Netdoctor Dr. F. Frank LeFever "flefever at ix.netcom.com" on 23 Jul 1998:
>>> >> /^^^^^^^^^^^\ The Architecture of a Robot Brain /^^^^^^^^^^^\
>>> >> /visual memory\ ________ / auditory \
>>> >>| /--------|-------\ / syntax \ | memory |
>>> >>| | recog-|nition | \________/---|-------------\ |
>>> >>| ___|___ | | | | _______ | |
>>> >>| /image \ | __V___ ___V___ | /stored \ | |
>>> >>| / percept \ | /deep \------/lexical\----|--/ phonemes\| |
>>> >>| \ engrams /---|---/concepts\----/concepts \---|--\ of words/ |
>>> >>| \_______/ | \________/ \_________/ | \_______/ |
>Thank you! Besides the problem of undefined terms, there is the
>problem I alluded to long, long ago: the lack of any correspondence
>between these line and box constructions and KNOWN cerebral
>connections, which are far more complex and interesting, and moreover
>divide what seem (to the untutored) obvious categories into non-obvious
>categories--for example, such "simple" dissections of visual input as
>those of Hubel and Wiesel, and dissections further along the visual
>stream (e.g. dorsal vs. ventral streams), alternative routes to
>phonology as shown by Deep Dyslexia, etc., etc.
>New York Neuropsychology Group
Ah, the Monday morning AI, Neuro, Robotics seminar rocks the Net.
The above-diagrammed Mind Model is entirely based upon Hubel and
Wiesel's discovery of mammalian (feline) feature-extraction, BUT:
For purposes of AI and Robotics (seminar groupings), the visual
memory channel needs only to deliver image-recognitions to the
rest of the associative mindgrid.
Surely, it is interesting to know HOW vision works, and the
knowledge of feature-extraction was a "conditio sine qua non"
for the development of this over-arching mind-model. Nottheless,
the Mentifex Model has not yet been found to be in disagreement
or in conflict with any known facts about the human visual system.
For instance, the notion that vision is broken down into many
wisely separate "planar" processing areas fits in. [End of post.]