Frameworks for understanding goal-directed agents

E. coli networks, as a model organism, can respond to a variety of input. E. coli's preferred food source is glucose, but if glucose is absent, E. coli is able to consume lactose instead by producing the protein lactase. To optimize the production of lactase, an E. coli network should predict its environment rather than just respond to it. The model of mRNA transcription in E. coli used in this research can be used to predict complex stimuli, such as naturalistic video. If a simple unicellular organism can do model-based reinforcement learning, it might be capable of other kinds of reinforcement learning. We therefore look to lessons from lower-level organisms for future reinforcement learning directions.