That claim matters because it reframes plasticity as one of several mechanisms the brain might use to be flexible. If a system is trained well enough, it may carry a repertoire of computational routines that can be recombined to meet new demands. For people who study learning and development, that suggests experiments should look not only for slow structural change but also for rapid shifts in how existing circuitry is deployed. The distinction affects how we think about rehabilitation, education, and inclusive design: strengthening computational modes could complement traditional approaches that aim to change connections.
Curious about what this means for human potential? The paper invites us to test whether human learners rely more on dynamic computation than previously thought and whether training regimes can cultivate that capability. Follow the link to explore the methods and examples the authors used, and consider how those ideas might reshape approaches to growth, adaptability, and access across different learners and communities.
Abstract
The human brain is an adaptive learning system that can generalize to new tasks and unfamiliar environments. The traditional view is that such adaptive behavior requires a structural change of the learning system (e.g., via neural plasticity). In this work, we use artificial neural networks, specifically large language models (LLMs), to challenge this traditional view and suggest that such adaptive behavior can be also achieved solely through computation if the learning system is sufficiently trained. We focus on statistical learning paradigms. These require identifying underlying regularities in seemingly arbitrary word sequences and are primarily considered to require neural plasticity. LLMs can capture arbitrary structures without requiring weight adaptation, even though these structures carry no semantic content and lie far from the natural language distribution on which the models were trained. Our work provides novel insights into the role of plasticity in learning, demonstrating that sufficiently trained learning systems are highly flexible, adapting to new tasks and environments solely through computation, much more than previously acknowledged. Furthermore, our work opens the door for future research to use deep learning models to conjure hypotheses about the brain.
