SYNAPSE: cognitive performance, EEG, wrist physiology, facial thermography and indoor environment under warm and cool LED lighting and urban soundscapes

SYNAPSE (SYnergistic Neurophysiological Assessment of Perception in Sensory Environments) is a multimodal dataset on how lighting colour and background sound affect the cognitive performance, physiology and perceived indoor environment of people doing office work. Study. Ten adults (6 female, 3 male, 1 not reported; age 28.0 ± 4.2 years, range 22–35) each completed two sessions on the same day in a simulated office at the Technical University of Denmark (24–26 June 2025): one under warm (2014 K) and one under cool (6250 K) tunable-white LED lighting. In each session they performed two rounds of computerised tasks (n-back, Stroop, visual counting or shape comparison, flanker), each round with a different soundscape over headphones (cafeteria babble or road traffic), and answered questionnaires before the tasks and after each round. Data recorded in each session: Cognitive performance: 8,172 trials with stimulus, response, accuracy and reaction time; EEG: 4 channels (Muse S), 256 Hz, 15.6 h in total; Wrist physiology (ActiGraph LEAP): photoplethysmography, acceleration and skin temperature, with device-derived heart rate and heart-rate variability (18 of 20 sessions); Facial thermography (FLIR AX8): relative head temperature every ~10 s, derived from the thermal video; Indoor environment (ATMOCUBE): air temperature, humidity, CO₂, particulate matter, VOC, NOx, sound level and illuminance at 1-min intervals; Questionnaires: alertness, energy, sleep and comfort before the session; perceived thermal, air-quality, lighting and acoustic conditions; task experience and affect after each round; a background questionnaire. All recordings share one session timeline (baseline, task rounds, task blocks), so occupant responses can be aligned with each other and with the room data. The dataset supports research on multi-domain indoor environmental quality and cognitive performance, and on building informatics, such as data-driven comfort and performance models and occupant-centric building operation. Files. The dataset follows BIDS 1.10 and is distributed as one archive, SYNAPSE_dataset.zip (214 MB unpacked), which unpacks into the folder SYNAPSE_dataset/. README.md, also available separately, gives the study design, a schema of all tables, data availability per session, missing data and data quality, and usage notes. Every table has a JSON file describing its columns, and DATA_DICTIONARY.tsv lists all columns in one place. Not included. Raw thermal video (it shows faces), the free-text operator log, health-screening answers and the identifiable background questionnaire are available from the authors under a data-use agreement, within the terms of the participants' consent. The task images are distributed with the SYNAPSE experiment platform (https://doi.org/10.5281/zenodo.23768059). The soundscape audio cannot be redistributed and is described in stimuli/soundscapes.json. Data quality. Every stream comes with quality-control tables. EEG quality differs strongly between channels (forehead channels AF7/AF8 largely usable; temporal channels TP9/TP10, under the headphone ear cups, often contaminated); facial temperatures are relative values; wrist data are missing for one participant. The README lists the availability of every stream per session and all known issues. Related resources. Experimental protocol: https://doi.org/10.17504/protocols.io.m7vznm25rvx1/v1. The dataset is described in a Data Descriptor submitted to Scientific Data.