Lifelike Robotic Collaboration

Requires

Lifelike Information Integration
 

Ron Cottam, Willy Ranson & Roger Vounckx

Abstract

           As we all (sic) were taught in school, the mammal eye works to create an inverted image of the viewed scene at the retina (whose orientation is rectified by the brain). Not so. There is no integrative capability at the retina to perform this function. Any “image” is generated much later, in the various layers and centers of the brain: it only “exists” within the (abstract) unification of high-level consciousness, and never in any “real” sense describable by science. If you are viewing this text via a computer screen or through the printed word, the same constraint holds: it does not exist at all as a unified entity outside your brain or imagination, merely as a collection of informational elements devoid of any implicit organization, which was transmitted through the Internet by a means which has been formally (scientifically) structured through the application of our imagination to achieve our aim of reproducing patterns across space and time. The same argument holds for the entirety of our environment: it is all beyond representation by (current) science. Not only does this argument apply to “objects”, it applies to equally well to any and every subject of discussion. Most particularly, in the current context of interest, we should not expect to find that a robot is capable of responding as a “black box” to external stimulus on the basis of an internally integrated “motive”, except where that “motive” is completely relatable to its formally unified degenerate representation – namely the binary “it exists” or “it doesn’t”! Such a quasi-hierarchical relationship (along with any “algorithmic” complexity it exhibits) is both nominally and functionally trivial when compared to the styles of real complexly-hierarchical operation which characterize living organisms. We should consequently beware of attributing anthropomorphic integrative unification to the internal workings of a “black box” robot unless it is entirely predictable (a character corresponding exactly to the quasi-hierarchical condition referred to above), in which case any resemblance of its actions to those of a human is far from likely, to say the least! So, can we describe and develop robots “in our own image” by the application of scientific techniques, or not? Or does the problem which must be addressed reside elsewhere? Descriptions of the natural world and the placing of robots within it which derive from Evolutionary Natural Semiotics (ENS) by way of signs are untouched by this dilemma. In the context of ENS, any formalized representation is derived pragmatically (but less-than-algorithmically) from its own scale-local grounding, and the various scale-localizations are coupled through and within the context of a global-to-and-from-local correlation which mediates between the scale-local groundings of a global grounding which it also creates. “Reality” (in a scientific reductionist sense) then refers to nothing more that the lowest level of description which we can be bothered to deal with, whether that be the atomic level, super-strings, membranes, … The descriptions which we habitually employ for “systems” which are internally structured in a network-like manner are suitable if, again, the network structure is amenable to complete (formal) integration reductio ad adsurdum, but for a “system” which exhibits “useful” complexity, they are worthlessly simple or simplified. Within ENS such representations (where we view the “system” as a whole and simultaneously its network-like internal structure) have the character of quasi-external representations, whose (cautious) applicability depends primarily on their degree of representational equilibrium. Much effort is currently being expended in developing “internalist” models of operational situations, rather than the “externalist” ones said to be characteristic of scientific endeavor. It is difficult to imagine, however, how a uniquely internalist representation of a “conscious” or aware state can or could be useful: its existence would imply not only the usually-quoted criterion of lack of knowledge of the causes of received stimuli, but also the complete absence of any attempt to investigate or imagine the origins of those stimuli. To do so requires the construction of an (imagined) externalist model of the situation: to not do so seems to imply lifelessness! Consequently, it makes more sense to describe living interactions as a negotiation between internalist and externalist representation, through a process which mirrors the internal-external negotiations which lie at the roots of human consciousness. Human consciousness is “singular”, in that it only exists as an individual unified “entity”. It is within the “sufficient interpretation” and correlation of a multiplicity of informational details that this text becomes (nothing more… just “becomes” itself) within our consciousness. Its existence emerges from the process of integrative interpretation (or interpretive integration, if you prefer). This process of the emergence of the informal from the formal (simplistically describable as emergence of the analog from the digital) is the very nature of living entities. It appears most obviously, but not uniquely, in the generation of analog protein folding from the digital code of DNA. Science does not merely omit this emergence from its confines; it expels it, as being too difficult to deal with. A lifelike nature is by definition external to a scientific development! So, how are we to develop “lifelike” robots? Ultimately, not uniquely by digital computation, although this can provide effective interfacing between a central information processor and the outside world. This is itself the nature of our own brains: a central really parallel processing style, whose operation is most closely related to the superposition-and-selection mechanisms of quantum mechanical interaction, and integration and differentiation of the results of this processing to serve localized output and input nodes. Currently this style of integration and differentiation is far beyond our constructional capabilities, and while a prime target must be to investigate and develop lifelike information integration, we can nevertheless achieve useful preliminary results if we couple our targets to the means which are available, so long as we do not fool ourselves into thinking that this will be sufficient.
 

______________________________________________
______________________________________________