Synthetic Cognitive Architecture

This is not my only moribund blog, I have two others.  My unfulfilled aspirations are like London buses.

Well here goes again, I may revive all three for a while, let's see.

My silence of late does not betoken a terminal change of direction on my part, but rather a further deterioration in the tenuous connection between what I am thinking, and what I am writing (mostly nothing), mitigated only by some wiki writing for github "projects" conceived as an alternative approach to my one principle objective (discussed here).
The direction of my intellectual pretensions remains: to add a chapter of my own to the story which began with Aristotle's conception of demonstrative science, and progressed first to Leibniz's idea of a calculus ratiocinator, providing a decision procedure for the whole of science and then to Carnap's ambition, armed with the more adequate modern logics available to him, to facilitate rationality in science through its formalisation.

Carnap pre-dated, and probably did not anticipate, the digital electronic computer, so he conceived of himself as, in some respects, following in the steps of Frege and Russell, rather than Leibniz.  His project for science failed for many reasons, which we need not go into right now.  It would have been sufficient that the complexity of formal proofs makes reasoning so arduous as to prevent its adoption by scientists, even had they felt the need for greater deductive rigour.  The advent of the digital computer changed all that.  Leibniz's project (within the limits which modern theory imposes upon it - at best a partial decision procedure) became more plausible, with the more advanced computational machinery continuously advancing leaps and bounds.  That machinery might make the burden of formality less onerous and its prospective benefits more alluring.

With the advent of the computer, a third centre of academic competence in logic, more vigorous and better funded than philosophy or mathematical logic, took over in its own way the thrust towards formality and its automation.  This new discipline had of necessity to embrace formal notations, and to mechanise their use.  Admittedly, most of these new formal notations were programming languages, suitable for describing algorithms, but not for the formalisation of science.   The complexity and unreliability of the resulting software begat the ambition to formally verify the software, and this lead some computer scientists (and some industrial corporations) into stronger formal systems of more general applicability.  The use of computers for formal deduction has continued to advance throughout almost the entire history of electronic computation, and the capabilities of the resulting systems have advanced considerably.   It nevertheless remains the case that formal reasoning about non-trivial problems, with the best machine assistance, takes considerably more time than traditional Journal style proofs.  The mass of extra detail which characterises formal proofs, in contrast with the extant tradition in mathematics, is easily coped with  by computers.  But finding a viable proof strategy requires mathematical intelligence, of a kind, so far, beyond the achievements of AI.

Hopefully, this defect of intelligence in mathematical software will be remedied, and my own project anticipates that remedy, as well as aspiring to contribute to it.  The stage is set, by these approaching advances, for the realisation, not only of Carnap's logicist project, but of something which eclipses in scope the ambitions of Aristotle, Leibniz and Carnap.

A particular feature of the three precedents I have cited is their breadth of scope, encompassing the whole of science.  Though there is much research in Computer Science which might contribute to such a project, there is nothing which encompasses the whole.   Thinking about that whole is my preoccupation, aspiring to write.  The advance of computer science, and of cognitive science, now make even those broader conceptions too narrow, and my own project embraces and advocates a wholesale paradigm shift from computation of data, to deduction over propositions.  This I now characterise as Synthetic Cognitive Architecture, an enterprise which belongs to a new kind of constructive philosophy which may be seen as a successor to Carnap's Logical Positivism.  This is philosophy as engineering.

At the core of this architecture is a logical system which facilitates inference across huge and diverse propositionally interpretable data.  A first essay at such a system was the ideas I christened X-Logic (the X, as in XML, connoting linguistic pluralism), ostensibly the subject matter of my blog of the same name.  In that same blog I now propose to discuss not only the logical core of the cognitive architecture, but more broadly, the epistemology underpinning the system.

Carnap's corner may be in the game too, where I hope to say something abut the relationship between this embryonic architecture and its distinguished (if ill fated) predecessors.