Feedback is a form of interdependence in which the output of a system affects its future output. Marshall Scoot Poole describes this in his encyclopedia article as a circular relationship between different components of a system [that] influence each other. His exact description is as follows:

"...a circular relationship in which (a) causal processes form a closed loop whereby each cause becomes an effect of the other factors in the loop or (b) information about the past or present influences future states of the system." (52).

The presence of feedback and feedback loops are not uncommon in systems; however, they are not ubiquitous. As Donella H. Meadows explains in Thinking in Systems, the structure of some systems is simple and straightforward enough that feedback and feedback loops do not exist (26). In these cases, influences from outside the system may influence the system's output, but the system's output itself influences neither these inputs nor the system's future output.

The influence of feedback is most obvious in systems wherein a feedback loop exists. These feedback loops can be either balancing or reinforcing, based upon whether or not the effect of the feedback loop is to stabilize the system or to amplify existing changes (Meadows 30). 

In cases like the stock market, whose viability as a system requires a certain amount of assured stability, a stable feedback loop reduces the likelihood of cascading, dramatic changes that have the potential to abruptly alter or destroy the system's efficacy. These sort of feedback loops are common in systems that have a stated goal shared by all or most of the participants in the system. They are also very common in biological systems, whereby any sort of cascading effect results in the death of the organism. Stabilizing feedback loops are not infallible; while they reduce the likelihood of destabilization, large shifts in external inputs can break the stabilizing effect.

In other cases, a reinforcing feedback loop encourages a cascading effect, encouraging a system to inevitably proceed further and further down a particular path of development. For instance, the demographics of a particular subculture is directly influenced by the subculture's present state, and the present state of a subculture will attract or repel individuals of different demographics. These sorts of loops are common in chaotic systems, wherein there is no shared goal amongst all individuals, and wherein dramatic changes in the system do not have a large or overburdening cost on the system's participants. Reinforcing feedback looks are not inherently destructive, though they do have the capacity to destroy a system by dramatically altering its components.