Hybrid neural–cognitive models reveal how memory shapes human reward learning

Journal article

A long-standing challenge for psychology and neuroscience is to understand the transformations by which past experiences shape future behaviour. Reward-guided learning is typically modelled using simple reinforcement learning (RL) algorithms. In RL, a handful of incrementally updated internal variables both summarize past rewards and drive future choice. Here we describe work that questions the assumptions of many RL models. We adopt a hybrid modelling approach that integrates artificial neural networks into interpretable cognitive architectures, estimating a maximally general form for each algorithmic component and systematically evaluating its necessity and sufficiency. Applying this method to a large dataset of human reward-learning behaviour, we show that successful models require independent and flexible memory variables that can track rich representations of the past. Using a modelling approach that combines predictive accuracy and interpretability, these results call into question an entire class of popular RL models based on incremental updating of scalar reward predictions.

Authors: Eckstein, MK; Summerfield, C; Daw, ND; Miller, KJ

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Nature Research
• Journal: Nature Human Behaviour
• Volume: 10
• Issue: 5
• Pages: 972-987
• Publication date: 2026-02-05
• Acceptance date: 2025-09-19
• DOI: 10.1038/s41562-025-02324-0
• EISSN: 2397-3374
• ISSN: 2397-3374
• Language: English
• Keywords: Reinforcement learning; Class (philosophy); Artificial neural network; Cognition; Simple (philosophy); Component (thermodynamics); Computational neuroscience