Eric Elmoznino

Can In-Context Learning Support Intrinsic Curiosity?

Eric Elmoznino

Sangnie Bhardwaj

Johannes Von Oswald

Rajai Nasser

Blaise Agüera y Arcas

João Sacramento

Rif A. Saurous

Guillaume Lajoie

Effective machine learning depends not only on how we model data, but also on what data we choose to collect. While large sequence models ha… (see more)ve revolutionized data modeling, the problem of automated data selection, or"intrinsic curiosity", remains a significant challenge. Classic approaches incentivize exploration by rewarding an agent based on its"learning progress", which measures how much a newly acquired observation improves a world model's predictive ability. However, evaluating these rewards traditionally requires expensive inner loops of gradient descent updates within each trajectory, rendering them computationally impractical at scale. In this work, we investigate whether the emergent in-context learning (ICL) capabilities of sequence models can eliminate this bottleneck by serving as immediate, update-free world models. Specifically, we evaluate whether an exploration policy can be trained to maximize learning progress, using solely the prediction errors and counterfactual context manipulations of an in-context learner. We first prove that in general Markov decision processes, this is in fact impossible in an unbiased way: the resulting intrinsic rewards either suffer from nuisance terms that bias their estimation of true learning progress, or they cannot be implemented using an in-context learner's prediction errors. Conversely, we prove a positive result for a broad subclass of non-temporal settings, encompassing active learning and Bayesian Experimental Design: here, ICL-derived rewards successfully bound and asymptotically converge to the true learning progress. We corroborate our theory with controlled experiments across continuous and symbolic environments, demonstrating that our ICL-driven framework successfully trains curious data-collection policies that explore optimally.

2026-06-16

arXiv (preprint)

doi.org

arxiv.org

Simplifying the Modeling of Arbitrary Conditionals in Natural Language

Causal Transformers model sequences through an autoregressive factorization of the joint distribution, which enables efficient left-to-right… (see more) decoding and conditional likelihood computation. However, they cannot tractably sample from or evaluate arbitrary conditionals -- e.g., a block of text conditioned on past and future tokens. Recent work aims to solve this problem through novel architectures, but they often lead to sub-optimal modeling of such conditionals and degraded generations. We propose Arbitrary Conditionals GPT (AC-GPT) which introduces a simple modification to standard causal Transformers to enable evaluating and sampling from arbitrary conditionals -- including past, future, and mixed contexts -- within a single forward pass. Unlike prior approaches, our method preserves the standard left-to-right ordering and next-token prediction objective essential for both strong performance and efficient training on natural language. Crucially, this compatibility allows existing LLMs to be fine-tuned for arbitrary conditioning. Our empirical results indicate that our method outperforms baselines on modeling arbitrary conditionals, without degrading standard left-to-right performance.

2026-06-11

arXiv (preprint)

doi.org

arxiv.org

A Compression Perspective on Simplicity Bias

Mizu Nishikawa-Toomey

Sarthak Mittal

Guillaume Lajoie

Dhanya Sridhar

Deep neural networks exhibit a simplicity bias, a well-documented tendency to favor simple functions over complex ones. In this work, we cas… (see more)t new light on this phenomenon through the lens of the Minimum Description Length principle, formalizing supervised learning as a problem of optimal two-part lossless compression. Our theory explains how simplicity bias governs feature selection in neural networks through a fundamental trade-off between model complexity (the cost of describing the hypothesis) and predictive power (the cost of describing the data). Our framework predicts that as the amount of available training data increases, learners transition through qualitatively different features -- from simple spurious shortcuts to complex features -- only when the reduction in data encoding cost justifies the increased model complexity. Consequently, we identify distinct data regimes where increasing data promotes robustness by ruling out trivial shortcuts, and conversely, regimes where limiting data can act as a form of complexity-based regularization, preventing the learning of unreliable complex environmental cues. We validate our theory on a semi-synthetic benchmark showing that the feature selection of neural networks follows the same trajectory of solutions as optimal two-part compressors.

2026-03-25

arXiv (preprint)

doi.org

arxiv.org

Identifying indicators of consciousness in AI systems

Patrick Butlin

Robert Long

Tim Bayne

Yoshua Bengio

Jonathan Birch

David Chalmers

Axel Constant

George Deane

Eric Elmoznino

Stephen M. Fleming

Xu Ji

Ryota Kanai

Colin Klein

Grace Lindsay

Matthias Michel

Liad Mudrik

Megan A.K. Peters

Eric Schwitzgebel

Jonathan Simon

Rufin Vanrullen

2025-11-09

Trends in Cognitive Sciences (published)

doi.org

Towards a Formal Theory of Representational Compositionality

2025-10-05

Proceedings of the 42nd International Conference on Machine Learning (published)

proceedings.mlr.press

Illusions of AI consciousness

Yoshua Bengio

Eric Elmoznino

The belief that AI is conscious is not without risk

2025-09-10

Science (published)

doi.org

In-context learning and Occam's razor

A central goal of machine learning is generalization. While the No Free Lunch Theorem states that we cannot obtain theoretical guarantees fo… (see more)r generalization without further assumptions, in practice we observe that simple models which explain the training data generalize best: a principle called Occam's razor. Despite the need for simple models, most current approaches in machine learning only minimize the training error, and at best indirectly promote simplicity through regularization or architecture design. Here, we draw a connection between Occam's razor and in-context learning: an emergent ability of certain sequence models like Transformers to learn at inference time from past observations in a sequence. In particular, we show that the next-token prediction loss used to train in-context learners is directly equivalent to a data compression technique called prequential coding, and that minimizing this loss amounts to jointly minimizing both the training error and the complexity of the model that was implicitly learned from context. Our theory and the empirical experiments we use to support it not only provide a normative account of in-context learning, but also elucidate the shortcomings of current in-context learning methods, suggesting ways in which they can be improved. We make our code available at https://github.com/3rdCore/PrequentialCode.

2025-07-14

International Conference on Machine Learning (Accept (poster))

doi.org

proceedings.mlr.press

Next-Token Prediction Should be Ambiguity-Sensitive: A Meta-Learning Perspective

The rapid adaptation ability of auto-regressive foundation models is often attributed to the diversity of their pre-training data. This is b… (see more)ecause, from a Bayesian standpoint, minimizing prediction error in such settings requires integrating over all plausible latent hypotheses consistent with observations. While this behavior is desirable in principle, it often proves too ambitious in practice: under high ambiguity, the number of plausible latent alternatives makes Bayes-optimal prediction computationally intractable. Cognitive science has long recognized this limitation, suggesting that under such conditions, heuristics or information-seeking strategies are preferable to exhaustive inference. Translating this insight to next-token prediction, we hypothesize that low- and high-ambiguity predictions pose different computational demands, making ambiguity-agnostic next-token prediction a detrimental inductive bias. To test this, we introduce MetaHMM, a synthetic sequence meta-learning benchmark with rich compositional structure and a tractable Bayesian oracle. We show that Transformers indeed struggle with high-ambiguity predictions across model sizes. Motivated by cognitive theories, we propose a method to convert pre-trained models into Monte Carlo predictors that decouple task inference from token prediction. Preliminary results show substantial gains in ambiguous contexts through improved capacity allocation and test-time scalable inference, though challenges remain.

2025-06-18

ArXiv (preprint)

doi.org

arxiv.org

Next-Token Prediction Should be Ambiguity-Sensitive : A Meta-Learing Perspective

2025-06-10

ICML.cc/2025/Workshop/ES-FoMo-III (published)

openreview.net

Does learning the right latent variables necessarily improve in-context learning?

Large autoregressive models like Transformers can solve tasks through in-context learning (ICL) without learning new weights, suggesting ave… (see more)nues for efficiently solving new tasks. For many tasks, e.g., linear regression, the data factorizes: examples are independent given a task latent that generates the data, e.g., linear coefficients. While an optimal predictor leverages this factorization by inferring task latents, it is unclear if Transformers implicitly do so or if they instead exploit heuristics and statistical shortcuts enabled by attention layers. Both scenarios have inspired active ongoing work. In this paper, we systematically investigate the effect of explicitly inferring task latents. We minimally modify the Transformer architecture with a bottleneck designed to prevent shortcuts in favor of more structured solutions, and then compare performance against standard Transformers across various ICL tasks. Contrary to intuition and some recent works, we find little discernible difference between the two; biasing towards task-relevant latent variables does not lead to better out-of-distribution performance, in general. Curiously, we find that while the bottleneck effectively learns to extract latent task variables from context, downstream processing struggles to utilize them for robust prediction. Our study highlights the intrinsic limitations of Transformers in achieving structured ICL solutions that generalize, and shows that while inferring the right latents aids interpretability, it is not sufficient to alleviate this problem.

2025-04-30

International Conference on Machine Learning (poster)

doi.org

proceedings.mlr.press

Multi-Agent Cooperation Through Learning-Aware Policy Gradients

Alexander Meulemans

Seijin Kobayashi

Johannes Von Oswald

Nino Scherrer

Eric Elmoznino

Blake A. Richards

Guillaume Lajoie

Blaise Agüera y Arcas

João Sacramento

Self-interested individuals often fail to cooperate, posing a fundamental challenge for multi-agent learning. How can we achieve cooperation… (see more) among self-interested, independent learning agents? Promising recent work has shown that in certain tasks cooperation can be established between learning-aware agents who model the learning dynamics of each other. Here, we present the first unbiased, higher-derivative-free policy gradient algorithm for learning-aware reinforcement learning, which takes into account that other agents are themselves learning through trial and error based on multiple noisy trials. We then leverage efficient sequence models to condition behavior on long observation histories that contain traces of the learning dynamics of other agents. Training long-context policies with our algorithm leads to cooperative behavior and high returns on standard social dilemmas, including a challenging environment where temporally-extended action coordination is required. Finally, we derive from the iterated prisoner's dilemma a novel explanation for how and when cooperation arises among self-interested learning-aware agents.

2025-01-21

ICLR.cc/2025/Conference (poster)

doi.org

openreview.net

A Complexity-Based Theory of Compositionality

2024-10-17

ArXiv (preprint)

doi.org