Sensorimotor Control and the Reverse Engineered Neurophysiology of the Brain:
A Robotic Model of the Human Body and Brain.

I. Introduction

Neural network models for the sensorimotor control functions of the brain have been studied by Stephen Grossberg (1987), Teuvo Kohonen (1988), (Ritter et al, 1989), (Guenther et al 2001), and many others. A new approach, presented in this paper, is a proposed connectivity of a sensorimotor control system by reverse engineering the biological modalities of mechanoreceptors1. The reverse engineered robotic system is designed to respond to the modality of “itch-type” activations. The motor response of the system is designed to generate a “scratch-type” trajectory for all possible end joints and all possible activation itch points.
The significance of this new approach is that it utilizes the experimentally observed neurological topographic mappings, sometimes referred to as “brain modules” (Purves et al, 1997, p 161) to design coordinate frames within the controller. Such a coordinate frame may lead to the robotic design of:

1. Biological perception by means of “tactile sensory montoring” of the external world.
2. “Self awareness” by locating and identifying all parts of the robotic body.
3. Self-knowledge by programming/teaching the robot the location of all bodily parts and the coordinate location of the space in the vicinity of the body parts, that is associated with all possible itch-scratch trajectories.
4. “Tactile-sensorimotor control” of all end-joints by executing itch-scratch trajectories for all possible itch-points. And
5. a receptor modality, often defined in terms the “subjective experience” or sensation evoked by the receptor (e.g. itch-type feeling). (Guyton, 1991; Kandel, Schwartz & Jessell, 1991; Gazzaniga, Ivry & Mangun, 2002; Bear,Connors,& Paradiso, 2001).

It is also significant that the reverse engineered brain (Controller) sheds light on portions of the neurophysiology of the brain. In this paper, the mechanoreceptors and their connectivity to the brain is assumed to adhere to the biological “labeled line” principle (Guyton, 1991), or the “Law of Specific Nerve Energy” (Haines, 2002), which ensure that each type of sensor responds specifically to the appropriate form of stimulus that gives rise to a specific sensation. In the biological system the specificity of each modality is maintained in the central connections of sensory axons. Thus the term stimulus modality encompasses the receptor, afferent axons, and the central pathways that are activated by the stimulus. It is noted that the central connections associated with sensory modalities often form neurological topographic mapping, or brain modules, in various regions of the brain. In this paper, it is proposed that those brain modules may form a coordinate frame in the brain that is utilized sensorimotor control. Furthemore, those same sensorimotor control structures may be identified in the experimental neurophysiological structure of the brain stem, thalamic and cortical regions.

The building path presented in this paper, for the itch modality, is expandable to the modalities of the visual sensors, the vestibular sensors, and ultimately to a complete sensorimotor control system of the total body and brain.

Topographic mappings and patterns of organization within the brain are essential for the formation of coordinate frames. Purves et al (1997, p. 161) writes that “Patterns of organization within the sensory cortices may be a fundamental feature of the cerebral cortex, essential for perception, cognition, and perhaps even consciousness.” In this paper the brain modules are used to form a coordinate frame within the controller that is utilized for tactile perception, robotic self knowledge, sensory motor control and modality generation.

III. Discussion

The discussion is divided into 3-parts. Part 1 discusses the extension of the reverse engineering procedure of the itch modality to other sensory modalities. Part 2 compares the neurophysiology of the brain with the connectivity of the robotic sensorimotor control sysem. And Part 3 compares a biological modality with a “mechanical modality.”

1. Extending the reverse engineering procedure of the itch-modality to other sensory modalities.

The primary characteristic of a reverse engineered modality is that it gives rise to a sensory form of “robotic self knowledge” in a coordinate frame defined by the sensor and located within the controller. It is proposed that the primary constraints imposed on the reverse engineered design of the modality of any additional sensor (such as the vestibular or visual sensors) is that the added sensor gives rise to a form of “robotic self knowledge” in a coordinate frame defined by the added-sensor and located within the Controller. Furthermore the coordinate frame defined by the added-sensor must be consistent with, and calibrated with the coordinate frame defined by the modalities of the mechanoreceptors. The generalized reverse engineering procedure may be applied to the modalities of the visual and vestibular sensors, Furthermore, the motor control connectivity of the robotic controller may shed light on the sensory motor control neurology of the brain (Guenther et al, 2001; Grossberg, 1998; Kohonen, 2001; Ritter, 1992) .

2. A Comparison of the connectivity of robotic sensorimotor control system with the neurophysiology of organic sensorimotor control.

The reverse engineered input and output portions of the RRC circuit are faithful functional emulations of some observed topographic distributions within the organic brain. The intermediate neurophysiology, not determined in this paper, between the input and output, so as to yield sensorimotor control by all the biological sensors, is a vibrant field of neuroscience research.

The input and output portions of the RRC circuit, by virtue of the fact that they reverse engineer portions of the neurophysiology of the brain, may shed light on the neuropysiological structure of portions of the biological brain.
With respect to the input to the RRC circuit, somatotopic feature maps associated with the modalities of mechanoreceptors and nociceptors distributed on the skin surface of the body, are maintained in the connections of receptors via afferent axons, through the brain stem, thence through the thalamic region of the brain, and thence to the somatosensory portion of the cerebral cortex.The most significant item is that the spinothalamic tract exhibits termination points and somatotopic organizations in the brain stem, thalamic region, and regions of the cerebral cortex (albeit those somatotopic organizations may differ with the modality of the receptor, and even with a given modality the somatotopic organization in the three regions may also difer one from the other (Brodal, 2004)6.

With respect to the output of the RRC circuit, somatotopic feature maps are observed in the motor cortex and brain stem and to a lesser degree in regions of the thalamus. The somatopic patterns are observed at various levels of the pyramidal tract. Efferent corticobulbular and corticospinal pathways (the pyramidal tract) descend from the motor cortex to the pons and medulla oblongata of the brain stem (Brodal, 2004).

The brain is a highly redundant organ. It is proposed that the neuropysiological structure of portions of the biological brain devoted to sensorimotor control may include three configured input and output circuits, represented by somatotopic distributions in regions of brain stem, thalamus, and cerebral cortex, and a distributed intermediate circuit that may serve the three input-output distributions.

In support of the hypothesis we note the following:

1. the afferent topographic organizations associated with the termination points in the brain stem, thalamus, and somatosenmsory cortex may be parts of a configured input circuit that forms coordinate frames in the brain.The input circuits represent the inputs to the three sensorimotor control circuits in the brain.
2. The efferent motor control pathways in the motor cortex are clearly adjacent to the somatotopic distribution of neurons in the somatosensory cortex. In the thalamic and brain stem regions, the somatotopic distribution of motor neurons are not clearly defined. The observed distribution of motor neurons may be consistent with a distributed output circuit that serves both the brain stem and the thalamic region.
3. The efferent corticospinal tract is mainly crossed in the lower medulla, whereas many of the cranial nerve nuclei receive crossed and uncrossed fibers from the motor cortex (Brodal, 2004, p 267). Thus any one of the three input distributions may control the output signal. The crossed and uncrossed fibers imply a distributed output circuit with output neurons distributed at various levels of the pyramidal tract including corticospinal neurons that may be utilizaed to open and close spinal reflex arcs.
4. The function of the three regions in perceiving sensation supports a distributed output circuit with the thalamus acting mainly as a relay. When the somatosensory cortex of a human is destroyed, that person loses most of the tactile sensibilities. However, the thalamus and the brain stem retain a slight ability to discriminate tactile sensations, even though the thalamus normally functions mainly to relay this type of information to the cortex (Guyton, 1991, p. 517).
   The intermediate organic neurophysiology is not detemined by the design of the intermediate circuitry of the RRC-robot. However, the generalized RRC-sensory motor control model presents a detailed functional building path of the intermediate circuitry, which is consistent with the input and output characteristics of the biological system. This building path may shed light on neuroscience research projects aimed at the determination of the intermediate neurophysiology.

3. Comparison of a biological modality with a “mechanical modality.”

Guyton (1991) writes, “Each of the principal types of sensation that we can experience- pain, touch, sight, sound, and so forth- is called a modality of sensation.” In this paper, the biological modality of sensation of mechanoreceptors is a subjective experience of “itch-feeling.”

3.1 Comparison of a Biological Modality with the Subjective Experience It Evokes: The long sought after Neuronal Correlate of a Modality (NCM)-circuit.

It is assumed that the biological modalities are correlated with the receptors, afferent axons, and the central pathways that they activate. Thus the authors take the view that the central pathways, activated by each modality, may be represented by a “neural-circuit” that has a “subjective experience,” namely a modality, correlated with it. That neural-circuit may be defined to be a Neural Correlate of a Modality (NCM) or a neural correlate of the “subjective experience.” It is proposed by the authors that this biological neural-circuit is the long sought-after “neuronal correlate of consciousness”-circuit7 (Metzinger, 2000: Crick & Koch. 2000 ) in the brain. Thus the biological sensations” of “touch feeling” or nociceptic “pain” may be correlated with a biological NCM-circuit.

3.2 Comparison of the Biological NCM-circuit With the “Mechanical Modalities” of the Sensors of the RRC-robotic Model

The internal coordinate frame of the itch-scratch RRC-robot constantly monitors the state of the pressure transducer/tactile sensors for itch-type activations. The following question is posed to the reader: Is there a mechanical-type subjective experience, analogous to the itch-sensation, correlated with the “itch-type” activation of the pressure transducers?

The mechanical itch-type activation is similar to the biological itch type activation in that a) the central connection represented by the connectivities in an internal coordinate frame, may be similar, b) Robotic location, identification and “self knowledge” are similar to biological self knowledge of itch type activations, and c) The itch-scratch motor control response of the robotic system is similar to the response of a biological system.

Based on the similarities enumerated above, the authors propose that there exists a “mechanical modality” correlated with the “itch” stimulus falling on the pressure transducer. The stimulus and the central connections of the robotic system form a circuit that is a “mechanical modality” generating mechanism. That mechanical modality is presented to the robot as a subjective experience of the itch-type stimulus that only the robot may feel. The RRC-circuit is thus viewed by the authors as an “itch-Sensation-generating Mechanism (SgM).

3.3 Designing Mechanical SgMs

The itch-scratch NCM-circuit may be expanded to all the modalities of all the somatic sensors. The modalities of the visual, auditory, olfactory and taste systems are the subjective experiences of “seeing,” “hearing,” “smelling,” and “tasting,” respectively. An RRC circuit that reverse engineers the biological NCM-circuit may also be viewed as a SgM that generates a mechanical “sensation” associated with the mechanical receptors and their central connections. The design of the NCM-circuit for visual perception is presented by Rosen & Rosen (2003b) in a paper titled “The design of the NCM for visual Perception: Solving the inverse optics problem of “seeing.”

Forum Discussion