Emergent Relations

Emergent Relations: Self-constructed/Hylomorphic

Carlos Castellanos
Keywords: emergence, self-organization, cybernetics, autopoiesis, autonomy, hylomorphism, self-construction, sensory evolution, artificial intelligence, artificial life, autonomous agents
Introduction Theories of emergence and self-organization derived from theoretical biology, cybernetics, autopoiesis and dynamical systems theory often stress the importance of structural autonomy and organiztional closure if a system is to exhibit emergent behavior. This entails the need for a system to have a capacity for initiating and evolving its own structural reconfigurations. This capacity for self-construction and modification allows for the development of new information linkages between the system and its environment; in effect creating a new system with new behaviors. These structural reconfigurations can be understood as occurring through self-organizing processes that arise from the emergent relations of existing elements. In other words, a truly emergent system must be able to evolve its own physical hardware for sensing and acting upon its environment. Peter Cariani notes how the ability to adaptively construct new sensors and effectors endows a system with the capacity to determine its own "observables" or feature primitives, thus modifying the relationship between its internal states and the physical environment.1 The system in essence has the ability to determine for itself which feature primitives are relevant and which action alternatives to take in order to influence the world. This stands in contrast to standard computational and connectionist/neural net approaches where adaptation to a changing environment relies on mappings between static perceptual and action categories that are determined in advanced by a designer.
While difficult to identify and nearly impossible to quantify, the exceedingly complex emergent behaviors of such systems can nevertheless be better understood if we adopt an ontology of hylomorphism that sees these behaviors as functionally-based processes that exist and occur within continuously shifting observational frames but which are nevertheless grounded in a physical (rather than abstract/platonic) world through their malleable material substrates. Whether or not a functionally-based hylomorphically emergent system has evolved new relevance criteria depends on how one chooses to observe and describe the given object, since it can "support radically different types of behavior, depending upon how one has chosen to observe it".2 Emergence and self-organization then, whether in the context of artificial life, strategic operations or social organization, can be understood as a continuously evolving set of relations between observers, the environment and the system's particular material instantiation; each contingent upon and co-determined with the other.
Methodology This experiment utilizes several properties related to theories of emergence and self-organization as a conceptual framework for the design of self-constructing systems capable of evolving and new informational linkages and emergent relations with their environment.
An electrochemical solution functions as an adaptive self-constructing, self-repairing entity capable of evolving its relationship with the environment so as to give rise to emergent relations characterized by complex patterns of behavior. Of particular interest in this experiment is the adaptive interaction of the unstable dendritic network with the many uncertainties of an external environment such as an architectural space.
The electrical signal from a microphone serves as an input stimulus to an alcoholic solution of stannous chloride [img (http://farm7 NULL.static NULL.flickr NULL.com/6100/6352451787_0561de6d0c_b NULL.jpg)]. The dendritic metallic threads that form are then turned into a sonic output stimulus that is sent to a pair of speakers. The sound from the speakers, as well as their vibrations and any other local environmental phenomena establish a continuous feedback loop that serves to stimulate new growth. The system is also "trained" to associate particular growth patterns with particular environmental inputs with outputs. By "rewarding" certain conductance changes produced in response to a particular environmental "perturbation", the system will adapt and become increasingly sensitive to that particular stimuli. Video recordings of the growth are also made for later offline analysis.
Additional experiments have also been conducted by transforming various sonic parameters (e.g. oscillator frequencies) from a sound synthesizer into lengths of electrical pulses that are fed into the solution. The resulting electrochemical growth is measured and used to alter the parameters of the synthesizer [img (http://farm8 NULL.static NULL.flickr NULL.com/7163/6805229627_87f2a53836_b NULL.jpg)]. Another experiment involves video data (e.g., from a motion tracker) being fed into the solution. The effect on the electrochemical dendrites is captured with a video camera, and turned into an output capable of altering the environment (and thus the sensor) in some way; the result of which is fed back to the solution to stimulate new growth [img (http://farm6 NULL.static NULL.flickr NULL.com/5304/5580342663_2f45a380a7_b NULL.jpg)]. In all experiments the same general feedback loop is created.
Biopoiesis sound/video implementation
Diagram of sound/video-based implementation of system
The shifting conductance/resistance patterns of the electrochemical network give rise to a complex self-organized learning process in which the thread growth is a physical manifestation of emergent relations characterized by dynamic, unstable equilibrium.
Emergent Unstable Equilibrium
Emergent Unstable Equilibrium
Working with such an unbounded state-space affords one the opportunity of conceptualizing relations with the environment in more open-ended terms by highlighting more complex systems of engagement with the world and emphasizing a vision of complex interplay between autonomous dynamic systems.
References 1. Cariani, Peter. "Emergence and Artificial Life." Artificial Life II, Santa Fe Institute Studies in the Sciences of Complexity, Vol X. Ed. C.G. Langton et al. Redwood City, CA: Addison-Wesley, 1992. 775-797 2. ibid., p. 778.