From wikipedia:

Stigmergy is a mechanism of indirect coordination between agents or actions.[1] The principle is that the trace left in the environment by an action stimulates the performance of a next action, by the same or a different agent. In that way, subsequent actions tend to reinforce and build on each other, leading to the spontaneous emergence of coherent, apparently systematic activity.

Stigmergy is a form of self-organization. It produces complex, seemingly intelligent structures, without need for any planning, control, or even direct communication between the agents. As such it supports efficient collaboration between extremely simple agents, who lack any memory, intelligence or even individual awareness of each other.[1]

Stigmergy is a core principle that infuses all the work that goes into building a machine consciousness.  Stigmergy implies representation.

There are 4 intertwined principles of organic life: [http://evolutionofcomputing.org/Multicellular/IntertwinedPrinciples.html]
Specialization, Polymorphic Messaging, Apoptosis, and Stigmergy.  Stigmergy is the key principle.  As we discover in systems biology: Specialization is driven by stigmergy.  Apoptosis is a a particular kind of behavior by agents (particularly agents of cellular chemistry) that is driven by stigmergic messages.  Polymorphic messaging implies that there are messages, that there are signs which are processed by stigmergic agents.  In a representational sense, there must be such agents.  Signs or messages, or signals drive the stigmergic behavior of cells, which are themselves driven by the action of agents such as proteins and other molecular level cellular structures.  

the behavior of organic cells follow stigmergic principles.  coupled with the systems biology approach, we will see how proteins and similar molecules form representation structures for the cell.  And from the stigmergic behavior of a cell's molecular processes we can develop a model to understand neuro-cognitive functioning. For instance,  how do neurons know to grow axons?  Where should the axons grow?  How does a cell know to keep growing an axon or dendrites until some concept or skill is acquired by the learning organism?  These are crude questions that illuminate the problem of molecular behavior as driven, and as driving the higher order representational problems of learning, memory, and action. 

As a model, stigmergy describes the aggregate behavior of cellular molecules in interacting with different molecules in the cells environment as signs to produce different kinds of behavior.  Cells are stigmergic.  

Stigmergy as a generic mechanism for coordination: definition, varieties and aspects Francis Heylighen (2011)
Stigmergic epistemology, stigmergic cognition  Action editor: Ron Sun Leslie Marsh a,*, Christian Onof (2007)
Expert Assessment of Stigmergy:A Report for the Department of National Defence  Tony White (2005)
http://arstechnica.com/science/2012/10/organism-without-a-brain-creates-external-memories-for-navigation/  (the slime molds create structures that demonstrate stigmergic behavior - maze solving)
Stigmergic Systems - summary of theory: http://www.stigmergicsystems.com/stig_v1/index.html?671932
Swarms, Stigmergy, and Collective Intelligence -slides  Informatics luis Rocha 2011  Lecture 19 Biologicallyinspired computing.
The concept of stigmergy  Timo Smieszek: http://www.scribd.com/doc/6447398/11/The-Concept-of-Stigmergy
Bio-inspired mechanisms for artificial self-organised systems. InformaticaJanuary 1, 2006 | Jean-Pierre, Mano; Christine, Bourjot; Gabriel, Lopardo; Pierre, Glize 
Bacterial Quorum Sensing in Pathogenic Relationships -- de Kievit and Iglewski 68 (9): 4839 -- Infection and Immunity



the idea of stigmergy is signs produce actions.  but stigmergy contains the more subtle notion that completely different processes of an agent can respond to the same sign.  And that completely different processes can produce the same output from different signs.  

A stigmergic agent contains some processes listed below:

sign x -> process m -> output y    x ; y

sign x -> process n -> output y     x ; y

sign z -> process m -> output y    z ; y

sign z -> process n -> output w    z ; w

sign x -> process p -> No Output

sign z -> process p -> No Output

the processes m and n accept certain inputs and produce certain outputs.  but how should m or n be selected as the process that produces y?   Because of stigmergy.  The inputs of an environment, in toto, produce process selection in agents.  There are hidden inputs to process m and process n, or we may think of m and n as possessing state which dictates which acceptance condition m and n are in (as automata).  [note: later we will see that state is, and must be treated as a kind of illusion we fabricate to explain process behavior]  

Process p takes x or z and does nothing with them.  Process P is dead end.  These kinds of processes are important to do nothing.   For instance, process p may hold sign x for a limited time and then release it.  process p in that scenario would act as a timing representation. 

In ordinary computational approaches, the programmer decides which process manages data.  In an actor model, the actor can change what it does with the next message it receives, but the actors are programmed.  However, the actor model of computation is a good computational corollary to manage complex stigmergy.   The issue is that the agents made of actors must adapt and develop.  - see Basic Model.  

[See Carl Hewitt and Actor Theory of Computation: 
#20110501 Hewitt, Meijer and Szyperski: The Actor Model (everything you wanted to know, but were afraid to ask)
Actor Model of Computation: Scalable Robust Information Systems (2011)
ACTORS: A Model of Concurrent Computation in Distributed Systems (1984?)
Other resources on the Actor Model and Unbounded Non-determinism by Carl Hewitt ]


previous next