To fully utilise the collected data, please read the following carefully.

A total of 119 respondents participated in the study. Their codes and anonymized personal information are provided in the XLSX file, Sheet 1. IDs correspond to the filenames of the CSV files. Each CSV contains Raw GSR (galvanic skin resistance) in ohms in column C, with UTC timestamps in column A. Column B indicates stimulus changes, alternating between "0" and "1"; each change denotes a new stimulus. The experimental scheme is available in the PNG file. Note that interstimuli (denoted IS in the PNG) were presented for a fixed 30-second period at the start of the experiment, and subsequently for flexible periods of ≥5 seconds, depending on the participant's GSR response.

A custom algorithm was used to transform raw data into relevant variables. The curated dataset is available in XLSX, Sheet 2. In the first round of curation, peaks of GSR responses are marked in the raw data. The onset of a peak is defined as a continuous segment of the GSR signal at least 200 samples long (for sampling frequency of 1000 Hz therefore 1/5 of a second), on which the first difference exceeds 50 ohms. The length of the difference step is 20 samples. If multiple peaks are detected, only the first is analysed. Extracted variables include:
- Latency: onset of the peak relative to stimulus, in seconds
- Lenght: interval from peak onset to peak maximum, in seconds
- Amplitude: difference between maximal and minimal GSR value within the peak, in ohms

The second round of curation removes peaks likely caused by participant movement. This information is not available in the data, it was manually noted during data collection.

The third round classifies remaining peaks as most likely relevant or irrelevant to the stimulus. Due to the physiological delay of GSR, only peaks exceeding a minimum latency of 1 s are considered stimulus-related.

The uploaded XLSX file reflects all three rounds of curation. Additionally, subjective fear ratings of the presented stimuli are included. Detailed curation procedures are described in the paper.