Introduction to living cortical networks and
multielectrode array technology
The brain is perhaps one of the most powerful and robust
computing machines in existance. It can recognize vastly
different patterns, store a lifetime worth of information,
and yet is more fault tolerant than any computer today. How
does the brain do this? To answer that question you would
need to study how neurons, which are the major computing
component in the brain, processes and encodes information.
In the past we we were limited to examining just a few
neurons at a time using single electrode "patch" style
recordings and from that we learned, and still learn, a
great deal about the inner workings of single neurons. But
we also know that much of the computational properties of
neurons occurs at the neural ensemble or "population"
levels (thousands to millions of neurons in concert).
Multielectrode arrays (MEA) were created to bridge the gap
between recordings of single neuron to and population
levels recordings by creating a system in which a large
grid of electrodes could be employed to collectively sample
the activity across a a small population of cells and yet
maintain the resolution to measure individual cells.
Figure 1 shows and example of a microelectrode array.
These MEAs were developed in the late 1970's by Jerry Pine
, Gunter Gross.
The MEA is essentially a
large grid of electrodes spread across the surface of
a dish in which neurons can be grown (cultured). These
arrays permit researchers to both study and stimulate
patterns of activity of neurons grown across the
surface of the MEA. For example, we can input various
stimulation patterns into the cortical network and
examine the response (output) of the network to learn
more about how these networks compute or store
In our lab, rat cortex obtained from Brain Bits
(http://www.brainbitsllc.com/), is mechanicly dissociated
and digested using the Papain Dissociation Kit from
(http://www.worthington-biochem.com/). The papain digestion
dissolves the connective tissue surrounding the neurons
leaving a suspension of neuron cell bodies (called soma)
which are then dropped onto the surface of the MEA and its
electrodes which have coated with polyethelyneamine and
laminin to improve adhesion and promote growth. (see link
for detailed protocols under the technology section of this
Neurons that are cultured in this matter will rapidly
begin to reconnect and form a dense neural network. In
other words, a living neural network that we can study in
detail. Figure 2 shows a time-lapse movie of the growth and
emergence of connectivity of neurons over the course of the
first 8 hours after being placed into the MEA. As you can
see, many of the neurons (which appear as small dark
spheres) send out "filaments" or processes to nearby
neurons. The movement that you are seeing are the neurons
in the process of forming a new neural network of cells. As
the network emerges, the neurons will begin spontaneously
send messages (action potentials) down these processes to
other neurons in the net. These action potentials are
electrical events which can be detected by the electrodes
on the MEA as large deviations or "spikes" in the voltage.
For example, during the first few days scattered individual
action potentials (spikes) begin to be detected by the MEA.
However after 10 days the network will begin to produce one
of the primary forms of network activity seen in these
networks in which neurons will begin to fire in synchronous
Figure 3 is a movie showing
the raw electrical activity recorded by an 8x8 grid of
electrodes on an MEA. Each window displays 200 milliseconds
of raw electrical activity including ambient electrical
noise (blue fuzzy line at the center of each window) and
spikes which appear as deviations from the noise. Note how
spontaneous spikes appear throughout the culture followed
by periods/bursts of activity. These bursts are
semi-periodic occurring every 1 to 15 seconds, dependent on
the culture measured.
Figure 4 illustrates the
fine spatial and temporal structure to these bursts. In
this movie, the grid of electrodes are depicted flat along
the bottom of the plot and the colorized waves represent
integrated activity (leaky integrator) over time. The move
is 10x slower than realtime and shows the burst activity
seen in Figure 3 as wells as some of the patterns that
occur during a burst. For example, the spikes can be seen
to propagate across the surface. In fact there is a great
deal of variability within each burst illustrate the
potentially rich amount of information that may be present
in the both the timing and location of spikes.
With this system we can measure this rich dynamical system
of neurons in realtime as well as stimulate patterns of our
own to understand how information is processed and encoding
in living neural networks.
A sample of areas of research using MEA technology: