7.2 Linear Causality Contrast

The contrast between linear causality and circular causality illuminates the fundamental difference between two distinct modes of understanding how things in the world influence one another. Linear causality describes a directional, non-recursive relationship: cause A produces effect B, and B does not in turn influence A. The causal chain may extend through many links—A causes B, B causes C, C causes D—but as long as no link feeds back to an earlier element, the structure remains linear, and each event can in principle be explained by the prior events that produced it. Circular causality, by contrast, describes a relationship in which effects return to influence their causes, forming a closed loop in which each element is simultaneously a cause and an effect of others.

The most fundamental distinction between the two causal structures lies in the relationship between time, causation, and explanation. In linear causality, the past fully determines the present through a one-directional flow: to explain why something is as it is, one traces back through the causal chain to prior causes. In circular causality, the current state of the system is jointly determined by all elements simultaneously through the circular relationships, and no single prior cause provides a complete explanation. The system's state is the solution to a set of coupled equations rather than the terminus of a causal chain, and explanation requires characterizing the entire circular structure, not just identifying the immediately preceding cause.

The behavior of a system governed by linear causality is fully decomposable: the effect of multiple causes is simply the sum of their individual contributions (when the relationships are linear). The causal chain from input to output can be analyzed step by step, and the causal responsibility of each link can be assessed independently. In mathematical terms, a linear causal chain can be represented as a composition of individual transfer functions:

Each G_i(s) represents a causal link in the chain, and the overall input-output relationship is simply their product. Analysis of the open-loop (non-feedback) chain can proceed one link at a time, and the system's response to any input can be determined without solving simultaneous equations.

In circular causality, this decomposition is not possible because the causal relationships are not independent. The output Y cannot be computed as a simple product of transfer functions because Y feeds back to influence the very input that drives it. The closed-loop transfer function that results from solving the circular causal equations introduces the denominator 1 + L(s) that is absent in the open-loop representation:

This denominator represents the effect of the circular causality on the system's behavior. For large |L(jω)|, the closed-loop gain approaches 1 regardless of the specific form of L, meaning the circular structure makes the output robust to variations in the plant dynamics—a property that is entirely absent in the open-loop, linear causal chain. Circular causality thus actively shapes system behavior in a way that the components alone, analyzed through their linear causal relationships, do not predict.

The consequences of this distinction for scientific explanation are profound. Linear causal thinking leads to explanations that identify a prior initiating cause as the "reason" for an observed effect: the thermostat turned on because the temperature dropped; the organism fled because it perceived a predator. Circular causal thinking reveals that these explanations are incomplete: the temperature dropped because the heater was off, and the heater was off because the temperature was at the set point, and the set point is what the loop is maintaining. The full explanation requires characterizing the circular causal system and its fixed points, not just the most recent link in the causal chain.

In therapy and conflict resolution, the contrast between linear and circular causality shapes the framing of interpersonal problems. A linear causal account of a conflict attributes it to one party's initiating action: "the argument started because X said something offensive." A circular causal account recognizes that the conflict exists within a system of mutual reactions in which each party's behavior is simultaneously a cause and an effect of the other's: X's remark is a response to Y's prior behavior, which was a response to X's earlier behavior, and so on through the circular causal history of the relationship. Effective resolution of conflicts embedded in circular causal systems requires interrupting or restructuring the circular pattern, not merely identifying and punishing the "original" cause—which is an artifact of where the observer chose to start tracing the circular causal chain.

Scientific methodology historically developed primarily around linear causal assumptions: controlled experiments isolate variables to determine which causes produce which effects, under the assumption that these relationships are unidirectional. When circular causality is present, however, standard experimental designs can produce misleading results, because the manipulation of a variable that participates in a feedback loop changes the entire circular dynamic, not just the direct causal relationship being investigated. Systems methods—structural equation modeling, agent-based simulation, system dynamics modeling—were developed specifically to extend scientific methodology to circularly causal systems, enabling the analysis of feedback-laden domains that resist reduction to linear causal chains.