## Definition **Bounded rationality** is the principle that agents in a system act rationally on the basis of the *local information available to them*, but because that information is always incomplete and filtered by their position in the system, their individually rational decisions routinely produce collectively suboptimal or actively destructive outcomes. Meadows draws on the concept (originally formalised by Herbert Simon) to explain why system pathologies persist even when every actor is acting "sensibly." ## The Local Information Problem Each agent in a system perceives only a small slice of it: a firm sees its own costs and margins but not the aggregate effect of all firms externalising pollution; a herder grazing the commons sees his own cattle but not the incremental damage of everyone's overgrazing; a drug cartel sees revenue and enforcement pressure but not the aggregate social cost of the addiction market. The feedback that matters for the health of the whole does not reach the agents who are driving the harmful dynamics. ## Why Systems Cannot Be Fixed by Better Intentions Meadows uses bounded rationality to make a structural argument: replacing the people in a system seldom changes its behaviour, because incoming agents face the same information constraints and the same incentive structure as their predecessors. A new CEO inherits the same quarterly reporting pressures, the same competitor signals, the same supplier relationships. The system produces the same decisions. This insight redirects intervention from *who* is acting to *how the system is structured* to channel information and incentives. ## Relationship to Delays Bounded rationality is compounded by delays. Agents not only see an incomplete picture of the system — they see a *lagged* picture. By the time the consequences of decisions become visible, the original decision has already locked in further commitments. Commodity cycles, political crises, and environmental collapses are all partly products of this combination: rational actors responding to delayed, partial signals. ## Design Implications A system designed for bounded-rational agents must: 1. **Close the information feedback loop** — make consequences visible to decision-makers at the point where decisions are made. Carbon pricing makes externalities legible to firms. Pollution monitoring makes aquifer depletion visible to farmers. Speed display signs make driving speed legible to drivers. 2. **Shorten delays** — reduce the lag between action and consequence. 3. **Align individual incentives with system goals** — so locally rational behaviour approximates globally beneficial behaviour (e.g., internalising externalities through taxation or tradeable permits). Meadows does not advocate naïve optimism that these fixes are easy — only that they are more effective than exhorting individuals to act against their perceived self-interest. ## Related - [[System]] - [[Feedback Loops]] - [[Leverage Points]] - [[System Traps and Opportunities]] - [[Resilience]] ## Sources - [[Thinking in Systems (Meadows 2008)]]