System memory is any mechanism that allows a complex system or its agents to retain information about past interactions and use that information to modify future behavior.
Memory is the fundamental requirement for learning, adaptation, and strong agency. Without a way to let the past influence the present, an agent is trapped in a purely reactive state, unable to build strategies or recognize patterns.
However, memory is not exclusive to nervous systems or digital hard drives. It can exist as structural traces in the environment. For example, rainwater carves grooves in a hillside. Future rain naturally flows through those same grooves because they offer the path of least resistance. The hillside “remembers” past rainfall through its physical structure. Similarly, a software team develops unwritten rules and conventions over years of shared mistakes—a cultural structure that acts as the team’s distributed memory.
In adaptive systems, memory is never just passive storage. It actively shapes the landscape of future decisions, shifting the probabilities of what actions the system will select next.