AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Default Mode Network: Simulating Experience for Better Predictions
The default mode network runs simulations on training data during idle time to generate better models for future environments and improve predictions. It uses spare cycles to become more efficient and refine the models. When navigating a given environment, attention and perception focus on what is necessary, with most of the information coming from predictions. This down-samples the large amount of available data into a manageable way of interacting. The system only encodes prediction errors, and the rest is filled in from memory. If something happens that is close to the prediction, it is considered good enough and not encoded. Similarly, if something happens that is far outside the prediction, it is also not encoded, as it is inconceivable and ignored.