TDE-R

And the slithy toves...

TDE = Chomsky's snowflake

You may have noted the rather unusual choice of letters (F,L,T,P) denoting the key parts of the heterodyne. This choice is hardly accidental. These labels reflect the four main types of lobe in the cerebrum, F(rontal), L(imbic), T(emporal) and P(arietal) - see figure 3(a). Their combination, called the TDE, functions as a hybrid (feedforward-feedback) 'motion control' governor. The TDE is the fractal pattern that acts as a generatrix (kernel function) for all biological intelligence. The TDE is the simplest ('Occam's Razor') mechanism that explains all of the major observations (interpolation) and then some (extrapolation beyond the data set - the gold standard in model making). In this page we describe how the TDE works. TDE theory integrates the most recent advances in research, including embodied, situated and embedded approaches to cognitive computing. TDE theory also makes several new claims, facts which can (and will) be verified.

Undergraduate neuroanatomy* presents us with the following basic facts**-

Frontal Lobe- associated with reasoning, planning, parts of speech, movement, emotions, and problem solving
Parietal Lobe- associated with movement, orientation, recognition, perception of stimuli
Occipital Lobe- associated with visual processing***
Temporal Lobe- associated with perception and recognition of auditory stimuli, memory, and speech
Cerebellum- regulation and coordination of movement, posture, and balance.
Limbic system- regulation of emotions, contains the thalamus, hypothalamus, amygdala, and hippocampus.

While this list is adequate for general educational purposes, it is theoretically unsatisfying. Medical science does not mention the existence of a unifying functional scheme. Such a scheme, were it to exist, would make it easier to claim evolutionary provenance for human intelligence, since we know that the paleohistorical increments in vertebrate intelligence form a series of genetic stepping stones over time. Therefore such lists can in fact be quite misleading, particularly for the ambitious engineer-philosopher, whose goal is as follows- "we can only claim to truly understand something if we can actually build it".

* Bryn-Mawr college website.

**The TDE fractal kernel is made from four lobes, Frontal, Limbic, Parietal and Temporal.

***The Occipital lobe is part of the Parietal lobe, and plays no major part in this level of discussion. Later, we will need it to explain the role of saccades in the TDE scheme.

Inverted 'U' data path

Noam Chomsky suggested that linguistic cognition arises because of a putative fractal pattern and/or process which he likened to a 'snowflake' (presumably, a reference to its fractal nature). It is the ambit claim of this research that the TDE is Chomsky's eponymous snowflake. The first chapter in this website demonstrates how the TDE forms the cybernetic basis of all intelligence. The first step in its discovery was analysis of the human CNS using the tripartite 'systems' approach - (a) where does the information come from, (b) how is it changed, and then (c) where does it go to.

Figure 3(a) depicts the main data flow paths superimposed upon the anatomy of the left hemisphere only, for simplicity. The situation in the right hemisphere is a mirror image of the left, considered from the anatomical (but not functional) point of view. This data path consists of a single loop:- {input(1) -->P(2)--> T(3)--> L(4)--> F(5)--> output(6)}. Specifically, each 'channel' (input 1-2-3 and output 4-5-6) contains a data hierarchy. The input channel data hierarchy is 'bottom-up', and acts like an information reducer/compressor. The output channel data hierarchy is 'top-down' and acts like an information expander/amplifier. These two hierarchies are united at the 'uppermost' point into an upside-down 'U' shaped path, as in figure 3(a) below.

The target is externally (third party, world) driven while the goal is internally (first person, self) guided
However, this data path's simplicity hides the main cause-effect relationship, which is a cybernetic goal-seeking DRIVE which acts BETWEEN the G(oals) in the left, motor channel and the sets of candidate T(argets) in the right sensor channel. If the reader loses sight of cause and effect, they should briefly consult the cybernetic view, figure 3(b).

For an intuitive 'grasp' of the meaning of these terms, imagine a gunsight (or camera lens) equipped with crosshairs. The target is the enemy combatant or vehicle (or photographic focus) you want to hit (snap), while the goal is the intersection of the crosshairs. The target is externally driven while the goal is internally guided. Now extend these concepts 'outward' and abstract them 'upward', if you will, and you have understood the key dynamic driver of the GOLEM. A subject-oriented programming paradigm called PEGS (or TEGS for cybernetic tasks*) has been constructed to more faithfully reflect cognitive operations - see figure 3(c).

*Although PEGS and TEGS seem identical on the diagram (figure 3), there is a subtle difference which gets bigger when the paradigms are implemented, see www.PTEGS.webnode.com

Global TDE-R / Local TDE

The next step in the investigation consisted of an abstract observation - that the LOCAL data processing relationship between the four lobes of each cerebral hemisphere seem to resemble the GLOBAL data processing relationship between the left and right cerebral hemispheres and the cerebellum and basal ganglia (the global pattern, as recognised by Endel Tulving, is called 'TDE-R' in www.tde-r.webs.com ). The similarity is not an exact one, but nonetheless, it is remarkable, and certainly worthy of further investigation (see figure 4 below). Note the subtle change that has been made to the cybernetic view 4(b), in the other views. This is because in reality, the L-lobe (which is the limbic system at TDE level 2) is the master of the other three- drawing the three circles for F, T and P lobes with one in the center for the L lobe expresses this unequal contribution to overall governance.

Tulving's episodic/ semantic/ procedural model

Recall that the local T-lobe functions as a store of static values (behavioural setpoints), while the local F-lobe functions as a store of dynamic values (iterated offset/bias values, or 'saccades')*. If we apply the local-global (ie TDE level 2 --> TDE level 3) transformation, the entire L(eft) C(erebral) H(emisphere) should function like a global F(rontal) lobe, while the entire R(ight) C(erebral) H(emisphere) should function like a global T(emporal)** lobe. These predictions are empirically supported by the work of the famous Estonian-Canadian scientist, Endel Tulving. Tulving demonstrated that at the top level the brain is a declarative knowledge processor, with episodic (autobiographical, narrative, dynamic) knowledge stored in the LCH, and semantic (general knowledge, universal facts, static) knowledge stored in the RCH. Since this transformation forms the very heart of the discovery of the TDE pattern, the reader should spend time with the diagrams until they are sure that they understand this claim thoroughly.

*This task subdivision is reflected in the two parallel data paths (at local TDE2 level) revealed by the bizarre phenomenon of 'blindsight'. Its pathology was found to rely on two data paths between eyes and brain- the 'what' (identity) pathway to the T-lobe, and the 'where' (trajectory) pathway to the F-lobe. What is even more remarkable is that this relationship, far from being unique, is actually ubiquitous. It is fractal- at many points in the brain, there exists a functional (complementary) tension between identity (what) and location (where) computational data properties.

**In one of the most unfortunate scientific naming mistakes, the Temporal lobe is named after the 'temples' (the flat areas on either side of our
forehead), and NOT because it is associated with time. As if that isn't bad enough, the Temporal1 (c.f. temples) lobe is more closely associated with spatial variations, not Temporal2 (c.f. time, chronology) ones.

Explains positive symptoms of psychosis
The reader should keep in mind that the truest functional view of the TDE at each of its three levels is the cybernetic view shown in figure 3(b). If the reader feels they are 'drowning' in detail, they should always return to figure 3(c). However, the views depicted in figures 3(a) and (c) and figures 4(a) and (b) are also necessary in certain explanatory contexts. Consider the description given of the uppermost level TDE3. This is the level of language (actually, narrative) and hence is the level of self. Note that it is an INTERsubjective level- it treats all subjects (ie selves) as equivalent. This may explain, in a very concrete and structural manner, why the schizophrenic's brain often 'replays' non-self voices in echoic memory, eg they 'hear voices', often instructing the subject to act destructively, eg self-harm. Note that this can only logically be the case if thought is indeed a form of internalised language, as was once believed, but is now theoretically out of favor.

The architecture of the TDE offers a mnemonically convenient entry point to subjectivity, by virtue of the choice of prefixes. TDE Level 1 (TDE1) is INFRAsubjective (BELOW subject level), TDE level 2 (TDE2) is INTRAsubjective (WITHIN the subject) , and TDE level 3 (TDE3) is INTERsubjective (BETWEEN subjects).

The INTERsubjective stance of the third level (TDE3) represents a pseudo-objective viewpoint, in which each language expression (eg sentence, paragraph) is processed equivalently by every subject, irrespective of whether they are its author and originator, or a third-party witness to its production.

Author-audience (eg writer-reader, or speaker-listener) valency is determined by second level (TDE2) analysis (see Chomsky's minimalist program). Chomsky's idea of SURFACE and DEEP structure of syntactic expressions is partially correct, according to TGT. Chomsky's SURFACE structure corresponds to TDE2 level, where it manifests as SYNTACTIC subject and object in the sentential form. Chomsky's DEEP structure, however, corresponds to TDE3 level, where it manifests as the matching SEMANTIC entities, usually called agent and patient in the sentential form. In figure 4.1, the order of the three language layers, from top to bottom respectively, is:- SEMANTIC -> SYNTACTIC -> SYMBOLIC.

TDE Level 3 INTERsubjectivity approximates/simulates the omniscience of true objectivity

How can changing one's viewpoint from outside to inside make much difference? Where do I start- the advantages are legion! Inside the box, we no longer possess the infinite access to global information that the objective viewpoint mandates. Instead, when we are 'inside the box', we must view the outside world ( = other subjects and objects) through some kind of information 'window'. Far from being limiting, such windows allow us to express self-control (eg of movement, of behaviour etc) in concrete form, in terms of cybernetics, typically goal/drive-state differentials (GSD's/DSD's) between projected representations of the outside world (situation images, or SI's) acting within the projective plane of the window.

The INTERsubjective nature of language is the property which gives language so much of its power, but all powerful things can go terribly wrong - this is what appears to happen when people with 'positive' schizophrenic symptoms 'hear voices' telling them to behave in an atypical manner, eg commit criminal and/or violent acts. In fact, hearing inner voices is all but predicted by TDE theory - specifically, it posits equal treatment of all linguistic forms at level 3, irrespective of subject. If language were only processed by level 3 (TDE3), we would have no inherent way of knowing where it came from.

This source-agnostic property arises from the duplex (counterflow) information transfer in TDE/GOLEM's linguistic computations. When people with paranoid schizophrenia hear inner voices, these voices seem all too real, often appearing to belong to familial authority figures, typically parents. TDE theory also claims that we model others by representing them internally as agents/subjects. We model them not only as static data structures (eg facial recognition) but as dynamic centres of narrative (episodic knowledge/ predicate logic) activity. Some scientists believe schizophrenia arises out of a flaw in the brain's efference copy processes. In the page titled 'efference copycats', efference copy is debunked as a falsehood, and a potential red herring.

TDE Level 3 INTERsubjectivity approximates/simulates the omniscience of true objectivity

Suggests key role for the cerebellum in cognition
Figure 4(a) depicts the main language path in the human CNS. This consists primarily of a duplex (counterflow) data path through Broca's Area (on the outgoing, syntactic channel) and Wernicke's Area (part of the incoming, semantic channel). The TDE model makes the following prediction - that there should be a contralateral global data path leading from Broca's/Wernicke's areas in the cerebrum to a matching language-critical region in the cerebellum. This is indeed what Silveri et al* found, a Broca's-like aphasia due to right cerebellar trauma.

Contralateral / Ipsilateral governance
At the behavioural level TDE2, there are two cerebral TDE patterns, each one exerting contralateral (opposite sided) governance** over its subordinate cerebellar image, which in turn exerts ipsilateral (same sided) somatic governance. This is the level (TDE2) at which 3D spatiotemporal functions apply. At the highest level (TDE3), the level of the self, the 'space' in which each subject's agency is embedded and applied is a socio-linguistic and not a spatiotemporal one. Note that at the topmost level (TDE3), neural representations are inherently subject-agnostic. When we construct internal data structures of our own (self-generated) speech, they are identical to those we construct for speech we hear (other-generated). TDE3 is an INTERsubjective layer.

Consciousness evolved for animals
Level 1 and 2 of the TDE are sufficient to generate animal behavior and consciousness. There are sixteen Level 1 (TDE1) heterodynes, each of which implement semantic embodiment - -see figure 4(a). These are grouped into four level 2 (TDE2) heterodynes, which implement semantic situatedness, via a subjective (perceptual control theory) mechanism. Up till level 2, the TDE architecture represents nothing more than a 'motion control' design, as in computer controlled digital movie cameras***, and other types of robot and telepresence drone.

Language evolved for humans
Level 3 of the TDE is the only level which humans do not share with most animals. True, higher primates like bonobos can understand and use 100's of human words, however, they don't understand grammatical (syntactic) variations. At level 3, knowledge is outputted for INTERsubjective communication, and inputted for INTRAsubjective storage. The purpose of language is semantic-knowledge interchange, storage and transfer. Language has the capability to enable one subject to communicate experience itself to another.

Figure 4(b) is included to remind the reader that adopting a nuanced interpretation of the functional geometry is at times unavoidable. In this diagram, the F3-P2 and L3-F2 are depicted as overlapping, forming the left thalamus. Similarly T3-P2 and L3-T2 overlap, forming the right thalamus. The overlapping of L3-P2 and P3-P2 represents the pons, the 'bridge' (a data pipe) between cerebrum and cerebellum.

*Silveri, M.C. Leggio, M. Molinari, M. (1994) The cerebellum contributes to linguistic production: A case of agrammatic speech following a right cerebellar lesion. Neurology 44(11):2047-50

**The generic term 'governance' or 'guidance' is preferred to the more commonplace 'control' because governance includes BOTH command (excitatory feedforward signals) AND control (inhibitory feedback signals) as prime constituents.

***In the early 1990's, the author was employed as control engineer (cyberneticist) at Digital Arts (SA), who designed and manufactured a portable 'mocon' (motion-control) rig for use by movie makers and TV advertisers in outdoor situations. The design was a straightforward PID implementation with DC servo motors.