Stefanos & Diaa
Since emergence is a concept with several (sometimes vague) definitions, it is useful to (a) review some of the definitions that are relevant to the context of our work, and (b) to converge to an operational definition we will use.
We have been reading a number of informative references on emergence in the fields of interactive and generative arts. All these references define emergence as the situation in which a new “macroscopic” form appears spontaneously from the arrangement of many “microscopic” parts that interact with one another via simple rules, in a way that cannot be straightforwardly anticipated by inspecting the rules and parts themselves. There is a lot to unpack in such a definition, and we will try to clarify some aspects of it here.
In generative art, emergence is most frequently related to artificial life processes used to generate the artwork. It consists of setting number of (re)production or (re)arrangement rules among a set of interacting artificial entities in motion and simulating the evolution of the system for many generations in a computer, until an unexpected pattern emerges. Examples of such processes from the physical world include termite mounds or microbial colonies (you can also see some examples of this in Montse and Jo’s blog). This type of emergence is often classified as physical emergence and its computational counterpart is called computational emergence. Note that the definitions of physical and computational emergence make no reference to an audience: they do not rely on the presence of an observer.
A contrasting definition is that of perceptual emergence: instances of emergence that rely on the perception of an observer. This is exemplified in the above picture, in which a shape emerges only upon observation. This is particularly relevant to interactive art, where there is a sense of “openness” in artworks, i.e., the participation of the audience is required for the artwork to be completed.
In this project, we will implement and study links between computational and perceptual emergence in an interactive and generative art project. The autonomous agents will be the participants, whose actions will determine the discretized agent states that will be used as inputs. These individual input states will interact in a virtual generative system that is sufficiently complex to exhibit computational emergence, such as, e.g., artificial neural networks. The output of the system will be presented as feedback to the participants, making the establishment of computational emergence manifest. The collective perception of the emergent pattern in the output will in turn accomplish the interactive perceptual emergence.
We have put together a 5-month action plan to develop and implement the project: