[MURG] Artificial Hippocampus

[C. Altman (QUIST)]

Chips Coming to a Brain Near You

In this era of high-tech memory management, next in line to get that 
memory upgrade isn't your computer, it's you.

Professor Theodore W. Berger, director of the Center for Neural 
Engineering at the University of Southern California, is creating a 
silicon chip implant that mimics the hippocampus, an area of the brain 
known for creating memories. If successful, the artificial brain 
prosthesis could replace its biological counterpart, enabling people 
who suffer from memory disorders to regain the ability to store new 
memories.

And it's no longer a question of "if" but "when." The six teams 
involved in the multi-laboratory effort, including USC, the University 
of Kentucky and Wake Forest University, have been working together on 
different components of the neural prosthetic for nearly a decade. They 
will present the results of their efforts at the Society for 
Neuroscience 's annual meeting in San Diego, which begins Saturday.

While they haven't tested the microchip in live rats yet, their 
research using slices of rat brain indicates the chip functions with 95 
percent accuracy. It's a result that's got the scientific community 
excited.

"It's a new direction in neural prosthesis," said Howard Eichenbaum , 
director of the Laboratory of Cognitive Neurobiology at Boston 
University. "The Berger enterprise is ambitious, aiming to provide a 
prosthesis for memory. The need is high, because of the prevalence of 
memory disorder in aging and disease associated with loss of function 
in the hippocampus."

Forming new long-term memories may involve such tasks as learning to 
recognize a new face, or remembering a telephone number or directions 
to a new location. Success depend on the proper functioning of the 
hippocampus. While this part of the brain doesn't store long-term 
memories, it re-encodes short-term memory so it can be stored as 
long-term memory.

It's the area that's often damaged as a result of head trauma, stroke, 
epilepsy and neurodegenerative disorders such as Alzheimer's disease. 
Currently, no clinically recognized treatments exist for a damaged 
hippocampus and the accompanying memory disorders.

Berger's team began its research by studying the re-encoding process 
performed by neurons in slices of rat hippocampi kept alive in 
nutrients. By stimulating these neurons with randomly generated 
computer signals and studying the output patterns, the group determined 
a set of mathematical functions that transformed any given arbitrary 
input pattern in the same manner that the biological neurons do. And 
according to the researchers, that's the key to the whole issue.

"It's an impossible task to figure out what your grandmother looks like 
and how I would encode that," said Berger. "We all do a lot of 
different things, so we can't create a table of all the things we can 
possibly look at and how it's encoded in the hippocampus. What we can 
do is ask, 'What kind of transformation does the hippocampus perform?'

"If you can figure out how the inputs are transformed, then you do have 
a prosthesis. Then I could put that into somebody's brain to replace 
it, and I don't care what they look at -- I've replaced the damaged 
hippocampus with the electronic one, and it's going to transform inputs 
into outputs just like the cells of the biological hippocampus."

Dr. John J. Granacki , director of the Advanced Systems Division at 
USC, has been working on translating these mathematical functions onto 
a microchip. The resulting chip is meant to simulate the processing of 
biological neurons in the slice of rat hippocampus: accepting 
electrical impulses, processing them and then sending on the 
transformed signals. The researchers say the microchip is doing exactly 
that, with a stunning 95 percent accuracy rate.

"If you were looking at the output right now, you wouldn't be able to 
tell the difference between the biological hippocampus and the 
microchip hippocampus," Berger said. "It looks like it's working."

The team next plans to work with live rats that are moving around and 
learning, and will study monkeys later. The researchers will 
investigate drugs or other means that could temporarily deactivate the 
biological hippocampus, and implant the microchip on the animal's head, 
with electrodes into its brain.

"We will attempt to adapt the artificial hippocampus to the live animal 
and then show that the animal's performance -- dependent in these tasks 
on an intact hippocampus -- will not be compromised when the device is 
in place and we temporarily interrupt the normal function of the 
hippocampus," said Sam A. Deadwyler , "thus allowing the 
neuro-prosthetic device to take over that normal function." Deadwyler, 
a professor at Wake Forest University, is working on measuring the 
hippocampal neuron activity in live rats and monkeys.

The team expects it will take two to three years to develop the 
mathematical models for the hippocampus of a live, active rat and 
translate them onto a microchip, and seven or eight years for a monkey. 
They hope to apply this approach to clinical applications within 10 
years. If everything goes well, they anticipate seeing an artificial 
human hippocampus, potentially usable for a variety of clinical 
disorders, in 15 years.

Overall, experts find the results promising. "We are nowhere near 
applicability," said Boston University's Eichenbaum. "But the next 
decade will prove whether this strategy is truly feasible."

"There is a big gap in making the microchip work in a slice preparation 
and getting it to work in a human being," added Norbert Fortin, a 
neuroscientist from the Cognitive Neurobiology Lab at Boston 
University. "However, their approach is very methodical, and it is not 
unreasonable to think that in 15 to 20 years such a chip could help, to 
some degree, a patient who suffered from hippocampal damage."

Research Group 
http://www.usc.edu/programs/pibbs/site/faculty/berger_t.htm
Wired Link http://wired.com/news/medtech/0,1286,65422,00.html

.......

C. Altman: Quantum topology, neural networks
http://www.umsl.edu/~altmanc/news.html

.......