Emergent Properties in Biology
Here's one of my old Usenet postings. I re-discovered this while restoring
a bunch of archived files to a new hard disk after a recent crash. After
ten years, I still think this was one of my more lucid efforts to explain what
was then and is still, a complicated field. The article has been quoted
several times in various settings. Feel free to quote it if you like, but don't
pretend you wrote it :-)
From: firstname.lastname@example.org (David Tinker)
Subject: Emergent Properties. I. Introduction
Organization: University of Toronto, Biochemistry
Date: Wed, 11 Nov 1992 15:50:22 GMT
The recent spate of articles in talk.origins on "emergent properties"
generated some heat, but not much light! My colleague, Larry Moran,
objects to the term because he suspects (wrongly) that it is based on
non-mechanistic or vitalistic superstition, and (rightly) that the term
implies there is an obstacle to the reductionist agenda in biology.
Other postings imply that the term "emergent properties" is tautologous,
and not unique to biology; it has been claimed that everything has emergent
properties, so the adjective "emergent" is meaningless.
I think the topic is worth further exploration, and submit this article to
stimulate discussion. I hope that we have not seen the last word on this
First, I do not think there is a satisfactory closed definition of the term
"emergent", nor is it universally used by the research school most concerned
with such properties, the 'Artificial Life' community. Nevertheless I
believe the concept is so well accepted in this community that the term can
be used casually with the assurance that it is understood. I turned to my
well-thumbed copy of "Artifical Life I" and searched in the annotated
bibliography and in the index for terms like "emergent" - see below for the
results. Nowhere did I find a definition that would satisfy Larry Moran, but
in re-reading the articles I found many clues to a definition. So, being
willing to be called a fool, I will essay a definition:
A system may be said to possess "emergent properties" when (a) it is composed
of a collection of entities, (b) it has global properties, obeying well-
characterised rules that may be used for predictive purposes, that arise from
non-linear combinations of local interactions among the entities, and (c) the
rules do not depend specifically on the chemical nature of the entities."
By "entities" I mean systems which may exist independently, and which
make up the system by simple addition to it. Thus the protein molecules
in a crystal are entities in this sense, but the atoms in a molecule are
not "entities" composing the molecule.
By "non-linear" I simply mean the mathematical connotation, as in "non-linear
function". I wished to use this term in the definition rather than the
less general adjective "non-additive".
Now some questions and tentative answers.
1. Do such systems exist?
Yes they do. Three systems with emergent properties that have been
well examined are (a) artificial neural networks, (b) organisms that
exhibit schooling or flocking behaviour, and (c) cellular automata.
2. You say that the properties are independent of chemical nature of the
entities. Does this mean you espouse a non-mechanistic view?
Not at all. Let's take flocking behaviour as an example. It appears
to arise when entities have a mechanism for detecting spatial proximity
of identical entities and a feedback mechanism for maintaining a range
of postions relative to their neighbours. Essentially identical
behaviour can arise in organism as diverse as fish, insects and birds.
It could also arise in collections of robots made out of Lego (tm) - all
that is required is there be physical mechanisms for _instantiating_
the local interactions. If I were studying sandpipers, say, I would
certainly hope to elucidate the physiological and biochemical mechanisms
of recognition and feedback, and to learn how the relevant genes have
evolved to optimise these interactions for efficient flocking behaviour.
The _instantiation_ of the behaviour does depend on mechanisms which
obey the laws of physics and chemistry, but the behaviour itself trans-
cends these laws.
3. Aren't these rules merely empirical inventions that will be unnecessary
when we understand the mechanisms fully?
I don't think so. The work in this area indicates many of the "rules"
governing such properties are universal, and have a formal logical
structure and grammar. In the sense that thermodynamics is a formal
system independent of any specific physical system, so are the laws
governing emergent properties. However, it is true that like
thermodynamics, "emergo-dynamics" will be ultimately related to
lower-level physical theories.
4. Aren't all properties of matter "emergent" - e.g. the properties of
Not in the sense I have defined. The properties of water depend
absolutely on the specific interactions of water molecules, whereas
the properties I have called "emergent" would arise no matter what
entities are involved.
5. Simple things like flocks and cellular automata don't convince me -
these are just computer games. Is there any evidence that *real*
biological behaviour can be 'explained' by such notions?
I'm glad you asked. See the amazing chapter by P. Hogeweg (cited below)
in which he models such high-level behaviours as bumblebee sociology.
The fact that successful models of living systems can be constructed
out of computer instructions or Lego indicates that the properties being
modelled are 'real' ones.
6. Wait a minute! I'm beginning to think you are a Moravecian (see Hans
Moravec, "Mind Children"). Do you really think biological properties
including (choke) consciousness could arise in machines? Is Data (in
Star Trek) really human after all? Do you think human beings are
Yes. In fact, my conviction that my humanity has "emerged" from the
properties of molecules contributes mightily to that emergent property
of me, that I call a "religious world-view".
7. How can I learn more about such area so that I can critically discuss
this topic on talk.origins?
Start with the "Artificial Life" volumes from the Santa Fe insitute,
published by Addison Wesley. In my next posting, I'll re-post a
*long* annotated bibliography that was prepared by G. Miller and P. Todd,
and posted in sci.bio a year ago.
Bibliography and Footnotes:
From the annotated bibliography, in C.G. Langton, editor, "Artificial
Life I", pp 527-625, Addison Wesley, 1989.
(a) Titles containing the word "_emergent_" or "_emergence_".
J.H. Holland. "Studies of the Spontaneous Emergence of Self-Replicating
Systems using Cellular Automata and Formal Grammars." In A. Lindenmayer
and G. Rozenberg, editors, "Automata, Languages, Development", pp 385-404,
North Holland, 1976.
J.J. Hopfield. "Neural Networks and Physical Systems with Emergent
Collective Computational Abilities." Proc. Nat. Acad. Sci. 79:2554-2558,
S.A. Kauffman. "Emergent Properties in Random Complex Automata."
Physica D, 10, 1984.
(b) Titles that are germane to this posting.
C.W. Reynolds. "Flocks, Herds and Schools: A Distributed Behavioural
Model". Computer Graphics: Proceedings of SIGGRAPH '87, 21(4):25-34,
S. Wolfram, editor. "Theory and Applications of Cellular Automata."
World Scientific, Singapore, 1986.
P. Hogeweg. "MIRROR beyond MIRROR, Puddles of LIFE". In C.G. Langton,
editor, "Artificial Life I", pp 297-316, Addison Wesley, 1979.
Towards a legitimisation of emergent behaviour?
From C.G. Langton, in "Artificial Life I" page 3:
" The "key" concept in AL is _emergent behaviour_. Natural life emerges
out of the organized interactions of a great number of nonliving molecules,
with no global controller responsible for the behaviour of every part.
Rather, every part is a behav_or_ itself, and life is the behav_ior_ that
emerges out of all the local interactions among individual behav_ors_. It
is this bottom-up, distributed, local determination of behaviour that AL
employs in its primary methodological approach to the generation of
lifelike behaviors. "
From R. Dawkins, _ibib._ page 209 (discussing the biomorphs produced by his
'Blind Watchmaker' program):
" ... Our watchword is that as much as possible must emerge rather than being
designed. But having seen the range of phenotypes that emerge from the
basic program, can we think of any modifications to the basic program that
seem likely to unleash opulent flowerings of new emergent properties? ... "
: Prof. David O. Tinker INTERNET: email@example.com :
: Dept. of Biochemistry uunet.ca!beltrix!blunile!dtinker :
: University of Toronto FAX: (416)978-8548 :
: Toronto, Ont. M5S 1A8 VOICE: (416)978-3636 :