GOLEM is a canalised model of mind
GOLEM theory is based upon a basic two-channel information flow model, where information flows in to the input channel from the creature's (or robot's) sensors, and information flows out from its muscles (motors, effectors). James' 19th century concept of common (perceptual) coding is incorporated, resulting in the following modified mental image of the model (see figure below). In this model, the problems associated with precise definition of canalised linguistic terms is resolved by proposing that, instead of semantics and syntax, cumulative and differential versions of hierarchical Semantic State be used. Cumulative (sigma) SSH is an ascending hierarchical measure of state computed in a bottom-up manner by the input channel hierarchy, while differential (delta) SSH is a descending hierarchical measure of state computed in a top-down manner by the output channel hierarchy
The research was based on some insights (ie inspired, educated guesses) about cognitive function and its relationship to brain architecture. The main two insights were as follows-
(1 ) The brain is a type of computer, just like the GOFPC (Good Old Fashioned Personal Computer) that sits atop most peoples' desks
(2) The lateralisation of brain function, a feature which is present to some degree in almost all[1] animals, seems most pronounced in humans, where it is implicated in external language production. The core research proposal is that the left-right function differentiation between hemispheres is a global 'version' of the fore-aft function differentiation between channels. In other words, the entire left cerebral hemisphere acts like a global output channel, while the entire right cerebral hemisphere acts like a global input channel - see Figure below. Also see www.tde-r.webs.com. This core insight continues to appear on a persistent basis in all subsequent web documents, eg see www
This feature has the following dramatic effect - in the animal GOLEM (see left diagram), the right hemisphere generates all the creature's 'self', or embodied, behaviour plans while the left hemisphere generates its 'other', or third party, behaviour plans. Since the creature is only able to execute its own 'self' behaviour from those plans (because it doesn't have direct governance over the body of any 'other'), we say that it can only 'predict' the 'other's' behaviour [5]. However, in the case of the human GOLEM (see right diagram), the evolution of speech makes that indirect 'prediction' link much more direct. The human GOLEM is able to generate behaviour in two direct ways- by executing its own plans (ie thoughts), OR by getting another human to execute its plans. Granted, the other human can veto the effect of the speech commands, but it cannot claim ignorance of their explicit content, so long as it belongs to the same language group, something that animals are able to do. Therefore externalised language (remember that all thought is an internalised linguistic form) permits a much tighter behavioural governance connection between individuals, which explains cities and to some extent civilisation. It is explicit linguistic mechanisms (eg tail wagging dances [3]) which allows bees, termites, and to a lesser extent, ants to form large compact million-member colonies.
1. The pairing of two paleoworms to make a single, laterally symmetric superworm probably occurred via a modification of the gene code in a pre-cambrian environment. All animals (including cephalopods) have two brain halves, two lungs, but only one heart, liver etc.
2. Marais, E.N. (1937) The Soul of the White Ant, see also The Soul of the Ape by the same author.
3. Barron, A.B. & Plath, J.A. (2017) The evolution of honey bee dance communication: a mechanistic perspective. Journal of Experimental Biology 220, 4339-4346
4. Wescott, R. W. (1995) "Language and Civilisation: Contributions of Linguistics to the Comparison of Cultures," Comparative Civilisations Review: 33 : 33 article 3
5. Minsky (1965): To an observer B, an object A* is a model of an object A to the extent that B can use A* to answer some questions about A.