Brain Architecture for an Intelligent Stream of Consciousness - Brain Theory from a Circuits and Systems Perspective - page 27

may be a source of enhanced computational power, particularly if formed at the
molecular level within a neuron. Such a discovery might actually be more likely
than real qubits, which require an isolated quantum system.
Conclusions
This chapter has presented a model in which the various attributes of a mental
image each have their own fixed locations. Also each attribute has one or more
backup locations, but physical locations are irrelevant in the logical brain architec-
ture of this topic.
Conscious STM is modeled as fed with appropriately processed signals from the
five senses, interspersed with successful recalls from subconscious LTM. A person
is aware of mental images mainly from the senses, but interspersed are recalls from
LTM that are extremely important to survival and intelligent behavior.
A cue editor helps ensure the availability of recalls when cues are inadequate.
This topic outlines a model in which a cue editor works with selected attributes
from a current image in conscious STM. When there is a mental block, the cue
editor immediately begins to modify cues in a random way to obtain returns from
LTM. A cue editor serves to resolve conflicting (overdetermined) cues, and cues for
which there are absolutely no memory matches within associative LTM.
Under this model, ineffective or conflicting cues are not discarded but are stored
in a special register composed of recursive neurons. Before searching again, a few
cues are randomly removed, and the remaining are used for a new search; and if still
there are no matches, then the removed cues are restored. The process of
randomizing is repeated again and again until after some point, returns are finally
obtained. This system can account for the common experience that forgotten
information often pops into one's head with delay, perhaps much later, at an
unexpected moment.
Parallel to the neural paths for memory search are neural paths that bring up
returns from LTM. A recall referee serves to deal with the common problem of
multiple returns from a given set of cues. Multiple returns are expected to occur in a
group shortly after cues are applied. Within a recall referee, each return undergoes a
quick but exact digital determination of priority. Most returns are rejected as being
unimportant, but one will have maximum priority and will be gated directly into
conscious STM, where it results in physical actions or additional memory searches.
Methods for priority calculations are presented in later chapters using recursive
neurons operating as controlled toggle devices, all operating in parallel. Controlled
parallel toggles calculate priority for each multiple return in a prompt manner.
Brain editors as above operate between short- and long-term memories to make
decisions and to produce a stream of consciousness, a stream that is infinitely more
intelligent than mere dumb acceptance of bad cues and unimportant returns.
Subsequent chapters are going to develop a structure with appropriate circuits for
short- and long-term memory as well as neural circuits for editing to accomplish the
abovementioned system.