Our progress in executing a large-scale replication study in sound brain science!
Wednesday, 1 April 2020
The COVID-19 pandemic has brought new, unforeseen risks to in-person data collection. While it is not yet clear how easily the virus transmits between people in a lab setting, all sites have temporarily paused data collection to protect our participants and researchers. As of now, 122 participants have fully completed the study! We are sad to leave the lab and will miss the large influx of new data, but we hope this strange time provides study personnel a chance to prioritize their personal health and well-being.
In the meantime, the UMN group is writing additional code in Matlab to load, error check, and plot the data, which will be made open to the public once data collection is complete.
Pictured above, Dr. Whiteford’s work-from-home co-workers, Arya (left) and Daenerys (right).
Kicking Off Data Collection
Tuesday, 27 August 2019
Preregistration is complete! Check out our project website on OSF to read the details of the study protocol. The code and pilot data on OSF are being updated regularly as sites complete their pilot. Once a site has confirmed their pilot is complete and all of their materials are linked to the project website, they may begin recruitment and data collection!
Example of what one might experience when beginning data collection after one year of setup.
Another possible internal experience of a scientist on this project.
Collecting a large amount of data on a complex protocol means there are a lot of logistics to iron out. All sites need to be consistent in how they name data files, how they log unanticipated issues that might occur during data collection, and how they independently quality check the data before sharing between sites. Tomorrow will be the first of several multi-site virtual meetings among data collectors to review study logistics and discuss any unanswered questions or issues that might have come up so far.
Finalizing Multi-Site Setup and Preregistering
Friday, 9 August 2019
It has been a busy last several months for the multi-site project! The project coordinator, Dr. Whiteford, visited all 6 sites to assist with setup and troubleshooting. Equipment measures were gathered at all sites to ensure that the experiments will work uniformly on different setups. In some instances, site-specific changes to the code were made or new equipment was purchased to ensure the experiments work as intended.
Pictured above, Dr. Anne Luebke and Dr. Ross Maddox perform equipment tests at the University of Rochester (UR). All UR EEG measures will be collected in the Maddox Lab in the soundproof booth pictured here.
Different Subjects, Different Places
All sites are working hard to incorporate any code changes and finalize their setup. Once the setup is finalized, testing a pilot subject on the full experimental protocol is an important last step to quality check the experiments. Each site is collecting data from at least one full pilot subject. The raw data will be made open and available to the public on the Open Science Framework. Our goal is to start full data collection in September 2019.
Same Brain, Different Places
What is the effect of different EEG equipment on our EEG measures? In order to answer this question, pilot EEG data were collected on Dr. Whiteford at each site and UMN twice. Pilot analyses show that the EEG measures between-sites are as reliable as measuring the same participant at UMN twice. Great news for aggregating data across sites! Stay tuned for our preregistration protocol for more details.
1. There have been multiple code changes since last November, either due to bug fixes or method changes. With lots of code changes, it can be easy to miss something! Sites that have confirmed they are ready for full data collection have shared their finalized code with the project coordinator on Box. She is in the process of combing all versions on GitHub, a tool used to share software and version control. This has been essential for double-checking for any unintentional between-site code differences. The finalized code will be open to the public and linked to the preregistration once complete.
2. A major challenge for collecting and analyzing multi-site EEG data is that each site has their raw data in a unique format. Dr. Whiteford has written code for each site so that they will be able to independently quality-check and format their raw data so that it is consistent across sites. This code will be shared with all sites before data collection begins and linked to the preregistration.
3. Sharing big files is slow, and the raw EEG data files for our study can be over 4 GB. This is a minor inconvenience but something all sites must keep in mind throughout the study, as the computer falling asleep could interrupt the process of file sharing.
A draft of the preregistration protocol is written, including methods and analysis plan, and all co-authors are currently reviewing it. Once the protocol is approved by all authors, it will be posted on the Open Science Framework and open to the public, along with the experiment code, pilot data, and pilot analysis code. After the completion of preregistration, full data collection can begin!
Beginning Setup for our Multi-Site Replication Study
Tuesday, 20 November 2018
We are conducting a large-scale replication study, across six collaborating institutions, examining the relationship between musical training and the neural processing and perception of sound. Our primary questions of interest include:
(1) Is long-term musical training related to enhanced neural processing and the perception of speech and non-speech sounds?
(2) Is musical training related to a reduction in the natural effects of aging on speech perception in complex environments?
This study is funded by the National Science Foundation, with a particular emphasis on increasing transparency and reproducibility in research science.
The experiment code has been tested at UMN and shared with each of the collaborating sites on Box. Before sharing the code, the University of Minnesota group tested out the code on their equipment to ensure the tasks were working as intended.
Pictured above, undergraduate research assistants Kara Stevens (left) and Angela Sim (right) practice using the EEG equipment at the University of Minnesota’s Multi-Sensory Perception Lab in the Center for Applied and Translational Sensory Science. Kara and Angela are both members of the Auditory Perception and Cognition Lab and will help with data collection at Minnesota.
First Multi-Site Visit: Purdue University
One challenge of conducting a multi-site study, where the same experiments are conducted in multiple locations, is to ensure the experimental setup is consistent across sites, despite minor differences in equipment. To help ensure consistency across labs, the project coordinator, Dr. Kelly Whiteford, is visiting each site. Last month, Dr. Whiteford made her first visit to the Systems Neuroscience of Auditory Perception lab (SNAPlab) at Purdue University, headed by Dr. Hari Bharadwaj. It was a productive visit! With the help of Dr. Bharadwaj and graduate research assistant, Bre Oakes, the setup in Purdue is nearly ready to go.
Dr. Hari Bharadwaj (left) and graduate research assistant Bre Oakes (right) lead the setup and data collection at Purdue University.
1. All of the behavioral tasks worked well, but the calibration of the sound levels remained an issue. The Purdue group did some excellent sleuthing to narrow down the culprit to a hardware/software interaction and resolved the issue.
2. To check for any major between-site differences, EEG recordings from Dr. Whiteford’s brain will be assessed for the same tasks at all 6 sites.
Back in Minnesota…
Kara and Angela went back to the MSP Lab this past week to record Dr. Whiteford’s brainstem responses. These pilot data will be compared to data collected on Dr. Whiteford at Purdue University.
Pictured is Dr. Whiteford after her first Minnesota EEG recording in a 32-electrode cap.