low frequency filtering of pupil data

[gkaguirre]

Currently, the code is filtering and removing the lowest 8 frequencies from the pupil signal prior to being placed in a packet. We would like to change this behavior.

First, we are working on measuring the size of the evoked pupil response, and relating this to the baseline pupil size. We need the un-filtered data to proceed with this.

Second, there is good evidence that the spontaneous low-frequency variation in pupil size strongly negatively correlates with the BOLD fMRI signal in early visual areas [Yellin 2016, NeuroImage].

Harry and I are going to branch and then modify the code to:

• calculate and return the pupil signal component at and below 0.05 Hz
• return the unfiltered pupil signal
• store the low-frequency component for use in subsequent modeling of the fMRI data.

[gkaguirre]

This has been implemented in the master branch.

[spitschan]

Closing this issue now.