to dynamic lighting with higher luminance. As luminance increases, dynamic lighting would appear brighter, warmer, but slower. Meanwhile, the atmosphere became cozier, safer, more formal and arousing.

Chroma significantly affected ‘dim–bright’ perception. As chroma increased, observers felt dimmer. When chroma increased, observers felt less comfortable, arousing, spacious, romantic and luxurious.

Hue significantly affected ‘cool–warm’ perception. An increase in warm-hue light resulted in a more romantic and lively feeling.

Speed positively affected the perception of fast, romantic, lively, attached and luxurious; however, negatively affected the perception of formal, spacious, comfortable, safe and urban.

Preference (‘ugly–beautiful’) was only affected by luminance. As luminance increases, observers evaluated the scene to be more beautiful. The ‘natural–artificial’ perception was affected by luminance, chroma and speed. As chroma decreased, speed decreased, or luminance increased, the atmosphere was felt to be more natural.

Table 5

Scales significantly affected by physical parameters (p value)

Scales	L*	C_ab*	S	HR
Scales	40, 70	18, 36	0.75, 1.5, 3	−1, 0, 1
Cool–warm	<0.01			<0.01
Dim–bright	<0.01	0.001	<0.01
Unsmooth–smooth	<0.01		<0.01
Slow–fast	<0.01		<0.01
Informal–formal	<0.01	<0.01	<0.01	0.014
Crowded–spacious	<0.01	<0.01	0.002	0.009
Uncomfortable–cozy	<0.01	<0.01	<0.01
Boring–interesting	<0.01	0.032	<0.01	0.002
Drowsy–alert	<0.01	0.002		<0.01
Agitated–tranquil	<0.01	0.003	<0.01
depressing–inspiring	<0.01
Tense–relaxed	<0.01	0.01	<0.01
Lifeless–lively	<0.01		<0.01	0.001
Unromantic–romantic	<0.01	0.001	<0.01	<0.01
Hostile–friendly	<0.01	<0.01	0.004
detached–attached	<0.01		<0.01	<0.01
Mysterious–familiar	<0.01	<0.01	<0.01
Personal–open	<0.01	<0.01	0.001	0.002
Plain–luxurious	<0.01	<0.01	<0.01	<0.01
Distant–intimate	<0.01			<0.01
Unpleasant–pleasant	<0.01	<0.01
Dangerous–safe	<0.01	<0.01	<0.01
Terrifying–delightful	<0.01	<0.01	<0.01
Natural–artificial	<0.01	0.019	<0.01
Ugly–beautiful	<0.01

Figure 3

Mean values of scales plotted against (a) lightness, (b) chroma, (c) hue range and (d) speed. Error bar represents 95% confidence interval

4. Atmosphere dimensions

4.1. Factor analysis

Factor analysis was used to uncover the underlying dimensions of the atmosphere perceptions of dynamic lighting. Before factor analysis, results from the ‘unsmooth–smooth’ and ‘natural–artificial’ scales were excluded because they were not included for atmosphere description. Finally, 23 scales with 46 terms were used. Principal component analysis and varimax orthogonal rotation were used to extract the dimensions. Four main factors were extracted, accounting for 62.4% of the variance. Table 6 lists factor loadings for all the 23 scales.

Table 6

Factor loadings of atmosphere dimensions

Account for 62.40%	Coziness 33.72%	Spaciousness 15.60%	Liveliness 8.37%	Warmth 4.72%
Scales	Coziness 33.72%	Spaciousness 15.60%	Liveliness 8.37%	Warmth 4.72%
Uncomfortable–cozy	0.752	0.288	−0.042	0.045
Tense–relaxed	0.746	0.156	−0.226	0.081
Hostile–friendly	0.740	0.249	−0.023	0.217
Terrifying–delightful	0.739	0.346	−0.108	0.161
Unpleasant–pleasant	0.730	0.264	0.131	0.129
Dangerous–safe	0.727	0.332	−0.093	0.099
Ugly–beautiful	0.706	0.108	0.267	0.036
Agitated–tranquil	0.693	0.208	−0.355	−0.085
Dim–bright	0.303	0.792	0.031	0.067
Personal–open	0.186	0.790	−0.135	0.067
Crowded–spacious	0.329	0.749	−0.043	0.021
Drowsy–alert	0.211	0.744	0.166	−0.017
Informal–formal	0.220	0.698	−0.291	0.025
Depressing–inspiring	0.468	0.502	0.280	0.178
Mysterious–familiar	0.385	0.423	−0.237	0.291
Boring–interesting	0.250	−0.013	0.754	0.024
Plain–luxurious	−0.151	−0.176	0.668	0.115
Slow–fast	−0.320	0.062	0.641	−0.029
Unromantic–romantic	0.230	−0.276	0.639	0.127
Lifeless–lively	−0.188	0.176	0.607	0.349
Detached–attached	−0.082	0.116	0.568	0.537
Cool–warm	0.145	0.056	0.163	0.808
Distant–intimate	0.407	0.014	0.161	0.684

Through reviewing the meaning of scales contributing to each factor, and comparing factors’ components with previous studies, factor names can be suggested. Factor 1, which highly loaded with ‘uncomfortable–cozy’ and ‘tense–relaxed,’ was denoted as coziness. Factor 2 was named as spaciousness, because of the high factor loadings of ‘crowded–spacious,’ ‘personal–open,’ ‘drowsy–alert,’ ‘informal–formal’ and ‘depressing–inspiring’; ‘dim–bright’ is expected to have positive relationship between luminance and spaciousness. Factor 3, which consists of ‘boring–interesting,’ ‘lifeless–lively,’ was named as liveliness; ‘slow–fast’ is expected to have high correlation between dynamic lighting and liveliness. And Factor 4, including ‘warm’ and ‘intimate,’ was named as warmth; ‘cool–warm’ is expected to have a strong relationship between hue range and warmth. Warmth accounts for a small part of the total variance, so it has been omitted. Figure 4 shows the plots of scales between combinations of the three factors.

Figure 4

Scales plotted within the atmosphere factors

Cronbach’s alpha values for coziness (mean = 0.88, min = 0.75, max = 0.94), spaciousness (mean = 0.75, min = 0.55, max =0.88), liveliness (mean = 0.71, min = 0.62, max = 0.87) and warmth (mean = 0.63, min = 0.58, max = 0.85) indicate acceptable internal consistency for each dimension.

4.2. Gender and academic background

ANOVA analysis was used to test gender difference and academic background difference on atmosphere perception. As a result, no perception difference except for the coziness(p < 0.01) factor was found between males and females. As shown in Figure 5(a), females were more likely to feel coziness. For academic background, significant differences existed in the spaciousness and liveliness factors (both p < 0.01). As shown in Figure 5(b), observers from a design background were more likely to feel spaciousness and liveliness.

Figure 5

(a) Gender effects on coziness, (b) academic background effects on spaciousness and liveliness

4.3. Relationship between lighting appearance and atmosphere perception dimensions

Effects of the ‘cool–warm,’ ‘dim–bright’ and ‘slow–fast’ scales on the lighting appearance dimensions were evaluated. Figure 6 shows mean values of Factor 1 to Factor 4 plotted against the lighting appearance ratings. Figure 6(a) shows that, the brighter the environment was, the higher coziness and spaciousness atmospheres were experienced. Figure 6(b) indicates warmer lighting would create a higher coziness, warmth and liveliness atmosphere. But when it got warmest, liveliness decreased slightly. Figure 6(c) shows that fast dynamic lighting helped to create liveliness, while reducing coziness.

Figure 6

Factor scores plotted against (a) ‘dim–bright,’ (b) ‘cool–warm’ and (c) ‘slow–fast’ rating. Error bar represents 95% confidence interval

Table 7 lists the effects of light appearance perceptions on atmosphere factors’ using the Spearman’s rho correlation method. Correlation coefficients are indicated by r values, and statistically significant differences are indicated by p values. It can be seen that perception of ‘dim–bright’ significantly affected coziness and spaciousness. It also affected warmth significantly yet weakly. All the four factors were significantly affected by ‘cool–warm,’ but only warmth was strongly affected. The dynamic perception ‘slow–fast’ markedly affected the liveliness factor, and had a negative relationship on the coziness perception.

Table 7

Effects of light appearance scales on each atmosphere dimension

	Factor 1 (coziness)		Factor 2 (spaciousness)		Factor 3 (liveliness)		Factor 4 (warmth)
	r	p	r	p	r	p	r	p
Dim–bright	0.312	0.000	0.793	0.000	0.030	0.321	0.071	0.019
Cool–warm	0.138	0.000	0.077	0.011	0.174	0.000	0.790	0.000
Slow–fast	−0.338	0.000	0.056	0.064	0.592	0.000	0.024	0.437

4.4. Effects of luminance, chroma, hue range and speed

The impact of physical changes in the dynamic lighting on atmosphere dimensions was investigated. Firstly, the effect of light parameters on the perception of each atmosphere factor was analyzed using ANOVA (Table 8). Secondly, correlations between physical parameters and factors with significant effects were plotted. The atmosphere of each light setting was described by the mean value for each factor. Factor scores are linear combinations of all scales, weighted by the corresponding factor loadings. The error bars show 95% confidence intervals.

Table 8

An overview of the effects of lighting parameters on the perception of atmosphere factors

	Coziness	Spaciousness	Liveliness	Warmth
L* (40,70)	p < 0.01	p < 0.01		p < 0.01
C_ab* (18, 36)	p < 0.01	p < 0.01	p < 0.01
S (0.75, 1.5, 3)	p < 0.01		p < 0.01	p = 0.034
Hue range (−1, 0, 1)		p < 0.01	p < 0.01	p < 0.01

Table 9

A comparison of atmosphere perception dimensions

Light type	Factors and factors’ components

Static lighting	Vogels¹³	Coziness			Liveliness
	Vogels¹³	delightful stimulating exciting lively inspiring uninhibited	spacious mysterious hospitable hostile personal relaxed	warm tranquil romantic cheerful safe intimate	detached terrifying musty threatening oppressive depressed exciting formal	chilly cool business hostile lethargic boring restless uncomfortable
		pleasant (‘gezellig’ in Dutch) pleasant (‘gemoedelijk’ in Dutch) cozy (‘behaaglijk’ in Dutch) cozy (‘knus’ in Dutch)
	Erp¹⁴	Coziness		Liveliness	Tenseness	Detachment
	Erp¹⁴	safe warm romantic personal relaxed tranquil	pleasant hospitable unhabituated mysterious luxurious intimate	stimulating exciting lively inspiring lethargic cheerful musty boring accessible	threatening terrifying oppressive hostile tense uncomfortable restless depressed detached	business formal spatial cool chilly
		cozy(‘behaaglijk’ in Dutch) cozy(‘knus’ in Dutch)
	Liu et al.¹⁶	Liveliness		Coziness
	Liu et al.¹⁶	drowsy–alert tired–energetic sluggish–stimulating sag–exciting dejected–inspiring hesitating–confident personal–open inactive–active dependent– independent	feminine–masculine drowsy–alert indifferent–interested informal–formal classical–modern business-like–artistic rural–urban lifeless–lively musty–fashionable	dangerous–safe terrifying–close tense–relaxed afflictive–cozy defiant–peaceful restless–tranquil inhibited–uninhibited	isolated(1)–involved isolated(2)–inviting unpleasant–pleasant hostile–hospitable uncomfortable–delightful disconcerting–reassuring

Dynamic lighting	Wang et al.⁹	Tenseness			Coziness			Liveliness
		threatening horrible oppressive	hostile uncomfortable tense		cozy living room tranquil	hospitable relaxed	lively stimulating		inspiring pleasant

	Present study	Coziness	Spaciousness	Liveliness	Warmth
	Present study	uncomfortable–cozy tense–relaxed hostile–friendly terrifying–delightful unpleasant–pleasant dangerous–safe ugly–beautiful agitated–tranquil	dim–bright personal–open crowded–spacious drowsy–alert informal–formal depressing–inspiring mysterious–familiar	boring–interesting plain–luxurious slow–fast unromantic–romantic lifeless–lively detached–attached	cool–warm distant–intimate

Table 8 shows the effects of lighting parameters on the perception of atmosphere factors. It can be seen that perceptions of coziness, spaciousness and warmth were all positively related to luminance. Chroma positively affected liveliness, but negatively affected coziness and spaciousness perceptions. Speed of dynamic lighting also had a positive effect on liveliness and warmth, however it negatively affected coziness. As for hue range, when there is more warm hue, the lighting atmosphere is perceived to have more liveliness and warmth, but less spaciousness.

Coziness is significantly related to luminance, chroma and speed. To generate coziness, the results indicate that one should increase luminance, or reduce chroma and speed. Spaciousness is significantly affected by luminance, chroma and hue range. A higher luminance, lower chroma and cool hue range could help increase a spacious atmosphere and the sense of arousal. For liveliness, effects of chroma, speed and hue range were all significant. Warmth was significantly affected by all the parameters except for chroma. Observers were inclined to feel warmer when the scene displayed a higher luminance, and warm hue range. Also, a medium speed reduced the perception of warmth compared to that produced by slow and fast speeds.

Major conclusions can be drawn from Figure 7. (1) By increasing luminance, a cozy, spacious and warm atmosphere was generated. Together with the positive relationship between ‘dim–bright’ and coziness; the relationship between luminance and coziness disagreed with those found in the previous studies.^14,15 This will be discussed in the next section. (2) Results concerning chroma partially agreed with the finding of Wang et al.⁹ that a higher chroma is perceived as more lively, but less detached. Contrary to Wang et al.⁹ we found that a higher chroma provided a less cozy atmosphere. (3) By changing the hue range from cool to warm, the atmosphere became lively and warm, but less detached and arousing. The hue ranges barely affected the coziness perception. This is in agreement with Liu et al.’s finding¹⁶ that coziness was hardly affected by the CCT of the sources used, but different from Vogels¹³ who found that CCT had a negative correlation with the coziness. (4) By increasing dynamic speed, the atmosphere becomes more lively but less cozy. This is contrary from Wang et al.’s finding that medium dynamic speed was coziest.⁹ However, spaciousness perception, which related to the sense of arousing and formal, has no significant relation with speed. Atmospheres created by fast speed are slightly warmer than those created by slow and medium speeds.

Figure 7

The effects of physical parameters on the atmosphere factors

To further reveal how physical parameters work on the atmosphere perception, linear regressions were calculated. Equations (3) to (6) show the linear regression equations for physical parameters against atmosphere factors. HR standing for hue range, values as −1, 0 and 1 for cool range, full range and warm range, respectively. S standing for speed, values as 0.75 ΔE_ab/s, 1.5 ΔE_ab/s and 3 ΔE_ab/s respectively. L* stands for lightness and C_ab* stands for chroma. The correlation coefficients applying to equations (3) to (6) are high with the values of 0.75, 0.97, 0.72 and 0.74, respectively.

5. Discussion

5.1. Atmosphere perception

Firstly, although light conditions and questionnaires were not same across the five studies, they all share the atmosphere factors of coziness and liveliness, also having highest loading factor in the present study. And they were similarly constructed. It confirms that coziness and liveliness are general atmosphere factors across dynamic and static lighting.

Table 9 lists the atmosphere perception dimensions from Vogels,¹³ Erp,¹⁴ Wang et al.,⁹ Liu et al.¹⁶ and this study. Vogels,¹³ Erp¹⁴ and Liu et al.¹⁶ researched atmosphere of static lighting, while Wang et al.⁹ and this study were on dynamic lighting. Both Vogels¹³ and Erp’s¹⁴ experiments were conducted in Dutch with Dutch people, while the other three were in Chinese with Chinese observers. Vogels,¹³ Erp¹⁴ and Wang et al.⁹ applied single words in questionnaire, while Liu et al.¹⁶ and this study applied word pairs.

Although light conditions and questionnaires were not same across the five studies, they all share the atmosphere factors of coziness and liveliness, also having highest loading factor in the present study. They were also similarly constructed. It confirms that coziness and liveliness are general atmosphere factors across dynamic and static lighting.

Secondly, tenseness, appearing in studies from Wang et al.⁹ and Erp¹⁴ study, is absent in Liu et al.¹⁶ and the current study. This may be caused by the difference of questionnaire forms. Liu et al.¹⁶ and this study used word pairs form, while Wang et al.⁹ and Erp¹⁴ used single word form. However, most words related to tenseness in Wang et al.⁹ and Erp¹⁴ can be found in Liu et al.’s¹⁶ word-pairs. This indicates that tenseness may be contained in this study paired with coziness.

Thirdly, spaciousness in this study is consistent with space impression of spatial impression studies from Flynn et al.^11,12 It also shares terms of detachment from the study of Erp,¹⁴ such as ‘business,’ ‘formal’ and ‘spatial.’ Both are considered to be associated with ‘less emotional,’ ‘isolation.’ So, in some way, spaciousness can also be interpreted as detachment.

Finally, a warmth atmosphere, which related to ‘cool–warm’ perception, was extracted as expected. However, low variance loading implied warmth may not be a primary atmosphere.

5.2. Discussion about light parameters’ effects

As shown in Figures 6 and 7 and Table 7, coziness is positively correlated with luminance. This is contrary to findings from previous studies.^14,15 The following explanations are possible.

It is due to the difference between pair-wise and single-term scales. Coziness in this study has an opposite meaning with ‘tense,’ as used by Wang et al.,⁹ Bronckers²⁸ and Erp.¹⁴ In other words, the negative relationships between luminance and tenseness found in Wang et al.⁹ agrees with our finding.

It is due to the difference between static lighting and dynamic lighting. In the studies by Custers et al.¹⁵ and Erp¹⁴ that used static white light, a brighter lighting produced less cozy or more tense atmosphere. In Bronckers²⁸ using static coloured light, luminance hardly affected coziness. This indicated that luminance effect on coziness for dynamic coloured lighting is opposite to that for static lighting.

The discrepancy may stem from cultural difference between Dutch and Chinese. This was also mentioned by Wang et al.⁹ and Liu et al.¹⁶ To confirm this, independent investigation is required.

To confirm luminance’s effect on coziness, further experiments are required. Coziness is also negatively correlated to speed and chroma, and hardly affected by hue range. This indicates that, the closer a dynamic lighting gets to static and neutral, the cozier the lighting atmosphere will be. A similar effect can also be found with static coloured light²⁸ that atmosphere will be less cozy if saturation increases.

For liveliness perception, the positive effects of speed⁹ and warm hues¹⁴ are in line with previous studies. However, Wang et al.⁹ pointed out that chroma hardly affected liveliness perception for dynamic coloured lighting over full hue range. We found that liveliness increased with chroma. This implies that chroma’s effect on liveliness may be strengthened to some degree by hue range (cool or warm).

Based on the former discussion, spaciousness can be interpreted as detachment.^9,14 In Erp’s research,¹⁴ detachment is positively related to luminance, and negatively related to ‘cool–warm’ perception. In Flynn and Spencer’s research,¹¹ cool white light is more spacious and clear. The same effect was also found in the present spaciousness perception. The consistency partially supports the corresponding relationship between spaciousness and detachment.

Warmth is mainly impacted by hue range, which is consistent with perception results under white light with different CCT¹¹ that low CCT white light is more visually warm. In addition, luminance and speed also help to create a warm atmosphere. Low luminance and fast speed help increase warmth.

The impact of luminance on preference is in keeping with Heerwagen’s finding²⁹ that luminance affected preference. However, chroma, speed or hue range had no significant impact on preference.

Some perception differences may exist between different genders. Males were more sensitive to dynamic lighting, but females were inclined to dynamic light but felt comfortable, relaxed and safe. Hence, dynamic lighting may create different atmospheres for different genders, which is consistent with gender effects of non-verbal communication,^30,31 emotional intensity³² and facial reactions to different emotional stimuli,³³ as well as the gender differences found in research from Wang et al.,⁹ and Knez and Enmarker.⁵ Atmosphere perception differences may also exist between observers with and without a design background. However, further investigations with larger group of samples on the atmosphere perception between genders and between type of expertise are required to exclude these possibilities.

6. Conclusion

This paper reports a 30-observer experiment investigating the atmosphere perception of dynamic lighting over warm and cool hue ranges. Four atmosphere factors were extracted: coziness, spaciousness, liveliness and warmth. Comparison with previous studies showed that, coziness and liveliness are common atmosphere factors across different light conditions and cultures (Dutch and Chinese).

Correlation analysis revealed the effects of lighting properties on atmosphere factors. High speed and high chroma are helpful in creation of liveliness, while low speed and low chroma can increase coziness. Hue range also weighs high on the atmosphere perception. Both cool hue range lighting and high luminance can produce spaciousness, which related to the attributes of alert, spatial and formal. Application of a warm hue range helps to create warmth as expected.

Coziness is positively correlated with luminance. This is contrary to findings from some previous studies. Dynamic lighting and culture differences (Dutch and Chinese) are possible reasons to explain the discrepancy. However, further experiments are required. Besides, the atmosphere perception between genders and expertise need to be clarified.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author Bing Li was supported by National Key Technology Research and Development Program of the Ministry of Science and Technology of China (Grant No. 2015BAF14B00), and Key Program of National Natural Science Foundation of China (Grant No. 61332017).

References

Boyce

. Human Factors in Lighting, 3rd. Boca Raton, FL: CRC Press, 2014.

Küller

. Physiological and psychological effects of illumination and colour in the interior environment. Journal of Light and Visual Environment 1986; 10: 33–37.

Knez

. Effects of indoor lighting on mood and cognition. Journal of Environmental Psychology 1995; 15: 39–51.

Flynn JE. Lighting-design decisions as interventions in human visual space. In: J. Nasar (ed.) Environmental Aesthetics: Theory, Research and Applications. New York: Cambridge University Press, 1988; pp. 156–170.

Knez

Enmarker

. Effects of office lighting on mood and cognitive performance and a gender effect in work-related judgment. Environment and Behavior 1998; 30: 553–567.

McCloughan

CLB

Aspinall

Webb

. The impact of lighting on mood. Lighting Research and Technology 1999; 31: 81–88.

Mehrabian

Russell

. The basic emotional impact of environments. Perceptual and Motor Skills 1974; 38: 283–301.

Watson

Clark

Tellegen

. Development and validation of brief measures of positive and negative affect: The Panas Scales. Journal of Personality and Social Psychology 1988; 54: 1063–1070.

Wang

Luo

Liu

Yang

Zheng

Liu

. A study of atmosphere perception of dynamic coloured light. Lighting Research and Technology 2014; 46: 661–675.

10.

Flynn

Hendrick

Spencer

Martyniuk

. A guide to methodology procedures for measuring subjective impressions in lighting. Journal of the Illuminating Engineering Society 1979; 8: 95–110.

11.

Flynn

Spencer

. The effects of light source color on user impression and satisfaction. Journal of the Illuminating Engineering Society 1977; 6: 167–179.

12.

Flynn

Spencer

Martyniuk

Hendrick

. Interim study of procedures for investigating the effect of light on impression and behavior. Journal of the Illuminating Engineering Society 1973; 3: 87–94.

13.

Vogels

Atmosphere metrics. In: Westerink

JDM

Ouwerkerk

Overbeek

Pasveer

Ruyter

(eds). Probing Experience, Dordrecht, Netherlands: Springer, 2008, pp. 25–41.

14.

Erp Tv. The effects of lighting characteristics on atmosphere perception. Eindhoven University of Technology, Eindhoven, The Netherlands, 2008.

15.

Custers

Kort

Ijsselsteijn

Kruiff

. Lighting in retail environments: Atmosphere perception in the real world. Lighting Research and Technology 2010; 42: 331–343.

16.

Liu

Luo

. A study of atmosphere perceptions in a living room. Lighting Research and Technology 2015; 47: 581–594.

17.

Kuijsters A, Redi J, Ruyter Bd, Seuntiëns P, Heynderickx I. Effects of ageing on atmosphere perception: Proceedings of Experiencing Light. Eindhoven, Netherlands, 2012.

18.

Sekulovski D, Seuntiens P, Hartog M. Measuring dynamic lighting atmospheres: Proceedings of the AIC 2011 Midterm Meeting, Zurich, 2011: pp. 147–150.

19.

Wan SH, Ham J, Lakens D, Weda J, Cuppen R. The influence of lighting color and dynamics on atmosphere perception and relaxation: Proceedings of Experiencing Light 2012, Eindhoven, Netherlands, 2012.

20.

Commission Internationale de I’Eclairage. Colorimetry, CIE Publication CIE 015:2004, Vienna: CIE, 2004.

21.

Post

Calhoun

. An evaluation of methods for producing desired colors on CRT monitors. Color Research and Application 1989; 14: 172–186.

22.

Sekulovski D, Vogels IM, Beurden Mv, Clout R. Smoothness and flicker perception of temporal color transitions: Color and Imaging Conference. Society for Imaging Science and Technology, 2007: pp. 113–117.

23.

Zhang Z, Zhu Y. Research on users' and designers' product color perception: Proceedings of the 2009 Second International Symposium on Computational Intelligence and Design - Volume 01. IEEE Computer Society, 2009: pp. 264–267.

24.

Hsu

Chuang

Chang

. A semantic differential study of designers’ and users’ product form perception. International Journal of Industrial Ergonomics 2000; 25: 375–391.

25.

Cohen

. Things I Have Learned (So Far). Methodological Issues and Strategies in Clinical Research, Washington, DC: American Psychological Association, 1992. : pp. 315–333.

26.

Fairchild

Reniff

. Time course of chromatic adaptation for color-appearance judgments. Journal of the Optical Society of America A 1995; 12: 824–833.

27.

Nunnally JC, Bernstein IH. Assessment of reliability. In: Psychometric Theory, 2nd Edition. New York: McGraw-Hill, 1978: pp. 245–246.

28.

Bronckers

. The effects of coloured light on atmosphere perception. Philips Research, Eindhoven, Netherlands: Eindhoven University of Technology, 2009.

29.

Heerwagen

. Affective functioning, “Light hunger,” and room brightness preferences. Environment and Behavior 1990; 22: 608–635.

30.

Hall

. Nonverbal Sex Differences: Communication Accuracy and Expressive Style, Baltimore, MD: Johns Hopkins University Press, 1984.

31.

Fujita

Diener

Sandvik

. Gender differences in negative affect and well-being: The case for emotional intensity. Journal of Personality and Social Psychology 1991; 61: 427–434.

32.

Buck

. The Communication of Emotion, New York: Guildford Press, 1984.

33.

Dimberg

Lundquist

L-O

. Gender differences in facial reactions to facial expressions. Biological Psychology 1990; 30: 151–159.