Structuring the World: Statistical Learning From the Top and From the Bottom.
8 affiliations
- 1Basque Center on Cognition, Brain and Language, Donostia-San Sebastian, Spain.
- 2Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany.
- 3Ikerbasque, Basque Foundation for Science, Bilbao, Spain.
- 4Department of Cognitive Science, Central European University, Vienna, Austria.
- 5Department of Engineering, University of Cambridge, Cambridge, UK.
- 6Department of Physiology, Development and Neuroscience, University of Cambridge, Cambridge, UK.
- 7Neuroscience Research Center, Charité Universitätsmedizin Berlin, Berlin, Germany.
- 8Bernstein Center for Computational Neuroscience, Berlin, Germany.
Abstract
Statistical learning is the brain's primary strategy for transforming continuous, noisy, and ambiguous sensory inputs into structured internal representations. This process is studied through a diverse array of perspectives spanning different species, sensory systems, levels of analysis, and methodological frameworks. Here, we present four such perspectives. Two of them approach the problem from the bottom, examining the neural substrates of the biological implementation. The first line describes implicit contextual learning in the rodent auditory midbrain, showing how neuronal frequency tuning adapts to predictable statistical regularities over days. The second line shows that explicit knowledge modulates belief updating as early as the human auditory midbrain and thalamus, revealing a functionally specialized subcortical inferential hierarchy. The other two research lines approach it from the top, examining the computations the brain performs. The third line compares statistical computations across species and human development, uncovering fundamental differences in probabilistic chunking strategies between honeybees, human infants, and adults. The fourth line provides a computational framework for continual learning in which experience is organized into structured internal models through ongoing contextual inference. We then examine what each line can contribute to the others. We argue that, if the brain evolved statistical learning under pressure to behave optimally, this optimality must have been achieved globally through the constraints and synergies that emerge from the interplay between subsystems and levels of analysis. The neuronal mechanisms implementing a computation are shaped by the computation they perform, while that computation is itself constrained by the available physiology: a mutual dependence that can only be understood by bridging across perspectives.
Related
- Language learning in a stressful world: individual differences in cardiac responses to stress and recovery predict statistical learning
- Investigating the role of the hippocampus in the implicit and explicit expression of statistical learning in patients with temporal lobe epilepsy
- Common attentional mechanisms, but distinct computational dynamics, across Statistical and Reward Learning
- Multi-Angle Lateral-View Breast Tissue Classification in Thermographic Images Using Deep Reinforcement Learning and Statistical Shape Modelling
- Deep-learning-based fMRI decoding of real-world size for hand-held objects
BibTeX
@article{tabas2026structuring,
title = {{Structuring the World: Statistical Learning From the Top and From the Bottom.}},
author = {Alejandro Tabas and József Fiser and Máté Lengyel and Livia de Hoz},
journal = {The European journal of neuroscience},
year = {2026},
month = oct,
doi = {10.1111/ejn.70712},
url = {https://europepmc.org/article/MED/42905041},
note = {CC BY},
}
RIS
TY - JOUR
TI - Structuring the World: Statistical Learning From the Top and From the Bottom.
AU - Alejandro Tabas
AU - József Fiser
AU - Máté Lengyel
AU - Livia de Hoz
PY - 2026
DA - 2026/10/01
JO - The European journal of neuroscience
DO - 10.1111/ejn.70712
UR - https://europepmc.org/article/MED/42905041
AB - Statistical learning is the brain's primary strategy for transforming continuous, noisy, and ambiguous sensory inputs into structured internal representations. This process is studied through a diverse array of perspectives spanning different species, sensory systems, levels of analysis, and methodological frameworks. Here, we present four such perspectives. Two of them approach the problem from the bottom, examining the neural substrates of the biological implementation. The first line describes implicit contextual learning in the rodent auditory midbrain, showing how neuronal frequency tuning adapts to predictable statistical regularities over days. The second line shows that explicit knowledge modulates belief updating as early as the human auditory midbrain and thalamus, revealing a functionally specialized subcortical inferential hierarchy. The other two research lines approach it from the top, examining the computations the brain performs. The third line compares statistical computations across species and human development, uncovering fundamental differences in probabilistic chunking strategies between honeybees, human infants, and adults. The fourth line provides a computational framework for continual learning in which experience is organized into structured internal models through ongoing contextual inference. We then examine what each line can contribute to the others. We argue that, if the brain evolved statistical learning under pressure to behave optimally, this optimality must have been achieved globally through the constraints and synergies that emerge from the interplay between subsystems and levels of analysis. The neuronal mechanisms implementing a computation are shaped by the computation they perform, while that computation is itself constrained by the available physiology: a mutual dependence that can only be understood by bridging across perspectives.
SN - 0953-816X
SN - 1460-9568
ER -