Systems, Man & Cybernetics - January 2017 - 31

Intermittent singular space-time
dynamics has been observed in brainimaging experiments by Freeman and
colleagues, demonstrating transitions
from metastable, high-dimensional
background states with large-scale
synchrony to brief episodes of lowdimensional states with the collapse
of synchrony. Neuropercolation helps
interpret the collapse of the highdimensional cognitive space to a lowdimensional subspace of attractors.
Such collapse demarcates the onset of
awareness and conscious experience,
the "aha" moment, as part of the intentional action-perception cycle.
Applications of neuropercolation
include novel brain-computer interfaces. The observed transient synchronization dynamics indicate the
formation of percepts and serves as
an indicator of high-level cognitive
experience. We analyzed electrocorticography/electroencephalography
(ECoG/EEG) data and identified
synchronization effects with sudden
transitions in spatiotemporal neurodynamics, which are interpreted as
cognitive phase transitions. Novel
scalp EEG designs are proposed with
high-density electrode arrays, which
provide the hardware and software
tools to drastically reduce the effect
of muscle artifacts.
Neuropercolation models can be
used to model and potentially predict
catastrophes with sudden, drastic
changes in the behavior of complex
systems, such as stock market crashes, solar f lares, earthquakes, and
other disasters. Our important fundamental theoretical result concerns the
prediction of the rapid switch from
one state to the next, described as
"dragon kings," as opposed to the unpredictable nature of "black swans,"
representing self-organized criticality
with self-similar dynamics.
Our results indicate that brains
are not computers in the sense of
our presently dominant digital computer paradigm. We propose a novel
computation paradigm that goes
beyond symbol manipulation by a
given rule set, as specified by the
Turing principles. Our model operates

Kozma at the naSa JPL with the Srr-2K mars rover prototype.

Kozma at the tucson 2008 consciousness conference, with (from left)
world-renowned neuroscientists Walter J. Freeman and Karl Pribram and
giuseppe Vitiello, an Italian quantum physicist.

sequentially, resembling Turing machines; however, it has several distinguished features exceeding those of
Turing computing with:
1) The computing sequences are selforganized through frequent phase
transitions between synchronized
and desynchronized states.
2) The symbols are not fixed but
dynamically emerge in the form of
spatially distributed, metastable
patterns.
3) Rather than following a predefined
instruction set that tells it how to
switch from one pattern to the
other, the system follows a dynamic
trajectory in a high-dimensional
Ja nu a r y 2017

state space, modified by learning
experience.
Some of our theoretical results
have been applied to develop new
intelligent algorithms for robot navigation and control. In collaboration
with colleagues at NASA JPL in Pasadena, California, under the leadership
of Homayoun Seraji, we developed
a novel autonomous control for the
prototype Mars Rover SRR-2K. Applications also include distributed sensor systems in the framework of the
AFRL's layered sensing paradigm, as
well as the development of decision
support systems under rapidly changing scenarios.

IEEE SyStEmS, man, & CybErnEtICS magazInE

Table of Contents for the Digital Edition of Systems, Man & Cybernetics - January 2017