SummaryLight triggers numerous non‐image‐forming, or non‐visual, biological effects. The brain correlates of these non‐image‐forming effects have been investigated, notably using magnetic resonance imaging and short light exposures varying in irradiance and spectral quality. However, it is not clear whether non‐image‐forming responses estimation may be biased by having light in sequential blocks, for example, through a potential carryover effect of one light onto the next. We reasoned that pupil light reflex was an easy readout of one of the non‐image‐forming effects of light that could be used to address this issue. We characterised the sustained pupil light reflex in 13–16 healthy young individuals under short light exposures during three distinct cognitive processes (executive, emotional and attentional). Light conditions pseudo‐randomly alternated between monochromatic orange light (0.16 melanopic equivalent daylight illuminance lux) and polychromatic blue‐enriched white light of three different levels (37, 92, 190 melanopic equivalent daylight illuminance lux). As expected, higher melanopic irradiance was associated with larger sustained pupil light reflex in each cognitive domain. This result was stable over the light sequence under higher melanopic irradiance levels compared with lower ones. Exploratory frequency‐domain analyses further revealed that sustained pupil light reflex was more variable under lower melanopic irradiance levels. Importantly, sustained pupil light reflex varied across tasks independently of the light condition, pointing to a potential impact of light history and/or cognitive context on sustained pupil light reflex. Together, our results emphasise that the distinct contribution and adaptation of the different retinal photoreceptors influence the non‐image‐forming effects of light and therefore potentially their brain correlates.