Naturalizing information

In this account of consciousness, "information" is a key concept -- it's the idea of information that allows the distinctive gap or "loose connection" between the two subcomponents of consciousness that in turn underlies its distinctive flexibility as a control system. But it's also a concept burdened with problems, including one major one for any use in a causal or mechanistic explanation -- it appears to require consciousness itself as generator and/or receiver of information. In an interesting post on Conscious Entities a while back (Sep 11/05), it was pointed out that "Information is a slippery but seductive term", at least partly because, though it can be treated in a very rigorous and well-defined manner, its more familiar use it carries with it the notion of "meaning", with hard-to-avoid overtones of conscious agency, for whom the information would be meaningful. Is information itself, in other words, not meaningful in a purely causal mechanism?

In trying to answer that, it might be useful to consider a couple of instances in which something that at least looks like "information" plays a key role in an operation that is without doubt mechanical. One such case is just that ubiquitous modern machine, the computer. Leaving aside, for now, the input and output of such devices (in which human agents are usually involved, though they needn't be), their data store and operational instructions (software) -- i.e., the means by which their behavior is controlled -- certainly have an information-like quality to them. In any case, another example would be that chemical machine, the living cell, in which the sequence of nucleotides on the long DNA coding molecule constitutes something that looks very much like information, used to construct the protein molecules that do the work of the cell. In both these cases, it's true that if you look closely enough, you can see the strictly causal (and/or stochastic, in the case of the cell) processes that are actually operating the mechanism, but such a close-up view also seems to lose an important or distinctive aspect of their functioning.

To see this, consider two kinds of machine: a basic mechanism involving straightforward causal processes and pathways; versus a mechanism based upon an information-like, intermediate structure that consists ofpatterns of small differences. The "small differences" may be differences in voltage, in magnetization, in nucleotide code, or some other content, but what's really determining the operation of the machine as such is thedifference, not the content of the difference. In this sense, we could say that the "semantics" or meaning of any such pattern of difference is simply the difference it makes in the operation of the machine. (Information has been defined -- see Floridi, quoting Bateson, in the Stanford Encyclopedia of Philosophy -- as "difference which makes a difference".) The advantage of such information-based machines is just their adaptability -- their "programmability", in a real sense, whether by an agent or by "natural selection".

Generally, then, I think it's possible to understand "information" in a kind of naturalized or "de-agentized" sense that still retains some notion of "meaning". In this blog, I've been using this same naturalized concept of information to suggest a mechanism for consciousness, in which the intermediate, information-like structure is just the phenomenal world as created and presented by consciousness. Qualia, in this understanding, are simply the tokens in that structure, or the bearers of the difference that "makes a difference".