Causality
Projects | | Links:
Causal cognition is the ability to acquire and exploit causal information about oneself and the world. We connect research on causal cognition in human and non-human animals with causal machine and reinforcement learning, where disentanglement is a candidate process for building causal representations.
Disentangled representations for causal cognition builds a unifying computational framework for causal cognition, organised around its explicitness, sources and integration, and identifies explicitness with disentanglement, the factorisation of a representation. A longer summary is in this post.
Ecological causal cognition needs disentanglement replies to commentaries suggesting that the framework does not take embodied and ecological perspectives on cognition enough into consideration.
