Experimental protocol.

(A) Overall timeline. After prior light history standardisation, participants performed executive (always first), emotional and attentional tasks (pseudo-randomly 2nd or 3rd, blue arrow). As the attentional task included fewer light conditions, it is not considered in the present manuscript (see methods for more details).

(B) Spectral power distribution of light exposures. Monochromatic orange: 0.16 mel EDI lux; Polychromatic, blue-enriched light (6500K); LOW, MID, HIGH: 37, 92, 190 mel EDI lux. For the present analyses, we discarded colour differences between the light conditions and only considered illuminance as indexed by mel EDI lux, constituting a limitation of our study. See Suppl. Table 2 for full details).

(C-D) Tasks procedures. Time is reported in seconds relative to session onset; participants were pseudo-randomly exposed to the 4 light conditions. (C) Executive task: alternation of letter detection blocks (0-back) and working memory blocks (2-back). (D) Emotional task: lure gender discrimination of vocalizations (50% angry (red), 50% neutral (white).

Illuminance impact on the hypothalamus subparts.

(A) Segmentation of the hypothalamus in five subparts in a representative participant. The nuclei encompassed by the different subparts are indicated in the right inset – according to (Billot et al., 2020). ARC: arcuate nucleus; DMH; dorsomedial nucleus; LH lateral hypothalamus; LTN: lateral tubular nucleus; MB: mamillary body; POA: preoptic area; PVN: paraventricular nucleus; PNH: posterior nucleus of the hypothalamus; SCN: suprachiasmatic nucleus; SON: supraoptic nucleus; TMN: tuberomammillary nucleus; VMN: ventromedial nucleus.

(B-C) Estimates (beta; arbitrary unit – a.u.) of the collective impact of illuminance variation on the activity of each hypothalamus subpart (Refer to Table 1 full statistics). (B) Executive task: significant main effect of hypothalamus subparts (p=0.002), no significant main of task type (p=0.4) or subpart-by-task-type interaction (p=0.61). (C) Emotional task: significant main effect of hypothalamus subparts (p<.0001), and of stimulus type (p=0.048) or subpart-by-stimulus-type interaction (p=0.74).

(D-G) Whole brain analyses of the collective impact of the variations in illuminance over the hypothalamus area - for illustration. A local positive peak (red; puncorrected<0.001) was detected over the posterior hypothalamus subpart (light blue) in executive (E) and emotional (G). A local negative peak (red; puncorrected<0.001) was detected over the inferior-tubular hypothalamus subparts (light orange) during the executive task (D), while local negative peak (red; puncorrected<0.001) was detected over the inferior-anterior (yellow) and superior-anterior (blue) hypothalamus subparts during the emotional task (F) – insets correspond to enlargements over the hypothalamus area. Arrows from panels B and C arise from and are colour coded according to the hypothalamus subpart that is displayed in panels D to G. These results indicate that our finding does not arise from a nearby “leaking” activation/deactivation.

(H-K) Estimates of the impact of each illuminance on the activity of the hypothalamus subparts. (Refer for Table 2 and Suppl. Tables 3/4/5/6 for full statistics) Activity dynamics across illuminance for each subpart (colour code as in A). Results are displayed per task or stimulus type although no interactions with task or stimulus type were detected. Significant illuminance-by-hypothalamus-subpart interactions were detected for (H-I) the executive task (p=0.041) and (J-K) the emotional task (p=0.041). Small letter indicate significant difference (p < 0.05) between the following subparts at illuminance: a. 92 mel EDI lux: posterior vs. superior-anterior & inferior-tubular; b. 190 mel EDI lux: posterior vs. inferior-anterior, superior-anterior & inferior-tubular; c. 0 mel EDI lux: posterior vs. superior-tubular; d. 92 mel EDI lux: posterior vs. superior-anterior; superior-anterior vs. superior-tubular; e. 190 mel EDI lux: posterior vs. inferior-anterior, superior-anterior & inferior-tubular; superior-tubular vs. superior-anterior, inferior-tubular & inferior-anterior.

Differences between hypothalamus subparts in the collective impact of the variation in illuminance on their activity.

Outputs of the generalized linear mixed model (GLMM) with subject as the random factor (intercept and slope), and task and subpart as repeated measures (ar(1) autocorrelation).

* The corrected p-value for multiple comparisons over 2 tests is p < 0.025. # Refer to Fig.2A for correspondence of subpart numbers

Statistical outputs of GLMM testing for differences between the activity of each subpart of the hypothalamus under each illuminance

* illuminance in mel EDI lux

# Only significant comparisons are reported in the main text. For the full table, including post hocs comparing light levels within a subpart, refer to Suppl. Table 3/4/5/6/7

Impact of illuminance on performance and relationships with the activity of the posterior hypothalamus subpart.

(A) Accuracy (percentage of correct responses) to the 2-back increased with increasing illuminance (p = 0.034).

(B) Accuracy to the 2-back task is negatively correlated to the activity of the posterior hypothalamus subpart (p = 0.0027).

(C-D) Accuracy to the 2-back task is not correlated to the activity of the inferior-anterior (C) and inferior-tubular (D) hypothalamus subparts (p > 0.4). Association between superior-anterior and superior-tubular subparts are not displayed but were not significant (p > 0.6). See Suppl Table 7 for full details.

(E) Accuracy to the 0-back task is not correlated to the activity of the posterior hypothalamus subpart (p = 0.45).

(F) Reaction times to the emotional stimuli did not significantly change with increasing illuminance (p = 0.41).

(G) Reaction times to the emotional stimuli are correlated to the activity of the posterior hypothalamus subpart (p = 0.04) with higher activity associated to slower reaction times. (H) Reaction times to the neutral stimuli are not correlated to the activity of the posterior hypothalamus subpart (p = 0.6). Solid and dashed lines correspond to the significant and not significant linear regression lines, respectively.