"These ideas [theories of Cosmos] cannot explain everything. They can explain the light of stars but not the lights that shine from planet earth. To understand these lights you must know about life, about minds [theories of Cognos]." - Stephen Hawkings
Abstract
While most scientists believe the brain is some kind of information processing system, there are others who think we need to discard the computational metaphor [1].
The division of the brain into its software and its hardware, however, is more than just a matter of naming convenience and scientific convention. These categories really do 'carve nature at its joints'-[2]. They are the result of using the type of reasoning known as retroduction (a.k.a. 'abduction'). If we assume that the mind is the brain's software, then most of what what we know about brains and minds becomes 'unremarkable'-[3].
Modelling the brain as an automaton such as a Turing Machine or a Von Neumann processor with shared memory space for both data and instructions is usually regarded as the next logical step to take. However there are good reasons for not doing this, and using a goal-oriented approach instead. The goal-oriented view of computation has several advantages, as follows:
(i) It is biologically plausible. Most organisms both simple and complex exhibit drive-based (goal-oriented) behaviours
(ii) It can subsume the Turing/VonNeumann model within its purview (ie include it as a subset), by making use of the dichotomy between declarative and procedural [4] computational paradigms. While the declarative paradigm maps naturally to goal-oriented processes, the procedural paradigm maps equally well to finite automata controlled by stored instructions. Indeed, many modern treatments regard declarative (explicit, goal-oriented) and procedural (implicit, script-based) paradigms as the two logical endpoints of a computational continuum.
(iii) It can in turn be subsumed by (ie expressed in terms of) a cybernetics framework, which must form the basis of any ultimately true model of biological governance.
This web-based dissertation describes the discovery of a unified model of brain, mind and self called the GOLEM (Goal-Oriented Linguistic Emulation of Mind). Its three main advantages over previous attempts to model human thought are-
(I) 'software'- its non ad hoc use of consciousness and emotionality as prevailing operational metaphors for thought. We know from introspection that this is the sine qua non, as well as the acknowledged gold standard, of models of mind (MOM). Any MOM that didn't include consciousness and emotionality as key constituents wouldn't pass the common sense (the 'pub') test.
(II) 'hardware/wetware' - its outright rejection of the modifiable synaptic view of neural plasticity (persistence of learning and memory consolidation). Instead of the internal, synaptic model of neuroadaptation, it suggests the use of an external, circuit-based model. All existing facts about pre-synaptic and post-synaptic physiology as well as the function and location of neurotransmitters are just as consistent with circuit-based neuroplasticity as they are for synapse-based neuroplasticity.
(III) The GOLEM theory of ABI (Artificial Biological Intelligence) uses the known facts about sleep-based memory consolidation to construct a complete explanation of sleep, its purpose (teleology) as primary mechanism of goal-based behavioural modification, and the mechanism which produces the patterns of neural adaptation required for new behaviours. The GOLEM model includes both a functional theory and a infralinguistic architecture from which the diurnal, circadian and ultradian phases of sleep can be derived directly, using non-circular, non-ad hoc arguments.
The GOLEM model of animal and human cognition may not ultimately prove to be true, but at the moment it is the only (to best of the author's knowledge) truly unified theory, one which-
(a) is complex enough so that all common behaviours, especially consciousness and emotionality, are modelled in significant, non ad hoc detail
(b) is simple enough so that all its constituent functions are mutually consistent (internal validity). Therefore it behooves [5] the (cognitive science) community to treat it seriously and devote significant analytical resources to its ultimate confirmation or refutation.
2. Plato , Phaedrus 265d-266a
3. Retroductive arguments involve an inference to the best explanation: their conclusions are supposed to be the best explanations for their premises. The logical form as one of the possible syllogisms was known to the classical world, but in its 'modern' form is usually ascribed to 19th Century American logician Charles Sanders Pierce.
4. some prefer 'imperative' to 'procedural', but they have the same meaning- the execution of externally supplied commands usually in the form of instructions to be followed
5. I would prefer to avoid a repetition of the experience of Alfred Wegener, plate tectonics founder, who did not live to see the full acceptance of his idea by the scientific establishment.
Acknowledgements & Background Events
Before listing those who provided formal academic assistance, I would like to thank my partner, Lesley, without whose love and forebearance this task would not have been completed.
I also wish to thank Professor Noam Chomsky for his informal role as mentor. We now know that language is the engine of the mind because of the pioneering work of Professor Chomsky, who has earned the right to be called the 'father of cognitive linguistics'. The knowledge that he was just an email away gave me the intellectual courage I needed to reject orthodoxy and create some genuinely new ideas, even when these ideas clashed with his.
Shortly after my project was convened in 2008, Flinders University underwent major staff changes, resulting in withdrawal of official access to key IT resources and laboratory facilities. I continued to schedule meetings with available sponsors as long as possible, being granted adjunct (post-doc equivalent) status for that purpose, including library card and multi-media access, to be reviewed on a three monthly basis. In 2009 I gave a public lecture in the Flinders Teletheatre, which was well attended and generously received. I formally enrolled in an honours science program in 2011, based on my research findings up to that point. My goal was to create a plausible honours study schedule, one that would allow me to attain my research degree, while still remaining on the topic of my choice, my new bioplausible AI called the Tricyclic Differential Engine (TDE). Finally in November 2012, I graduated with 2nd class honours [1].
1. Professor Paul Calder assisted me with the last minute production of the three copies of my thesis officially required, stepping into the gap left by the unexpected reassignation of my original supervisor. I am deeply thankful for his last minute assistance.