Re-analysis of Cui, M. E., & Herrmann, B. (2023) Eye movements decrease during effortful speech listening. Journal of Neuroscience, 43, 5856โ5869.
-
subj
-
trg: trigger code (cond * 10 + snr)
-
trial: # 28s segment per condition
-
id: unique trial id (per participant)
-
block: 1 or 2 in temporal order
-
snr: 1 to 5 (-4, 1, 6, 11, 16 dB SNR; note: max. two-step rule described in manuscript was apparently not implemented)
-
condition: 1 (intact) or 2 scrambled
-
pa: pupil area (used for re-analysis)
-
pd: pupil diameter
-
pa_pfe: pupil area corrected for pupil foreshortening error by regression (PFE; used in original publication)
-
pd_pfe: pupil diameter corrected for PFE
-
fd_int: fixation duration with interpolation
-
fd: fixation duration without interpolation (used for re-analysis and in original publication)
-
sgd_int: spatial gaze dispersion with interpolation
-
sgd_i: spatial gaze dispersion without interpolation, excluded if > 10% NaN per time window (used for re-analysis and in original publication)
-
sgd: spatial gaze dispersion without interpolation, excluded if any NaN per time window
The scripts contain the following parts of the analyses:
- 01_import_data: convert csv into a data.frame in R with readable labels etc.
- 02_estimate_models: estimate initial Mixed Models in lme4 (all outputs available in respective HTML file)
- 03_estimate_bayes_models: estimate the more complex Mixed Models in brms (the large results file available on request)
- 04_report_bayes_models: extracts the model outputs from the bayes models (all outputs available in respective HTML file)
- fig: scripts to recreate all results figures and illustrative figures
- I have tried to fully replicate the analysis except for PFE correction. PFE correction by regression appeared to introduce rather than reduce noise. Correction maps were implausible for some participants. Possibly the sampled display area was too small?