Page 114 - Biomimetics : Biologically Inspired Technologies
P. 114

Bar-Cohen : Biomimetics: Biologically Inspired Technologies DK3163_c003 Final Proof page 100 21.9.2005 11:40pm




                    100                                     Biomimetics: Biologically Inspired Technologies

                    considerably. (This is much as in electronics: the same logic circuit can be implemented with
                    electromechanical relays, in silicon CMOS circuits, using vacuum tubes, or even using fluidic
                    devices. While these implementations are physically dissimilar; they are mathematically identical.)
                       Although not enough is known to create a definitive list of specific human cognitive functions,
                    the following items would certainly be on such a list:

                    .     Language
                    .     Hearing
                    .     Seeing
                    .     Somatosensation
                    .     Action (movement process and thought process) origination
                      This Appendix focuses upon the implementation of cognitive knowledge links, confabulation,
                    and action command origination by the human cerebral cortex and thalamus. It is assumed that the
                    reader is familiar with the concepts, terminology, and mathematics of elementary confabulation
                    (e.g., as discussed in Hecht-Nielsen, 2005) and with elementary human neuroanatomy and neuro-
                    physiology (e.g., as presented in Mai et al., 2004; Mountcastle, 1998; Nicholls et al., 2001; Nolte,
                    1999; Paxinos and Mai, 2004; Steward, 2000). The theory hypothesizes that all human cognitive
                    functions, including those listed above, are implemented using the basic confabulation machinery
                    sketched in this Appendix. To keep this Appendix focused, the manner in which confabulation can
                    be used to carry out specific cognitive functions (such as those listed above) will not be discussed
                    here, as this is essentially the material covered in the main body of the chapter.
                       To keep the size of this Appendix reasonable, and to avoid speculations about fine details, the
                    treatment will avoid extensive discussion at the level of individual neurons, synapses, and axonal
                    signals. For example, only simplified gross or summary aspects of interneuronal signaling processes
                    and neurodynamic processes will be discussed. Yet, the theory contends that the fine details jibe
                    with these slightly larger-scale functional descriptions. Consensus building (dynamically interact-
                    ing confabulations taking place contemporaneously in multiple lexicons), which the theory hy-
                    pothesizes is the dominant mode of use of confabulation in human cognition, will only be briefly
                    mentioned, as a detailed treatment would go beyond the introductory scope of this sketch of the
                    theory’s biological implementation. At the current time, the theory presented in this Appendix is the
                    only existing detailed explanation of the operation of cerebral cortex or thalamus, and of human
                    cognition.


                    3.A.2 Summary of the Theory

                    The fundamental hypotheses of the theory are summarized in this section. Subsequent sections
                    elaborate.
                       All information processing involved in human cognition is hypothesized to be carried out by
                    thousands of separate thalamocortical modules; each consisting of a particular small localized patch
                    of cortex (possibly consisting of disjoint, nonadjacent, subpatches), and a particular, uniquely
                    paired, small localized zone of thalamus, which are reciprocally connected axonally. These feature
                    attractor modules (of which human thalamocortex has many thousands) are hypothesized
                    to each implement a list of (typically thousands of) discrete symbols (which is stable over time,
                    and can be added to) and to carry out a single symbolic information processing operation called
                    confabulation. Each symbol is represented by a specific collection of neurons within the module.
                    These collections are all about the same size within any single module; but this size varies
                    considerably, from tens to hundreds of neurons per symbol — a genetically determined value,
                    between modules located in different parts of the cortex. Any pair of such neuron collections of the
                    same module, representing two different symbols, typically have a few neurons in common. Each
                    neuron which participates in such a collection typically participates in many others as well. When
   109   110   111   112   113   114   115   116   117   118   119