I use models to understand behaviors of both biological and artificial agents and explore how computation could be implemented in neural networks. Following the release of GPT-4, my research interest has partially shifted towards understanding the emergence of intelligence within large language models.
- Broadly, understanding the representations used by both biological and artificial agents to perform cognitive tasks
- How feedforward and recurrent connections contribute to encoding prior information in working memory.
- Exploration strategies across species.
- Trade-offs between complexity and effectiveness in the formation of task representations by biological agents.
- How to perform path integration without the use of a metric representation.
- To what extent does language contribute to intelligence?
- Xiong, H.D., Ji-An, L., Mattar, M. G & Wilson, R. C. (2023). Neural network modeling reveals diverse human exploration behaviors via state space analysis. Conference on Cognitive Computational Neuroscience 2023.
- Xiong, H.D. & Wei, X. X. (2022). Optimal encoding of prior information in noisy working memory systems. Conference on Cognitive Computational Neuroscience 2022