Riccardo Fusaroli

cool! I had somehow missed it, I think, I'll read it carefully :-)

🌍Introducing BabyBabelLM: A Multilingual Benchmark of Developmentally Plausible Training Data!

LLMs learn from vastly more data than humans ever experience. BabyLM challenges this paradigm by focusing on developmentally plausible data

We extend this effort to 45 new languages!

Science as a Conversation When do you know that your paper has done enough?

Thinking of your academic paper as less your magnum opus and more as a line of dialogue in a greater conversation can take the pressure off scientific publishing. Here are my most recent thoughts on "Science as a Conversation":

niklaselmqvist.medium.com/science-as-a...

What Using Our Hands While Speaking Reveals About Our Brains Researchers now have fascinating insights into how our brains function while we communicate — and our hand gestures hold the key.

Happy with this popular science article about our work (with @lindadrijvers.bsky.social and @judithholler.bsky.social), showing that listeners use co-speech hand gestures to predict upcoming meaning ✋ www.medscape.com/viewarticle/...

For an older larger perspective on the research and a talk: bsky.app/profile/fusa...

w @chrismmcox.bsky.social, @yngwienielsen.bsky.social, S.Cho, @rockberta.bsky.social, A.Simonsen, A.Knox, M.Lyons, M.Liberman, C.Cieri, S.Schillinger, A.Lee, A.Hauptmann, K.Tena, @chchatham.bsky.social, J.Miller, J.Pandey, A.Russell, R.Schultz, J.Parish-Morris 11/

Our work calls for a more nuanced approach. Let's move beyond simple group differences and explore the rich interplay of factors that make conversation possible. It's not about being "better" or "worse" at the dance, but about having different styles of dancing. 10/

This means we need to re-evaluate how we study conversational skills. Simply measuring average response latency in a lab with a stranger might emphasize differences, but it tells only a fraction of the story about a child's everyday communicative abilities as they engage their context. 9/

The takeaway: Turn-taking isn't a single, monolithic trait that reflects social competence. It’s a multicomponential skill that emerges from the interplay of individual differences, moment-to-moment predictions, and interpersonal adjustments—all shaped by social context. 8/

Interpersonal mechanisms: Perhaps more crucially, adults tend to speak more predictably when interacting with autistic children. The children might simply be reacting efficiently to this more predictable input, rather than being "interrupting". The dynamic is a property of the dyad. 7/

So why the faster responses in autism? We see two possible complementary explanations. 1) Individual mechanisms: Differences in executive function or impulsivity might lead to a lower threshold for jumping into the conversation, resulting in more overlaps and shorter gaps. 6/

We found that autistic kids adjusted their response timing to their parent's tempo, slowing down or speeding up to match them. But this interpersonal adaptation disappeared with the unfamiliar experimenter. 5/

This "unfamiliarity effect" also revealed how individual skills are deployed. In TD kids, higher social cognition skills were linked to faster responses, but only with the stranger. It seems these skills are recruited more heavily when the interaction is less predictable. 4/

But here's where it gets nuanced. This "faster" pattern was most pronounced when children talked with an unfamiliar experimenter. In the comfort of a conversation with their own parent, the group differences were much smaller. Familiarity scaffolds interaction. 3/

The Development of Turn‐Taking Skills in Typical Development and Autism Social interaction depends on turn-taking and adapting to one's conversational partner, yet little is known about the typical and atypical development of these abilities. We investigated this in a lo...

As in a previous paper (onlinelibrary.wiley.com/doi/10.1111/...) autistic children responded faster than TD children. They produced more overlaps & shorter pauses between turns. These finding directly challenges the common idea that a social competence deficit leads to slower, delayed responses. 2/

Social Context Matters for Turn‐Taking Dynamics: A Comparative Study of Autistic and Typically Developing Children Engaging in fluent conversation is a surprisingly complex task that requires interlocutors to promptly respond to each other in a way that is appropriate to the social context. In this study, we dise...

Conversational turn-taking feels effortless, but it's a complex dance. We find social context—who you're talking to and what you're talking about—fundamentally changes conversational dynamics in both autistic & TD children. 1/

onlinelibrary.wiley.com/doi/10.1111/... w @chrismmcox.bsky.social

really cool work, nice to see the work is starting to get out :-)

I haven't given any news a while, I've been nose deep into this novel preprint with my excellent collaborators @stepalminteri.bsky.social, @urihertz.bsky.social and Bahador Bahrami: "Uncovering the semantics of teaching in
experiential learning with Large Language Models".
doi.org/10.31234/osf...

The first publication of the #ERC project ‘LaDy’ is a fact and it’s an important one I think:

We show that word processing and meaning prediction is fundamentally different during social interaction compared to using language individually!
👀 short 🧵/1

psycnet.apa.org/fulltext/202...
#OpenAccess

YouTube video by Wim Pouw EnvisionBOX overview2025

www.envisionbox.org has been shortlisted for the Leo Waaijers Open Science price: ukb.nl/en/news/shor...

@babajideowoyele.bsky.social @jamestrujillo.bsky.social @sarkadava.bsky.social @DavideAhmar @acwiek.bsky.social

Amazing Markus Küpper made an animated video:
www.youtube.com/watch?v=HduI...

The proper looks of autumn

In Ch 19 (nyu-cdsc.github.io/learningr/as...) of his 2nd edition, Kruschke used *residual* SD as a standardizer for group differences from a multilevel ANCOVA. Is there any precedent for using a *residual* SD as a standardizer for a standardized mean difference effect size? #RStats

because the internet can still be fun, I found out that this was a swiss watch meant for the US marker (uncalibrated, to avoid US tariffs (ah!) on precise watches, apparently), and more recent than it looks, but mimicking older styles.

The kids’ find of the day was my great great grandfather’s clock in silver. About 40 years ago I inherited it, it didn’t work and I tried to open it to fix it. Still open since and not working :-)

The social, decoupled self: interpersonal synchronization of breathing alters intrapersonal cardiorespiratory coupling People synchronize their periodic behavioural and physiological rhythms with each other during social interaction. While this interpersonal synchronization has largely been associated with positive ef...

🫁❤️New preprint out: The social, decoupled self

We show effects of interpersonal synchronization of physiological rhythms on intrapersonal cardiorespiratory coupling: when we sync our breathing, our breathing–heart rhythms decouple, with a perturbed phase-relationship
www.biorxiv.org/content/10.1...

thanks! very interesting indeed! :-)

Oh, super cool! So, you’d imagine language mix specific cues/strategies? So that’d eg be different in Vanuatu? Or do you think there could be more generalizable curs?

How do multilingual babies acquire their languages? In Africa, multilingualism is the norm. In our new study with 9-11 month old Ghanaian babies learning up to 5 languages simultaneously, we found that they were able to recognize words in text passages in Akan, one of their languages.

I second that!

tl;dr Collinearity is a form of lack of information that is appropriately reflected in the output of your statistical model. When collinearity is associated with interpretational difficulties, these difficulties aren’t caused by the collinearity itself. Rather, they reveal that the model was poorly specified (in that it answers a question different to the one of interest), that the analyst overly focuses on significance rather than estimates and the uncertainty about them or that the analyst took a mental shortcut in interpreting the model that could’ve also led them astray in the absence of collinearity. If you do decide to “deal with” collinearity, make sure you can still answer the question of interest.

Was asked about collinearity again, so here's Vahove's 2019 post on why it isn't a problem that needs a solution. Design the model(s) to answer a formal question and free your mind janhove.github.io/posts/2019-0...