Nonlinear Feedback Loops

Nonlinear Feedback Loops

A feedback loop could be defined as a channel or pathway formed by an ‘effect’ returning to its ’cause,’ and generating either more or less of the same effect. An example of this might be a dialogue between two people. What one person says now will affect what the other person will say, and that will in turn feed back as the input to what the first person will say in the future. If we were to draw a model of a linear system, it would look something like this. There would be an input to the system, some process, and an output. As we can see, the input and the output to the system are independent of each other. The value that we input to the system now is not in any way affected by the previous output. There are of course phenomena where this holds true, such as the flipping of a coin. The value I will get from flipping a coin now will not be dependent in any way on the value I got the last time I flipped it. In mathematics, this is called the Markov property. But the fact is that many things in our world do not behave like this, meaning that current input variables to the system are dependent on previous outputs, and current outputs will affect future inputs. The state of the weather yesterday will affect the state of the weather today. The amount of money I have in my account today will through interest affect the amount I have tomorrow and so on. This phenomenon where the output of a system is “fed back” to become inputs as part of a chain of cause-and-effect that forms a circuit or loop is called a feedback loop.

Positive & Negative Feedback Loops

Feedback loops are divided into two qualitatively different types, what are called positive and negative feedback. A negative feedback loop represents a relationship of constraint and balance between two or more variables. When one variable in the system changes in a positive direction the other changes in the opposite, negative direction, thus always working to maintain the original overall combined value to the system. An example of this might be the feedback loops that regulate the temperature of the human body. Different body organs work to maintain a constant temperature within the body by either conserving or releasing more heat. Through sweating and capillary dilation, they counter-balance the fluctuations in the external environment’s temperature. Another example of negative feedback loops might be between the supply and demand of a product. The more demand there is for a product the more the price may go up which will in turn feedback to reduce the demand.


We can note the direct additive relationship here. When one component goes up the other goes down in a somewhat proportional fashion, with the end result being a linear system that tends toward equilibrium. The idea of equilibrium plays a very important role in linear systems theory. When we have these additive negative feedback loops, the net result is a zero-sum game. The total gains and losses combined are zero, and we can then define this as the system’s equilibrium or “normal” state, with our models then being built on this assumption of there being an equilibrium. This concept of equilibrium holds well for isolated systems and systems in negative feedback loops, but this assumption about there being an equilibrium breaks down in nonlinear systems, and thus we describe them a being “far-from-equilibrium.”

Positive Feedback

Positive feedback in contrast to negative feedback is a self-reinforcing process. The increase in the values associated with one element in the relation are correlated with an increase in the values associated with another. In other words, both elements either grow or decay together. Examples of this are numerous, such as compound interest where last year’s increases result in an increase in this year’s input, or chain reactions such as cattle stampedes are another example. The result is always a self-reinforcing process that leads to exponential outcomes of growth or decay. The total gains and losses are non-additive and do not sum up to zero, thus there is no equilibrium and these nonlinear systems are said to exist far-from-equilibrium. These positive feedback loops are of course unsustainable, requiring the input of energy from their environment. The exponential growth in human industrial activity over the past few centuries could be cited here. The more developed our industrial technologies are the better we are able to process and access petroleum which again feeds back to result in more energy and more industrial development, and so on. And this is all the product of some input of energy from the system’s environment that will eventually reach some limit.


As soon as we put our system into its environment, its output will in some way affect that environment. And this will, in turn, affect the future input to the system through what is called a time delay feedback loop. If the new input produces a result in the opposite direction to previous results, then it is a negative feedback and their effects will stabilize the system towards some equilibrium point. If these new inputs facilitate and accelerate the development of the system in the same direction as the preceding results, they are positive feedback resulting in nonlinear exponential growth or decay. Lastly, whereas negative feedback will lead to the system converging around some equilibrium state, positive feedback leads to divergent behavior as it rapidly moves away from an equilibrium and we describe them as being “far-from-equilibrium.”


Feedback loops are an example of the premise within complexity theory that complex phenomena can be the product of simple rules. Almost all phenomena that you would consider not normal are nonlinear. Positive feedback loops are behind a very many processes of change within complex systems. A social riot would be an example of positive feedback, when a riot begins with few people these individuals are vulnerable but with every extra person that chooses to partake in the riot it makes it more likely that it will be successful and less likely that any one individual will be reprimanded. Thus more will beget more, as this positive feedback cascades through the individuals aligning their states. Conflict escalation can involve positive feedback, given some act of aggression an opposing agent will be threatened, becoming less tolerant and more likely to react which will in turn feedback to effect the same action on the behalf of the other. An example of this would be an arms races between two nations, where the two sides continue to try and outcompete the other leading to all losing and growing potential for conflict.

Likewise, the phenomenon of irrational exuberance is another example of positive feedback. When the value of a trader’s stock goes up this feeds back to boost the trader’s self-confidence in their decision making and encourages them to make more investments that may be even riskier. Another good example would be what is called the Matthew Effect within sociology, which describes the fact that advantage tends to beget further advantage. Thus this phenomenon is also known as the “rich get richer” as these feedback loops tend to increase initial inequalities. We might think about the fact that bank managers are more likely to lend money to people who already have lots of money. Likewise, those who are already well connected within society will have greater potential for making more influential connections. This accumulative effect is described within network science by the concept of preferential attachment, which explains that those nodes that initially acquire more connections than others will increase their connectivity at a higher rate, and thus an initial difference in the connectivity between two nodes will increase further as the network grows.

Systems Innovation

  • LinkedIn
  • YouTube
  • Twitter
  • Facebook