Steve HutchinsonBig Pines

Working memory

A short-lived store for active context during a cognitive session: recent memories, active plans, and unresolved contradictions. Managed by the attention engine's budget constraints. Implemented in Redis or in-process cache.

Working memory is the active surface of cognition - the small set of items that are simultaneously available for deliberation during a cognitive loop iteration. Where the associative memory layer holds millions of memories on disk, working memory holds five primary items and ten background items in fast in-memory storage. The attention engine is responsible for deciding what fills those slots: it scores every candidate (incoming events, retrieved memories, flagged contradictions, active goals) by salience and places the highest-scoring items in primary slots. Items in primary slots generate richer activity traces, receive stronger reinforcement signal, and are more likely to influence arbitration outcomes. Items in background slots are monitored but not actively deliberated - they can be promoted to primary if their salience rises during the session. Working memory persists between turns within a session (managed by the session manager) so that context from an earlier turn influences retrieval and reasoning in later turns. At session close, the working memory state is flushed: any items that should survive into future sessions must be explicitly flagged for the next session's hydration step.

This site collects anonymous usage data to understand how people read and navigate the blog. Accepting enables persistent reader preferences across visits.