Burbeck
 on
 Computing

Multicellular Computing:
Stigmergy -- the secret of complex organization
 

 
Web evolutionofcomputing.org


Paper on Multicellularity (pdf) Presentation (pdf) Home Page

Four Principles logo

The Four Principles

Summary table
   Specialization
      in computing
   Polymorphic Messaging
      in computing
           Loading code
           Interpreted code
      in biology
  Stigmergy
     and "self"
     in computing
     in the Internet
  Cell Suicide (Apoptosis)
     in computing
Intertwined principles



Complexity
The problem
Out of control
Characterizing complexity
Dynamic complexity


Why the Biology Metaphor
Parallels with computing
Information processing
Encapsulation


Emergence
Example emergent systems
Multi-level emergence
   in computing
   in biology
Scale and emergence


Evolution
of computing
of multicellularity


Conclusions

Discussion & Comments

Stigmergy is an organizing principle in which individual parts of the system communicate with one another indirectly by modifying and sensing their local environment. Termite mounds are a classic example but cells do it and now computers are doing it too.

The third of the four principles of multicellular systems is that much of the communication between cooperating entities (cells, social insects or computers) is indirect. The entities deposit long-lived cues in external structures -- connective tissue, termite mounds, or databases as the case may be.  The persistent information provide by these cues helps to organize the collective behavior of the cells/computers/insects. The term stigmergy was coined in the 1950’s [1] to put a name to these sorts of reciprocal relationships between social insects and the structures they build, e.g., termite mounds, ant hills, beehives and even the pheromone-marked ant trails of nomadic social ant species such as army ants.  More recently, the term has been adopted by other disciplines including computer science.

A termite moundAlthough the term was coined to describe social insect behavior, the phenomenon itself arose much earlier. The very bodies of all multicellular organisms are stigmergy structures.  Cells in a multicellular organism create a growing body whose shape and boundaries are defined by a non-living extracellular matrix created by the cells. The cells deposit all sorts of cues, in the form of messenger molecules, in or on this matrix.  In the earliest and simplest multicellular organisms the extracellular matrix may be nothing more than a “slime” excreted by the cells that forms a clump or thin film in which the cells live and through which messenger molecules diffuse from one cell to its neighbors.  More complex multicellular organisms have much more complex extracellular matrix structures that support more subtle complex communication.  Plants create rigid stigmergy structures made largely of cellulose and other complex sugars.  Most complex animals create connective tissue that gives structure to their various organs and generally holds their bodies together.  In addition, mollusks (snails, clams, etc.) create shells, insects create chitinous exoskeletons, vertebrates create bone that is akin to coral (itself a stigmergy structure). Unlike coral, bone is constantly reshaped by the cells that create and maintain it to adapt to the changing stresses it encounters.

Social insects, cooperating cells, and cooperating computers communicate both with signals and cues.  The distinction is that signals are active communication events in real-time whereas cues are information embedded in the stigmergy structure to be read and reread many times. Both are specialized messages in the sense that they mean different things to different specialized receiving cells (or insects or computers). However cues are further specialized by their location -- that is, in addition to the information intrinsic to their molecular form or digital content, there is also information inherent in their location in the stigmergy structure. Because cues have both a message content and a location, cues support more complex kinds of communication than do signals and hence tend to support more complex social organizations. For example, complex ant societies rely more on cues whereas simple ant societies rely more on signals[2]

As is the case with social insects, cells in multicellular organisms communicate both by signals (polymorphic messenger molecules moving indiscriminately through blood, lymph or other liquids) and by cues (polymorphic messenger molecules attached to the extracellular matrix).  For example, bone, when stressed, provides cues to osteocytes and other bone cells for its own reshaping to better handle the forces placed upon it.  And smooth muscle cells in the walls of blood vessels modulate their contractility according to cues from the extracellular matrix.[3]  Not surprisingly, as with social insects, simple multicellular organisms communicate primarily by signals whereas complex multicellular organisms communicate more by cues.

Analogously, computing systems in complex human organization such as businesses rely on records (cues) deposited in databases (stigmergy structures), whereas loose organization, e.g., file-sharing, can work with real-time peer-to-peer messaging. Here again, multicellular computing recapitulates biology.  Stigmergy is ever present in complex computing systems and many novel stigmergy structures are emerging in the Internet.

Finally, it is worth noting that stigmergy is intimately related to the notion of "self."  A multicellular organism's "self" is more about the body than about the cells of which it is made.  The body includes both the cells of the organisms and the nonliving extracellular matrix.  However the extracellular matrix persists whereas most kinds of cells die and are replaced many times over during the lifespan of the body.  Similarly an ant colony or bee hive outlives all the insects in it except for the queen, and the Internet, whose primary stigmergy structure is its root DNS servers, outlives all of its constituent parts. So, for example, protecting a computing system is more about protecting its stigmergy structure(s) than about protecting its individual computers.


[1] See “Self-organization in social insects.”  Bonabeau, E., Theraulaz, G., Deneubourg, J.L.,  Aron, S. & Camazine, S., Trends in Ecology and Evolution, vol 12, pp. 188-193, 1997.

[2] “Individual versus social complexity, with particular reference to ant colonies,” Anderson, C & McShea, D. W.  Biol. Rev., vol 76, pp. 211-237, 2001.  p. 228

[3] Extracellular matrix controls myosin light chain phosphorylation and cell contractility through modulation of cell shape and cytoskeletal prestress.” Polte, TR, Eichler, GS, Wang, N, & Ingber, DE.  Am J Physiol Cell Physiol 286: C518-C528, 2004.





Last edited 4/10/2009