5D Library › Architectural Acoustics, Reverberation Time and the Environmental Psychology of Built Space · 中文

Chapter 5 Illuminance, Spectrum and Ceiling Height

Relation to Book 14

The author’s framework said that this chapter “echoes Book 2”. Checking the 5D Library catalogue calls for a correction: Book 2 is Turning the Hearing Inward, a bioacoustic study of listening; it is Book 14, Dynamical Systems of Neurotransmitters and Endocrinology, whose fourth chapter deals with light, intrinsically photosensitive retinal ganglion cells (ipRGCs), melanopsin and circadian rhythms. That chapter has already covered the discovery of ipRGCs, the human phase-response curve to light, suppression of melatonin by light at night, the camping experiment and a simplified circadian phase model. This chapter only cross-refers to it, repeating neither the material nor the program, and asks an architectural question instead: by what quantity should the light in a room be measured?

From “colour temperature” to “melanopic equivalent daylight illuminance”

The familiar lighting quantities are illuminance (lux) and correlated colour temperature (CCT, kelvin). Illuminance is weighted by the photopic spectral sensitivity of the eye, V(λ)V(\lambda); colour temperature describes whether light looks “warm” or “cool”. But ipRGCs are most sensitive near blue-green light (around 490 nm), unlike V(λ)V(\lambda), so two sources can have the same illuminance and colour temperature and still stimulate ipRGCs differently.

Evidence

The 2018 CIE standard. CIE S 026/E:2018 defines effective quantities for each of five photoreceptor types (three cones, rods and ipRGCs); for ipRGCs this is the melanopic equivalent daylight illuminance (melanopic EDI, in lux): the illuminance of standard daylight (D65) that would give the same melanopic stimulation as the light in question (International Commission on Illumination 2018). A source’s melanopic daylight efficacy ratio (melanopic DER) is its melanopic EDI divided by its illuminance, with D65 equal to 1 by definition.

Evidence

The 2022 expert consensus. In 2022 Brown and 17 colleagues proposed recommendations for indoor light exposure in healthy adults, all expressed as melanopic EDI at the eye (measured in the vertical plane at about 1.2 m): at least 250 lux during the day, using daylight first where available; no more than 10 lux from at least 3 hours before bedtime; no more than 1 lux in the sleep environment, which should be as dark as possible; and no more than 10 lux for unavoidable night-time activities. Daytime light should ideally be enriched, and evening light depleted, in short wavelengths (Brown et al. 2022).

Another widely used measure is the circadian stimulus (CS) of the Lighting Research Center at Rensselaer Polytechnic Institute, based on a model of melatonin suppression at night, with a design criterion of CS 0.3 at the eye during the day; in 2019 Figueiro and colleagues used desk or ceiling luminaires to deliver CS ≥ 0.3 at the eye in four office settings and found lower self-reported sleepiness and higher vitality and energy on intervention days, though these were subjective measures without a randomised control (Figueiro et al. 2019). The two measures rest on different models; this book uses the internationally standardised melanopic EDI.

Calculation

# Melanopic EDI = photopic illuminance at the eye x melanopic daylight efficacy ratio
# (DER; D65 = 1 by definition). Targets from the 2022 consensus (Brown et al.): daytime
# >= 250 lx, evening (last 3 h before bed) <= 10 lx, sleep <= 1 lx, all melanopic EDI at
# the eye. DER values here are a range to look up for a real lamp, not measurements of
# any product.
import math
for der in (0.3, 0.45, 0.6, 0.8, 1.0):
    print(f"DER {der:.2f}: day needs >= {math.ceil(250 / der):4d} lx at the eye;"
          f" evening <= {10 / der:4.1f} lx; sleep <= {1 / der:.1f} lx")
print("same melanopic EDI from colour temperature or from quantity:")
print(f"  'cool' lamp, DER 0.9, 100 lx -> {0.9 * 100:.0f} lx mEDI")
print(f"  'warm' lamp, DER 0.45, 500 lx -> {0.45 * 500:.0f} lx mEDI")
SEATS = {  # illustrative overcast-day vertical daylight at the eye (lx), DER 1.0
    "1 m from window": 2000, "3 m": 600, "5 m": 220, "7 m": 90}
ELECTRIC = (200, 0.5)                     # vertical lx from ceiling lights, their DER
for seat, day in SEATS.items():
    medi = day * 1.0 + ELECTRIC[0] * ELECTRIC[1]
    print(f"{seat:16s}: daylight {day:4d} + electric {ELECTRIC[0]} lx"
          f" -> {medi:5.0f} lx mEDI"
          + ("  meets 250" if medi >= 250 else "  BELOW 250"))
# DER 0.30: day needs >=  834 lx at the eye; evening <= 33.3 lx; sleep <= 3.3 lx
# DER 0.45: day needs >=  556 lx at the eye; evening <= 22.2 lx; sleep <= 2.2 lx
# DER 0.60: day needs >=  417 lx at the eye; evening <= 16.7 lx; sleep <= 1.7 lx
# DER 0.80: day needs >=  313 lx at the eye; evening <= 12.5 lx; sleep <= 1.2 lx
# DER 1.00: day needs >=  250 lx at the eye; evening <= 10.0 lx; sleep <= 1.0 lx
# same melanopic EDI from colour temperature or from quantity:
#   'cool' lamp, DER 0.9, 100 lx -> 90 lx mEDI
#   'warm' lamp, DER 0.45, 500 lx -> 225 lx mEDI
# 1 m from window : daylight 2000 + electric 200 lx ->  2100 lx mEDI  meets 250
# 3 m             : daylight  600 + electric 200 lx ->   700 lx mEDI  meets 250
# 5 m             : daylight  220 + electric 200 lx ->   320 lx mEDI  meets 250
# 7 m             : daylight   90 + electric 200 lx ->   190 lx mEDI  BELOW 250

The first part turns the recommendations into numbers an ordinary lux meter can read: for a source with a DER of 0.45, about 556 lux at the eye is needed during the day to reach 250 lux melanopic EDI, and in the evening it should stay below about 22 lux. The second part shows that “cool” versus “warm” does not settle everything: a source with DER 0.9 at only 100 lux gives 90 lux melanopic EDI, while one with DER 0.45 at 500 lux gives 225 lux. The quantity of light matters as much as its spectrum. The third part is an illustrative office on an overcast day: a seat 1 m from the window easily exceeds 250 lux, while one 7 m from the window reaches only about 190 lux even with the ceiling lights. The DER values in the program are a range to look up or measure for a real lamp, not measurements of any product.

Evidence

Evening light and sleep. Chang and colleagues in 2015 had 12 adults live in a laboratory and read for 4 hours before bed on a light-emitting tablet on five consecutive evenings and from a printed book on five others (in random order): after reading on the tablet, participants took nearly 10 minutes longer to fall asleep, had less REM sleep, showed suppressed melatonin and a delayed circadian rhythm, and were sleepier and slower to become alert the next morning (Chang et al. 2015).

Evidence

Daytime light and alertness. Souman and colleagues in 2018 systematically reviewed 68 studies that manipulated light intensity or spectrum and measured alertness: raising the intensity of white light increased subjective alertness in most studies, though many found no significant effect; the effect of raising colour temperature was inconsistent; and few studies using reaction times found significant effects. The authors considered sample sizes generally too small for firm conclusions (Souman et al. 2018).

Note

Qualification of the author’s framework. “High daytime illuminance rich in blue light, warm low-CCT light in the evening” is consistent in direction with the 2022 consensus, but should be expressed as melanopic EDI at the eye rather than colour temperature; and the evidence that bright daytime light raises alertness consists mainly of subjective ratings, with objective measures inconsistent. The lighting figures in this book apply only to the everyday surroundings of healthy adults and are not light therapy; people with bipolar disorder, eye disease or taking photosensitising medication should consult a doctor before changing their light habits.

The cathedral effect: ceiling height and modes of thought

Evidence

Meyers-Levy and Zhu, 2007. This paper in the Journal of Consumer Research reports three experiments in rooms with 10-foot (about 3.0 m) or 8-foot (about 2.4 m) ceilings. In Experiment 1 (30 analysed), participants under the higher ceiling reported more bodily feelings of “freedom” and fewer of “confinement”, and solved freedom-related anagrams faster and confinement-related ones more slowly. Experiment 2 (100 participants) also manipulated whether the ceiling was salient (lanterns hung from the ceiling or at eye level): only when the ceiling was salient did participants under the high ceiling use more, and more abstract, dimensions in a categorisation task and rate two products that were sleek overall but had a few crude details (a coffee table and a wine rack) as more “sleek”; when it was not salient these differences vanished. Experiment 3 (34 participants) used memory tests and found more clustering by category in free recall (relational processing) under the high ceiling and better cued recall (item-specific processing) under the low one (Meyers-Levy and Zhu 2007).

These results are interesting, but the weight of evidence needs a clear look. The three experiments had samples of 30, 100 and 34, with only 22–29 people per cell in Experiment 2, and the key effects depended on whether the ceiling was “salient”. We found no direct replication of these experiments. Later research asked different questions: Vartanian and colleagues in 2015 used functional MRI and found that participants judged rooms with higher ceilings more beautiful and were more willing to “enter” them, but they measured aesthetic judgements and approach–avoidance decisions, not abstract thought (Vartanian et al. 2015). The program below shows what a small 2 × 2 design can detect.

Calculation

# How much can a 2 x 2 between-subjects study with about 25 people per cell show?
# Ceiling high/low x ceiling salient/not; assume a true simple effect d = 0.5 when
# salient and 0 when not (an interaction of size 0.5 SD). Normal approximation for
# power; simulation (fixed seed) for the average size of the effects that come out
# "significant" (the winner's curse).
import math, random
from statistics import NormalDist
P = NormalDist().cdf

def power(z):
    return P(z - 1.96) + P(-z - 1.96)

d = 0.5
print(" n per cell | power: simple effect | interaction | mean d if significant")
rng = random.Random(2007)
for n in (25, 50, 100, 125, 200):
    z_simple, z_inter = d / math.sqrt(2 / n), d / math.sqrt(4 / n)
    sig = []
    for _ in range(20000):
        est = rng.gauss(d, math.sqrt(2 / n))
        if abs(est) / math.sqrt(2 / n) > 1.96 and est > 0:
            sig.append(est)
    print(f"    {n:4d}    |        {power(z_simple):.2f}         |"
          f"    {power(z_inter):.2f}     |"
          f"    {sum(sig) / len(sig):.2f} (true 0.50)")
n80 = math.ceil((2 * (1.96 + NormalDist().inv_cdf(0.8)) / d) ** 2)
print(f"80 % power for the interaction needs about {n80} per cell ({4 * n80} people)")
#  n per cell | power: simple effect | interaction | mean d if significant
#       25    |        0.42         |    0.24     |    0.76 (true 0.50)
#       50    |        0.71         |    0.42     |    0.60 (true 0.50)
#      100    |        0.94         |    0.71     |    0.52 (true 0.50)
#      125    |        0.98         |    0.80     |    0.51 (true 0.50)
#      200    |        1.00         |    0.94     |    0.50 (true 0.50)
# 80 % power for the interaction needs about 126 per cell (504 people)

Assume a true effect of d=0.5d = 0.5 when the ceiling is salient and 0 when it is not. With 25 people per cell, the power to detect the simple effect is only 0.42, and to detect the interaction only 0.24; and among studies that “happen to be significant”, the estimated effect averages 0.76, about half as large again as the true 0.5 (the “winner’s curse”). Testing this interaction with 80 % power would need about 126 per cell, about 504 people in all.

Note

Qualification of the author’s framework. “Higher ceilings promote abstract thinking, lower ceilings concrete detail” should read: a set of small experiments observed this pattern when the ceiling was salient, and there is as yet no direct replication; it cannot be used to guarantee that any spatial design “boosts creativity”. Chapter 7 reads the Qing writer Li Yu on ceiling height from a different angle: winter cold and “shivering without being cold”.

Hypothesis · not established

The ceiling-height effect appears only when the space is noticed, and affects the choice of processing style rather than task accuracy (not established). A testable consequence: a pre-registered multi-site replication (at least 126 per cell) should reproduce the effect of ceiling height on the number of categorisation dimensions when the ceiling is salient and find none when it is not, with no difference in task accuracy between ceiling heights. If there is no effect even in the salient condition, the original finding is weakened.

Chapter summary

The effect of light on ipRGCs should be measured as melanopic EDI at the eye rather than by colour temperature; the 2022 consensus recommends at least 250 lux during the day, no more than 10 lux from 3 hours before bed and no more than 1 lux during sleep. The illustrative calculation shows that quantity matters as much as spectrum, and that seats far from a window may not reach the daytime recommendation. Evening light from screens delays sleep and circadian timing (a laboratory study of 12 people); bright daytime light has fairly consistent effects on subjective alertness and inconsistent ones on objective alertness. The ceiling-height effect of Meyers-Levy and Zhu (2007) appears only when the ceiling is salient, rests on small samples and has not been directly replicated; with 25 per cell the power to detect the interaction is only about 0.24. The next chapter discusses an invisible environmental variable: the air.

References

Brown, Timothy M., George C. Brainard, Christian Cajochen, et al. 2022. “Recommendations for Daytime, Evening, and Nighttime Indoor Light Exposure to Best Support Physiology, Sleep, and Wakefulness in Healthy Adults.” PLoS Biology 20 (3): e3001571. https://doi.org/10.1371/journal.pbio.3001571.
Chang, Anne-Marie, Daniel Aeschbach, Jeanne F. Duffy, and Charles A. Czeisler. 2015. “Evening Use of Light-Emitting eReaders Negatively Affects Sleep, Circadian Timing, and Next-Morning Alertness.” Proceedings of the National Academy of Sciences 112 (4): 1232–37. https://doi.org/10.1073/pnas.1418490112.
Figueiro, Mariana G., Michael Kalsher, Barbara C. Steverson, Judith Heerwagen, Kevin Kampschroer, and Mark S. Rea. 2019. “Circadian-Effective Light and Its Impact on Alertness in Office Workers.” Lighting Research & Technology 51 (2): 171–83. https://doi.org/10.1177/1477153517750006.
International Commission on Illumination. 2018. CIE s 026/e:2018 CIE System for Metrology of Optical Radiation for ipRGC-Influenced Responses to Light. CIE Central Bureau, Vienna. https://doi.org/10.25039/S026.2018.
Meyers-Levy, Joan, and Rui (Juliet) Zhu. 2007. “The Influence of Ceiling Height: The Effect of Priming on the Type of Processing That People Use.” Journal of Consumer Research 34 (2): 174–86. https://doi.org/10.1086/519146.
Souman, Jan L., Angelica M. Tinga, Susan F. te Pas, Raymond van Ee, and Björn N. S. Vlaskamp. 2018. “Acute Alerting Effects of Light: A Systematic Literature Review.” Behavioural Brain Research 337: 228–39. https://doi.org/10.1016/j.bbr.2017.09.016.
Vartanian, Oshin, Gorka Navarrete, Anjan Chatterjee, et al. 2015. “Architectural Design and the Brain: Effects of Ceiling Height and Perceived Enclosure on Beauty Judgments and Approach-Avoidance Decisions.” Journal of Environmental Psychology 41: 10–18. https://doi.org/10.1016/j.jenvp.2014.11.006.

Fan, Y. (2026). Architectural Acoustics, Reverberation Time and the Environmental Psychology of Built Space: From Sabine's Equation and Fractal Visual Dimensions to Cognitive Performance, Read Beside Classical Chinese Spatial Design [建筑声学、混响时间与空间环境心理学的物理模型] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23782090 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).