“P.15: Wiener’s problem: What is the ontological status of information?
Most people agree that there is no information without (data) representa- tion. Following Landauer and Bennett 1985 and Landauer 1987, 1991, and 1996, this principle is often interpreted materialistically, as advocat- ing the impossibility of physically disembodied information, through the equation ‘‘representation5physical implementation.’’ The view that there is no information without physical implementation is an inevitable assumption when working on the physics of computation, since computer science must necessarily take into account the physical properties and limits of the carriers of information. It is also the ontological assumption behind the Physical Symbol System Hypothesis in AI and cognitive science (Newell and Simon 1976). However, the fact that information requires a representation does not seem to entail that the latter ought to be physically implemented. Arguably, environments in which there are only noetic entities, properties, and processes (for example, Berkeley and Spinoza), or in which the material or extended universe has a noetic or nonextended matrix as its ontological foundation (for example, Pythag- oras, Plato, Leibniz, and Hegel), seem perfectly capable of upholding the representationalist principle without also embracing a materialist inter- pretation. The relata giving rise to information could be monads, for example. So the problem here becomes: Is the informational an indepen-
dent ontological category, different from the physical/material and (assuming one could draw this Cartesian distinction) the mental? Wiener, for example, thought that ‘‘information is information, not matter or energy. No materialism which does not admit this can survive at the present day’’ (1948, 132). If the informational is not an independent ontological category, to which category is it reducible? If it is an independent ontological category, how is it related to the physical/ material and the mental? Answers to these questions determine the orientation a theory takes with respect to the following problem:
P.16: The problem of localization: Can information be naturalized?
The problem is connected with P.4 – Namely, the semanticization of data. It seems hard to deny that information is a natural phenomenon, so this is not what one should be asking here. Even elementary forms of life, such as sunflowers, survive only because they are capable of informational processes. The problem here is whether there is information in the world independently of forms of life capable of extracting it and if so, what kind of information is in question (an informational version of the teleological argument for the existence of God argues both that information is a natural phenomenon and that the occurrence of environmental informa- tion requires an intelligent source). If the world is sufficiently information rich, perhaps an agent may interact successfully with it by using ‘‘envi- ronmental information’’ directly, without being forced to go through a representation stage in which the world is first analyzed informationally. ‘‘Environmental information’’ still presupposes (or perhaps is identical with) some physical support, but it does not require any higher-level cognitive representation or computational processing to be immediately usable. This is argued, for example, by researchers in AI working on animats (artificial animals, either computer simulated or robotic). Ani- mats are simple reactive agents, stimulus driven. They are capable of elementary, ‘‘intelligent’’ behavior, despite the fact that their design excludes the possibility of internal representations of the environment and any effective computation (see Mandik 2002 for an overview; the case for nonrepresentational intelligence is famously made in Brooks 1991).
So, are cognitive processes continuous with processes in the environ- ment? Is semantic content (at least partly) external (Putnam)? Does ‘‘natural’’ or ‘‘environmental’’ information pivot on natural signs (Peirce) or on nomic regularities? Consider the typical example provided by the concentric rings visible in the wood of a cut tree trunk, which may be used to estimate the age of the plant. The externalist/extensionalist, who favors a positive answer to P.16 (for example, Dretske or Barwise), is faced by the difficulty of explaining what kind of information and how much of it saturates the world, what kind of access to, or interaction with, ‘‘in- formation in the world’’ an informational agent can enjoy, and how
information dynamics is possible. The internalist/intentionalist (for example, Fodor or Searle), who privileges a negative answer to P.16, needs to explain in what specific sense information depends on intelli- gence and whether this leads to an antirealist view.
The location of information is related to the question of whether there can be information without an informee, or whether information, in at least some crucial sense of the word, is essentially parasitic on the meanings in the mind of the informee, and the most it can achieve, in terms of ontological independence, is systematic interpretability. Before the discovery of the Rosetta stone, was it legitimate to regard Egyptian hieroglyphics as information, even if their semantics was beyond the comprehension of any interpreter? I’ve already mentioned that admitting that computers perform some minimal level of protosemantic activity works in favor of a ‘‘realist’’ position about ‘‘information in the world.’’ Before moving to the next problem, it remains to be clarified whether the previous two ways of locating information might not be restrictive. Could information be neither here (intelligence) nor there (natural world) but on the threshold, as it were, as a special relation or interface between the world and its intelligent inhabitants (constructionism)? Or could it even be elsewhere, in a third world, intellectually accessible by intelligent beings but not ontologically dependent on them (Platonism)?
P.17: The It from Bit hypothesis (Wheeler 1990): Can nature be informationalized?
The neologism informationalized is ugly but useful to point out that this is the converse of the previous problem. Here too, it is important to clarify what the problem is not. We are not asking whether the metaphorical interpretation of the universe as a computer is more useful than mislead- ing. We are not even asking whether an informational description of the universe, as we know it, is possible, at least partly and piecemeal. This is a challenging task, but formal ontologies already provide a promising answer (Smith 2003). We are asking whether the universe in itself could essentially be made of information, with natural processes, including causation, as special cases of information dynamics (for example, information flow and algorithmic, distributed computation, and forms of emergent computation). Depending on how one approaches the concept of information, it might be necessary to refine the problem in terms of digital data or other informational notions.
Answers to P.17 deeply affect our understanding of the distinction between virtual and material reality, of the meaning of artificial life in the ALife sense (Bedau 2003), and of the relation between the philosophy of information and the foundations of physics: If the universe is made of information, is quantum physics a theory of physical information? More- over, does this explain some of its paradoxes? If nature can be informa-
tionalized, does this help to explain how life emerges from matter, and hence how intelligence emerges from life? Of course these questions are closely related to questions listed in section 5: ‘‘Can we build a gradualist bridge from simple amoeba-like automata to highly purposive intentional systems, with identifiable goals, beliefs, etc.?’’ (Dennett 1998, 262).”
OPEN PROBLEMS IN THE PHILOSOPHY OF INFORMATION
LUCIANO FLORIDI
METAPHILOSOPHY Vol. 35, No. 4, July 2004
pp. 554-582
