Cheng-Yu Hsieh

Join our journal club next Monday! We're happy to have @cyhsieh.bsky.social to discuss "What Chinese Compounds Tell Us About Reading"! Scan the QR code to register or click here: forms.gle/JV5cp1RHHaH2...

Compositionality in the semantic network: a model-driven representational similarity analysis Abstract. Semantic composition allows us to construct complex meanings (e.g., “dog house”, “house dog”) from simpler constituents (“dog”, “house”). Neuroim

Important fMRI/RSA study by @marcociapparelli.bsky.social et al. Compositional (multiplicative) representations of compounds/phrases in left IFG (BA45), mSTS, ATL; left AG encodes constituents, not their composition, weighing the right element more, vice versa IFG 🧠🧩
academic.oup.com/cercor/artic...

Diagram showing three ways to control brain activity with a visual prosthesis. The goal is to match a desired pattern of brain responses. One method uses a simple one-to-one mapping, another uses an inverse neural network, and a third uses gradient optimization. Each method produces a stimulation pattern, which is tested in both computer simulations and in the brain of a blind participant with an implant. The figure shows that the neural network and gradient methods reproduce the target brain activity more accurately than the simple mapping.

👁️🧠 New preprint: We demonstrate the first data-driven neural control framework for a visual cortical implant in a blind human!

TL;DR Deep learning lets us synthesize efficient stimulation patterns that reliably evoke percepts, outperforming conventional calibration.

www.biorxiv.org/content/10.1...

Thanks for the invite. Looking forward to seeing many of you online!

2 Lecturer (Assistant Prof) positions available @yorkpsychology.bsky.social! Come join our department!

#neuroskyence #cognition #psychscisky #neurojobs

jobs.york.ac.uk/vacancy/lect...

JOB ALERT: Computational Cognitive Neuroscience Postdoc position in Osaka, Japan! Possible start in October 2025 (contact me ASAP), or from April 2026. PLEASE REPOST! #postdocjobs #neuroskyence #neuroscience #psychscisky #compneurosky #neurojobs 1/

📣 Save the date 🗓️ to present your exciting statistical learning research at the 6th Interdisciplinary Advances in Statistical Learning Conference June 10-12 2026 in San Sebastián 🇪🇸

Keynotes by
@jennysaffran.bsky.social
@noranewcombe.bsky.social
@pyoudeyer.bsky.social

More info to follow #IASL26

I'm sharing a Colab notebook on using large language models for cognitive science! GitHub repo: github.com/MarcoCiappar...

It's geared toward psychologists & linguists and covers extracting embeddings, predictability measures, comparing models across languages & modalities (vision). see examples 🧵

Psycholinguistic Databases, Stimuli, Utilities — Concepts & Cognition Laboratory

Updated our lab's psycholinguistic database page to include Kathy Rastle et al's new web interface for the Children and Young Peoples Books Lexicon (CYP-LEX). Check it out! Give me a holler if you want us to link to your dataset or know of others I've missed. www.reilly-coglab.com/data

** New resource ** We analysed the characteristics of words in 1200 books suitable for children and young people. Properties of each word (frequencies, etc) are now available in an interactive website.
cyp-lex.rastlelab.com

What can children learn about morphology from reading for fun? A key part of becoming a skilled reader is understanding how words are built — that is, how small parts of words that carry meaning come together to form words. For example, the word unhappy is made u...

What can children learn about morphology when they read for pleasure? We analysed the words in 1200 books suitable for children and young people to find out! Read the blog post here: www.rastlelab.com/post/what-ca...

A new preprint by myself!

I drafted a journal club article arguing that intrinsically photosensitive retinal ganglion cells (ipRGCs) are responsible for more than just non-image-forming functions

osf.io/preprints/os...

Language

Thanks to Evan Kidd, @mcxfrank.bsky.social & @asifamajid.bsky.social (Eds) for organizing the awesome *Open* Encyc. of Cog. Sci. and entrusting me with/ the honor of penning the entry on "Language" (in < 3500 words, so do see other entries for much, MUCH more!)
oecs.mit.edu/pub/ho9e9c80...

3 hour open-science workshop I gave at Vin University Hanoi now online:
"Credible science, evaluating science, and the need for a science reform"

UG class level super basic research methods intro and my take on some of the main challenges we're facing.

Details 👇🧵

Tens of thousands of psychological measures have been created, but 79% are not reused more than twice.

In a new article in NHB lead by @faridanvari.bsky.social and @ruben.the100.ci we quantify this fragmentation in psychology's measures and propose solutions.

www.nature.com/articles/s41...

Let your research do the talking!
We’re awarding 5 🦍 grants ($1,770 each)—could one be yours?

Submit a 500-word research proposal + overview on how one could support your next study for a chance to win!

If you're exploring #language, #linguistics or anything in between—apply now!
➡️ bit.ly/3FeV9rX

Conceptual Combination in Large Language Models: Uncovering Implicit Relational Interpretations in Compound Words With Contextualized Word Embeddings Large language models (LLMs) have been proposed as candidate models of human semantics, and as such, they must be able to account for conceptual combination. This work explores the ability of two LLM...

1/n Happy to share a new paper with Calogero Zarbo & Marco Marelli! How well do LLMs represent the implicit meaning of familiar and novel compounds? How do they compare with simpler distributional semantics models (DSMs; i.e., word embeddings)?
doi.org/10.1111/cogs...

But we can do something! If you also work on language processing or first language acquisition focussing on cross-linguistic variation, then submit an abstract to *Crosslinguistic Perspectives on Processing and Learning 2025* #X_PPL2025!

Call for abstracts: easyabs.linguistlist.org/conference/X...

In politics, people are not always truth seekers. Often, we reach conclusions because they fit with preferred political narratives: we are biased by ideological motivations. But are some people better equipped to overcome such bias than others? 🚨 Find out in the first pre-print from my PhD 🚨

Thank you Olaf ☺️☺️

More important, we speculate this compositional process may help readers segment words in the writing system that lacks spacing to indicate word boundaries (e.g. Chinese, Japanese, Thai).

The successful implementation of the computational model suggests that readers can learn meaningful functions of constituents for meaning composition, through their experience with compound words.

The findings suggest that constituent meanings are routinely combined in Chinese compound processing, even in tasks that do not explicitly require meaning combination (i.e. lexical decision).

We investigated this issue by testing whether the ease of integrating constituent meanings impacts compound processing, using computational modelling.

We found this measure predicted sensibility judgements on novel compounds, lexical decision on familiar and unfamiliar compounds.

Research on compound word processing has focused primarily on Germanic languages, where the functions of the constituents are very predictable. Over 85% of the compounds are structured such that the 1st constituent determines the nature of the 2nd (e.g. teacup=cup FOR tea).

Compositional processing in the recognition of Chinese compounds: Behavioural and computational studies - Psychonomic Bulletin & Review Recent research has shown that the compositional meaning of a compound is routinely constructed by combining meanings of constituents. However, this body of research has focused primarily on Germanic ...

1st post here! Excited to share this work with Marelli & @kathyrastle.bsky.social. We've found readers "routinely" combine constituent meanings for Chinese compound meaning, despite variability in constituent meaning and word structure, even when they're not asked to. See threads👇 for more details: