## Definition A **feedback loop** is a closed chain of causal connections through which a change in a stock affects the flows into or out of that same stock. Feedback loops are the engines of all persistent system behaviour. Meadows identifies two fundamental types: **reinforcing (positive) loops** that amplify change, and **balancing (negative) loops** that resist it. ## Reinforcing (Positive) Feedback Loops A reinforcing loop occurs when a change in a stock generates a flow that produces more change in the same direction. The word "positive" refers not to beneficial effect but to the direction of amplification: the loop reinforces its own signal. - A fox population grows → more foxes reproduce → population grows faster (exponential growth). - Money earns interest → larger balance earns more interest → compound growth. - A virus spreads to new hosts → more infected hosts spread to others → epidemic curve. Reinforcing loops produce **exponential growth** when unchecked. Because humans intuitively expect linear change, exponential dynamics — whether in populations, epidemics, or financial markets — routinely surprise us. Every reinforcing loop will eventually be constrained by a balancing loop or physical limit; nothing grows forever. ## Balancing (Negative) Feedback Loops A balancing loop seeks a **goal state**: it senses the gap between current stock level and a desired level and generates flows to close that gap. - High fox population → prey scarcity → fox starvation and lower reproduction → population declines toward sustainable level. - A thermostat senses room temperature below target → heater activates → temperature rises toward target → heater switches off. - A central bank raises interest rates when inflation exceeds its target → borrowing contracts → price pressure eases. Balancing loops produce **goal-seeking** and **oscillating** behaviour. When the feedback signal is delayed, the loop overshoots and then overcorrects, generating oscillations that can amplify into instability. ## Dominant Loops At any moment one loop may be stronger than others, making it the **dominant loop** — the one shaping the system's current trajectory. Dominance can shift over time: a reinforcing growth loop may dominate early in a system's life, while a balancing resource-limit loop becomes dominant as the stock approaches its ceiling. Understanding which loop dominates — and when dominance shifts — is the key to predicting system behaviour. ## Delays Inside Loops Delays between a cause and its feedback effect are among the greatest sources of system surprise. A delay in a balancing loop causes the corrective action to continue past the point where it was needed, producing overshoot. A delay in recognising an exponential trend means the reinforcing loop has already built formidable momentum before action is taken. Meadows treats delays as one of the most important and underappreciated features of real-world systems. ## Feedback as the Source of Behaviour Meadows' key insight is that it is not *events* but the underlying *feedback structure* — which loops are present, which dominate, where the delays fall — that generates the characteristic behaviour of a system over time. To understand why a system does what it does, find its feedback loops. ## Related - [[System]] - [[Stocks and Flows]] - [[Leverage Points]] - [[Resilience]] - [[System Traps and Opportunities]] ## Sources - [[Thinking in Systems (Meadows 2008)]]