December 2018 – Connecting to people

December was a shorter month than the previous ones, as I left for the Christmas break on the 19th. However, there was good progress made in research, collaborations, student supervision, invited talks, and outreach. The reason I called this month connecting to people, is that I got two new students, I have a closer collaboration with Thanos, and we got a lot of resonance from the Maastricht people volunteering for Pint of Science. I even got called “the pint of science lady” once! But, let’s get started.


In December, I got a bit more training into the TMS methodology, by Teresa Schumann. We also filled in the project proposal form for the Non-Invasive Brain Stimulation committee to approve. This made me think about the exact parameters of the stimulation, the equipment and the analysis pipeline of the EEG data. Some formal details:

  1. NIBS lab required: NIBS1
  2. Equipment required:
    • TMS-compatible EEG caps by EASYCAP (sizes 54, 56, and 58cm)
    • R30
    • MC-B70 coil
    • Sham B70 coil
    • Neuronavigation
  3. Number of participants: 40 (20 per group)
  4. Number and length of sessions per participants:
    • Number of sessions: 2 (sham and real stimulation)
    • Length per session: 1h in NIBS1, 3 hours in total
    • Minimal amount of days between sessions: 1 week
  5. Stimulation parameters:
    • Intensity: cTBS (600 pulses) at max 100% resting motor threshold,
    • Site: right pre-SMA
    • Localisation: using structural MRI scan, identifying the pre-SMA for each participant individually based on TAL coordinates.
  6. EEG Recording
    • Number of EEG electrodes: 64
    • Hardware low pass filter setting (Hz): 0.1 Hz
    • Hardware high pass filter setting (Hz) or time constant (s): 100 Hz
    • Sampling rate: 500 Hz
    • Amplifier gain:1 microV
    • Measurement reference electrode location: left mastoid
    • Additional electrodes used for off-line rereferencing: right mastoid
    • Locations of electrodes recording eye movements and blinks: Fp1, Fp2
    • Location ground electrode: AFz
  7. Analysis
    • Software package(s) used: MNE Python
    • Procedure for handling eye artifacts : Artifact correction with ICA
    • Describe intended analysis (ERP/Time-frequency/frequency bins/MVPA/source localization/connectivity etc): ERP analysis time-locked to the beat onset, and time-frequency analysis to find out which frequencies are more sensitive to beat perception
    • Are you intending to do any analysis on data recorded during the stimulation? No

Research collaboration

My collaboration with Thanos Lykartsis is going very well. Thanos delivered the first audio analysis of the stimuli. The beat tracking algorithm can classify very well between male vs. female voices and story vs. poem.


I received 4 applications for master thesis internships, of which I accepted 2 and I am very happy to have two new people. These new people brought me hope that I will have help to conduct the experiment and analysis together.

Invited talk in Marburg

On December 13th I gave a talk in the Neurolinguistics Colloquium of Philipps-University Marburg, where I defended my PhD three years ago. It was moving to go back after these few years, and explain what I have learned on the way. My talk was focused on the how to “go natural” in the neurobiology of language. I explained in detail how I created the stories for my PhD research and what were the objectives of our approach at Imperial College London. I also gave a sneak peak into how I am planning to model the responses in NERHYMUS.


Pint of Science is going very well, we have a couple of volunteers as event managers and we sent out a big university-wide call to ask for more volunteers. This call was, to my surprise, presented also in the narrowcasters at Oxfordlaan 55, in the cafe banditos, see picture.

Last, I leave you with my favorite view of my home town in Greece, Heraklion.


November 2018 – Finally the EEG experiment

So, it’s beginning of January and only now I have time and energy to write the wrap up of November (and December). In general November has been very creative in research, teaching and outreach. But let’s take it from the top.


November started with unresolved issues in the Matlab script, related to stimulus presentation. I have extensive notes of the first couple of days, because I had to keep track of the solutions I’ve already tried, in order to avoid losing time on re-trying things that didn’t work out. Looking into Matlab code generally makes me dizzy, but at least I managed to navigate and eventually solve the issues. I ran the whole experiment on myself and discovered some missing files and some typos that needed to be fixed. One of the audio files needed to be re-recorded, as it included a mispronunciation of a name in only one part of the story. Luckily, the speaker was available in Maastricht, so he passed by and we wrapped up the recording in 15 minutes. Then I normalized that file and included it to the list of audio files. I also fixed the typos and balanced the correct answers to half option a and the other half option b.

My initial experimental script was doing the normalization in the stimulus presentation script and the resulting normalized files were only peak-normalized. This ended up not having the same loudness across files. With the valuable help of my collaborator Thanos, who explained how to do a two-step normalization, I made a short script which normalized the files in peak and loudness externally from the experimental script. Then, in the experimental script I only load and play the files without needing to normalize them anymore. In mid November I sent the normalized files to Thanos, so that he runs his auditory analysis algorithms on the signal.

We then determined how to analyze the audio signal, based on Thanos’s previous research and its relevance for cognitive functions. In my email to Thanos I wrote:

I see that the most prominent features given the theory on speech rhythm, such as the period of the strongest peak (beat), do not perform so well in the automatic language identification task in all the corpora you tested. This might at first be surprising, but the difference should be viewed, as you already discussed in the paper, under the task performed by the classification algorithm and the fact that the algorithm might not be reflecting cognitive processes.

Given these, I would like to suggest that we take all the beat histogram features for our analysis and use them in two ways:

  1. to test whether the algorithm can discriminate between conditions, poem vs. story and which features are the most prominent in this classification.
  2. then, to take the most cognitively motivated features, e.g. amplitude and F0, and compare their fit to the EEG data.

We decided to compute all the novelty functions based on the beat histograms, as laid out in Thanos’s 2015 paper. We would then discuss that we can do with features like spectral flux or spectral flatness, which do not seem to be relevant for cognition, at least for the moment, but seemed to contribute very much to the classification between languages.

On November 20th, I had a second project proposal meeting with the TMS group, in which I outlined the details of the experimental setup. The group recommended me to be cautious about: preparation of the subject (the gel would dry out when applying the TMS), how to optimize procedures so that we are fast and effective, use a TMS compatible cap, how to localize the pre-SMA for each participant. In my meeting with Teresa on November 23rd, she stressed that the most important thing is how we localize the pre-SMA, whether we need a structural MRI and look in each brain for the specific co-ordinates, or whether it is an easy area to localize, such as other motor regions, and does not need an MRI scan. She advised me to look into previous studies which have targeted the pre-SMA and check the methodological details. I have to complete and send back the NIBS Project Proposal Form before we start measuring for NERHYMUS.


Most of the students of the writing skills course submitted their Introduction on time on November 9th. I was amazed at the general low quality of the essays. However, after I visited the course on how to evaluate and give feedback for writing skills and theses, I realized that the students are just starting to understand how scientific writing is expressed and how to write in this form themselves. So, I re-evaluated my initial evaluation, and made an effort to give my students meaningful feedback on how to improve, rather than dismissive general comments. From the individual short meetings for providing this feedback I had the impression that the students truly appreciated my effort and that it helped them understand what they did wrongly and how to improve.

With regard to the mentoring course, I had the impression that the second mentor meeting was a disaster. Even though I did my best to follow the instructions of the course administrator, I felt that the students were not interested, as the session was very much just repeating the content of the first session. The general impression was that what we did in the scheduled 2 hours, could have easily been done in 1 hour without needing to overtire the students with repetition. Maybe it was also the wrong time of the day for such a task, as 4-6pm in the late afternoon is quite late. I have forwarded this feedback to the colleagues who planned the course and hope that this session will be planned differently for next year.


We’ve made a lot of progress with Pint of Science Maastricht. In mid November, Lena and I met with Danielle Vogt and Flore Clerx from the Marketing and Communications of the FHML, who really helped us to pitch our call for volunteers in a fun manner. They were also very supportive of Pint of Science Maastricht in general and they suggested to hold a general info session in the Turnzaal in January and have media coverage and possibly interviews to spread the word about the festival. Lena worked with them on the text for the university-wide call for participation and we sent it out at the end of November.

On November 27th I held the first info meeting in a small circle with people from NP&PP and CN. The goal of this meeting was not only to inform, but also to recruit volunteers for the taster event, which will introduce Pint of Science to Maastricht on February 27th, 2019.

On November 28th, I went to the Societal Engagement event, organized by the University of Maastricht, representing Pint of Science Maastricht with my own stand in the “market”.

Many people showed interest in the initiative, especially because they found the idea of bringing science to the pubs cool. I also got some important contacts, such as the organizers of TEDx Maastricht Gonny Willems and Sophie Kells. Sophie works in the press office of the university and came to my stand to offer her enthusiastic support to Pint of Science Maastricht. I approached Gonny at the bar and we decided to meet in January in order to discuss collaborating on the level of training speakers for public speaking and on organizing an event with the topic “Redefining Europe” for the Pint of Science festival in May.

Last, on November 13th I went to the launch event of the Female Empowerment Maastricht initiative. This album has some amazing photos.

Photo by: Brian Megens Photography (

I took part in the Work-Life balance workshop, together with my colleague and friend Alex Emmendorfer, see us both in the picture.

Photo by: Brian Megens Photography (

That was a decent summary of November.