Program

09:10–09:20 Opening Remarks

09:20–10:20 Invited Talk: Dr. Claire Stevenson

Learning to solve analogies: why do children excel where AI models fail?

Recent work with large language models (LLMs) concludes that analogical reasoning, using what you know about one thing to infer knowledge about a new, somehow related instance, has emerged in these systems. My lab has conducted a series of behavioural and mechanistic‑interpretability studies to investigate whether analogical reasoning has indeed emerged, and, if so, whether the developmental phenomena resemble those of humans or follow a different trajectory. We provide evidence of similarities in children's and LLMs' development in learning to solve analogies, while also highlighting key differences. I will propose a theory of how LLMs' reasoning abilities are developing and conclude with a discussion of developmental insights that could help AI models achieve human‑like analogical reasoning.

10:30–11:40 Poster Session 1

• • • • • • Transformer Attention as a Unified Model of Encoding and Retrieval in Human Sentence Processing
            Dan Parker

          • ChineseDevBench: A Chinese Developmental Benchmark for Language Development
            Shaonan Wang, Yiwen Wu, Na Li, Zesheng Chen, Gan Wang, Shuchen Zhang, Xin Sun, Luan Li and Yaran Chen

          • Predicting Sentence Acceptability Judgments in Multimodal Contexts
            Hyewon Jang, Nikolai Ilinykh, Sharid Loaiciga, Jey Han Lau and Shalom Lappin

          • ToM in LLM is not ToM, but a Pragmatic Effect
            Agnese Lombardi and Alessandro Lenci

          • What Kind of Language is Easy to Language-Model Under Curriculum Learning?
            Nadine El-Naggar, Tatsuki Kuribayashi and Ted Briscoe

          • Modeling semantic association in self-paced reading with language model embeddings
            Sara Møller Østergaard, Kenneth Enevoldsen, Afra Alishahi and Bruno Nicenboim

          • Quantifying the Pragmatics of Discourse Markers: A Surprisal-Based Analysis of Taiwanese "Ah"
            Ri-Sheng Huang

          • Acoustic Representations Support Statistical Learning of Syllable Sequences: A Computational Study
            Nika Jurov and Thomas Schatz

          • Simulating Cross-Linguistic Influence in Bilingual Reading - a Knowledge Distillation Approach
            Irene E. Winther and Stefan L. Frank

          • Investigating Brain-LLM Alignment with Category-Deprived Text
            Shantanu Nath, Marco Tettamanti and Roberto Zamparelli

11:40–13:00 Oral Session 1

• (11:40–12:00) Character-aware Transformers Learn an Irregular Morphological Pattern Yet None Generalize Like Humans
  Akhilesh Kakolu Ramarao, Kevin Tang and Dinah Baer-Henney

• (12:00–12:20) Comparing Transformer Model Interpretability with Human Cognition: A Dual Analysis of Attention and Attribution
  Lingchen Kong, Jinnie Shin and Pavlo Antonenko

• (12:20–12:40) Correlating Language Model Surprisal With Cloze and Plausibility: Getting the Best of Both Measures
  Kate Rebecca Belcher and Matthew Crocker

• (12:40–13:00) Is Cross-Lingual Transfer in Bilingual Models Human-Like? A Study with Overlapping Word Forms in Dutch and English
  Iza Škrjanec, Irene Elisabeth Winther, Vera Demberg and Stefan L. Frank

13:00–14:00 Lunch Break

14:00–15:00 Invited Talk: Dr. Cory Shain

Language processing at the computational level and beyond

In this talk, I will survey insights from my group into language processing across Marr's levels of analysis: computational, algorithmic, and implementational. I will first show that the computational-level notion of optimality correctly predicts key empirical properties of human language learning and processing, and I will argue that these results place a burden of proof on algorithmic-level theories. I will then review emerging evidence of deviations from optimality, which plausibly reveal the inner workings of the language processing system at an algorithmic level. Finally, I will review insights from the implementational level about the functional brain networks that likely support these processing abilities.

15:00–16:00 Oral Session 2

• (15:00–15:20) Brain-to-Text Decoding with Brain Atlases and Brain Foundation Models
  Haruka Akama, Ryo Yoshida, Max Müller-Eberstein and Yohei Oseki

• (15:20–15:40) Floating or Suggesting Ideas? A Large-Scale Contrastive Analysis of Metaphorical and Literal Verb–Object Constructions
  Prisca Piccirilli, Alexander Fraser and Sabine Schulte im Walde

• (15:40–15:50) Social Meaning in Large Language Models: Structure, Magnitude, and Pragmatic Prompting
  Roland Mühlenbernd

• (15:50–16:00) Semantic Norms and Structural Relations in Chinese Semantic Composition: A Dataset and Analysis
  He Zhou, Emmanuele Chersoni and Yu-Yin Hsu

16:00–17:00 Poster Session 2

• Headlines You Won’t Forget: Can Pronoun Insertion Increase Memorability?
  Selina Meyer, Magdalena Abel and Michael Roth

• Aggregated Transformer Attention Measures Predict Reading Times Beyond Surprisal
  Lukas Mielczarek and Laura Kallmeyer

• ‘Layer su Layer’: Identifying and Disambiguating the Italian NPN Construction in BERT’s family
  Greta Gorzoni, Ludovica Pannitto and Francesca Masini

• Teaching LLMs to unveil tendentious implicit contents of Italian political communication
  Walter Paci, Lorenzo Gregori and Alessandro Panunzi

• Examining Algebraic Recombination for Compositional Generalisation
  Joaquin Cardona Ruiz, Antske Fokkens and Lucia Donatelli

• Can LLMs Simulate Human Behavioral Variability? A Case Study in the Phonemic Fluency Task
  Mengyang Qiu, Zoe Brisebois and Siena Sun

• Translation from the Information Bottleneck Perspective: an Efficiency Analysis of Spatial Prepositions in Bitexts
  Antoine Taroni, Ludovic Moncla and Frederique Laforest

• Causal Inferences Are Driven by Noun Concept Specificity Evidence from Self-Paced Reading and Large Language Model Surprisal
  Fabian Schlotterbeck and Raphael Barth

• Polar Questions in SPA–TTR: Linking Dialogue, Acquisition, and Neurosemantics
  Jonathan Ginzburg, Shiyun Dong, Robin Cooper, Andy Luecking and Staffan Larsson

• Generics as Useful Defeasible Rules of Inference
  Justin Helms

17:00–17:50 Oral Session 3

• (17:00–17:10) Topic-context Dependency on Continuous Semantic Reconstruction of Language from fMRI Signals
  Fermin Travi, Agustin Delmagro, Diego Fernandez Slezak, Bruno Bianchi and Juan E. Kamienkowski

• (17:10–17:20) Disentanglement and Compositionality of Letter Identity and Letter Position in Variational Auto-Encoder Vision Models
  Bruno Bianchi, Aakash Agrawal, Stanislas Dehaene, Emmanuel Chemla and Yair Lakretz

• (17:20–17:30) Plausibility as Commonsense Reasoning: Humans Succeed, Large Language Models Do not
  Sercan Karakas

• (17:30–17:40) Benchmarking Source-Sensitive Reasoning in Turkish: Humans and LLMs under Evidential Trust Manipulation
  Sercan Karakas and Yusuf Şimşek

• (17:40–17:50) Toward Cognitive Alignment in Large Language Models: Integrating Linguistic Theory and Human Data
  Wajdi Zaghouani

17:50–18:00 Closing Remarks