NeuroAI is an emerging interdisciplinary field that seeks to bridge neuroscience and artificial intelligence (AI) to mutually advance both domains. It operates on a two-way street:
-
Neuroscience for AI: Using insights from the brain's structure, function, and learning mechanisms to inspire the development of more capable, energy-efficient, and robust AI systems.
-
AI for Neuroscience: Applying powerful AI tools and computational models (such as deep neural networks) to analyze vast amounts of complex neural data, leading to a deeper understanding of how the brain works.
This reading list aims to keep updating on latest NeuroAI papers.
This version selects some entries by theme/topic for easier reference. For the more frequently updated chronological version, see README-chronological.md.
- Neural Representation, Geometry, and Manifolds
- Memory: Working, Episodic, and Associative
- Transformers, Attention, and Large Language Models
- Predictive Coding, Energy Efficiency, and Neural Computation
- Causality, Reasoning, and Cognitive Science
- Reinforcement Learning, Planning, and Control
- Robotics, Embodiment, and Spiking Networks
- Neuroscience Reviews, Critiques, and Meta
- Mathematics, Statistics, and Methodology
- Interpretability, Mechanistic and Explanatory
- Benchmark Datasets, Cognitive Tests, and Evaluation
- Books, Textbooks, and Lecture Notes
- Miscellaneous, Philosophy, and Other
- Neural tuning and representational geometry
- Three aspects of representation in neuroscience
- Distributed representations of words and phrases and their compositionality
- Shared Representational Geometry Across Neural Networks
- The Geometry of Concepts: Sparse Autoencoder Feature Structure
- Tracking the topology of neural manifolds across populations
- Manifolds: A Gentle Introduction
- Dimension Reduction using Isomap
- A neural manifold view of the brain
- A formal model of capacity limits in working memory
- Representation and computation in visual working memory
- The capacity of visual working memory for features and conjunctions
- The Distributed Nature of Working Memory
- Theories of Error Back-Propagation in the Brain
- On prefrontal working memory and hippocampal episodic memory: Unifying memories stored in weights and activity slots
- Adaptive chunking improves effective working memory capacity in a prefrontal cortex and basal ganglia circuit
- Attractor dynamics with activity-dependent plasticity capture human working memory across time scales
- A generative model of memory construction and consolidation
- Abstract representations emerge in human hippocampal neurons during inference
- Memory in humans and deep language models: Linking hypotheses for model augmentation
- Are Emergent Abilities of Large Language Models a Mirage?
- Mechanism for feature learning in neural networks and backpropagation-free machine learning models
- Transformer as a hippocampal memory consolidation model based on NMDAR-inspired nonlinearity
- Attention is not all you need anymore
- The Annotated Transformer
- Attention and Memory in Deep Learning
- Transformer Mechanisms Mimic Frontostriatal Gating Operations When Trained on Human Working Memory Tasks
- Scaling Laws for Neural Language Models
- Emergent Abilities of Large Language Models
- Representational Strengths and Limitations of Transformers
- On the Emergence of Position Bias in Transformers
- Transformers: a Primer
- Toy Models of Superposition
- A Mathematical Framework for Transformer Circuits
- Transformers, parallel computation, and logarithmic depth
- Mastering Decoder-Only Transformer: A Comprehensive Guide
- Transformers as Support Vector Machines
- Do Language Models Have a Critical Period for Language Acquisition?
- RNNs Implicitly Implement Tensor Product Representations
- Tensor product variable binding and the representation of symbolic structures in connectionist systems
- Memory Networks: Towards Fully Biologically Plausible Learning
- Upper and lower memory capacity bounds of transformers for next-token prediction
- TransformerFAM: Feedback attention is working memory
- STRIDE: A Tool-Assisted LLM Agent Framework for Strategic and Interactive Decision-Making
- Chain of Thought Empowers Transformers to Solve Inherently Serial Problems
- Theoretical Limitations of Self-Attention in Neural Sequence Models
- Self-attention Does Not Need $O(n^2)$ Memory
- Hierarchical Reasoning Model
- Energy-Based Transformers are Scalable Learners and Thinkers
- LLMs are Bayesian, in Expectation, not in Realization
- Lost in Embeddings: Information Loss in Vision-Language Models
- Predictive Coding: a Theoretical and Experimental Review
- Predictive coding is a consequence of energy efficiency in recurrent neural networks
- On Neural Differential Equations
- Neural Ordinary Differential Equations
- Continuous Attractor Neural Networks: Candidate of a Canonical Model for Neural Information Representation
- Hopfield Networks is All You Need
- Learning Attractor Dynamics for Generative Memory
- Statistical mechanics of complex neural systems and high dimensional data
- Wake-sleep transition as a noisy bifurcation
- On the Paradox of Learning to Reason from Data
- CRAB: Assessing the Strength of Causal Relationships Between Real-World Events
- Passive learning of active causal strategies in agents and language models
- SPARTQA: A Textual Question Answering Benchmark for Spatial Reasoning
- A Critical Review of Causal Reasoning Benchmarks for Large Language Models
- Predictive Coding: a Theoretical and Experimental Review
- Cognitive computational neuroscience
- Testing theory of mind in large language models and humans
- RAVEN: A Dataset for Relational and Analogical Visual rEasoNing
- Compositionality Decomposed: How do Neural Networks Generalise?
- Compositional architecture: Orthogonal neural codes for task context and spatial memory in prefrontal cortex
- Reasoning ability is (little more than) working-memory capacity?!
- Distributional Reinforcement Learning in the Brain
- Reinforcement Learning: An Overview
- Reinforcement Learning: An Introduction
- Deep reinforcement learning can promote sustainable human behaviour in a common-pool resource problem
- Feed-Forward Networks with Attention Can Solve Some Long-Term Memory Problems
- Prefrontal cortex as a meta-reinforcement learning system
- Bridging Neuroscience and Robotics: Spiking Neural Networks in Action
- Combined Sensing, Cognition, Learning, and Control for Developing Future Neuro-Robotics Systems: A Survey
- AI, Robotics & Neuroengineering at Ken Kennedy Institute
- Special Issue : Applications of Neural Networks in Robot Control
- Embodied AI Workshop
- A call for embodied AI
- Spiking neural networks: Towards bio-inspired multimodal perception in robotics
- PsychRNN: An Accessible and Flexible Python Package for Training Recurrent Neural Network Models on Cognitive Tasks
- Neuroscience needs behavior
- No Free Lunch from Deep Learning in Neuroscience: A Case Study through Models of the Entorhinal-Hippocampal Circuit
- NeuroAI critique: What have we learned about artificial intelligence from studying the brain?
- A Review of Neuroscience-Inspired Machine Learning
- The new NeuroAI
- Trainees’ perspectives and recommendations for catalyzing the next generation of NeuroAI researchers
- Future views on neuroscience and AI
- Averaging is a convenient fiction of neuroscience
- Circular and unified analysis in network neuroscience
- Circular analysis in systems neuroscience: the dangers of double dipping
- From circuits to behavior: a bridge too far?
- The Matrix Calculus You Need For Deep Learning
- Notes on Quadratic Forms
- Vector Calculus Notes
- Modern Quantum Mechanics.pdf
- Book Of Proof
- Everything You Wanted To Know About Mathematics
- Handbook of Mathematics
- Probabilistic Machine Learning
- The Urgency of Interpretability
- Interpretability, Mechanistic and Explanatory
- A Mathematical Philosophy of Explanations in Mechanistic Interpretability -- The Strange Science Part I.i
- Evaluating Explanations: An Explanatory Virtues Framework for Mechanistic Interpretability -- The Strange Science Part I.ii
- Mechanistic Interpretability for AI Safety -- A Review
- Layer by Layer: Uncovering Hidden Representations in Language Models
- Transformer Interpretability Beyond Attention Visualization
- Testing theory of mind in large language models and humans
- RAVEN: A Dataset for Relational and Analogical Visual rEasoNing
- SPARTQA: A Textual Question Answering Benchmark for Spatial Reasoning
- DevBench: A multimodal developmental benchmark for language learning
- The little book of deep learning
- The Handbook of Brain Theory and Neural Networks
- Textbook: Introduction to Machine Learning
- Infinite Powers: How Calculus Reveals the Secrets of the Universe
- The Self-Assembling Brain: How Neural Networks Grow Smarter
- Handbook of Mathematics
- Probabilistic Machine Learning
- How To Build Conscious Machines
- Why Is Anything Conscious?
- Is gravity evidence of a computational universe?
- Intelligence at the Edge of Chaos
- A theory of consciousness from a theoretical computer science perspective: Insights from the Conscious Turing Machine
- The Brain–Cognitive Behavior Problem: A Retrospective
- Against the Epistemological Primacy of the Hardware: The Brain from Inside Out, Turned Upside Down
If you have suggestions for better topic groupings, or want to contribute, feel free to open a PR!