New paper - Facial expressions mimicry is not just politeness
Jan 13
1 min read
Updated: Jan 30
Our recent paper is receiving media attention!
Read about how our facial EMG electrodes a nd data analysis pipeline are helping psychologists study facial expressions.
Article
Continuous 30fps with no gap doesn’t make a no-stimulus block useless. Hanein’s averaging step still needs an uncompromised reference, and post hoc artifact tagging in a trial-by-trial log only labels overlap after the fact; it can’t supply the clean baseline. For Subway Surfers City near-misses, jaw clenching on the swipe contaminates the zygomaticus window, but a pre-task block still gives something to subtract—like keeping a tier list for Aion 2 that updates instead of freezing one ranking. Maybe I’m wrong?
A no-stimulus block assumes the stimulus ever stops. Mine doesn't — continuous 30fps play with no gap between swipes, so jaw clenching and the zygomaticus response overlap from first frame to last. I tag artifact windows post hoc with a trial-by-trial log for baseline, stimulus, and artifact windows keyed to each swipe, because Hanein's averaging step has nothing clean left to subtract.
I added a no-stimulus baseline block before any averaging, and now I log each trial as baseline, stimulus, or artifact. That one step keeps the zygomaticus signal from a screen edge out of the mimicry count, especially in Hanein's pipeline for Subway Surfers City near-misses. Using a baseline-and-artifact window calculator mid-study made the jaw-clench confound visible before it contaminated the reflex average.
A no-stimulus baseline block before averaging would also help UX researchers testing endless runners and physical therapists tracking jaw clenching in stroke recovery. The zygomaticus signal from a chin rest or screen edge is the same confound whether the subject is playing Subway Surfers City or doing facial rehab. Keeping a trial-by-trial log for baseline, stimulus, and artifact windows separates reflex from artifact for anyone using facial EMG outside psychology.
Hanein's pipeline needs one step before averaging: a no-stimulus baseline block, otherwise the zygomaticus signal from a screen edge or a chin rest reads as mimicry. In the Subway Surfers City example, the reflex to a near-miss is confounded by jaw clenching on the swipe. A trial-by-trial log, like a worksheet for tagging baseline, stimulus, and artifact windows, would separate the two.