Overview of research project - Author's Preface
In spite of not being published in a peer-reviewed journal, this research project (2008 - present) is a serious investigation into the relationship between human thought (including consciousness and emotionality) and computer technology (hardware and software). Each stage of the investigation is published using one of the free, open-access website hosts. If another researcher's work is discussed critically (eg Bernard Baars GWT), an email is sent to that person, asking them if I have satisfactorily understood their work.
The project was initiated at Flinders University in South Australia* in 2008. At the time, Flinders official policy was that the solution to the problem of human brain and mind function was best approached in a multi-disciplinary manner. This project was given informal but official imprimatur during a series of meetings with key staff from Flinders School of Psychology (Prof. Leon Lack), Flinders School of Computer Science (Assoc. Prof. David Powers) and Flinders Medical Centre (Prof. Marcello Costa, head of Neuroscience). The first step was my achieving Honours level in an official research program. This was successfully completed in 2012. In my Honours thesis, the brain's fractal architecture was modelled as a bio-plausible Turing equivalent hybrid automaton with a circular memory 'tape'. This machine was called the TDE or TDE Differential Engine (a recursive anagram). After my Honours year, the structure of FUSA* changed radically, as funding pressures began to impact on Academic standards. All of the three original sponsors of my program have since retired.
Introduction - Qualiate Systems
'Science' (ie an academic consensus of working scientists) still does not know how to study subjective ('qualiate') systems like the human self and mind. Indeed, it has not even reached agreement on basic terminology. Should we just call a spade a spade and refer to such entities simply as machines (which is what they are, unless one is a vitalist or religious, of course). Or should we pull back slightly, and choose from the following list of commonly deployed adjectives- qualiate, psychophysical, phenomenal, subjective? In this research, I refer to something called 'The Subjective Stance', a gedankenexperiment whose purpose is to show how radically the Cognitive Science term 'subjective' differs from the everyday use of the word.
While the self and mind resist mainstream analysis, the human brain ('wetware') is more amenable to conventional forms of scientific and mathematical analysis. Consequently large amounts of money and technology continue to be applied to the problem, albeit with little to show for it. The image that appears on the website's HOME page is a notable exception. It represents the cortex as a map of perceptual categories, confirming one of the GOLEM model's axioms.
Nevertheless, the worst possible choice that any scientist in my position can make is to turn their back on the difficult, poorly defined parts of the problem. The truth is, this is scientific research into just what constitutes a 'self', what in the last century was sometimes called the 'soul'. To avoid the hard bits is tantamount to giving up. This research project fits neatly into the 'too hard basket'.
Proof of Concept- a conscious machine
The conventional approach to modelling is to emphasize external veracity- ie to increase the correlation between observed data and model predictions. This research project places primary emphasis on internal validity, ie aims to construct a model that is in ‘agreement with itself‘, one in which all the parts fit together, which is, in the words of AI pioneer Al Newell, a ‘unified design’. The biggest need at the moment is to demonstrate a machine that is convincingly self-conscious. This machine will in all likelihood be simpler by several orders of magnitude than the real human self and mind, but it will provide a proof of concept.
Contrary to both popular belief and scientific orthodoxy, the human brain and the digital computer share key architectural features. The number of common features and their core functions is sufficient that it is credible to describe human brains as computers. In particular, the way that behaviour is created in humans closely resembles the process of source code compilation, linking and execution in computers.
Brains differ from computers in several significant ways. Crucially, their on-line operation mode is one we call ‘conscious’. The matching off-line state of unresponsiveness also has a special term -sleep. Like humans, animals that are much simpler in plan, such as nematodes and fruit fly, also sleep. Furthermore, if an animal sleeps, it does so in response to a homeostatic drive cycle. Animals strictly denied sleep will die, suggesting that sleep serves a vital function. This research is based on a hypothetical model of conscious computation, one which relies on sleep for long-term memory consolidation. During waking periods, the model uses attentional processes to create short-term memories, by uniting pre-existent features present in long term memory.
GOLEM - cybernetic & linguistic cognition
The GOLEM model enhances the basic I/O system model (ie input->system->output) with a revised version of cybernetics, one in which familiar homeostatic setpoints are augmented by heterodynamic ‘offsets’. The model requires goals or drives to be defined linguistically, a modelling feature that requires more precise and unified definitions of semantics and syntax than those currently available. Part of this work therefore consists of redefining semantics and syntax as synchronous and asynchronous finite automata called Moore and Mealy machines, respectively.
Previous work
This latest phase of the project concentrates on constructing a credible model of machine consciousness. Previous phases (listed below) have revealed other aspects, on which this current phase builds on.
1.
2.
3.
4.
5.
6.
7.
8.
9.