Lightnews — Scholar-powered news

Reposted by Marco Ciapparelli

Giosuè Baggio @giosuebaggio.bsky.social · 13d

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...

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

academic.oup.com

4 9

Marco Ciapparelli @marcociapparelli.bsky.social · 27d

Happy to share that our work on semantic composition is out now -- open access -- in Cerebral Cortex!

With Marco Marelli (@ercbravenewword.bsky.social), @wwgraves.bsky.social & @carloreve.bsky.social.

doi.org/10.1093/cerc...

3 11

Reposted by Marco Ciapparelli

Melissa Franch, PhD @mfranch.bsky.social · Feb 27

I am incredibly proud to share my first, first-author paper as a postdoc with @benhayden.bsky.social . How does the human hippocampus, known for encoding concepts, represent the meanings of words while listening to narrative speech?
www.biorxiv.org/content/10.1...

12 79 310

Reposted by Marco Ciapparelli

Marc Coutanche @marccoutanche.bsky.social · Aug 26

Here's a set of new results from my lab asking how the brain combines different ideas (concepts)! Now in press at J of Cog Neuro, we looked at how semantic composition (combining different concepts together) shapes brain activity. Preprint here: www.biorxiv.org/content/10.1... #neuroskyence

The Neural Consequences of Semantic Composition

Humans can create completely new concepts through semantic composition. These ‘conceptual combinations’ can be created by attributing the features of one concept to another (e.g., a lemon flamingo mig...

www.biorxiv.org

2 13 34

Marco Ciapparelli @marcociapparelli.bsky.social · Jul 18

Compare concept representations across modalities in unimodal models, using the AlexNet convolutional neural network to represent images and an LLM to represent their captions

Marco Ciapparelli @marcociapparelli.bsky.social · Jul 18

Perform representational similarity analysis to compare how the same concepts are represented across languages (in their correponding monolingual models) and in different layers of LLMs

1

Marco Ciapparelli @marcociapparelli.bsky.social · Jul 18

Replace words with sense-appropriate and sense-inappropriate alternatives in the WiC annotated dataset and look at the effects of context-word interaction on embeddings and surprisal

1

Marco Ciapparelli @marcociapparelli.bsky.social · Jul 18

Extract word embeddings from BERT and inspect how context can modulate their representation. For example, what happens to "fruitless" when we place it in a sentence that points to its typical metaphorical meaning ("vain") as opposed to one where its meaning is literal ("without fruits")?

1

Marco Ciapparelli @marcociapparelli.bsky.social · Jul 18

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 🧵

1 4 9

Reposted by Marco Ciapparelli

Qihong (Q) Lu @qlu.bsky.social · May 26

I’d like to share some slides and code for a “Memory Model 101 workshop” I gave recently, which has some minimal examples to illustrate the Rumelhart network & catastrophic interference :)
slides: shorturl.at/q2iKq
code (with colab support!): github.com/qihongl/demo...

hidden state representation during training

1 10 31

Marco Ciapparelli @marcociapparelli.bsky.social · Apr 28

7/7 Additional compositional representations emerge in left STS and semantic (but not compositional) representations in the left angular gyrus. Check out the preprint for more!
Link to OSF project repo (includes code & masks used):
osf.io/3dnqg/?view_...

Characterizing semantic compositions in the brain: A model-driven fMRI re-analysis

Hosted on the Open Science Framework

osf.io

1

Marco Ciapparelli @marcociapparelli.bsky.social · Apr 28

6/7 We find evidence of compositional representations in left IFG (BA45), even when focusing on a data subset where task didn't require semantic access. We take this to suggest BA45 represents combinatorial info automatically across task demands, and characterize combination as feature intersection

1

Marco Ciapparelli @marcociapparelli.bsky.social · Apr 28

5/7 We conduct confirmatory RSA in four ROIs for which we have a priori hypotheses of ROI-model correspondence (based on what we know of composition in models and what has been claimed of composition in ROIs), and searchlight RSAs in the general semantic network.

1

Marco Ciapparelli @marcociapparelli.bsky.social · Apr 28

4/7 To better target composition beyond specific task demands, we re-analyze fMRI data aggregated from four published studies (N = 85), all employing two-word combinations but differing in task requirements.

1

Marco Ciapparelli @marcociapparelli.bsky.social · Apr 28

3/7 To do so, we use word embeddings to represent single words, multiple algebraic operations to combine word pairs, and RSA to compare representations in models and target regions of interest. Model performance is then related to the specific compositional operation implemented.

1

Marco Ciapparelli @marcociapparelli.bsky.social · Apr 28

2/7 Most neuroimaging studies rely on high-level contrasts (e.g., complex vs. simple words), useful to identify regions sensitive to composition, but less to know *how* constituents are combined (what functions best describe the composition they carry out)

1

Marco Ciapparelli @marcociapparelli.bsky.social · Apr 28

🚨New preprint!🚨
1/7 Happy to share new work with Marco Marelli, @wwgraves.bsky.social & @carloreve.bsky.social! We re-analyze fMRI studies to identify which regions represent semantic information compositionally, and what operations best describe composition.

doi.org/10.1101/2025...

Characterizing semantic compositions in the brain: A model-driven fMRI re-analysis

Semantic composition allows us to construct complex meanings (e.g., ''dog house'', ''house dog'') from simpler constituents (''dog'', ''house''). So far, neuroimaging studies have mostly relied on hig...

doi.org

1 6

Reposted by Marco Ciapparelli

Kanishka Misra 🌊 @kanishka.bsky.social · Apr 2

another day another minicons update (potentially a significant one for psycholinguists?)

"Word" scoring is now a thing! You just have to supply your own splitting function!

pip install -U minicons for merriment

from minicons import scorer
from nltk.tokenize import TweetTokenizer

lm = scorer.IncrementalLMScorer("gpt2")

# your own tokenizer function that returns a list of words
# given some sentence input
word_tokenizer = TweetTokenizer().tokenize

# word scoring
lm.word_score_tokenized(
["I was a matron in France", "I was a mat in France"],
bos_token=True, # needed for GPT-2/Pythia and NOT needed for others
tokenize_function=word_tokenizer,
bow_correction=True, # Oh and Schuler correction
surprisal=True,
base_two=True
)

'''
First word = -log_2 P(word | <beginning of text>)

[[('I', 6.1522440910339355),
('was', 4.033324718475342),
('a', 4.879510402679443),
('matron', 17.611848831176758),
('in', 2.5804288387298584),
('France', 9.036953926086426)],
[('I', 6.1522440910339355),
('was', 4.033324718475342),
('a', 4.879510402679443),
('mat', 19.385351181030273),
('in', 6.76780366897583),
('France', 10.574726104736328)]]
'''

3 7 21

Reposted by Marco Ciapparelli

Matilde Ellen Simonetti @matildellen.bsky.social · Mar 25

🚀 My first PhD paper is out! 🚀
"How do multiple meanings affect word learning and remapping?" was published in Memory & Cognition!
Big thanks to my supervisors and co-authors (Iring Koch & @troembke.bsky.social).
Curious? Read it here: rdcu.be/eeY9o
#CognitivePsychology #WordLearning #Bilingualism

How do multiple meanings affect word learning and remapping?

rdcu.be

2 4 14

Marco Ciapparelli @marcociapparelli.bsky.social · Mar 19

13/n In this context, LLMs flexibility allows to generate representations of possible/implicit meanings, which lead to representational drifts proportional to their plausibility.

Data + code available: osf.io/s5edx/?view_...

Conceptual Combination in Large Language Models: Uncovering Implicit Relational Interpretations in Compound Words with Contextualized Word Embeddings

Hosted on the Open Science Framework

osf.io

Marco Ciapparelli @marcociapparelli.bsky.social · Mar 19

12/n Overall, our approach is consistent with theoretical proposals positing that word (and compound word) meaning should be conceptualized as a set of possibilities that might or might not be realized in a given instance of language use.

1

Marco Ciapparelli @marcociapparelli.bsky.social · Mar 19

11/n Also, bigger model != better: the best layer of BERT consistently outperformed the best layer of Llama. Results align with NLP/cognitive findings showing that LLMs are viable representational models of compound meaning but struggle with genuinely combinatorial stimuli.

1 1

Marco Ciapparelli @marcociapparelli.bsky.social · Mar 19

10/n Expectedly, LLMs vastly outperform DSMs on familiar compounds. Yet, unlike DSMs, LLM performance on novel compounds drops considerably. In fact, looking at novel compounds, some DSMs outperform the best layer of BERT and Llama! (image shows model fit; the lower the better).

1 1

Marco Ciapparelli @marcociapparelli.bsky.social · Mar 19

9/n As predicted, the closer a paraphrase CWE is to the original compound CWE, the more it was deemed plausible by participants. This holds for familiar compounds rated in isolation, familiar compounds rated in sentential contexts, and novel compounds rated in sentential contexts.

1 1

Marco Ciapparelli @marcociapparelli.bsky.social · Mar 19

8/n We reanalyzed possible relation task datasets using BERT-base (widely studied) and Llama-2-13b (representative of more recent, larger, and performant LLMs). As baselines, we used simpler (compositional) DSMs.

1 1