Research in human preference lighting
Independent studies conducted in North America and Asia draw similar conclusions as to what quality of light appeals most to human beings.
The Pacific Northwest National Laboratory (PNNL) published the first study in 2015 at about the same time IES released the TM-30 technical memoranum (Get more information on the research by reading, “Human perceptions of colour rendition vary with average fidelity, average gamut, and gamut shape” by M. Royer, A. Wilkerson, M. Wei, K. Houser, and R. Davis, which was published in 2016 in Lighting Research & Technology.). Observers were exposed to a space filled with recognizable branded objects, such as soda cans, fruits and other organic items, and clothing. A mirror was also present in the space so participants could observe their skin tone under different light conditions. The spectral power distribution of the light source was changed throughout the day.
The test was done in a manner similar to an eye examination, where participants were asked: “Do you like this, or do you like that?” as light sources were modified throughout the test. They were exposed to high fidelities such as 100 CRI, and then it went down to 80, 70, and even 50 CRI, all at 3227 C (3500 K). The sources were also over- and under-saturated.
When the research team ran the results against statistical models, interesting trends emerged. They discovered the fidelity of the light source, contrary to common belief, had little correlation to preference. On the other hand, saturation, especially in the red spectrum, made a light source preferable. They defined preference with the following simple equation:
Preference = Fidelity + Red Saturation
The research team also recommended a fidelity greater than 74, and a red oversaturation between two and 16 per cent, with an oversaturated gamut index.
Second came a study from Penn State University (PSU) employing a different methodology, but concluding with similar findings (Read “Models of colour quality over a wide range of spectral power distributions” by T. Esposito and K. Houser, published in Lighting Research & Technology in 2018.). Researchers conducted an absolute test as they outfitted a viewing area with common, colourful branded objects and organic items and showed participants one light source per day. This study was done at 3227 C (3500 K), and observers made judgments on many aspects of each light source, including preference.
Several spectrums—some with high fidelity and good gamut, others oversaturated, and the rest undersaturated—were tested. Although the format and methodology was different, preferred light sources fell in the same range as
the PNNL study. Once again, the results did not show a strong correlation to fidelity but rather a robust connection to saturation.
Two more studies have since been published on the topic. PNNL did a follow-up study employing the same methodology but showing light sources in five different chromaticity groups, between 2427 C (2700 K) and 4027 C (4300 K), and found similar preference results consistent between the various colour temperatures (Consult “Human Perceptions of Color Rendition at Different Chromaticities” by M. Royer, A. Wilkerson, and M. Wei, published in 2017 in Lighting Research & Technology.).
The last study was conducted by researchers at Zhejiang University in China (For more information, read “Toward a unified model for predicting colour quality of light sources” by F. Zhang, H. Xu, and H. Feng, published in 2017 in Applied Optics.). They showed a total of 164 lighting scenes in four chromaticity groups, from 2527 C (2800 K) to 6227 C (6500 K), and drew similar conclusions relative to preference. As the first study to take place outside of the United States, it also demonstrated preference is not conditioned by culture or the habitual built environment and transcends borders.
Defining a model for colour preference
By, using the parameters of TM-30 and the conclusions of the aforementioned studies, simulation spaces were set up in a private office in Chicago for several months in 2017 and 2018. While not scientific, this study allowed observers to spend extended periods of time under several preference models in their normal working environment. It also made it possible to collect feedback from visitors about three light sources, all falling within the spectrum defined by the PNNL, PSU, and Zhejiang studies.
The three preference models tested modulated the red saturation, hue bins one and 16, between –1 and 14 per cent, while maintaining an oversaturated gamut index (Rg) of 103 to 110. Conversely, the colour fidelity index (Rf) went down from 92 to 83 because Rg and red saturation increased as the light source was deviating more and more from the blackbody radiator.
The middle target, with the following parameters, was ultimately preferred:
- Rf = 91;
- Rg = 107;
- Hue bin one = four per cent; and
- Hue bin 16 = seven per cent.
While it has a fidelity index of 91, very close to CRI 90, the latter would fall short of a preferred light source. It is important to note a preferred light source needs to not only have fidelity to the reference illuminant, but also be oversaturated (Rg 107), specifically in the red content—hue bins one and 16. This is why TM-30 is superior to CRI because it takes more parameters into consideration to more precisely describe a light source. The additional parameters provided by TM-30 allows one to specify how this fidelity index is actually achieved: With an oversaturation of red of four and seven per cent, respectively in hue bins one and 16, and resulting in an overall gamut index of 107, indicating an oversaturation of the light source by seven per cent.
This preferred spectrum gives skin tones a more natural look by adding a little red content. It also enhances organic elements and wood tones, giving them a warmer colouration, and it makes brightly coloured items, whether a fruit, vegetable, or manmade, appear more vibrant.