Not so abstract ideas: Understanding abstraction from an emergent systems perspective

Thought, reason, language: These cognitive gifts of the Muses are what help to make human beings special. In much of traditional cognitive science, these higher level cognitive functions are believed to be the basis of abstraction, or the processing of abstract concepts. Unfortunately, our understanding of abstraction also biases our understanding of cognition.


Abstraction: the act of obtaining or removing something from a source; a general idea or quality rather than an actual person, object, or event; the state of someone who is not paying attention to what is happening or being said.

How useful is a definition? They’re concise, objective, easily accessed from a mobile phone or tablet. We make a compilation of them and call it a dictionary. But just like our experience of reality is not the truth, a word’s definition is not its meaning. While we can understand a definition, it is not the same as the sense of meaning generated when we transform information from the physical world into a cognitive representation.

To work with an emergent cognition framework, it is necessary to define abstraction from an emergent systems perspective. This shift in perspective means abstraction is not primarily an extractive process but a generative one. Extractive processes focus on removing unnecessary and peripheral information, on deriving a general meaning out of multiple instances. Generative processes focus on integrating available and relevant information, on synthesizing multiple instances into an emergent meaning.

Each type of abstraction focuses on information in different ways, similar to how we use the information in pictures in different ways. With extractive abstraction, we take a picture of the person and then mentally remove everything but their face, much like the function of an ID photo. As a result, our concept of the person is fixed and well-defined. In contrast, with generative abstraction, we take several pictures and then mentally superimpose those in a way that the person is foregrounded against a collage of rich but relevant details, more like the function of a photo album. The result here is that our concept of the person is emergent and mutable. Both convey information about the person, but for different purposes and at different levels of scale.


Generative abstraction: the act of something emerging from a process; a general idea or quality synthesized from the interactions or relationships of multiple representations; the state of someone who is transfixed by what is happening or being said. A generative abstraction is abstract in terms of being a greater level whole differentiated/sourced from its lesser level parts, and generative in terms of being a dynamic effect caused by the interactions of these parts.

Generative abstraction characterizes the distinction between one level of cognitive processing and the next. At the level of Perceptual processing, a Perception is a whole produced by the integration of sensory pieces of information. At the level of Associative processing, a Perception is a piece that can be integrated with other Perceptions to produce a higher level whole, an Association. We can think of generative abstraction as a metaphorical shifting of one’s point of view from a zoomed-in focus on an individual unit as the sum of its part to a zoomed-out focus on the individual unit as one of many units.

This version of abstraction is basically another way of describing cognitive representation as an emergent process. It means that conceptual information is essentially generated from, embedded within, and informed by our experience of perceptual (the sense of) and associative (the behavioral/situational contexts of) information. Our ability to generalize a concept is not developed by being able to understand its meaning independently of any contextualized relationships. Instead, meaning emerges from the integration of multiple contextualized relationships. While this meaning is a form of information that exists at a different level than the relationships which give rise to it, our ability to generalize a concept is developed by an understanding of the interdependencies between these levels of information.

The traditional ways of thinking about cognition have been based upon abstraction as a process that involves removing, subtracting, separating meaning and context. To understand cognition from an emergent systems perspective requires redefining abstraction as a generative process. In doing so, we are better able to describe the relationships among Perceptual, Associative, and Conceptual representations of meaning.