We can define adaptation as the capacity for a system to change its state in response to some change within its environment.1 An adaptive system then is a system that can change given some external perturbation, and this is done in order to optimize or maintain its condition within an environment by modifying its state.2 One of the simplest examples of an adaptive system might be a mousetrap. It is designed to respond to some perturbation that triggers a mechanical reaction within the system, and it is simply executing on a logic that was built into its mechanical structure by design.
The growth of a plant or fungus towards a source of light, what is called phototropism, is another example of adaptation. The cells on the plant that are farthest from the light release a chemical causing them to elongate and thus move the plant towards the light source.3 This adaptability gives the organism some flexibility that improves its performance and chances of survival. Flexibility simply means it can generate an optimal response to a limited set of changes within the state of its environment.
In both of these previous examples, the capacity for adaptation is simply embedded within the physical mechanics of the system, but entities that are capable of more advanced forms of adaptation have specialized subsystems dedicated to regulating this process of adaptation. We call these specialized components regulatory or control systems.4 For example, an animal like a cat has a nervous system dedicated to sensing, processing, and responding to information it receives from its environment. With an electrical nervous system the creature is able to respond very rapidly and also capable of a much more complex algorithmic processing of information.
With this control system it is able to generate a wide variety of responses to deal with a rapidly changing environment where it may be presented with a large number of different scenarios; as a creature like our cat might encounter during the dynamic activity of hunting. Beyond this type of algorithmic logic that governs basic control systems, adaptive systems can have a much more complex conceptual framework for representing their environment and we call this a schema. Schemata, which is the plural for schema, are mental frameworks or concepts used to organize and structure information. With schemata an adaptive system has a full model for classifying and correlating different information about its environment, which can then be used to interpret new information, learn and generate novel responses to a very wide variety of input stimulus.
Within complex adaptive systems theory, these adaptive systems are called agents, and they are so called because they have agency. Agency can be defined as an action or intervention designed to produce a particular effect.5 An agent then is an entity that takes an active role to produce a specific outcome. Thus, agents do not act in a random fashion but actions are performed in order to produce a particular effect. That is to say, all adaptive systems have a goal, whether we are talking about a plant that adapts its state by moving towards the direction of the sunlight or a trader who buys a particular security to diversify her portfolio. These agents are acting according to a set of rules that are specifically formulated or designed to achieve the desired outcome, although these desired outcomes may be very diverse from the plant requiring more sunlight to the trader wanting more capital.6
Adaptive systems have a particular internal order or structure that enables them to intercept and transform energy and resources of some kind. We may be talking about our plant intercepting photons, combining and transforming them into sugars through the process of photosynthesis, or we may be talking about a business within an economy that takes in some input and transforms it into an output that generates revenue. The aim or goal then of an adaptive system with agency is to maintain and develop this internal order and capacity to process resources. They can only do this by importing energy or resources and exporting entropy to and from their environment. In other words, these systems are dependent upon their environment to ensure their continued functioning and they adapt in order to maintain and optimize their status within this environment, with this whole process being described as homeostasis.
The capacity for adaptation gives rise to autonomy, whereas much of our science is focused on studying deterministic systems – where we search for linear cause and effect interactions that govern them and then encode these in mathematical equations as laws of nature – the capacity for adaptation though gives an element a certain capability to act autonomously from these deterministic linear cause and effect laws. We say to a certain extent because most simple adaptive systems like a thermostat are essentially deterministic they are determined to respond to some cause in their environment with a given effect. It is only when we have advanced adaptive systems with internal agency that we get autonomy and the capacity for a variety of responses given any cause. The more complex the logic governing the adaptive system is the more capable it is of producing a variety of responses to any given input and the more it is able to operate sustainably in a broader complex environment.