The Design of the NCC-circuit for Audition and Sound Generation:
Comprehension, Verbalization, Conceptualization and Declarative memory.
The Dawn of Human Intelligence.

Introduction

The auditory system is organized tonotopically such that the frequency of a stimulating sound is mapped onto a location along a basilar membrane within the cochlea. (Kandel et al, 2000). The tonotopic organization is maintained along the neuronal pathway from the cochlea and auditory nerve through the initial stages of the central auditory system, such as the cochlear nuclei, inferior colliculus and medial geniculate nucleus, to the primary auditory cortex (Rauchecker, 1999). The tonographic maps and the neuronal pathways associated with them are strikingly similar to the retinotopy in the visual system and the development of collective modalities and sub-modalities described by Rosen et al in the design of the NCC for visual perception (Rosen, 2003b). In this paper it is proposed that auditory perception is mediated by an auditory-NCC that is designed in a manner analogous to the visual NCC. The auditory NCC is made up of a frequency-time (f-t) domain function space that is an adjunct to the tactile and visual spaces defined by Rosen et al in their theory of consciousness and visual perception (Rosen, 2003a b). The auditory f-t world map has been designed into the array of Relational Robotic Controllers (RRC) shown in figure 1. The auditory-q nodal map module and the p-sound generators are part of the “self” circuit and occupy a central position in figure1. The world map configuration of the nodal maps, forming the input circuit to the RRC, has been described by Rosen (Rosen, 2003a,b). Navigating through the arm and leg nodal maps shown in figure 1 is performed by p-muscle/joint control signals. In an analogous manner, it will be shown that navigating through the f-t space is performed with p-sound control signals generated by the sound-generating musculature of many animals. The p-sound signals that navigate through the auditory-NCC-space are unique in that the signals may also generate a sound that is “heard” by the auditory NCC. The auditory NCC that “hears” all the sounds present in the external environment is part of the Darwinian search engine described by Rosen et al for the operation of the tactile and visual search engine (Rosen, 2003ab). The Darwinian search for TT-generators within the NCC, the Neuronal Correlate of Emotions (NCE) and the operation of the hedonic motivational system have been described by Alan Rosen and David B Rosen (Rosen, 2003e).

The Darwinian Utility of audition and sound generation.

The overriding principles used in the design of the auditory NCC are Darwinian survival and reverse engineering (Dennett, 1995) the evolutionary adaptation and known connectivity of the human and animal body and brain. The functional utility of the auditory NCC was assumed to be supportive of Darwin’s laws of natural selection and survival of the fittest.

Sound generation and sound perception have been used as a Darwinian survival mechanism by lizards, birds, mammals and primates. In all cases, with the exception of humans, the sound generating organic structures are limited to a capability to generate very few phoneme sounds. The innate capability of Homo Sapiens to generate and distinguish over 100 different phonemes arose with the development of complex vocal chords, and the capability to control breathing, mouth, tongue, lips and facial muscles used to generate sound phonemes (Clements, 1999; Nespor, 1999). Concurrent with the development of physiological vocalization structures, the brain developed the functional capabilities to control them and to perceive the sounds generated by them (Rauschecker, 1998; Eisenberg, 1976).

In the lower animals, sound generation is used as a defensive or mating mechanism (for example, rattling snake, barking dog, singing bird). The received sound may be recognized by the hedonic motivational system and thereby generate the emotion of fear or attraction (love) in the receiving animal. The emotion may then generate a Task initiating Trigger (TT) on a muscle-joint control nodal map module that leads to defensive or mating tasks on the DHTD (Rosen, 2003e). The TT may also operate to trigger the sound-generating task (sound-generating-muscles) of the receiving animal (for example the response of a dog to the barking sounds of other dogs). This leads to a simple form of communication that is limited by the small number of phoneme generators that are physiologically designed into the organism. The unique human physiological vocalization design and the capability to generate over 100 different phoneme p-sounds gives rise to language and a complex form of verbal communication.

Sound generating p-signals (including verbal communication in humans, and barking, singing, roaring, hissing, etc in animals), may be applied to a NCC-auditory-sound generating nodal map module. The received q-sound, also associated with the nodal map module, may operate on the NCE to trigger a Task Initiating Trigger (TT) that generates a somatic p-signal that controls the various bodily muscles and joints or a p-sound generating signal that responds to the applied auditory q-signal. The design of such a nodal map module is presented in the following sections by reverse engineering the NCC-q-signals and the p-sound generating control signals.

A NCC Theory for Auditory Perception and Comprehension:

The present day view is that “auditory perception” is transformational. The transformations operate on the tonotopic organization that is preserved in the basilar membrane, the cochlear (mechanoreceptor) hair cells, the auditory nerve, the medial geniculate nucleus and the auditory cortex (Rauschecker, 1999, Kaing 1965). The transformations are performed by matching acoustic input (represented by the tonotopic organization) against some internal representation of the sound generator (words in the language). For example, Gerry TM Altmann (1990) extracts meaning from a representation by constructing a message level interpretation of the representation (in speech processing the representation is a string of lexicon items).

A mechanistic design, an NCC for auditory perception is presented in this paper that dispenses with most, if not the entire auditory signal processing that is directed towards the perception of sound. The design of the NCC for auditory perception is reverse engineered by tracking the pathways associated with the auditory collective modalities through various brain relays to the “self” location and identification circuit defined by Rosen et al (Rosen, 2003ab).

The NCC-circuit for auditory perception gives an animal or robot the capability to “hear” (in high fidelity) an orchestral symphony or auditory noise present in the environment. However it does not, by itself, yield any comprehension or identification of the auditory signal. In order to comprehend or identify the auditory signal it is necessary to “emotionally” perceive it. That is, the perceived signal must give rise to a TT and upset the homeostasis of one or more autonomic homeostatic systems (Rosen, 2003e).

A solution to the problem of “comprehension” of sound patterns by animals is presented. Comprehension and emotional perception is demonstrated by reverse engineering the functional flow of sound-TTs through the hedonic motivational system in the brain of a dog. There is a high degree of “comprehension” associated with emotionally triggered sound generators (vocalization) and the simultaneous “self” perceived (heard) sound patterns. In both the sound generation and perception “self” knowledge and “self” survival (contingency-TTs) are important elements, possibly the necessary and sufficient conditions required to achieve “comprehension”.

The Darwinian Language Model: A neuroanatomical-connectionist model for the brain organization of language.
The Darwinian adaptation of human language is described in terms of the same circuit components and functional flows used to describe comprehension and sound generation in animals. The complexity of human language is due to the combinations and permutations of the large number of phoneme sounds that humans are capable of producing and hearing.

The Darwinian language model described in this paper is consistent with Wernicke’s neuroanatomical connectionist model for the organization of language (Wernicke, 1874). Wernicke’s model, which was based on studies of patients suffering from Broca’s aphasia (Broca, 1861, 1865) and Wernicke’s Aphasia (Wernicke, 1874), has been criticized nearly from its inception as an oversimplification that did not capture the cognitive and conceptual complexity of the behavioral disruptions found in even Broca’s and Wernicke’s aphasia (e.g. Jackson, 1878, Luria 1966). The modern theories are non-connectionist characterizations, with movement towards linguistically and cognitively relevant characterizations. Major modern proponents of this movement were Zurif, Caramazza, and Myerson (1972), and Swinney et al (1996).

The Darwinian language model overcomes the oversimplifications and captures the cognitive and conceptual complexity of behavioral disruptions of both aphasic and normal persons. The operation of the auditory NCC, the NCE, and the Darwinian search engine in verbalization and comprehension of speech is fully consistent with experimental observations; the most prominent may be enumerated as follows:

a) The observed unilateral cerebral dominance for language (Swinney, 1999) is consistent with a recent Darwinian adaptation observed only in humans, contrasted with the evolutionary adaptation of the human primitive brain and the reduced cerebrum of the mammalian and primate brain (Rauschecker, 1999).

b) The babbling sequence stepper module described in section 4.4 produces speech that is fluent, effortless, and rapid, however totally empty of comprehension. This is comparable to patients with Wernicke’s aphasia that may produce speech that is fluent, effortless, and rapid, filled with inappropriate verbal paraphernalia, however remarkably “empty” of meaning (Swinney, 1999)

c) Damage to the motor generating p-phoneme nodal map is analogous to Broca’s aphasia, which results in difficulty in articulation and production of speech, with relative sparing of comprehension. Physiologically, Broca’s area lies next to the motor area for the muscular control of speech (lip, palate, vocal chords, jaw) (Swinney, 1999).

d) The auditory NCC and the Darwinian search engine generate a comprehensible relationship between the p-phoneme generators of linguistically relevant representation (words syntax, semantic), and a combination of visual, auditory, tactile, olfactory and taste TTs. Thus the word syntax and sentence grammar are linguistically and cognitively relevant characterizations of experiential-sensory TTs. This dispenses with the modern movement towards non-connectionist characterizations described in terms of transformational processing (access, integration). (Zurif et al and Swinney et al)

The evolution of language is a cultural adaptation:

Different languages of equal complexity but with different lexicons evolve by conditioned learning of word-TTs with other sensory-TTs (especially visual pattern-TTs). This leads to the conclusion that the cultural evolution of language must adhere to a non-genetic form of Darwin’s laws of natural selection and survival of the fittest. In the same manner as emotionally generated TT give rise to procedural memory and procedural tasks on the DHTD, so do emotionally generated p-phoneme motor-TTs (verbalization) give rise to declarative memory and verbal-discourse tasks on the DHTD. Pinker in The Language Instinct, also argues that language is an evolved adaptation (Pinker 1994).

The formation of family-tribal-societal groups: Evolutionary biology and Sociobiology.

Emotionally generated TT give rise to procedural memory and procedural tasks on the DHTD. In order to direct those tasks towards the formation of family-tribal-societal groups, the procedural tasks on the DHTD must be coupled with emotionally generated p-phoneme motor-TTs verbal-discourse tasks on the DHTD. Just as procedural memory facilitates the achievement of a task objective on the DHTD, so does declarative memory facilitate the communication and guidance of others (friends as well as enemies) towards the performance of tasks on the DHTD. Thus language and declarative memory is a cultural adaptation that is supportive of evolutionary biology (Tooby and Cosmides, 1987), whereas procedural memory may lead to cultural adaptations that are supportive of sociobiology (Wilson, 1975). Note that evolutionary biology is often said to include the “mind” since psychological (declarative) factors are used in the research procedures. Sociobiology, on the other hand is often said to be more independent of the “mind” (verbal language?), since procedural-behavioral factors are used in the research procedures (sociobiological research is generally independent of human psychology).

The Dawn of Human Intelligence

Language is a Darwinian adaptation that extends by means of declarative tasks all the procedural DHTD tasks that mankind has learned to perform. Generational-cultural learning of language is mankind’s “entry ticket” to the cognitive niche. The dawn of human intelligence is the development of mankind’s capability to conceptualize and form abstractions of their observations and experiences in the external world, and the capability to communicate (discourse) these experiences. The reverse engineered relational brain (RRC) relates rather than computes. The basis for human intelligence is an RRC-brain with the capability to conceptualize and form abstractions by relating the actions associated with procedural memory to the vocalization-action of words associated with declarative memory.

Artificial Intelligence: How to build an intelligent brain that exhibits characteristics of human intelligence?

In the Philosophy of Artificial Intelligence Margret Boden defines Artificial Intelligence (AI) as the study of how to build and/or program computers to do the sorts of things that minds can do (Boden, 1990). However, since minds exhibit characteristics of “consciousness”, “self knowledge” and experientially learned knowledge, characteristics that have never been designed into a machine, most scientists work with a curtailed definition of Artificial Intelligence. The curtailment consists of side stepping the “mind” and defining AI as the development of computers whose observable performance has features, which in humans are attributable to mental processes. A large number of the studies and publications in the field of Artificial Intelligence make use of this curtailed definition of AI (see for example, Aleksander, 2001; Bechtel and Abramson, 2002; Boden, 1990; Cummins, 2000). In this paper the authors favor a “more controversial” definition of AI, namely AI is the science of intelligence that is at the intellectual core of cognitive Science (Boden, 1990). Therefore the primary tool for the study of artificial Intelligence should be reverse engineering the “intelligence” exhibited by the human mind.

Thus one may ask the question, how does a cognitive scientist reverse engineer or build an intelligent brain-computer that exhibits characteristics of human intelligence?

The answer is- build, design/program/teach a RRC-robot a language of discourse with which to describe verbally the experiential procedural and declarative knowledge that it gains in the various TSM-pattern recognition circuits that make up the memory system of the robot. The procedure for programming human-like intelligence is by reverse engineering the teaching process of human infants as they grow from infancy to adulthood. A well-programmed RRC-robot may possess super-human intelligence for 3 reasons,

1. A well-designed RRC-controller may be designed by reverse engineering the adult human brain and avoiding the brain-development phases of infancy and puberty. The prototype-robot must be an adult-size obstacle avoiding locomotive robot equipped with a procedural and declarative memory system that is prepared to gain experiential knowledge via its tactile, visual, auditory, olfactory and taste sensory systems.

2. With the exception of the mating-reproductive task-objective, a full complement of emotional-TTs associated with the food-energy, shelter, defense, and bodily maintenance task-objectives should be designed into the autonomic involuntary control system. The emotional-TTs associated with human mating and reproductive task-objectives do not contribute significantly to the development of intelligence.

3. The programming-teaching procedure may be more much more efficient than human teaching procedures because of the deep knowledge of the system-designer of exactly how the pattern recognition circuits learn and remember the various TT combinations that the RRC robot is exposed to via its sensory system.

Super-human intelligence is trained-programmed into the robot by teaching it the verbal proficiency of a medical doctor, nuclear physicist, cognitive neural scientist, chemist engineer, historian, and all the proficiencies of every college professor at a major modern university. This form of Artificial Intelligence is part and parcel of the study of, or the science of intelligence in general or as the intellectual core of cognitive science.

The challenge of building such a super-intelligent RRC-robot is formidable, as formidable as building and designing organs and organisms by means of genetic code-blue prints, or colonizing the planets of the solar system. It is a multi-disciplinary, system-engineering project that has been identified and defined by Machine Consciousness Inc. (www.mcon.org) as yielding the most “societal-payoff” with a potential for transforming the 21st century into the age of humanoid robotics.

Forum Discussion (last updated 00/00/200x)