tripleC 9(2): 295-304, 2011
ISSN 1726-670X
http://www.triple-c.at
CC: Creative Commons License, 2011.
Kinds of Information in Scientific Use
John Collier
collierj@ukzn.ac.za, School of Philosophy and Ethics, University of KwaZulu-Natal, Durban, South Africa.
Abstract: There are many different mathematical definitions of information that have their various uses, but I will be con-
cerned with notions of information used in applications in various branches of science that are distinguished by their topic,
i.e., what they apply to. I describe the major uses information, and show their relations to each other. I will argue that the
various uses form a nested hierarchy, in which each is a restriction on the previous, inheriting the properties of its predeces-
sor, but adding in new features that make it a special case. The lowest level is physical information determined by distinc-
tions and the highest is explicit representation in linguistic social communication. Is there anything common to information at
all these levels? I will argue that there is, and that information in each case is what Donald MacKay (1969) called a distinc-
tion that makes a difference. What distinguishes the use of information at each level is what distinctions make a causal
difference at that level. At each successive level distinctions that make a difference at a previous level make no difference at
that level. In order to create this sort of filter new levels have to be formed by cohesion peculiar to the identifying character-
istics at that level. A consequence of this view is that information must have causal powers, and that there is a tight connec-
tion between information and causation.
Keywords: Information, causation, “it from bit”, entropy, natural hierarchy, function, representation, intentionality, meaning,
reference, communication
Acknowledgement: Acknowledgement text acknowledgement text acknowledgement text acknowledgement text acknow-
ledgement text acknowledgement text acknowledgement text acknowledgement text acknowledgement text acknowledge-
ment text acknowledgement text acknowledgement text
Information has proven to be an elusive concept. I will argue that this is at least in part because it
has a variety of uses in the sciences, and various parties tend to claim that their own use refers to
the “real” information. There are other reasons for the elusiveness of the concept, not the least
being that there seems to be no single concept that ties various usages of information concepts
together. We might therefore conclude, a popular trend these days with diverse concepts, that a
pluralistic approach to information is the best that we can do, and even that this reflects not limita-
tions on our abilities, but objective reality. Furthermore there is sometimes confusion between vari-
ous ways of representing or measuring information and the disparate ways in which information
can be said to exist in our world. Information has a number of mathematical definitions that are
useful, often in different contexts. It is an interesting question to what degree these definitions can
be unified into a single mathematical theory of information. However, this is not my subject here,
nor is the issue of how various mathematical definitions of information can be used to represent or
measure information as it exists in the world.
I am interested here in what I call substantive information, i.e., information not as it is perceived
or measured (e.g., through instruments, or to speak more generally, instrumentally), but as it would
make a difference even if it were not perceived or measured (see Collier, 2008 for an application of
this distinction in biology). Basically, substantive information is distinguished by its causal powers
independently of whether these are observed. I will argue that there are various kinds of substan-
tive information to be found in various sciences, and that they do indeed differ substantially from
each other (pun intended). However they are related to each other such that each kind is a causal
restriction on a more general kind, so that they form a succession, or hierarchy of restrictions. What
they share, however, is that each kind of information makes a causal difference at its own level
(and by inheritance to the preceding levels), but not necessarily to successive levels. So each bit of
information at a given level makes at least a potential causal difference at this level. The explana-
tion for how this level is formed, though, involves the preceding levels (usually just the first preced-
296 John Collier
CC: Creative Commons License, 2011.
ing) through cohesive processes at the lower level that lead to the unification of the higher level. So
we get unification of kinds of substantive information by common principles applying at each level,
but also by common causal processes across levels at each level that allow level formation. Al-
though I believe that the unification of the sciences best follows this level oriented approach, rather
than through reduction, I will not argue that general point here.
1. Kinds of substantive information
Information can be used as a tool in various ways, for example to estimate channel bandwidth of
sensory pathways and various internal processes, the diversity of the genome and many others. In
1987 I searched the past three years of Biological Abstracts and found only seven out of several
hundred that I judged to use the information concept substantively. The others either used the con-
cept gratuitously (it was completely unnecessary) or used it as a measurement only. In order to
determine substantive use in science it is important to judge whether the concept has referred to
information in the world rather than just as a tool for measurement, or in some sort of metaphorical
sense that can be eliminated.
1.1. It from bit
The most liberal and inclusive view is the “It from bit” view. It has originated independently from so
many people that it is pointless to attribute an origin, though it probably goes back to Leibniz’ view
that the world has a logical structure of perceptions based in the ability to discriminate. The term is
due to John Wheeler (1990), supported by Murray Gell-Mann (1995), and the view has recently
been powerfully if controversially championed by Stephen Wolfram (2002). On this view, any caus-
ally (dynamically) grounded distinction makes a difference, thereby ensuring that it is information
(MacKay, 1969, see also Bateson, 1973). On this view information is objective, and there is nothing
else.
Although the idea applies to classical physics, its motivation comes primarily from quantum me-
chanics (QM). Each interaction in QM involves a sort of choice, and the choice can be represented
in terms of bits. The equations governing QM, therefore, can be represented in terms of bits trans-
ferred. In QM the wave function is conserved (unitarity), so information is conserved (see Collier,
1999a; 2010a for more general way to apply the notion of information transfer to causation that
does not imply preservation of information).This presents a real problem for the physics of “black
holes”, which seem to destroy information. Stephen Hawking has spent a long time considering this
paradox, which is not yet completely resolved (he reversed his original position that information is
lost irretrievably). The importance to the present paper is that it cannot be stated properly without
invoking the concept of information. Interestingly, the black hole paradox is intimately connected to
thermodynamics.
1.1.1. Where does information come from?
Physical information is an asymmetry in a system (Muller, 2007). This follows from the principle of
indifference, according to which if there is nothing to distinguish two states, then they are effectively
the same. Muller argues that, contrary to widespread thinking (e.g. Edwin Jaynes, who used a sub-
jective principle of indifference grounded in Bayesian approaches to the definition of probability), it
is possible in principle to determine the intrinsic information in a system in terms of its asymmetries.
His work is based on some of my work arguing that information originates in symmetry breaking
(Collier, 1996). In particular, nonequilibrium processes can produce symmetry breaking through
splitting in phase changes, leading to greater diversity and the production of new cohesive struc-
tures (Collier, 2010b). I will have more to say about this in what follows, but it is the basis of cos-
mogenesis (Layzer, 1990) and the origin of much of the diversity we see in biology (Brooks &
Wiley, 1988). It is also the origin of new hierarchical levels of information in many kinds of systems
(see section 1.3 below).
tripleC 9(2): 295-304, 2011 297
CC: Creative Commons License, 2011.
1.2. Negentropy
After the development of the statistical ground of thermodynamics (which was not fully accepted
until Einstein’s explanation of Brownian motion in 1905), James Clark Maxwell (1871) investigated
the nature of the Second Law, which says that entropy cannot decrease. To do this he invoked a
demon residing in a box that could locate fast molecules and let them go through a gate into an-
other box originally in thermodynamic equilibrium with the box the demon was in. If the demon
could do this, then it could reduce the entropy of the two-box system. He soundly concluded from
the demon thought experiment that the Second Law was statistical in nature (see Collier, 1990a for
an explanation of his argument and some further consequences of the demon thought experiment).
The interesting aspect of the thought experiment from the perspective of this paper is that to do
its job the demon must be able to obtain information about the state of the system without increas-
ing the net entropy of the demon/box system. This cannot be done without creating information out
of nothing. This violates a basic intuition about causation that it must involve resources that already
exist (see Collier, 1990a for the full argument; the intuition is also used in Collier, 1999a). The ar-
gument does not show that the Second Law is necessary, but it does show that a) it is statistical,
and b) that the entropy rate of a system cannot be intentionally reversed. The issue has led to a
great deal of work on the physics of computation and the relation between physics and computa-
tion (for references see Collier, 1990a; 1999a; 2010a).
Leon Brillouin (1962) conjectured that information is equivalent to the complement of entropy, or
negentropy, though the idea goes back to Erwin Schrődinger (1945). Brillouin clearly connected the
idea to measurement, as in Maxwell’s thought experiment, but also suggested that it was a physi-
cal value. This is how it is used in studies of the physics of computation. The basic idea here is that
if we take the actual entropy of a system, H
Act
, and imagine all of the internal constraints released,
the state of the “relaxed” system has the entropy H
MAX
, the equilibrium entropy (note that this is not
typically the entropy of the system if it is allowed to progress to equilibrium as an open system).
The basic ideas are spelled out in (Collier; 1986; see also Collier, 1990b; Collier & Hooker, 1999),
though they can also be found in (Wiley & Brooks, 1981), Landsberg (1984) and writings of a num-
ber of physicists. The importance of the idea is that in an expanding phase space both entropy and
order (negentropy) can increase together. Furthermore, the process is able to drive the formation of
even more order and organization, producing hierarchically organized systems (Layzer, 1990).
1.3. Hierachical Information
Donald MacKay (1965) asked the question of the nature of information in hierarchies, which he
thought was fundamental to understanding organized information. This basic idea has been devel-
oped in a number of ways, including attempts to get an informational measure of organization
(Bennett, 1985). However in the spirit of this paper, I am more interested in how information hierar-
chies might exist in nature.
There are two types of hierarchies found in nature. One is abstract, and refers in general. For
example we have the physical, the chemical, the biological, the psychological and the social. These
are not entities, and they have no direct causal powers – they don’t do anything. On the other hand
we can talk of the atoms that make up a particular molecule, the molecules that make up a particu-
lar cell, the cells that make up a particular body, and the bodies that make up a particular social
group and a particular ecology (the former is part of the latter). These are particular entities and
they have causal powers. They do things.
It is widely agreed in taxonomy that species are real entities (though some would take the real
entities to be smaller groups like populations), whereas higher taxa are not. This does not mean
that the higher taxa are fictional, however. We can’t just classify species into higher taxa any way
we please. They form what Duhem (1954) called a natural classification, the aim of scientific the-
ory. However, if we are interested in substantive information, information that plays a role in what a
thing is and does, we need to restrict ourselves to particular things.
298 John Collier
CC: Creative Commons License, 2011.
I have argued (Collier, 2003) that hierarchical information in particular systems must be arranged
such that the levels in the hierarchy are cohesive. That is the levels are distinguished by causal
closure that makes the level an entity. Basically, cohesion is a relation among the parts of some-
thing that make the parts parts of the same thing. So cells, for example, are distinguished by the
organizational processes and chemical forces that make the cell more contained in a centripetal
than a centrifugal direction (Ulanowicz, 1997). When we examine the closure of these processes
and forces, the balances is inwards, holding the cell together and maintaining its existence. So the
cell is at a higher, or more inclusive, than chemical processes and their organization.
1.3.1. Cohesion
Cohesion is the dynamical basis of individuation. It is both what gives things their identity, and what
makes them different from other things (Collier, 2010b). Cohesion is not absolute, but comes in
degrees, so there will be borderline cases. Furthermore, cohesion does not have just one dimen-
sion, but many, with the resulting cohesion being a sum of the many aspects. In general, systems,
objects and properties will show a mixture of cohesive and noncohesive properties – not all the
properties of something contribute to its cohesion, just the properties that keep it together. Cohe-
sion is not merely the presence of interactions; the relevant interactions must be stabilizing for the
thing of interest. Lastly, cohesive forces and processes need not be purely internal, but may involve
external processes and forces that contribute to the closure of cohesion.
1.3.2. Informational hierarchies
In biological organisms the DNA is a code that can be transcribed and processed to produce both
internal and external traits of the organism. To do this the information is transmitted through chan-
nels
1
in which it is modified as it passes from one form to another until it is expressed in what we
typically call traits (Collier, 2008). The information in the trait will contain information in the DNA,
but it may also contain more information that came from other sources, or was produced by sym-
metry breaking during development. The whole organism is selected (by surviving to pass on its
genes – and perhaps other forms of information as well – or not). This creates a feedback between
environmental and genetic information that shapes the genetic information. Seldom is there any
direct selection for genes (thought it happens).
Brooks and Wiley (1988) argued that there are several levels of information relevant to biological
evolution. There is a genetic level, an organismal level and a population or species level (there may
be further significant levels between these, but their approach is painted with broad strokes). Not all
genetic information is expressed at the organism (phenotypic) level, and not all phenotypic informa-
tion is expressed at the species level. However each of these levels is closed in each case by co-
hesive forces and processes. So we have a real and concrete informational hierarchy, assuming
that the case can be made that the information at each level is substantive. I will return to this issue
in section 2.2 below.
1.4. Functional information
In general a function is a property (more accurately a process) that has some end. The weakest
form of function is what is called teleonomic: the end is implicit in the organization of whatever per-
forms the function, and it leads to the maintenance of the function. Some functions are teleological,
which requires some sort of representation of the end that is generated internally, and thus has
intentionality. I will return to these later.
Biology is characterized by teleonomic processes that focus on preservation and reproduction.
There are two basic approaches to biological function, the etiological and the organizational (Col-
lier, 2004). The etiological approach defines the function of a trait in terms of what it was selected
1
Throughout I use ‘channel’ in precise sense of Barwise and Seligman (1997) that I explain in some detail in (Collier
2008).
tripleC 9(2): 295-304, 2011 299
CC: Creative Commons License, 2011.
for (being selected for means that the trait was selected because of what it does – just being se-
lected can be accidental, e.g., if the trait is connected to another trait that is selected for). The
heart, for example has the function of pumping blood, since that is what it was (and is) selected for.
Calming babies by holding them over the heart is not the function of the heart because it was not
selected for this calming effect. The effect is an accident. There are a number of problems with the
etiological approach that I won’t go into here, but one aspect of the account is that it implies that
there is a specific “proper” function for each trait. The other main account of function is to define
function in terms of organizational role. Unlike the etiological approach, a trait is functional not on
its own, but together with other traits. We would say that it has functionality rather than a proper
function. So the role of the heart is its contribution to the circulation of blood, which in turns contrib-
utes to nutrition and oxidation, which in turn contributes to survival and reproduction. The heart is
functional just to the extent that its role is to contribute ultimately to survival and reproduction.
There is a clear connection between these two accounts. A trait that is functional in the organiza-
tional sense will make it more likely that an organism that has it will survive, all other things being
equal. However the etiological account requires actual survival and reproduction. Traits that con-
tribute to survival but aren’t quite good enough are not functional. This is workable, but perhaps a
bit odd. There is also the fact that a trait has to work in some sense in order for it to be selected for.
The etiological approach seems to put the selective cart before the functional horse. Nonetheless,
looking for traits that are selected for is a good strategy for finding functional traits. The ones that
aren’t good enough are probably long gone.
1.4.1. Autonomy
In order to understand organizational role better, the notion of autonomy is useful. In biological
systems autonomy is self-regulation that leads overall to a greater likelihood of survival and repro-
duction. In fact it is a form of cohesion, but rather than being based mostly on forces (such as the
cohesion of rocks and motorcars), it is based on organizational forces that typically involve rather
small energy wells, and so, relatively small forces. The reason for this is that autonomy in an un-
predictable or even changing environment requires shifting from one strategy to another, and
switching is easier if it doesn’t require a lot of force. Autonomy, then, is the organizational form of
cohesion that leads to the persistence of the entity that has it (Collier, 1999b; 2002). Like all forms
of cohesion, autonomy involves closure (Collier, 2000), and comes in degrees. The closure in this
case is primarily the closure of processes. A trait is functional inasmuch as it contributes to auton-
omy. If we are speaking of an organism, then it is what leads to survival. If we are talking of a ge-
netic lineage, then it is what promotes reproduction. Notice that these two ends can and often do
compete with each other. This is typical of hierarchical or nested autonomy. Tragic perhaps, but
more or less inevitable.
1.4.2. Functional information
Why would we think that function involves information in any way? First of all, traits are an expres-
sion of genetic information that are exposed to selective forces, so they fit the information model of
section 1.3. However we are now in section 1.4 dealing with function, and unless we are to reduce
functional information to hierarchical information, there must be some special role for information in
at least some functions.
Marcello Barbieri (2001) has argued that much of biological functioning involves codes. The ge-
netic code, no matter how important, is just one of these. The notion of a code seems to presup-
pose a notion of information: a code is a transformation from “one world to another” as Barbieri
puts it, in which the encoded form acts as a stand-in for what it encodes. That is, there has to be an
information channel between the two. So a code embodies an information channel, and we can
employ information theoretic concepts to understand what it is that the coded processes do.
Whereas Barbieri thinks that codes are enough to understand many complex life processes,
others think that a code alone is not enough. Various biosemioticians (too many to list here) think
300 John Collier
CC: Creative Commons License, 2011.
that these codes must be interpreted as well as being some form of mechanistic transcription. For-
tunately, interesting though the issue is, it does not need to be resolved in order to justify the claim
that at least some functionality involves substantive information. A somewhat worrying aspect of
the interpretation requirement, though, is that it is often stated in terms of meaning. Indeed, inter-
pretation itself strongly suggests meaning. However, like Barbieri, I prefer to keep meaning re-
served for intentional systems, although I agree with the majority of biosemioticians that something
is lacking in the code account alone.
1.5. Intentionality and representation
This section will be brief, since it is obvious to most people that our inner representations are
meaningful and contain information, perhaps even about something that does not exist at all. In the
case of codes, if there is a misrepresentation the code is malfunctioning, but this is not so for inten-
tional representations. This suggests strongly that intentionality is something over and above the
biological processes that support it.
Intentionality is clearly functional, so it inherits informational aspects of functionality. I have ar-
gued (Collier, 1999c) that intentional representations must have a certain degree of autonomy in
order to serve their function.
… the main distinguishing feature of genuine semiosis appears to be the existence of a
complex of anticipations related to other such complexes within which an interpretant can be
formed. Representations permit complex anticipations and relations to other representations
so as to allow intentional activity. Causal relations may look representational when they
causally carry a signal about something, but this is not enough for them to be representational
(Collier, 1999a, p. 128).
In other words, intentional representations must be integrated in to their context. Furthermore, it is
this integration that gives them their meaning, not the underlying processes. So the cohesion of
meaningful ideas is based in their integration with other ideas, as well as sensory inputs and motor
outputs constrained by their flexibility in being applied as a unit in other contexts. The latter is re-
quired for them to be distinguishable from their contexts.
1.6. Linguistic information
The case that language involves substantive information is so obvious it hardly seems worth men-
tioning. It is worth noting, though, that language is necessarily social, and takes us into the social
realm. It is more than just symbolic thought (pace Terrance Deacon), though symbolic thought is
necessary for language. I would argue that all social activity that cannot be understood at the
purely biological level (1.4) must be in a linguistic content to make sense, but establishing that is
not necessary for justifying the substantive role of information in language. A more interesting
question is whether the role of information in language can be reduced to the resources required
for intentionality and representation. This is a thorny issue that I will not try to resolve here.
2. Relations between the kinds of information
Broadly, the kinds of information I have discussed above form an abstract hierarchy: fundamental
physics, thermodynamics, hierarchical thermodynamic systems, functional systems, and intentional
systems. Each type is nested within the previous one. Intentional systems are a type of functional
system, which in turn are a type of hierarchical thermodynamic system. The latter are thermody-
namic systems, which are physical systems. I will justify these claims in the subsection 2.1.
Furthermore, each successive kind is not just a restriction of the previous kind, but it is formed
out of that kind by similar self-organizing processes, like those involved in Layzer’s cosmogenesis
(Layzer, 1990), Brooks and Wiley’s unified approach to biology (Brooks & Wiley, 1988), and Kauff-
man’s origins of order (Kauffman, 1985; 1991; 1993). These processes produce new levels not just
in the abstract, but in real and particular systems. I will argue this in section 2.2.
tripleC 9(2): 295-304, 2011 301
CC: Creative Commons License, 2011.
2.1. Nesting
Stan Salthe (1985; 1993) has argued that hierarchies involve a series of focal levels with each in-
teracting primarily if not exclusively with the next and preceding levels. If my account in section 1 is
coherent and basically correct, then information levels, both within the broad levels I describe as
well as between them, are good candidates for this sort of hierarchical arrangement. In particular
entities, there is a relationship of containment, corresponding to what Salthe calls a scalar hierar-
chy. Salthe also picks out a specification hierarchy, which is a relation between kinds such that
each is contained within another, as in taxonomic systems. As I pointed out above, this sort of hier-
archy is not necessarily substantive, though it may be natural. I once
suggested to Stan that the interesting cases come when the two hierarchies correspond, and I still
believe this to be the case. Inasmuch as substantive information and its relations between kinds is
“interesting”, there should be a nesting that involves both specification and scale. Scalar nesting is
guaranteed if each level in a system is contained physically and temporally within another higher
level. If cohesion occurs at each level, then scalar nesting is guaranteed. Explicit recognition of the
significance of cohesion is missing from Salthe’s early accounts, though it was a central part of my
account of informational hierarchies from the beginning (Collier, 1986).
Figure 1: Nesting of major kinds of information (Collier, 2008)
To complete the picture it is necessary to show that each level in particular systems involves a
cohesive constraint on the above level. It is fairly obvious that negentropy is nested within basic
physics. I will defer its cohesive nature to section 2.2. Hierarchical information is nested within
thermodynamic information (Collier, 2003). In the case of code based functionality at least, informa-
tion is a restriction on hierarchical information (only some hierarchical information is functional).
Furthermore, functionality occurs only under the far from equilibrium conditions that allow organiza-
tion based autonomy. The move from functionality to intentionality, however, presents some diffi-
culties. In order to make the case for nesting, intentional information must be a restriction on coding
based functionality. This is not obvious. I will address this again in the conclusion of this paper.
IT FROM BIT
NEGENTROPY
HIERARCHICAL
FUNCTIONAL
INTENTIONAL
tripleC 9(2): 295-304, 2011
ISSN 1726-670X
http://www.triple-c.at
CC: Creative Commons License, 2011.
2.2. Cohesion formation
Each level is a restriction on the level preceding it (outside it in Figure 1). This restriction is created
by the formation of cohesion through self-organization within the preceding (outer) level. For exam-
ple, negentropy arises in the early universe as matter and energy go out of equilibrium as the uni-
verse expands. Prior to that there are merely local concentrations and dispersions of matter, but no
steady separation. As the universe expands to the point that the expansion rate is greater than the
equilibration rate matter condenses out, and the local entropy H
ACT
is lower than H
MAX
, creating
local negentropy. Further hierarchical levels are formed when subsystems grow or expand, permit-
ting bifurcations that produce new cohesive levels.
It is generally agreed that living systems formed through bifurcations within chemical systems,
followed by selection. The formation of codes may have followed basic metabolism, or have come
first, but stable living systems (lineages in this case) had to wait for the sort of reliable transmission
provided by a code. So it is reasonable to argue that stable living things emerged out of dissipative
processes feeding on negentropy (Schrödinger, 1945) through the stabilization of codes, a new
form of information. It should be recalled that each of these new levels inherits the information of its
predecessors, and that functional levels, for example, are subject to it from bit, negentropic and
hierarchical dissipative processes within their own processes at their own level.
As mentioned above, however, the transition from functional processes to intentional processes
is not very clear. I have noted that intentional processes rely on some degree of autonomy of those
processes, inheriting characteristics of functionality, but at a new level. What more is required for
intentionality to emerge will have to wait for further work, though I have made a suggestion in my
conclusions.
3. Conclusions
For each kind of substantive information used in the sciences there is a distinct level formed by
bifurcations that form cohesive structures at the next higher level. This is reflected in the informa-
tion at each level, which inherits the properties of the lower level, but produces new asymmetries at
its own level through the formation of new cohesions peculiar to the level. The only really problem-
atic transition with current knowledge is from the biological to the intentional level. This is because
it is not obvious that the information in the intentional level is based on biological codes that make
up at least part of the functional level (see Barbieri, 2010 for a possible approach). A possible solu-
tion is a better understanding of the nature of neural transmission and the nature of neural transmit-
ters, which may work together like a code. It seems obvious that the functioning of the neural sys-
tem depends on the transmission of substantive information. Perhaps there is another level be-
tween the functional and the intentional – the nervous system. This would make sense in the over-
all picture, since the nervous system differentiated from the rest of the body both evolutionarily and
developmentally. This in itself is not enough because other organs have also differentiated, but do
not constitute a separate information system. (The immune system might be a notable exception. It
might also have its own cohesion and informational properties (Matthen & Levy, 1984)) It would be
required to show that the nervous system was autonomous to some degree prior to intentionality
(which presumably emerged out of it). Alternatively, it might be possible to show that intentionality
can be fully reduced to nervous system activity, but I doubt this.
References
Barbieri, M. (2001). The Organic Codes: The Birth of Semantic Biology. Acona, Italy: peQuod. [Reprinted in 2002 by Cam-
bridge University Press].
Barbieri. M. (2010). On the Origin of Language: A Bridge Between Biolinguistics and Biosemiotics. Biosemiotics, 3, 201-223.
Barwise, J. & J. Seligman (1997). Information Flow: The Logic of Distributed Systems. Cambridge: University of Cambridge
Press.
tripleC 9(2): 360-369, 2011 303
CC: Creative Commons License, 2011.
Bateson, G. (1973). Steps to an ecology of mind. London: Paladin.
Bennett C.H. (1985). Dissipation, information, computational complexity and the definition of organization. In D. Pines (Ed.),
Emerging Syntheses In Science: Proceedings of the Founding Workshops of the Santa Fe Institute. Redwood City,
CA: Addison West Publishing Company.
Brooks, D.R. & E.O. Wiley. (1988). Evolution as entropy: Toward a unified theory of biology. Chicago: University of Chicago
Press.
Collier, J (1986). Entropy in evolution. Biology and Philosophy, 1, 5-24.
Collier, J. (1990a). Two faces of Maxwell's demon reveal the nature of irreversibility. Studies in the History and Philosophy
of Science, 21, 257-268.
Collier, J. (1990b). Intrinsic information. In P. P. Hanson (Ed.), Information, language and cognition. Oxford: Oxford
University Press.
Collier, John (1996). Information originates in symmetry breaking. Symmetry: Culture & Science, 7, 247-256.
Collier, J. (1999a). Causation is the transfer of information. In H. Sankey (Ed.), Causation, natural laws and explanation (pp.
279-331). Dordrecht: Kluwer.
Collier, J. (199b). Autonomy in anticipatory systems: Significance for functionality, intentionality and meaning. In: D.M.,
Dubois (Ed.), Proceedings of the American Institute of Physics from CASYS '98 - Second International Conference on
Computing anticipatory systems (pp. 75-81). Woodbury, NY.
Collier, J. (199c). The dynamical basis of information and the origins of semiosis. In E. Taborsky (Ed.), Semiosis. Evolution.
Energy towards a reconceptualization of the sign (pp. 111-136). Aachen: Shaker Verlag.
Collier, J. (2000) Autonomy and process closure as the basis for functionality. In J. L. R. Chandler & G. van de Vijver (Eds.),
Closure: Emergent organizations and their dynamics. Annals of the New York Academy of Science, 901: 280-291.
Collier, J. (2002). What is autonomy? Partial proceedings of CASYS '01: Fifth International Conference on Computing An-
ticipatory Systems, International Journal of Computing Anticipatory Systems, 12, 212-221.
Collier, J. (2003). Hierarchical dynamical information systems with a focus on biology. Entropy, 5, 100-124.
Collier, J. (2004) Interactively open autonomy unifies two approaches to function. In D. M. Dubois (Ed.), Proceedings of the
American Institute of Physics from CASYS '03 - Sixth International Conference on Computing anticipatory systems
(pp. 228-235). Melville, NY.
Collier, J. (2008). Information in biological systems. In P. Adriaans & J. van Benthem (Eds.), Handbook of philosophy of
science, Volume 8: Philosophy of information (pp 763-787). Dordrecht: North-Holland.
Collier, J. (2010a). Information, causation and computation. In G.D. Crnkovic & M. Burgin (Eds.), Information and
computation: Essays on scientific and philosophical understanding of foundations of information and computation
(World Scientific Series in Information Studies). Singapore: World Scientific 2010.
Collier, J. (2010b). A dynamical approach to identity and diversity in complex systems. In P. Cilliers & R. Preiser (Eds.),
Complexity, difference and identity: An ethical perspective (pp. 83-97). Berlin: Springer.
Collier, J. & Hooker, C.A. (1999). Complexly organised dynamical systems. Open Systems and Information Dynamics, 6,
241-302.
Duhem, P. (1954) The aim and structure of physical theory. Princeton: Princeton University Press.
Giddens, A. (1984). The constitution of society. Berkeley: University of California Press.
Gell-Mann, M. (1995). The quark and the jaguar: Adventures in the simple and the complex. New York: Owl Books.
Kauffman, S. A. (1985). Self-organisation, selective adaptation, and its limits. In Depew & Weber (Eds.), Evolution at a
crossroads. Cambridge, MA: MIT Press.
Kauffman, S., A., (1991). Antichaos and adaptation. Scientific American, 265, 64-70.
Kauffman, S. A. (1993). Origins of order: Self-organisation and selection in evolution. New York: Oxford University Press.
Landsberg, P.T. (1984). Can entropy and "order" increase together? Physics Letters, 102A, 171-173.
Layzer, D. (1990). Cosmogenesis: The growth of order in the universe. New York: Oxford University Press.
MacKay, D. M. (1969). Information, mechanism and meaning. Cambridge, MA: MIT Press.
Matthen, M. & E. Levy (1984). Teleology, error, and the human immune system. Journal of Philosophy, 81(7), 351-372.
Maxwell, J.C. (1871). Theory of heat. London: Longmans, Green and Co.
Muller, S. J. (2007). Asymmetry: The Foundation of information. Berlin: Springer.
Salthe, S.N. (1985). Evolving hierarchical systems. New York: Columbia University Press.
Salthe, S.N. (1993). Development and evolution: Complexity and change in biology. Cambridge, MA: MIT Press.
Schrödinger, E. (1945), What is life? Cambridge: Cambridge University Press.
Ulanowicz, R.E. (1997). Ecology: The ascendant perspective. New York: Columbia University Press.
Wheeler, J.A. (1990). Information, physics, quantum: The Search for links. In W. Zurek (Ed.), Complexity, entropy, and the
physics of information. Redwood City, CA: Addison-Wesley.
Wiley, E.O. & Brooks, D.R. (1981). Victims of history: A nonequilibrium approach to evolution. Systematic Zoology, 31, 1-24.
Wolfram, S. (2002). A new kind of science. Champaign, IL: Wolfram Media.
304 John Collier
CC: Creative Commons License, 2011.
About the Author
John Collier
John Collier was educated at MIT, UCLA and Western Ontario in Planetary Science and Philosophy. For a number of years
he went back and forth between the two until he settled into Philosophy. He has had lecturing and/or research positions in
five countries and four continents, but now seems to have settled in Durban, South Africa, where he has now lived longer
than he has lived anyplace else in his life. His current interests are metaphysics, complexity, information theory, evolution-
ary ethics and the way they overlap. His doctoral dissertation was on meaning and reference in theory change.