Eukaryotic cells are capable of performing very complex information
processing tasks, and can be thought of as a computational device
capable of performing perception-action behavior. This behavior
takes the form of the cell extracting information from the local
environment, integrating the information relative to the current
internal state, and producing an action to enhance the cell's
fitness in the current environment. To facilitate this sort
of information processing the cell would need to work in a coordinated
fashion requiring a long-range signaling mechanism that can
integrate information from across the cell quickly. Many short-range
signaling mechanisms have been identified in the eukaryotic
cell biology, but a long-range signaling mechanism has yet to
be conclusively established. However, evidence indicates that
the cytoskeleton could fill this long-range signaling role,
specifically the microtubule network because of its organizational
characteristics and preliminary evidence for information transmission
cpacitiy.
To explore this proposed natural signaling medium a computer
learning model is used that combines a biologically motivated
growth simulation with an abstract signaling mechanism to create
an adaptive signaling medium. This signaling medium is molded
by a process we call adaptive self-stabilization, which is essentially
a feedback mechanism that translates network fitness into regulatory
signals for modulating the growth dynamics. Ultimately a goal
of this work is to harness the model's inherent oscillatory
dynamics for controlling a biomimetic robot that captures the
interdependent nature of the eukaryotic cell's various components.
|
