Working Language Memory & Emergence of Self

"..whereof one cannot speak, thereof one must be silent,.."
Wittgenstein, Tractatus Logico-Philosophicus

7. Optic chiasma 8. Limen insulae 9. Tuber cinereum with infundibulum 10. Anterior (rostral) perforated substance 11. Claustrum 12. Putamen 13. Lateral part of globus pallidus 14. Medial part of globus pallidus 15. Basis pedunculi 16. Mamillary body 17. Optic tract 18. Posterior (interpeduncular) perforated substance 19. Cortex of insula 20. Superior cerebellar peduncles 21. Substantia nigra 22. Mesencephalic (cerebral) aqueduct 23. Medial geniculate body 24. Lateral geniculate body 25. Temporal genu of optic radiation 26. Pulvinar of thalamus 27. Sagittal stratum 28. Splenium of corpus callosum 29. Upper lip of calcarine sulcus



The role ‘language’ plays in the efficient, adaptive integration of external and body-internal elements of the newborn’s environment is a necessary condition for the subsequent emergence of both awareness and consciousness as was briefly sketched before: “The body internal receptors assess the physiological status of the infant’s internal milieu, then a conducting neural pathway transmits the information to a developing integrative neuronal pool (somatic or autonomic) which filters, sorts out and relays the information to a neuron motor pool that gets set to execute adaptive responses by activation of the appropriate somatic or autonomic effectors (muscles, glands).  This way, elements of the internal and external environment begin to get simultaneously incorporated and commingled within the context of the pleasure / aversive emotions they generate, code-for and store in the process. Originally the responses are stereotyped and become as efficient and sophisticated as the reflex structure complexity (as modified) allows it to be at a given moment.

 Eventually, the ongoing interactive process slowly evolves a primitive psychological reflex structure with elements of awareness. The driving force behind this development is both need (Freudian diminution or extinction of urge) and an instinctual infant curiosity or need for stimulation. Contingent upon the subsequent attainment of the relevant intellectual, emotional, pre-linguistic and psychological structures, the concept of an external object in the environment slowly emerges. At this stage such objects are no longer considered by the infant in association with the satisfaction of his primitive needs or as an extension of self, now the objects  have an existence and permanence of their own. This post-awareness signals  the culmination in the development of an input > output sensorimotor reflex system, one that arguably an idealized robotic structure can, in principle, achieve. But the intercalation and elaboration of an experiential abstract event between body / external input and brain output (i.e., a conscience of thought) requires  a linguistic structure in place as a necessary but not sufficient element for the generation of that quality of thought. Past the first year of existence, that ‘feeling’ event heralds the dawning of a primitive conscience of self……

 Other than vision, none of the special senses discussed  have a more credible relation to language generation than audition. It is generally believed that the newborn’s body language and babbling noises are an attempt to reach the mother and repeat the mothers cooing sounds. However, during lactation the newborn is essentially cortically blind as we have mentioned earlier and it has been demonstrated that children born deaf and / or blind are also capable of displaying this behavior (William Wang, The Emergence of Language, 1991 Freeman); it will disappear after the child starts talking. If we accept the reasonable premise that such behavior represents primitive attempts to communicate by body language movements and primitive speech, we have to conclude that body input (visceral, hypothalamic, etc.) into language generating neuronal circuitry have preceded any other subsequent coding for external sensory input into the same anatomical loci where inherited generative grammar processing (along with proto-semantic development) may be developing. Any future codification into language structures during subsequent development will have to reckon with these permanent neuronal circuits daily reinforced by visceral and emotional parameters and consolidated into permanent memory.” See chapter on Visceral Brain, Language and Thought.

 Because of its gross anatomical location in the brain the parietal angular gyrus or ‘gnostic area’ seemed like the best candidate for a central rallying point bringing together somatic, autonomic, special sensory and kinesthetic information to bear upon the elaboration of language, thought and adaptive muscular and behavioral responses. We will now bring additional brain imaging evidence using ‘working language memory” in defense of our view on the reciprocal role of language on thought and the associated emergence of self. See Diagram 1 and Fig. 2 below (zoom for details).

Fig. 2 Left Postero-lateral Dissection, Human Brain

Memory is more than the static retention of previously acquired or inherited information about facts, feelings or actions. It requires, in the dynamic sense, the simultaneous presence of both present, concurrent state or event and a recorded past, plus the capacity to identify internal and mnemic causal relations between the two states.(B. Russell, Analysis of Mind). It is important at this point to distinguish between the different conceptual (modular?) subsets of memory. Immediate memory is concurrent with the precise moment the triggering object or event is present, the focus of current attention. It is of limited capacity (can hold ca. 7 items) and will quickly degenerate (in 30 seconds) unless the stream of thought is rehearsed along various reverberating cortico-thalamic neuron loops (see Fig.1), becoming now a ‘working memory’ able to, within a short time span, correlate and sort out the various cortical and sub-cortical contents visited (Fig.1), integrating into a propositional code the best adaptive response to the triggering stimulus, including a delayed or no response at all. This time span for the duration of re-entrant loops between cortical and sub-cortical sites comprises the short term memory. When it becomes in the perceived best biosocial interests of the subject to stabilize this experience, then it becomes ‘long term memory’ (Memory, Squire & Kandel, Scientific American 2000). Notice that there is no unique storage site for working memory and it is best conceived as a temporary memory capacity distributed among various loops operating in parallel (Fig.1).

The most important in our context is the ‘phonological loop’ that temporarily stores spoken words or meaningful sounds. Unfortunately, our conclusions will be based on data derived from the lexical loop activated by language comprehension tests (see Diagram 1). One most important participant in language working memory is the frontal lobe to guide ongoing cognition and behavior (Goldman Rakic) at conscious levels. In a previous chapter we described an ‘implicit memory’ loop controlled by the amygdala operating at unconscious levels (see Emotional Variable in the Logic Equation). We postulate that as long as the working language memory is active the frontal lobes will hold the information in spite of distracting concurrent events. We hold that while the brain is ‘talking’ (audibly or as 'inner speech') there will be thought production and a possible experience of self-consciousness with qualia, a distinctive human quality.



        Any interference with the inter-neuronal connectivity surrounding the ‘gnostic area’, whether established by abnormal development or disease, will result in varying degrees of dyslexia, blindness or semantic deafness. Wernicke’s area, located in the postero-superior temporal lobe provides a converging point for audio-visual sensory information thus providing the auditory and visual images of language transmitted directly to Broca’s speech area (motor image of language) in the left infero-lateral frontal lobe via the arcuate fibers connecting both. A quick view at Diagram 1 below will show that an interference with a proper elaboration of a motor image of language may occur with or without semantic concomitants.

Diagram 1

Diagram 1 (Legend)

Phonological Routes. Tested by reading aloud, even words not understood (semantic deficiency) or not pronounceable (syntactic deficiency). Other deficiencies are: anomia where subjects can not find names for things, or can not match spoken words with pictures or produce speech spontaneously .

Tests (where vision and audition are normal):

A. Spontaneous talking (Route 8>>6)

Only with syntaxis (3>>6)

Only with meaning (2>>6)


B. Repeating sounds (1>>6)

Only with syntaxis (5>3>>6)

Only with meaning (2>>6)


C. Reading loud (4>>7)

Only with syntaxis  (5’>3>>7)

Only with meaning (2>>7)


Lexical Routes. Requires prior convergence (8) in a putative frontal cortex where Motor image is assembled (Broca’s?)


A)    Spontaneous writing (8>>7’)

Only with syntaxis (Memory>Visual image>5’>3’>>7’)

            Only with meaning (Memory>Angular gyrus>>2’>>7’)


            B) Copying from text (1’>>7’)

            Only with syntaxis (5’>3’>>7’)

            Only with meaning (2’>>7’)


A)      Writing from dictation (4’>>7’)

Only with syntaxis (5>3’>>7’)

            Only with meaning (2’>>7’)


        Compare route Wernicke > 5’ > Lexicon’s memory > 3 > Broca with Wernicke > Angular Gyrus > 5A > Lexicon’s memory > 3 > Broca. We can suggest that radiologically demonstrated lesions to Wernicke’s area producing clinical aphasia with a significant semantic deficit are interfering with its connections with the angular gyrus. Aphasic patients are characterized by a measurable varying deficit in their ability to incorporate the meanings of words into propositional representations requiring their conscious discernment.

         However, the most important conclusion in support of our brain model of consciousness is the reciprocal correlation between language, thought and the associated higher order consciousness in these patients as will be summarized below. We remind the reader that our innate 'protosemantic potential, unlike Chomsky's ‘Universal Grammar’, in our view, is NOT a product of a general reasoning process and consequently, the properties of a language so conceived are more prone to be influenced by the bio-psycho-sociological (bps) states of the speaker than the truth value content of the situation it is encoding for.What institutional structure will eventually guide humanity? That based on the nomothetic constructs of the natural sciences (discovery of natural laws) or the idiographic tenets of the historical sciences (self-affirmation guided by individual, unique aspects of reality)? Is the world composed of historical facts or natural things?, Wittgenstein asked in his Tractatus Logico-Philosophicus. Back to Kant or existentialism? In this context, we may also ask, can semantically or syntactically correct logical assertions ever represent truth-functional composites of objects in reality? Or do they just represent meaningless (non-absolute) reflections on the currently changing historical scenery? These considerations should be kept in mind when we assign degrees of credibility to the evidence to be discussed below. We should ask, like the positivists of the Vienna Circle, whether the semantic meaning of a sentence depends on  the methodology of its verification or simply its use?. It has been the dream of mathematical logicians since Frege and Russell to design a data base of meaningful sentences with a precise logical structure of cardinality corresponding to the structure of the facts about objects in nature they hope to encode for. In this respect, we have long argued about the severe limitations of the human sensorium to intuit the facts (reality in itself) of the external world being observed and the observer’s subsequent neuronal combinatorial capacity limitations to represent them as meaningful truth-functional composites. The lesson we have learned thus far is that metaphysics can not always be successfully grounded on logic.

           Before we discuss the data it seems proper to acquaint the reader with some additional, but relevant, philosophical and technical aspects of our argumentation and conclusions. We have had a difficult time identifying the correct algorithm to explain the dynamic equilibrium there must exist between the input / output kinetics (I/O) of the visceral brain (Module I), the multi-modal right hemisphere (Module II) and the ‘talking’ dominant brain (Module III). It is not self-evident how to reconcile Markov’s stochastic (Module I), analog / parallel (Module II) and digital / linear (Module III) processing strategies. Stan Franklin’s (Univ. of Memphis, Tennessee) view of these processors as: “…coalitions of processors clamor(ing) for the attention of the spotlight……..will choose a winning coalition to attend……..the value of a single variable, not much like "executive control (by a working memory, within the context of the Global Workspace theory)." Parenthesis provided.

         Working memory is operationally defined as the temporary storage of relevant contextual information from immediate past events, about general or specific aspects of an object or event that is not currently (on-line) being perceived. Its main feature being its "dynamic adaptive content." Baars' attractive concept on the plurality of self-systems notwithstanding, our personal experience is definitely more consistent with that of an individualized self, "me", not "we". The “me” is an ongoing update of a store of rational impressions both complex and dynamic, the result of both conscious and unconscious processes continually updating themselves, repairing or deleting genetic / social survival inconsistencies in it. It is an abstract of the totality of our narrative memory. As Pat Hayes says, it is “One important special aspect of the representational language or code used by the narrative memory, a term which plays the role of the "essential indexical"; that is, it stands for the individual itself in the system's view of its own history.” Indexical stands for a kind of expression whose meanings were largely described by the Spanish philosopher Ortega: “ is himself and his existential circumstance”. Those expressions include the use of pronouns (I, you, etc.), demonstratives (this, that), temporal expressions (now, yesterday) and locative expressions (there, here, etc.) The most crucial conclusion about the evaluation made on a subject’s use of indexicals is that they arguably require a subject’s glimpse, on the spot, introspection into the content of the plurality of his ‘self-systems’, not a mere awareness of internal / external environmental temporal variations and the programmed adjustments to pre-set values thereof (first order consciousness) but an ability to ‘observe’ himself as an active part of a coordinated plan / response he is simultaneously a constituent part thereof, the hallmark of a higher order consciousness, a bonafide self-consciousness nobody has been able to reduce to a physical object of study. The ability to sort out the best adaptive course of action in response to a changing internal / external challenge, including delayed responses, altruistic responses or responses contrary to self-interests requires an ‘objective’, simultaneous, panoramic overview of self as observer and actor in the ongoing vital drama of life. One of the interesting practical consequences of this view is that it will allow a window for the observation of situations where there may or may not be a ‘collegiate’ consensus among the various participating neural ‘selfs’ in deciding the best adaptive response to an ensuing circumstance demanding a prompt response. To complicate matters further, any of the participating self-modules may have functional ‘lockouts’ thereby restricting control to either one of the various competing self-systems, often giving rise to neuro-psychiatric disorders. A search for access to viable lockout switches for therapeutic control purposes will require a familiarity with the equilibrium algorithm we have failed thus far to identify, one showing how the overall system behaves.

While we grope to develop further the dynamic equilibrium algorithm aspect of our model we continue to look for every opportunity to document our intuition that ‘thinking is a silent talking to ourselves’ (inner language). Language production humanizes in the interest of species survival in society, not in the interest of an understanding of ‘reality in itself’, that lofty but evanescent goal of our intellect, as we have argued and documented in several articles.

          The measurable effects of syntactic and / or semantic complexity on comprehension accuracy registered as the word-by-word reading times during sentence comprehension in normal and aphasic subjects of different language ability (Just & Carpenter, 1992, Haarman,1996) correlates well with data obtained from PET Scan brain imaging studies. Aphasics share a common area of chronic hypometabolism in Wernicke’s area (superior posterior temporal lobe) its extension varying inversely with the level of syntactic sentence comprehension (Hanson, 1991). These investigators have all used computerized ‘connectionist’ models emphasizing some of the critical postulates of the ‘Global Workspace’ theory  (Baars,1993) like thalamo-cortical re-entrant loop mechanisms. As we mentioned before, an ongoing image of self  is necessary for a successful semantic novelty detection during language comprehension because it requires an overview and sorting out of serially processed word meanings before their logical integration into an overall propositional representation. In our opinion it is not necessary an adherence to the Baars’ Gobal Workspace theory to realize the importance of a self representation by the subject in detecting, sorting out and interpreting subtle semantic variations before integrating all the implied contributions of a plurality of neural selves into a propositional sentence ready to be decoded into action at Broca’s area. While we recognize the importance of the roles played by the various memory modules (implicit, explicit and working) in the successful integration of their contributions into a logic proposition during a language comprehension effort one can not ignore the fact that novel detection, creation and production of language structures heretofore unfamiliar to the subject reader represents another exclusively human dimension, a higher order consciousness of self. A careful experimental design maintaining a constant indexical content and a controlled variable syntax, the only independent variable that remains to be measured and assessed is the semantic content during a language comprehension test. While we may agree with Harman in his general interpretation of language comprehension evaluation, we beg to differ with his analogy comparison to the blindsight experiments within Baar’s Global Workspace theory. See Crick, F. (1993), "The Astonishing Hypothesis: The scientific search for the soul". New York: Scribner's, as is explained in a previous chapter (See also our Amazon Book Review, 2001). Sentence comprehension in the aphasic patient may be viewed as a selective deficit in the underlying neuro-anatomical connectivity between Wernicke’s temporal area and the parietal ‘gnostic’ area (angular gyrus) as revealed independently by the lesions. In general normal and aphasic patients predictably vary in the degree and extent of the various ‘self systems’ they can summon to bear upon the processing of syntactic, semantic, and referential knowledge in language working memory. It is not clear if Wernicke’s hypometabolic status represents the neuronal site of integration properly or an interference with an activation of the angular gyrus where lesion-related deficiencies in integration is known to have occurred. A corroboration of these results is needed utilizing the ‘phonological’ route, outlined in Diagram 1, above.


            We doubt there will ever be an opportunity to empirically demonstrate, in an explicit way that a skillful auto mechanic, following the written instructions for the assembly of a new model engine, requires two different levels of conscious language comprehension. The main difficulty is to distinguish between first order (automatic) and higher order (creative) levels of consciousness. Even well published neuroscientists have difficulty in ascertaining that the ultimate computer can not create purposeful, novel, adaptive solutions not contemplated in the software. The difficulty stems from the unconscious operation of a working memory pre-programmed for the automatic processing in support of technical sentence comprehension such as access to lexicon memory, grammatical associative priming and computation of syntactic relations. At this first order high level of operation  a good software program in Artificial Intelligence (AI) can arguably perform.

            However, when the technical language comprehension requires an ongoing detection of novelty approaches in design and a word by word scanning and integration (by lexical and parsing processes) of their meanings into an effective propositional representation that the pre-frontal lobe can now translate into a relevant narrative accompaniment to the appropriate muscular motor responses, is something the ultimate computer can not do even outside the time frame constraints of a working memory. In our opinion lexical processing by sentence context and syntactic processing by semantic-pragmatic and priming cues requires the continued temporal assessment of alternative strategies by an observer who now simultaneously doubles up as an actor in the selection, a task exclusively human. (See Carpenter, 1992). It is of interest that the same reasoning has been used in defense of Barr’s Global Workshop theory of consciousness and Edelman’s conception of consciousness as a “Remembered Present” (see our Amazon Book Review, 2001) both of which we find incomplete in that they are dealing essentially with first order consciousness or thought generation. Any interpretation of their results as evidence of a higher order or meta-consciousness would imply agreement with Heidegger’s concept of reality as a ‘present-pastness’  when he said "for the most part we ARE our own having-been" (emphasis supplied), a concept we find counterintuitive and un-necessary for our concept of meta-consciousness.

          It is very difficult to design an experiment that would credibly require the inclusion of meta-consciousness to obtain quantifiable measurable results. Language comprehension tests, as described, provide those elements that make the conclusion credible. By using uniquely designed combinations of semantic relations expressed using the same set of nouns and verbs in all possible orders or using the same words in object-relative instead of subject-relative sentences (in right- branching instead of center-embedded relative sentences) increases the number of novel meaning representations in the normal subjects (new combinations the subjects were never exposed to) where the intact inter-neuronal connectivity required by the Global Workspace theory is viable as a generator of first order consciousness (awareness) in addition to the exclusively human capacity to experience meta consciousness with qualia. Stroke patients where the general locus of hypo-metabolic activity (as measured by PET Scan imaging techniques) score a proportional diminished capacity in syntactic sentence comprehension and creative language performance as explained. Subjects are unable to use semantic-pragmatic or priming cues to assist in their comprehension (i.e., subject, predicate, object), only aided by on-line analysis syntactic cues such as word order and structural embedding. Notice that in these tests (Caplan, Baker, & Dehaut, 1985) the sentences used are semantically reversible, i.e., all nouns and verbs can switch roles without creating an ungrammatical sentence, they are reversible sentences. As we said above, the word by word reading requires first to ascertain their individualized meaning, formulate a proper syntax arrangement where the elements get now integrated into propositional semantic representations of the  whole sentence (all of this within the context of the paragraph it belongs to!), quite some task for an aphasic; and for a serial, sequential computer processor also! It is not only a temporal accumulation of information from various sources, it requires knowledge about how the constitutive elements are related to each other, their individual content (meaning), their best possible group formation arrangements consonant with collective meaning (verb argument configuration), all within the context of the final grammatical role. The important thing to consider is the successful, meaningful integration of qualitatively different data from their various locations in space and time  into novel configurations not experienced before by the subjects.

          Curiously, this experimental data has been offered in support of a theory (Global Workspace) of consciousness we do not support, as noted, for being incomplete and counterintuitive. The suggestion that the experienced or remembered ‘present is immediate past’ is like saying that our existence is a perpetual watching of our own instant replay!  No way!

End Chapter 9