Sapient Structures for Sapient Control
 

Ron Cottam, Willy Ranson & Roger Vounckx

Abstract

Sapience is a direct result of hierarchical structure.

Formal information processing systems operate mono-rationally and within temporal completion: living systems do not. Living systems are functionally hierarchical; formal information processing systems are not, although they may often appear to be so. Living systems can achieve sapience: others can not. Formal processors consist of predictable elements at every level of their organization: living organisms most ably demonstrate that the most resilient stable systems are composed of nominally unstable elements. Formal information processing systems may be able to provide enough processing power for an aware entity to survive in a restrictedly hostile unaware environment, but it will certainly not be sufficient to guarantee survival in the multiply aware environment of a biological ecosystem. Life is, and must be, the most relevant paradigm for sapience.

The biggest block to synthesizing high-level sapient processors is the formally mono-rational sequential nature of current hardware. Large processing systems run up against the capacitive and relativistic capping of communication speed. It is the spatial information processing density which must be maximized, and not simply the quantity of processing. This is why the distinction between formal complication and natural complexity is so important. Rationally constructed and operating computational machines can exhibit extremely complicated behaviour, but their information processing density is irrevocably coupled to the physical size of their individual processing elements. Complex living systems are not automatically subject to the same limitation, and they can explore their phase spaces and generate new information in a manner which is more related to their characteristic Lyapounov exponents than to their characteristic elemental size.

(Conventional) Boolean logic gates throw away the physical component of information and reformulate it from their power supply: this approach corresponds conceptually to the reductive destruction of information, whose re-use requires restitution from pre-established non-local memory: this does nothing to aid the rapid processing of large volumes of data. The extension of mono-rational sequential processing into its parallel counterpart does not change this. Formal parallel processing systems are sequential in everything other than name. Formal parallel processing implements sequentially: 1. separation of a process into independent threads, 2. separate processing of those threads, and 3. formal reassembly of the thread outcomes. True parallelism exploits continuous simultaneous interaction and separation of its different threads, in the manner of a quantum interaction rather than that of a linear superposition. In the world of computers, hardware/software co-designed concurrent processors are fast closing in on this ideal, but they fall short because of their limitation to Boolean logic.

Formal information processing systems are by their very nature and construction single-leveled: change in the bit output of a single (low-level) gate will of necessity influence directly a (high-level) decision output. Consequently, “everything has to wait until the smallest guy has done his job!” Thus - the imposition of maximal clock speeds for processors, which define how fast you can run them while still being absolutely certain that there is no discrepancy across organizational levels. Biological systems, however, have evolved to satisfy a major requirement of survival: it must not be necessary to wait for each and every bodily cell to react before jumping out of the way of a speeding train! (a classic case of this is the cortex-bypassing fear-learning pathway through the amygdala in our own brains). This is the central meaning of hierarchy – massively complex systems are provided with multiply-scaled adaptable reactions and choice in the face of diverse stimuli.

Evolution requires that expanding systems split into functional sub-units in a sensible integrated manner, if they are to persist. The materialization of a higher level single organization from a lower level multiply populated one is most popularly characterized by the emergence of novelty at the higher level. The vast majority of their relationship, however, depends on the cross-level transport of order, and not on novelty. Systems which are stable are precisely that – stable. The transport of order between system levels freezes a multilevel structure into a quasi-permanent form. Stability is not an “added value” which can be obtained from multilevel structuring; it is its very nature.

Natural hierarchical information processing structures do not necessarily rely on loop-like architectures, although they may use these at a motor control level. A more useful paradigm is that of an irreversible process, which may indeed be modeled to some (small) extent as a loop, but only by dropping out fundamental intimacy of interaction between the different operational directions. Simplistically, rather than "sense > perceive > act", which is formulated from individually formally-modellable "sense", "perceive" and "act" modules, somehow tied together, it makes far more sense to concentrate on the two tying-together ">" processes, whose interaction is principal. The question is not primarily one of "constructing knowledge out of information", but one of "propagating information through knowledge" (Nils Langloh) as a way of modifying it.

Hierarchical systems, then, are assemblies of different levels of organization which are more-or-less tied together in ways which promote simultaneously level isolation and level integration : their overall correlation is one of context and temporally dependent negotiated compromise, and not one of formal relationships. However, this cross-scale correlation itself makes up the defining unity of the system: it is as much a system property as are the operational parameters of the smallest constructional element.

True hierarchical systems can be decomposed into two quasi-independent mono-rational hierarchies, one consisting of the more usually noticed scaled levels, the other consisting of the inter-level negotiational transition regions. In a world where Newtonian physics provides a good approximation across a wide range of scales, the former consists of a set of Newtonian potential wells, reductive in terms of local level description towards the highest level. The latter then consists of set of inter-Newtonian level quantum-entanglement-like regions, each of which provides a rational ecosystem for its adjacent Newtonian level, and the set is reductive towards nonlocality. The complementarity of the two hierarchies is functionally symmetrical: the quantum-like system provides a rational ecosystem for the Newtonian one; the Newtonian system provides a rational ecosystem for the quantum-like one. Newtonian and QM descriptions are not parallel scale-related physical models, they are complementary representations of our surroundings.

The cross-scale correlation of levels in the (more normal) Newtonian system is mirrored by a cross-scale correlation of levels in the quantum-like one, whose nature is again a fundamental system property. These two hyperscalar correlations each provide ultra-high level entity-ecosystemic exchanges, much as we ourselves use each of logic and emotion to resolve the others’ decision-making dead-ends. One of the two hyperscale systems corresponds to “information which is accessible at specific scales of the global system”; the other corresponds to “information which is inaccessible at specific scales but which complements that which is accessible”. Segregation of these two fuels the generation of fluctuating asymmetries in the inter-scalar negotiations, much as our desires fuel our own beliefs and actions through the generation of conscious-unconscious asymmetries.

Inorganic systems can develop very restricted materializations of hierarchical character: even single crystals of the electronics-important zinc blende structure materials (diamond, Ge, Si, GaAs, InSb, HgTe, …) show cross-scale information transport which does not precisely correspond to the spatial symmetry of the lattice structure. However, the informational differences between small and large scales remain minimal, and although higher organizational levels may be marginally different from lower ones there is a high degree of self-similarity across scales. Living systems, however, expand to far greater system complication, segregate into more numerous organizational levels, and the closure of a specific level with respect to its neighbors is characteristically more complete. A consequence is that living systems are capable of generating extraordinary richness in their highest levels of their organization. It is also notable that, while the higher levels of an inorganic system remain extremely “physical” in nature, those of a living system are often far less tangible. Our own society of individuals is a case in point.

Intelligence may be said to be “the ability of a system to adapt its behaviour to meet its goals in a range of environments” (David Fogel). Sapience, or wiseness, goes much further than this, in its wide ranging considerations and cross-scale harmonizing and resolution of conflicting goals, causes and effects. We propose that sapience is a fundamental system property, but one which only manifests itself within systems whose hierarchical structure is sufficiently complex for hyperscalar interactions to be able to dominate the crude automatisms of inter-scalar slaving. Sapience is a direct result of hierarchical structure, and manifests itself as an aspect of life.
 
______________________________________________
______________________________________________