#1.1 Cognition is Linguistic as well as Cybernetic

This research project was motivated by a desire to explain why some important brain functions are lateralised, using no ad hoc reasoning. In other words, what inherent architectonic features can we invoke to explain functional lateralisation (internally) and language use (externally)?

Using the 'paucity of input' argument to explain how human infants learn language so quickly, easily and undeniably, Noam Chomsky suggested that the infant is born with inherent linguistic capability. Later, Steven Pinker suggested that we are all born with a 'language acquisition device' (LAD) in our brains.

If we apply the well-known parsimony [1] heuristic known as Ockham's Razor, the simplest model that explains this fact is if the brain uses language internally, ie if cognition itself is inherently linguistic. In other words, the facts suggest that brains use internalised counterparts of words, syntax and semantics to do our thinking for us. 

Each animal brain consists of two hemispheres. This is because, except in the simplest case [2] all animals (including humans) are laterally symmetric. The canonical GOLEM is a duplex (bidirectional) two channel (bicanalised) model of finite state computation, as shown in Figure 1.1a below. It is also a model of the cognitive processes in each brain hemisphere. Therefore, each brain can be modelled by a pair of GOLEMs. This is not such a crazy idea- the clearly defined and delineated binary structure (two branches) of the Reticular Activation System represent a de facto proof of cerebro-hemispheric bicanalisation. 

Figure 1.1a The GOLEM model uses cerebro-hemispheric bicanalisation to model thought processes including emotionality and consciousness. The input channel (INCH) computes cognitive semantics (eg from observations of other creatures), while the output channel computes cognitive syntax (eg behaviour plans). The pairing of the data paths through Broca's and Wernicke's areas in the human brain forms a de facto GOLEM-like structure. 

Figure 1.1b The diagram above depicts the major stages in the development of the GOLEM model. Top left: Each channel contains a set of overlapping data hierarchies which, taken together, form a duplex linguistic heterarchy. Top right: the input channel (INCH) forms representations, or data types. It consists of 'has-a', or constituency relations. The output channel (OUCH) forms reproductions, or tokens of the INCH's data types. It consists of 'is-a', or inheritance relations. Bottom right: the many-to-few relations which characterise global linguistic ordering also exist within each of the six GOLEM segments. Bottom left: The synchronous relationship crreated in the transition from automation to animation layers provides one aspect of qualiate experience, while the allocation of intra-subjective experience to the embodied hemisphere (and, therefore, inter-subjective experience to the situated  hemisphere) provides the other aspect, the aspect noted by Gilbert Ryle.

According to the principle of linguistic cognition (PLC is one of the basic canons established by this research project), computation inside the brain proceeds along similar lines to language use outside it. This still leaves considerable room for uncertainty, since there is no clear consensus on how language itself functions. Even Chomsky appears to contradict himself. He is the one who drew our attention to the pre-existing capacity of new born infants to acquire language. Presumably, the infants have pre-existing framework, and they 'hang' lexical elements (ie words) and other syntactic 'sugar' on to this semantic base. The alternative, in which syntax precedes semantics, is nonsensical, yet it seems to be what Chomsky wishes us to accept. 

GOLEM theory uses the above 'semantics first' principle, as well as the principle of combinational semantics, namely that the semantics of a combination of words is the combination of the words' semantics. 

But where do semantics come from in the first place? The answer must be from the brain's embodiment abilities, ie its instinctual and acquired proprioception templates and neural plans for 'most useful' and 'gravitationally equilibriate' postural combinations. Figure 1.1c below depicts the computation of offset values (ie setpoint changes) from bias values (ie tonic setpoint measures).

Figure 1.1c Computation of phasic (behavioural) offsets from tonic (structural) biases are simple arithmetic operations (addition, subtraction). The interpolation observed in 'phi' illusion is therefore easily explained. The compound propriocept values (eg combination of joint angles #1 and #2) form a semi-finite state machine, because, while there are only a finite number of cybernetically preferred  propriocept values, dynamic, inertia-driven interpolation between these 'poles' cannot be avoided.

1. parsimony is taken to mean a higher order type of simplicity in this context

2. jellyfish are axially symmetric, while some other multicellular 'cooperatives' seem to lack any cospecific symmetries at all

3. In 'The Concept of Mind', 50's wunderkind Gilbert Ryle identifies one of the key principles of flexible, subjective cognition, namely the ability to transform an internal (intrasubjective) viewpoint to its external (intersubjective) equivalent.