From Aaron Sloman Tue Mar 3 19:56:14 GMT 1998 To: psyche-d@listserv.uh.edu Subject: pre-conscious meaning (was Re: Causation and Awareness) I tried posting to the group sci.psychology.consciousness but I got no acknowledgement, so I guess this did not get sent. Apologies if you get it twice. Aaron Newsgroups: sci.psychology.consciousness References: <1867.199802261004@twain> From: A.Sloman@cs.bham.ac.uk (Aaron Sloman) Subject: pre-conscious meaning (was Re: Causation and Awareness) Chris Malcolm wrote > Date: Thu, 26 Feb 1998 10:04:39 In response to Walter Freeman [WF] > >Intentionality in higher animals differs from tropisms and homeostatic > >processes shared with their lowly predecessors. The difference is manifested > >in the development of meaning, which is an integral part of consciousness > >that is not adequately defined or understood. [CAM] > I would like to change two words in the last sentence, so that it reads > "The difference is manifested in the development of meaning, which AS > an integral part of consciousness is not adequately defined or > understood." > > I agree that meaning AS an integral part of cosnsciousness is not > understood. There is, however, an interesting non-conscious analogue of > meaning which we are currently learning a lot about, and which may help > to throw light on its mysterious and grander cousin, conscious > meaning. What is this non-conscious analogue of meaning? It occurs > anywhere that information is encoded in a language of some kind which > has a grammatical structure, and where this encoded information is used > to control a mechanism of some kind so that it produces behaviour > appropriate to what the encoded information is *about*, and the encoding > and decoding are performed by purely automatic machinery. I think this is absolutely correct. It's for good reasons that the phrases "information processing", "information technology", "information engineering" are now more and more widely used. A simple case is a computer processor using a bit pattern as an address: it takes that pattern to refer to a location in the virtual memory of the current process, and it demonstrates that it does that by performing some action on the location specified by the address. A more complex case is a file-store manager program keeping track of which parts of a disk are occupied and which are not, and reasoning about whether a new file can fit into one of the available gaps, and if not compacting the store to make room. Yet more complex examples are flight control systems and plant control systems which receive information from various external transducers and take complex control decisions on the basis of that information and previous information of various kinds. All this is commonplace in software engineering, where the word "information" is used to talk about semantic contents transmitted, stored, manipulated and used (as opposed to the Shannon/Weaver mathematical notion of ``information'' which is purely syntactic, and therefore badly misnamed.). This shows that meaning, reference, information can be found in systems which do not have architectures rich enough to support all mentalistic notions, e.g. desire, belief, intention, fear, etc. Dennett (e.g. in Brainstorms) missed out an important intermediate type of stance when he talked about the physical stance, the design stance and the intentional stance, for his intentional stance presupposed the existence of a complete more or less rational agent, whereas the design stance referred to some lower level of functional organisation which (if I understand him correctly) does not involve any meanings, only structural and functional relationships such as might be described in a circuit diagram. If we allow that part of the explanation of what's going on inside an operating system, or an automatic landing system, or an office management system, or a brain, is the manipulation of meanings, then the analysis of how such systems work involves taking up an ``information level design stance''. Not only is this commonplace in engineering I think it is spreading among brain scientists. (Which sorts of information about approaching objects reach the amygdala? How is such information used?) This information-level design stance makes no presumptions about rationality, unlike Dennett's intentional stance. The CPU which fetches the contents of a specified memory location and then copies it to another specified memory location is neither rational nor irrational: questions of rationality do not arise for such a simple system, which lacks the architecture to support desires, beliefs, etc. Only when the architecture is rich enough to contemplate alternative ways of achieving goals does a distinction between choosing rationally and choosing irrationally begin to get a grip. A file system manager which frequently compacts the file store even when there is a large amount of unused filespace, and thereby slows everything down unnecessarily, might be described as irrational, though this could be misleading if the system has no way of reflecting on its reasons for doing things, and considering different reasons. (How many animals can do that?) (I've written papers on this topic, available at the Cognition and Affect ftp directory ftp://ftp.cs.bham.ac.uk/pub/groups/cog_affect/ ) [CAM] > It is that *aboutness* of the information, and its useful employment by > the mechanism, that gives us the weak (non-conscious) form of Brentano's > Intentionality, and hence the weak (non-conscious) form of meaning. yes. (In the new subject-line I've used "pre-conscious" in case "non-conscious" suggests a system which could be conscious of things but happens not to be conscious of the particular item of meaning.) > If you believe that conscious meaning of the kind we humans enjoy is an > evolutionary development not present in more primitive forms of life, > then it is hard to resist the conclusion that conscious meaning > developed *after* non-conscious meaning, and quite possibly *from* > it. yes. > If that is the case, then studying non-conscious semantics is > clearly a useful route towards developing an understanding of conscious > semantics. yes. (That's what brain science and AI are both doing.) Moreover human brains still have powerful mechanisms of the oldest kind providing the infrastructure for, and to some extent controlling, the more recent architectural extensions which support various forms of consciousness (and also more sophisticated forms of non-conscious information processing). In other words, the vast majority of the meaning processing (or if you prefer information processing) in a normal human nervous system is not conscious processing. I recently put together some slides relevant to this for a talk at a literary society in london (on how machines could love). If anyone's interested they are available in postscript format at http://www.cs.bham.ac.uk/research/projects/cogaff/96-99.html#31 (prints as 11 pages of A4). Aaron === Aaron Sloman, ( http://www.cs.bham.ac.uk/~axs/ ) School of Computer Science, The University of Birmingham, B15 2TT, UK EMAIL A.Sloman@cs.bham.ac.uk Phone: +44-121-414-4775 (Sec 3711) Fax: +44-121-414-4281