Design of the NCC

Introduction

The Neuronal Correlate of Consciousness (NCC) described by Rosen et al (2003a,b) is a powerful sensory-motor control tool for all locomotive actions performed by all bodily parts. The Consciousness Mechanism (CM) generates “self” knowledge, a “measure” of the near space (with the “self” in the center), and “self” awareness of the internal reactions of the body to external stimuli. The NCC-circuit is also trained to determine the numerical size of each scale division (measure) of the coordinates of the world map, by means of the “itch-scratch” response and the calibration of the visual FOV space with the somatic “itch-scratch” “near” space. Thus “Learning a Body-centered Representation of a Three-dimensional Target Position” (Guenther, Bullock, Greve and Grossberg, 2001), is automatically achieved by the CM without any further calculation.

The application of the CM to locomotive behavior is an integrated form of sensory-motor control where all the biological sensors are directly associated with the control of the somatic motor system (the tactile, visual, auditory, olfactory, taste, and vestibular sensors). The building path for the NCC would not be discovered without the very prolific scientific publications of a large number of cognitive neural scientists and engineers. The authors are particularly indebted to Stephen Grossberg (1972-2004) for solving problems (without the benefit of the CM) that were somewhat instrumental in the deduction of biological control systems. In addition Teuvo Kohonen (1991) and Helge Ritter (1989) generated the Neural network equations that were directly applicable to the design of “self” knowledge via the “itch- scratch” response.

A robotic NCC-based motor control system may be designed by reverse engineering the functional utility of the biological motor control system. A reverse engineered design must adhere to Dennet’s reverse engineering requirement “No sound functional analysis is complete until it has confirmed that a building path has been specified” (Dennet, 1995, p.194). Therefore, a building path for the design of a volitional, obstacle avoiding multi-tasking NCC-based robot is presented in the following sections.

The building path for the NCC- brain circuit that controls the biological somatic motor system is specified by designing a reverse engineered hybrid electronic circuit-controller that emulates the control functions taking place in the mammalian brain. The design of the robotic body, upon which the controller operates, is not discussed in this paper. The “controller” electronic circuit, called a Relational Robotic Controller (RRC) is a hybrid circuit made up of electronic neural networks and algorithmic sequential software programs. The neural networks emulate the biological feature maps, or brain modules (Purvis et al, 1997) that are present in the biological brain, whereas the software programs work in conjunction with, and as an adjunct to the neural network circuits. The design and operation of the RRC is unique because it is NCC-based and reverse engineered to operate like the biological brain. Constraint data for the reverse engineered design was obtained by reference to the cognitive neuroscience research data for the control of the operation of mammalian limbs (Gazzaniga, 1998; Kandel et al,1991; Pinker,1984; Purvis et al,1997).

The NCC-based constraint on the reverse engineered RRC circuit was that it “build a model of the external world and place its “self” in the center”. A brain organ, with a model of the “self” and external world incorporated within it, may then implement locomotion, “self” survival and reproductive success through a broad range of environmental contingencies. Two fundamental constraints were derived from the cognitive neuroscience literature. The first is that all motion should be pre-planned and goal directed. The second is a volitional constraint that allows a brained organism the option of re-planning a pre-planned trajectory whenever a obstacle is detected along the path of the pre-planned trajectory (Kandel et al, 1991; Gazzaniga,1998). Note that by the author’s definition, the controller or brain of a robot or organism is said to be a volitional organism or robot if it is designed with the capability of re-planning a pre-planned trajectory of motion.

Conclusion

The NCC is as powerful sensory-motor control tool for a volitional, obstacle avoiding, multi-tasking, and procedural memory retaining robot. The application of NCC to a robotic system, such as the sensory-motor control of a reverse engineered, biological-somatic motor system, is uniquely different from all previous publications:
Learning a body centered representation of a 3D target position is automatically achieved by the CM, without any further calculations.

“Self knowledge and a world map in the brain with the self in the center” yields a numerical size of each scale division (measure) of the coordinates of the world space.

The Consciousness Mechanism operates as a feedback mechanism that facilitates learning precise sensory-motor control.

Volition and free will.

The concept of volition, as described in the design of the NCC-robotic controller, is characterized by a short response time, one frame period, during which a sensory pattern is recognized and a complete pre-planned trajectory generated in the sequence stepper module. Thus a robotic controller may be designed to yield a deterministic response within a time frame of approximately 20 milliseconds. There are many publications (see Schwartz, 2002) that differentiate between “free will” and volition, generally on the basis of the “deterministic” response time of the biological human brain. If a human response to an observation occurs within an interval shorter than a frame period, the response is said to be a “free willed” response. If on the other hand the response occurs in an interval longer than a frame period, the response is said to be a “deterministic”-response. The NCC-robotic controller described in this paper exhibits “deterministic”-response characteristics. The authors of this paper define a robot or biological brain to be volitional if it exhibits a “deterministic”-response to NCC-sensory data.

Robotic volition, as discussed in this paper, is related to obstacle avoidance and the instinctive action that may be taken to avoid environmental contingencies. A volitional capability is designed into a sensory-motor control system by adhering to two cognitive neuroscience constraints. First is that all action must be pre-planned and goal directed. And second is that the biological organism must be given the capability to re-plan any pre-planned goal directed action on the basis of environmental contingencies that may suddenly appear along a pre-planned trajectory.

Thus the response of a volitional brain is deterministic, generally determined by a “contingency” that manifests itself within a pre-planned trajectory. Is the response of a “free willed” brain “non-deterministic”? Possibly more random?

Volition in the Mammalian brain

In Mammals, volitional obstacle avoidance is most likely programmed in the lower brain stem regions of the brain as well as in higher thalamic regions. In those regions obstacle avoidance is performed involuntarily and the programming is most likely implemented in the analogue to the Sequence Stepper Module. Obstacle avoidance is also innately programmed in the pattern recognition circuit located in the motivational system of the mammalian brain. It is therefore likely that obstacle avoidance occurs in both the sequence stepper module and the pattern recognition circuit of the motivational system.

A Procedural Memory system in the Brain.

Learning and memory is generally classified as reflexive (procedural) or declarative on the basis of how the information is used. In the multi-tasking robot, procedural TTs and procedural memory is the basis for all control functions of the somatic motor system. Declarative memory relates to the verbal ability to express verbally in declarative sentences (Kupfermann, 1991). Therefore, only humans or auditory-verbal nodal map robots may exhibit characteristics of declarative memory.

In the connectionist multitasking brain the nodal map module and the pattern recognition system, is also a reverse engineered procedural-memory module of the biological brain. The advantage of the connectionist brain over the biological brain is that once a set of TTs are recorded and recognized the self circuit does not “forget”. The biological brain, on the other hand, suffers from a process known as “extinction” wherein over the long term, TTs recognized by the pattern recognition circuit, are lost. The advantage of the biological brain is a greater learning speed, and the presence of long-term potentiation (LTP), which is a strong function of environmental contingencies. “All animals that exhibit associative conditioning, from snails to humans, seem to learn by detecting environmental contingencies rather than detecting the simple contiguity of a conditioned stimulus and unconditioned stimulus”, (Kupfermann, 1991). Both the advantages and disadvantages of the connectionist brain result from the simulation of the biological brain neuron that are stimulated by hormonal secretions, by crude and imperfect electronic neurons, which are used as the basic building blocks of the neural networks of the multi-tasking controller.

Forum Discussion