iiinova — where your digital self keeps evolving

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 06/10/2026

3/3 t.co/n3qEEq6cvM Here the open repository! We're also designing hyper-realistic VR environments specifically to reduce therapy dropout rates. A true joint effort between Information Engineering, Bioengineering, and Psychology! #VR #Neuroscience #Psychology #MentalHealth #UniPisa

t.co

https://doi.org/10.17605/OSF.IO/YN29T

010

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 06/10/2026

2/3 Beyond quantifying avoidance subtypes (vr-BAT), we are now implementing closed-loop biometric feedback. By tracking skin conductance and heart rate in real time, the VR environment dynamically adjusts if anxiety gets too high.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 06/10/2026

1/3 Exciting update! Regional TV visited our lab at the University of Pisa to cover the next phase of our phobia research, building on our latest paper in Communications Psychology. 🧬✨ doi.org/10.1038/s442...

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 22/09/2026

7/7 Scenario, data and code are open on OSF: doi.org/10.17605/OSF... Protocol on protocols.io. Not spider-specific: anything built around avoiding a feared stimulus can use the same structure. Paper: doi.org/10.1038/s442...

doi.org

OSF

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 22/09/2026

6/7 Clustering gave two behavioural subtypes that only partly match questionnaire groups. Low-fear people were always low-avoidance; high-fear people showed up in both clusters. Fear and avoidance are complementary constructs, not interchangeable ones.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 22/09/2026

5/7 Together those dimensions explained more variance in self-reported fear (adj. R² = .49) than the classic distance-only model (.32). Some of them have no counterpart in traditional metrics at all.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 22/09/2026

4/7 But VR records the whole approach, not just the endpoint. A PCA gave five dimensions: spatial distancing, hesitant approach, action latency, initial freezing, pre-contact freezing. Two people can stop at the same distance and get there in totally different ways.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 22/09/2026

3/7 Distance, time and velocity all tracked self-reported fear in both versions, so the VR task works as a standardised stand-in for the real one, with much less manual measurement error and far better temporal precision.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 22/09/2026

2/7 75 volunteers, from zero spider fear to severe. Two sessions in randomised order: approaching a real spider down a 12 m corridor, and approaching a virtual one in the same corridor rebuilt in VR.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 22/09/2026

1/7 New in Communications Psychology: phobic avoidance is usually measured with one number, the distance you keep from what scares you. We rebuilt that test in VR and found at least five different behaviours hiding behind it. 🧵 doi.org/10.1038/s442...

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

7/8 Caveats, because it's a pilot: 9 of 15 met the per-protocol criterion, session order wasn't randomised, and we picked the target band ourselves. Peak anxiety and its latency didn't differ between conditions.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

6/8 It also worked as an emotional container: the variance of the total anxiety load dropped significantly (p = .011). Not lower for everyone, but everyone pulled toward the same window.

Graphic representing the Area Under the Curve (AUC)

Graphic representing the Linear Mixed Model

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

5/8 After the trigger (~24% into the session), scripted anxiety kept climbing while adaptive anxiety flattened out, accumulating more than three times more slowly.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

4/8 In the adaptive session those ratings silently drove the room: above 0.66 avatars left, below 0.33 more came in and started interacting. Anxious enough for habituation, calm enough to stay in the chair.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

3/8 15 volunteers with clinical social anxiety sat in the same virtual waiting room twice, months apart, rating their anxiety continuously with the controller. They were never told the two sessions worked differently.

Screenshots during the BRAVE scenario, and a picture from the experimental setting showing a participant wearing the VR headset

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

2/8 Most VR exposure replays a fixed script: the room fills the same way for everyone. The one thing VR can do that a real room can't is change while you're inside it, in response to you.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

1/8 Exposure therapy for social anxiety works, but it's aversive enough that the people who need it most often refuse it or drop out. At ECCE 2026 in Siena (ecce2026.unisi.it) I presented a VR waiting room that adapts in real time to how anxious you are: doi.org/10.1145/3822... . 🧵

doi.org

Adaptive exposure therapy for Social Anxiety Disorder: a preliminary validation in virtual reality | Proceedings of the 2026 European Conference on Cognitive Ergonomics

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 16/09/2026

Wonderful time at ECCE 2026 (ecce2026.unisi.it) in Siena (Italy) presenting a preliminary but promising study (doi.org/10.1145/3822...) from the BRAVE project (braveproject.unipi.it), while meeting old friends and colleagues. Kudos to @martinabriglia.bsky.social for her first paper!

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 14/09/2026

Recently came back from Belgium, where @circleu.bsky.social organized a matchmaking event at UC Louvaine to design an educational project on migration: amazing experience! www.circle-u.eu/events/acade... Absolutely recommended for other academics in the alliance willing to join its future events 💪

circle-u.eu

Academic Matchmaking Event at UCLouvain - September 2026

Are you an academic from one of the Circle U. universities looking to develop courses on the Open Campus of our alliance? Would you like to explore how to design such courses and find partners to co-c...

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 24/06/2026

6/6 This framework helps catch real-time incongruencies between how people think they feel and how their body actually reacts—opening a new era of tailored psychometrics. Models using ONLY behavioral data from VR controllers performed surprisingly well. A cost-effective 'lite' version is feasible!

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 24/06/2026

5/6 📈 The big takeaway? While in-the-moment self-rating is the strongest individual predictor, late integration of bodily markers (like the tvsymp index) was crucial to accurately spot non-anxious controls. Biology and conscious appraisal work hand in hand.

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 24/06/2026

4/6 On the ML side, we built a custom Late Fusion Transformer Encoder framework. By processing the behavioral and engineered physiological time-series in independent, specialized branches first, the model hit a sweet 82.5% overall accuracy (F1-score = 0.853).

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 24/06/2026

3/6 🕹️ While immersed, 63 participants continuously rated their anxiety in real-time using a joystick slider. No memory recall bias here. Meanwhile, we simultaneously tracked their physiological data: ECG (for HRV) and EDA (for sympathetic activation)

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 24/06/2026

2/6 Most VR setups for anxiety focus on performance fears (like public speaking). But what about everyday social dread? We built an ecologically valid VR waiting room that gradually fills with avatars engaging social interactions (more details here: doi.org/10.1109/Metr...)

doi.org

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 24/06/2026

1/6 🚨 New paper alert! Freshly accepted in IEEE Transactions on Affective Computing (TAFFC) 🧠✨ How can we measure Social Anxiety Disorder (#SAD) dynamically, skipping the biases of old-school, static questionnaires? Check it out 👇 ieeexplore.ieee.org/document/115...

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 21/04/2026

Some time ago I stressed the importance of sharing scientific failures & rejects to normalize them. However, today I'm proud to celebrate a success! 💪 The Circle U. post-doc mobility program will represent an occasion to further link with Frank Scharnowski's lab www.circle-u.eu/news/circle-...

circle-u.eu

Circle U. Postdoc Mobility Programme

Seed funding for your next cooperative research project! Deadline: 15 March 2026

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 02/03/2026

If you plan to attend at the 5th edition of IEEE 𝐌𝐞𝐭𝐫𝐨𝐗𝐑𝐀𝐈𝐍𝐄 (hosted in Chemnitz, Germany, October 20-22 of 2026) consider submitting your abstract at our thematic session: "Advances in Affective Computing and Social Interaction Technologies"!😎 www.metroxraine.org/thematic-ses...

010

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 03/12/2025

Socials profiles (including this) are full of posts disseminating or celebrating achievements, but sharing negative news could normalize the daily academic routine of rejection. Today I'm sharing that, so far, I was rejected by 4 of the 7 funding opportunities I applied for 😵‍💫

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 10/10/2025

I'm not always being interviewed by Italian national television with BRAVE's PI Alberto Greco, but when I do I'm sweating every drop of water from my body because of the hottest day in months! Nevertheless, it was funny to show off some research activity 😎 www.youtube.com/watch?v=c--Z...

youtube.com

TheBRAVEproject

YouTube video by Sergio Frumento

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 08/10/2025

Vienna is always amazing, but with such wonderful friends and colleagues is priceless! Thanks again Stanisa Raspopovic for the invited talk and the opportunity to have deep conversations about shared research interests 🦾 ... and best wishes to Pietro Palopoli for his promising career!

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 10/04/2025

Good luck! Submission parties should become normal in academics, since rejections hurt so much the first times... Mentors should reward submission more than publication

010

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 07/03/2025

This could sound weird, but feel free to PM me if you are planning to visit Miracle's square in Pisa (that with the leaning tower)... I live 5-minutes walking from there and I would be glad to have a chat about your research on consciousness while admiring some inspiring piece of art! : )

010

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 30/11/2024

I will use this place to 1) lurk on what other scientists are doing, 2) share our research and data, 3) have interesting conversations about hot scientific topics. Hated Twitter and hope in BlueSky, BTW it's still not totally open so just in case I use Mastodon too! fediscience.org/@sergiofrume...

fediscience.org

Sergio Frumento PhD (@sergiofrumento@fediscience.org)

82 Posts, 45 Following, 19 Followers · Research fellow at BodhyLab of University of Pisa (Italy), deepening consciousness and exposure therapies through subliminal stimuli. I'm a psychophysiologist al...

000

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 30/11/2024

Hello BlueSky folk! I'm a post-doc researcher in Psychophysiology of emotions and consciousness, actively doing research on specific phobias 🕷️ social anxiety 😰 PTSD 🫣 and emotionally-subliminal &/or perceptually-subliminal stimuli 😵‍💫 via VR, EEG, EDA, HRV. Here to discuss scientific data & methods!

100

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 30/11/2024

Not even its name is standard! There is no agreement about BAT meaning "Behavioral Avoidance Test" or "Behavioral Approach Task". Virtual reality is often over-rated, but in this very sector it can really be a game changer.

010

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 30/11/2024

Very interesting topic. I'm personally struggling with standard measures of specific phobias and social anxiety, the two disorders I'm mainly working on: for phobic avoidance in particular, I found so many versions of the same (supposed to be) "standardized" protocol! (from doi.org/10.1109/Metr...)

120

Sergio Frumento, PhD https://fediscience.org/@sergiofrumento @sergiofrumento.bsky.social · 26/11/2024

I'm not into the topic and if I was I would be interested in all researchers, regardless of gender... but the name WomWoM sounds great 😆

000