@[toc] # Overview Welcome to the repository for *Expectations boost the reconstruction of auditory features from electrophysiological responses to noisy speech*, a combined EEG-pupillometry study of degraded speech processing in a non-clinical adult population. This repository contains data and stimulus materials generated during the course of the study. The experimental procedure involved a *training* and a *test* phase. Training involved listening to a narrated story of ~11 min duration. This was followed by 96 test trials, in which participants were presented two repetitions of identical speech stimuli that had been degraded by noise-vocoding or sine-wave synthesis. These two auditory stimulus presentations were separated by a 4 s interval, during which time participants saw either: 1. the written sentence corresponding to the sentence they had just heard (**correct prior; P+**); or 2. a written sentence different from the one they had just heard (**incorrect prior; P-**); or 3. four hash symbols (**no prior; P0**). 64-channel EEG and pupil diameter were recorded throughout both the training and test phases of the session. In addition, participants reported the subjective clarity of degraded sentence stimuli after each repetition. For trials in which written sentence information was presented (P+ and P- conditions), participants also reported whether the written sentence corresponded to the auditory stimulus. Further details about experimental design, data collection and analysis procedures can be found in the accompanying [*bioRxiv* preprint](https://www.biorxiv.org/content/10.1101/2021.09.06.459160v1). Code for reproducing the analyses presented in the manuscript may be accessed from the associated GitHub repository. # File structure The contents of each file directory are briefly summarised below. **Please note, altering the file structure / naming conventions may render files incompatible with analysis scripts archived on GitHub.** Files with the `.mat` extension are in the binary data container format used by *MATLAB*. ## `behav` `SWS_NVS_results_{subject}_{date}.mat` contains numerous variables pertaining to stimulus parameters and behavioural (key-press) responses in each experimental session. Trial-level information about stimulus identity, experimental manipulation, and behavioural response are collated in the `res_mat` array. Key columns from this array are identified below, in conjunction with the entries stored in the corresponding `res_mat_header` array: |column|header |definition| |-:|-|-| |1|nBlock|Block number (presentation order; 1-6)| |2|nTrial|Trial number (nested within block; 1-16)| |4|nOriStim|Stimulus number (nested within list; 1-16)| |5|Cond|Prior condition (1 = P+, 2 = P-, 3 = P0)| |6|StimType|Stimulus type (1 = SWS, 2 = NVS)| |7|nList|Stimulus list number (1-5)| |8|Pair|Stimulus pair number (1 = A, 2 = B)| |15|Resp1|Clarity rating #1 (1 = I did not understand anything, 2 = I understood some of the sentence, 3 = I understood most of the sentence, 4 = I clearly understood everything)| |16|Resp2|Clarity rating #2 (coded as per Resp1)| |17|Resp3|Sentence congruency judgement (1 = Yes, 2 = No)| `SWS_NVS_triggers_{subject}.mat` contains information about the timing of stimulus onsets and offsets (expressed as EEG sample points) for training and test trials. ## `eyeTrack` Eye-tracking (pupil diameter) data were acquired using a Tobii Pro TX300. `SWS_NVS_tobii_{subject}.mat` contains data that have been converted from their native format ready for preprocessing in *MATLAB*. These files comprise information about the acquisition sampling rate (`fs`) and stimulus event times (`events`), in addition to the main `data` array. The contents of the latter are detailed below: column|definition -:|- 3| X gaze coordinate of the left eye 4| Y gaze coordinate of the left eye 5| X gaze coordinate of the right eye 6| Y gaze coordinate of the right eye 7| Left eye position (x coordinate) 8| Left eye position (y coordinate) 9| Right eye position (x coordinate) 10| Right camera eye position - y coordinate 11| Left eye validity 12| Right eye validity 13| Diameter of left eye pupil 14| Diameter of right eye pupil 15| Distance of the camera from left eye 16| Distance of the camera from right eye 17| Time (s) ## `preprocEEG` Contains files required to run either of the time-frequency or stimulus reconstruction pipelines in *MATLAB* independently of the EEG preprocessing pipeline. (Note that running any of these pipelines will populate this directory with additional intermediary files.) `ICAcomp_rej_ft_test_SWSNVS{subj}.mat` and `ICAcomp_rej_ft_training_SWSNVS{subj}.mat` comprise ICA-corrected EEG `data` structures for test and training trials, respectively. `nlhie_ft_test_SWSNVS{subj}.mat` contains preprocessed EEG `data` from test trials prior to ICA correction. These files also include stimulus `event`, electrode `layout`, and raw data header file (`hdr`) information. The rank of the data (which may have altered due to the interpolation of bad channels) is stored under `rankICA`. ## `rawEEG` The EEG was acquired from 64 Ag/AgCl EasyCap mounted active electrodes using a BrainAmp system. `*.eeg` files contain EEG data; `*.vhdr` files contain metadata about the EEG acquisition. Note that `*.vmrk` marker files do not contain information about stimulus events; this information was captured via an independent EEG channel. ## `stimuli` `list_stimuli_102018.csv` contains all sentence stimuli used in the experiment, along with information encoding stimulus list/number/pair codes necessary to map each instance onto an experimental event (test trial). `training_tale3.wav` is the audiobook recording that participants heard during the training phase of the experiment. Test stimuli are provided as original `{code}.mp3` and converted `{code}.wav` files, named according to the unique code assigned in column 6 of `list_stimuli_102018.csv`. The noise-vododed (`{code}_voc_7Bds.wav`) and sine-wave (`{code}SWS.wav`) variants of these files (i.e. the stimuli presented during test trials) are also provided. `filtEnv_*.mat` files are bandpass-filtered stimulus envelopes produced by the *MATLAB* stimulus reconstruction pipeline. # Citation If the materials archived here are useful for your own research, please cite this repository per the `Citation` panel on the main page. Please also cite the accompanying manuscript: >Corcoran, A. W., Perera, R., Koroma, M., Kouider, S., Hohwy, J. & Andrillon, T. (2022). Expectations boost the reconstruction of auditory features from electrophysiological responses to noisy speech. *Cerebral Cortex*, bhac094. doi: [https://doi.org/10.1093/cercor/bhac094](https://doi.org/10.1093/cercor/bhac094) # License The contents of this repository are distributed WITHOUT WARRANTY; without even the implied warranty of merchantability or fitness for a particular purpose. Click on the License information icon on the main page for further details.