M
Matthew Botvinick
Researcher at University College London
Publications - 249
Citations - 57443
Matthew Botvinick is an academic researcher from University College London. The author has contributed to research in topics: Reinforcement learning & Computer science. The author has an hindex of 77, co-authored 224 publications receiving 48206 citations. Previous affiliations of Matthew Botvinick include Princeton University & University of Pennsylvania.
Papers
More filters
Book ChapterDOI
Cognitive Control as Cost‐Benefit Decision Making
TL;DR: In this article, a reward-based decision-making problem is framed, where the exertion of cognitive effort is determined by the output of a decision that considers both the costs and benefits of mobilising cognitive control at a given moment.
Posted Content
Deep Reinforcement Learning and its Neuroscientific Implications
TL;DR: A high-level introduction to deep RL is provided, some of its initial applications to neuroscience are discussed, and its wider implications for research on brain and behavior are surveyed, concluding with a list of opportunities for next-stage research.
Posted Content
Been There, Done That: Meta-Learning with Episodic Recall
Samuel Ritter,Jane X. Wang,Zeb Kurth-Nelson,Siddhant M. Jayakumar,Charles Blundell,Razvan Pascanu,Matthew Botvinick +6 more
TL;DR: This paper propose a meta-learning architecture that combines the standard LSTM working memory with a differentiable neural episodic memory, and explore the capabilities of agents with this episodic LSTMs in five meta learning environments with reoccurring tasks, ranging from bandits to navigation.
Journal ArticleDOI
Evidence integration in model-based tree search
TL;DR: Results from two experiments are presented, providing the first evidence to the authors' knowledge that the standard integration model of choice can be directly extended to multistep decision making.
Journal Article
Learning to reinforcement learn
Jane X. Wang,Zeb Kurth-Nelson,Dhruva Tirumala,Hubert Soyer,Joel Z. Leibo,Rémi Munos,Charles Blundell,Dharshan Kumaran,Matthew Botvinick +8 more
TL;DR: This work introduces a novel approach to deep meta-reinforcement learning, which is a system that is trained using one RL algorithm, but whose recurrent dynamics implement a second, quite separate RL procedure.