Marcelo Mattar

Learning distributed representations with complementary components - A generative approximation framework for episodic memory encoding

This paper proposes a generative framework for memory encoding that uses neural networks to approximate inputs through frozen weights, demonstrating effects similar to mnemonic precision and expertise in image generation models.

Stochastic Decision Policy Is Optimal Under Costly Control

This paper shows that stochastic decision policies can be optimal when considering the cost of control, using sequential sampling and recovering policies in macaques.