Table 1.

Settings for the identification by two-means clustering (I2MC) algorithm and their suggested values for data at 300 Hz

Setting	Used Value(s) at 300 Hz	Impact When Changed
Interpolation window	100 ms	Increase will lead to interpolation of blinks, decrease will lead to less periods of data loss being interpolated.
Interpolation edge	6.7 ms (two samples)	Increase will require more data points at data loss edge, and will not interpolate in the event of flicker (i.e., repetition of short period of data loss and data points). At least two samples are required.
Clustering window size	200 ms	Increase will lead to clustering procedure being more readily carried out over saccade-fixation-saccade sequence.
Downsampling	150, 60, and 30 Hz	Removal of downsampling steps will lead to more susceptibility to short bursts of noise.
Window step size	20 ms	We observed no difference between 3.3 and 20 ms.
Clustering-weight cutoff	2 standard deviations above the mean	Increase will lead to fewer fixation candidates (more conservative), decrease to more fixation candidates (more liberal).
Merge fixation distance	0.7°	Increase will lead to more fixation candidates being merged.
Merge fixation time	40 ms	Increase will lead to more fixation candidates being merged.
Min. fixation duration	40 ms	Increase will lead to more short fixation candidates being excluded.