Can Morning Light Shift Human Melatonin Rhythms in Under 24 Hours?

A literature search in the PubMed database was performed in May 2025 using combinations of keywords such as ‘light’, ‘circadian’, ‘melatonin’, ‘phase advance’. Inclusion criteria were: a. sample of healthy adult participants, b. study includes at least one manipulation of light intensity, timing, duration, or spectral composition, c. study reports pre‐ versus post‐light intervention DLMO or reports dim‐light DLMO versus bright light DLMO and d. phase advances are measured within the same circadian cycle as the light intervention, and sleep deprivation is not involved.

In total, five papers met the inclusion criteria, which together investigated 16 experimental light intervention manipulations.

Based on changes in reporting standards for light interventions, melanopic equivalent daylight illuminance (mEDI) values were calculated using an Excel worksheet: the ‘Human Centric Lighting Toolkit’, as described previously [17]. This yields identical results to those obtained using the toolkit from the International Commission on Illumination (e.g., CIE S 026). When available, spectral data reported by the authors were used directly. If spectral data were not reported, mEDI was estimated using the closest matching built‐in spectrum provided in the toolbox, based on the described light source and correlated colour temperature (CCT).

Among the reviewed studies, only Ohashi et al. [18] reported mEDI directly. For Gabel et al. [19], spectral composition data were available, resulting in two calculated mEDI values (103.59 and 109.45), which were averaged. For Danilenko et al. [20], mEDI was estimated based on the spectral composition of a halogen lamp with a standard CCT of 3000 K, as only the lamp type was specified. For Kozaki et al. [21, 22], mEDI values were derived using the reported light source type and CCT. For dawn simulation protocols (e.g. [19, 20]), mEDI was estimated based on the maximum light intensity reached during the intervention, rather than the average intensity across the full simulation.

A consistent finding across studies was the importance of light intensity in same‐day phase advances. For instance, Ohashi et al. [18] administered 1 h of bright light at 8000 lx shortly after waking and found a statistically significant average phase advance of 11 min. This raises the question: what is the minimum intensity required to induce a measurable phase shift? Kozaki et al. [22] addressed this issue using a within‐subjects design. Participants were exposed to 3 h of morning light at various intensities, beginning 1 h after waking. Melatonin onset was measured that same evening and compared with the previous evening's DLMO. Significant phase advances of 15, 26, and 27 min were observed for intensities of 3000, 6000, and 12 000 lx, respectively. By contrast, lower intensities of 750 and 1500 lx did not result in significant shifts.

These findings suggest the existence of a threshold or saturation effect in the dose–response relationship between the intensity of morning light and circadian phase advancement. In other words, while low‐intensity light may be insufficient to shift circadian melatonin rhythms within a single cycle, light above approximately 3000 lx appears capable of eliciting a measurable phase advance. This is consistent with other dose–response models of the non‐visual effects of light in human chronobiology, which suggest that while responsiveness increases with intensity, the effect plateaus beyond a certain threshold [23, 24]. Interestingly, to our knowledge, the relationship between dose and response for resetting human circadian melatonin rhythms by light has only been studied in relation to phase delays, rather than phase advances [25, 26].

Kozaki et al. [21] extended this work by examining the impact of morning light exposure, with the original goal of testing whether it was protective against light‐induced melatonin suppression at night. While they did not report phase shift data following the light interventions, the timing of melatonin onset advanced from dim to bright light conditions, ranging from 100 to 2700 lx. Specifically, the average timing of DLMO after dim‐light exposure was 22:15, while the average timing of DLMO after 2700 lx exposure was 21:51. Although this 24‐min average difference was not statistically significant, it suggests that even moderate‐intensity morning light may promote earlier circadian timing.

Two studies [19, 20] investigated the effects of dawn simulation, a protocol involving gradually increasing light intensity over time after waking. Although the total light intensity achieved in these studies was lower than that used in bright light protocols, the exposure lasted longer and began earlier in the circadian day.

In Gabel et al. [19], the dawn simulation began 30 min after waking. Light gradually increased from 0 to 250 lx, and this level was maintained for an additional 20 min. Participants in the comparison (dim light) condition were exposed to very low light levels throughout the same period. This relatively low‐intensity and brief exposure to the dawn simulation protocol did not result in a significant phase shift, with average DLMO timings of 21:50 and 21:38 for the dim and dawn simulation conditions, respectively. It could be argued that the low light intensity and duration levels in this dawn simulation protocol were insufficient in producing a phase‐shifting response.

By contrast, Danilenko et al. [20] conducted a more extended dawn simulation, starting 30 min after wake‐up time continuing for 2 h. This simulation reached light intensities of 1000 and 2000 lx in two separate experiments. These protocols resulted in significant phase advances of 19 and 20 min, respectively. Interestingly, these intensities are lower than those producing phase shifts in fixed morning light protocols (e.g. [22]), suggesting that longer duration and earlier timing may compensate for lower intensity. This pattern is consistent with human PRC studies, where longer‐duration light pulses (e.g., 6.7 h [10]) produce larger phase shifts than short pulses (e.g., 1 h [7]). This suggests that longer light exposure with an earlier timing in the biological morning can produce phase advances even when light intensity is moderate.

In addition to timing and intensity, the spectral composition of light may influence whether a phase advance can be achieved within the same cycle. Gabel et al. [19] found that 20 min of monochromatic blue light (470 nm) at just 100 photopic lux, delivered 2 h after waking, resulted in a 30‐min significantly earlier timing of DLMO (21:20 ± 19 min) compared to a dim‐light condition (21:50 ± 23 min). This finding suggests that monochromatic short‐wavelength light, even at a low intensity and duration, can produce circadian phase advances when administered in the morning. This is consistent with the well‐established spectral sensitivity of non‐image‐forming photoreception, which peaks in the short‐wavelength (blue) range [27, 28]. Thus, blue‐enriched morning light might be especially effective for inducing a same‐day phase advance.

Three out of the five reviewed studies using fixed morning light protocols employed fluorescent lighting [18, 21, 22], which typically emits less power in the short‐wavelength range compared to modern light sources such as blue LEDs or full‐spectrum light sources [29]. In fact, the reported spectral distributions in these studies confirm relatively low irradiance in the blue range, which may have reduced the effectiveness of their light interventions despite high overall lux levels. This further reinforces the importance of considering spectral characteristics of light when designing light‐induced phase‐advance interventions.

For this reason, recent guidelines [30, 31] recommend reporting light exposure using melanopic Equivalent Daylight Illuminance (mEDI), a metric that quantifies the circadian impact of light by integrating both its spectral power distribution and overall intensity. Specifically, mEDI represents the illuminance of a standard daylight spectrum (D65) that produces the same melanopic stimulation as the test light source. This biologically relevant measure offers a more accurate indication of light's potential to influence the circadian system compared to traditional photopic lux alone.

Given the role of spectral composition and intensity in circadian phase shifting, we plotted mEDI values calculated from reported spectral data against the magnitude of phase advances observed in various studies (Figure 2). Phase advances were measured either directly (pre‐ vs. post‐light intervention DLMO) or relative to a dim‐light control condition (dim‐light DLMO vs. bright light DLMO). Overall, a Spearman correlation between mEDI and average phase advance is modest and only approaching significance, but shows a positive trend (r = 0.51, p = 0.06). This suggests that higher melanopic stimulation might generally promote greater phase advances. Interestingly, phase advances appear larger when calculated relative to a dim‐light intervention (cyan data points in Figure 2), rather than measured directly (orange data points in Figure 2). This is likely due to the intrinsic average drift of circadian rhythms during prolonged dim‐light exposure, exaggerating the magnitude of the phase advances following the light intervention.

While these studies highlight the potential for same‐day phase advances, their findings also highlight a major limitation: interindividual variability. Although not all studies reported the standard deviation of the phase shift, those which did showed high variability. For instance, a significant phase average phase advance of 11 min in one study [18] showed a standard deviation of 22 min, with around one third of participants presenting a small phase delay. The standard deviations of phase shifts measured in Kozaki et al. [22] ranged from 13 to 21 min across conditions. This suggests that the effects of morning bright‐light exposure on circadian rhythms are highly variable across individuals and will not consistently cause phase shifts across all participants.

Indeed, this high interindividual variability is common in the general chronobiology literature, with non‐image‐forming light effects having been shown to be influenced by prior light history [32], caffeine intake [33], and trait factors like lens transmittance and genetics [34]. Variability in intrinsic circadian period can affect the net measured phase shift because differing amounts of endogenous drift occur between assessments, altering the apparent magnitude of light‐induced phase advances across individuals. Specifically, individuals with longer circadian period would experience a shorter net phase advance in response to morning light.

Interindividual variability in light responses is also affected by protocol design components such as non‐personalised intervention timing. For example, baseline DLMO timing in Kozaki et al. [21, 22] had a standard deviation of over an hour and their intervention was fixed at 9 a.m. Therefore, the timing of the intervention relative to DLMO will vary greatly across participants and might correspond to different regions of each individual's PRC, producing variable (and occasionally opposite) phase‐shift responses.

Taken together, the reviewed studies suggest that light‐induced circadian phase advances can occur within the same circadian cycle, particularly when bright light (> 3000 lx) is administered shortly after wake‐up time. Lower intensities may also be effective if delivered earlier (e.g., before wake through dawn simulation), or if the light is spectrally enriched in the short‐wavelength (blue) range. However, the magnitude of phase shifts is typically modest (10–30 min), and responses are highly variable across individuals, limiting the predictability of outcomes without personalised parameters.