Learning visual preferences in indoor spaces can be inherently challenging, particularly in daylit environments. Visual preferences extend beyond visual comfort and depend on both qualitative and quantitative attributes which cannot be represented by linear and isolated scale points. For that reason, Likert-type scales alone are insufficient for learning human preferences. This article discusses the issues with Likert-type scales, introduces the benefits of relative comparisons and presents the most recent research on pairwise daylighting preference learning, including conceptual methods and algorithmic approaches. It also presents the challenges and limitations of pairwise preference learning methods related to lighting and describes practical implications in real-world applications.
Get full access to this article
View all access options for this article.
References
1.
AndersenM.Unweaving the human response in daylighting design. Building and Environment2015; 91: 101–117.
2.
ThurstoneLL.Psychophysical analysis. The American Journal of Psychology1927; 38: 368–389.
3.
ThurstoneLL.A law of comparative judgment. Psychological Review1994; 101: 266–270.
4.
AlbaumG.The Likert scale revisited. Market Research Society Journal1997; 39: 1–21.
5.
GreenleafEA.Measuring extreme response style. Public Opinion Quarterly1992; 56: 328–351.
6.
KentMG.Correspondence: What is discomfort from glare and how should we measure it?Lighting Research and Technology2025; 57: 665–667.
7.
JamiesonS.Likert scales: how to (ab)use them?Medical Education2004; 38: 1217–1218.
8.
SchweikerMFuchsXBeckerS, et al. Challenging the assumptions for thermal sensation scales. Building Research and Information2017; 45: 572–589.
9.
FuchsXBeckerSSchakib-EkbatanK, et al. Subgroups holding different conceptions of scales rate room temperatures differently. Building and Environment2018; 128: 236–247.
10.
SidneyS.Nonparametric statistics for the behavioral sciences. The Journal of Nervous and Mental Disease1957; 125: 497.
KingsleyDCBrownTC.Preference uncertainty, preference learning, and paired comparison experiments. Land Economics2010; 86: 530–544.
13.
LockheadGR.Absolute judgments are relative: a reinterpretation of some psychophysical ideas. Review of General Psychology2004; 8: 265–272.
14.
ConitzerV.Eliciting single-peaked preferences using comparison queries. Proceedings of the 6th International Joint Conference on Autonomous Agents and Multiagent Systems, Honolulu, HI, USA, 14–18 May2007: 1–8.
15.
SungYTWuJS.The visual analogue scale for rating, ranking and paired-comparison (VAS-RRP): a new technique for psychological measurement. Behavior Research Methods2018; 50: 1694–1715.
16.
KentMGFotiosSAltomonteS.Discomfort glare evaluation: the influence of anchor bias in luminance adjustments. Lighting Research and Technology2019; 51: 131–146.
17.
FotiosS.Using category rating to evaluate the lit environment: is a meaningful opinion captured?Leukos2019; 15: 127–142.
18.
HoeijmakersEJMartensBHendriksBM, et al. How subjective CT image quality assessment becomes surprisingly reliable: pairwise comparisons instead of Likert scale. European Radiology2024; 34: 4494–4503.
19.
GaculaMCJrSinghJ.Statistical Methods in Food and Consumer Research, 1st edition. London, UK: Academic Press, 2013.
20.
GaculaMCJr.Design and Analysis of Sensory Optimization, 1st edition. Trumbull, CT: John Wiley & Sons, 2008.
21.
O’MahonyMIMasuokaSIshiiR.A theoretical note on difference tests: models, paradoxes and cognitive strategies. Journal of Sensory Studies1994; 9: 247–272.
22.
CourcouxPSemenouM.Preference data analysis using a paired comparison model. Food Quality and Preference1997; 8: 353–358.
23.
DaneelsRVan de PerreLMeuretY, et al. Reading comfort in relation to monitor and ambient luminance levels: a study using continuous rating scale and paired comparison methodology. Lighting Research and Technology2024; 56: 866–885.
24.
BradleyRA.Science, statistics, and paired comparisons. Biometrics1976; 32: 213–239.
25.
Rajae-JoordensREngelJ.Paired comparisons in visual perception studies using small sample sizes. Displays2005; 26: 1–7.
26.
ZhuSXieXYeZ, et al. Comparing point-wise and pair-wise relevance judgment with brain signals. Journal of the Association for Information Science and Technology2024; 75: 957–971.
27.
ClarkAPHowardKLWoodsAT, et al. Why rate when you could compare? Using the ‘EloChoice’ package to assess pairwise comparisons of perceived physical strength. PLoS One2018; 13: 0190393.
28.
MantiukRKTomaszewskaAMantiukR.Comparison of four subjective methods for image quality assessment. Computer Graphics Forum2012; 31: 2478–2491.
29.
KendallMG.Further contributions to the theory of paired comparisons. Biometrics1955; 11: 43–62.
30.
KendallMGSmithBB.On the method of paired comparisons. Biometrika1940; 31: 324–345.
31.
KahnemanD.Prospect theory: an analysis of decisions under risk. Econometrica1979; 47: 278.
32.
JoachimsTGrankaLPanB, et al. Accurately interpreting clickthrough data as implicit feedback. ACM SIGIR Forum2017; 51: 4–11.
33.
FangerPO.Thermal Comfort: Analysis and Applications in Environmental Engineering, 1st edition. Copenhagen, Denmark: Danish Technical Press, 1970.
34.
AlfanoFROlesenBWPalellaBI.Povl Ole Fanger’s impact ten years later. Energy and Buildings2017; 152: 243–249.
35.
WienoldJChristoffersenJ.Evaluation methods and development of a new glare prediction model for daylight environments with the use of CCD cameras. Energy and buildings2006; 38: 743–757.
36.
OsterhausWKBaileyIL.Large area glare sources and their effect on visual discomfort and visual performance at computer workstations. Proceedings of the 1992 IEEE industry Applications Society Annual Meeting, Houston, TX, USA, 4–9 October1992: 1825–1829.
37.
QuekGJainSKarmannC, et al. Comparison of questionnaire items for discomfort glare studies in daylit spaces. Lighting Research and Technology2023; 55: 730–758.
38.
MaZTzempelikosA.Thermal sensation vs preference: quantifying the consistency and informativity between questions and scales in thermal comfort assessment. Building and Environment2026; 291: 114170.
39.
WienoldJIwataTSarey KhanieM, et al. Cross-validation and robustness of daylight glare metrics. Lighting Research and Technology2019; 51: 983–1013.
40.
RicciFRokachLShapiraB.Recommender Systems: Techniques, Applications, and Challenges, 2nd edition.New York, NY: Springer, 2021.
41.
OpenAI. ChatGPT [large language model]. Retrieved 25 February 2026 from https://chat.openai.com/
42.
KarmannCChinazzoGSchülerA, et al. User assessment of fabric shading devices with a low openness factor. Building and Environment2023; 228: 109707.
43.
VeitchJANewshamGR.Preferred luminous conditions in open-plan offices: research and practice recommendations. International Journal of Lighting Research and Technology2000; 32: 199–212.
44.
de BakkerCAartsMKortH, et al. Preferred luminance distributions in open-plan offices in relation to time-of-day and subjective alertness. Leukos2021; 17: 3–20.
45.
Van Den WymelenbergKInaniciMJohnsonP. The effect of luminance distribution patterns on occupant preference in a daylit office environment. Leukos2010; 7: 103–122.
46.
NicolFWilsonMChiancarellaC.Using field measurements of desktop illuminance in European offices to investigate its dependence on outdoor conditions and its effect on occupant satisfaction, and the use of lights and blinds. Energy and Buildings2006; 38: 802–813.
47.
SadeghiSALeeSKaravaP, et al. Bayesian classification and inference of occupant visual preferences in daylit perimeter private offices. Energy and Buildings2018; 166: 505–524.
48.
XiongJTzempelikosABilionisI, et al. Inferring personalized visual satisfaction profiles in daylit offices from comparative preferences using a Bayesian approach. Building and Environment2018; 138: 74–88.
49.
FotiosSLogadóttirÁChealC, et al. Using adjustment to define preferred illuminances: do the results have any value?Light and Engineering2012; 20: 46–55.
50.
BradleyRATerryME.Rank analysis of incomplete block designs: I. The method of paired comparisons. Biometrika1952; 39: 324–345.
51.
TuaycharoenNTregenzaPR.View and discomfort glare from windows. Lighting Research and Technology2007; 39: 185–200.
52.
BoycePRVeitchJANewshamGR, et al. Lighting quality and office work: two field simulation experiments. Lighting Research and Technology2006; 38: 191–223.
53.
da SilvaPCLealVAndersenM.Occupants interaction with electric lighting and shading systems in real single-occupied offices: results from a monitoring campaign. Building and Environment2013; 64: 152–168.
54.
De KorteEMSpiekmanMHoes-van OeffelenL, et al. Personal environmental control: Effects of pre-set conditions for heating and lighting on personal settings, task performance and comfort experience. Building and Environment2015; 86: 166–176.
55.
MeerbeekBte KulveMGrittiT, et al. Building automation and perceived control: a field study on motorized exterior blinds in Dutch offices. Building and Environment2014; 79: 66–77.
56.
BakkerLGHoes-van OeffelenEC, et al. User satisfaction and interaction with automated dynamic facades: a pilot study. Building and Environment2014; 78: 44–52.
57.
InaniciMN.Evaluation of high dynamic range photography as a luminance data acquisition system. Lighting Research and Technology2006; 38: 123–134.
58.
MahDTzempelikosA.Inferring personal daylighting preferences using HDRI and deep learning techniques. Building and Environment2024; 266: 112128.
59.
SafranekSDavisRG.Sources of error in HDRI for luminance measurement: a review of the literature. Leukos2021; 17: 187–208.
60.
LindelöfDMorelN.Bayesian estimation of visual discomfort. Building Research and Information2008; 36: 83–96.
61.
XiongJAwalgaonkarNMTzempelikosA, et al. Efficient learning of personalized visual preferences in daylit offices: an online elicitation framework. Building and Environment2020; 181: 107013.
62.
LuSMahDTzempelikosA.Toward AI-assisted, human-centered daylighting operation: non-invasive daylighting preference evaluation using deep learning. Building and Environment2026; 287A: 113761.
63.
FotiosSAHouserKW.Using forced choice discrimination to measure the perceptual response to light of different characteristics. Leukos2013; 9: 245–259.
64.
LeeSKaravaPTzempelikosA, et al. A smart and less intrusive feedback request algorithm towards human-centered HVAC operation. Building and Environment2020; 184: 107190.
65.
YePDoermannD.Active sampling for subjective image quality assessment. Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Columbus, OH, USA, 23–28 June2014: 4249–4256.
66.
Perez-OrtizMMikhailiukAZermanE, et al. From pairwise comparisons and rating to a unified quality scale. IEEE Transactions on Image Processing2020; 29: 1139–1151.
67.
LuSMahDTzempelikosA.Non-invasive assessment of the visual environment using Conditional Generative Adversarial Networks. Building and Environment2026; 287A: 113798.
68.
WardGJ.The RADIANCE lighting simulation and rendering system. Proceedings of SIGGRAPH ‘94: 21st Annual Conference on Computer Graphics and Interactive Techniques, New York, NY, USA, 24–27 July1994: 459–472.
69.
IsolaPZhuJYZhouT, et al. Image-to-image translation with conditional adversarial networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA, 21–26 July2017: 1125–1134.
70.
MahDTzempelikosA.A Convolutional Autoencoder-based method for learning and ranking personal daylighting preference. Building and Environment2025; 285C: 113595.
71.
KoWHKentMGSchiavonS, et al. A window view quality assessment framework. Leukos2022; 18: 268–293.
72.
KonstantzosIChanYCSeiboldJC, et al. View clarity index: a new metric to evaluate clarity of view through window shades. Building and Environment2015; 90: 206–214.
73.
Abd-AlhamidFKentMWuY.Quantifying window view quality: a review on view perception assessment and representation methods. Building and Environment2023; 227: 109742.
74.
LinTYLeAVChanYC.Evaluation of window view preference using quantitative and qualitative factors of window view content. Building and Environment2022; 213: 108886.
75.
Te KulveMSchellenLSchlangenLJ, et al. The influence of light on thermal responses. Acta Physiologica2016; 216: 163-85.
76.
Te KulveMSchlangenLvan Marken LichtenbeltW. Interactions between the perception of light and temperature. Indoor Air2018; 28: 881–891.
77.
BelliaLAlfanoFRFragliassoF, et al. On the interaction between lighting and thermal comfort: an integrated approach to IEQ. Energy and Buildings2021; 231: 110570.
78.
WangZFuYGaoW.The crossed and interaction effects of indoor light and thermal factors on human perceptions. Developments in the Built Environment2024; 17: 100339.
79.
ChinazzoGWienoldJAndersenM.Daylight affects human thermal perception. Scientific Reports2019; 9: 13690.
80.
ChinazzoGWienoldJAndersenMJ.Influence of indoor temperature and daylight illuminance on visual perception. Lighting Research and Technology2020; 52: 350–370.
81.
Van Den WymelenbergKG.Visual comfort, discomfort glare, and occupant fenestration control: developing a research agenda. Leukos2014; 10: 207–221.
82.
JungBInaniciM.Measuring circadian lighting through high dynamic range photography. Lighting Research and Technology2019; 51: 742–763.
