16.11.2012 Views

Brain–Computer Interfaces - Index of

Brain–Computer Interfaces - Index of

Brain–Computer Interfaces - Index of

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

214 D.M. Taylor and M.E. Stetner<br />

person wouldn’t be able to tell the difference. The artificial limb would respond just<br />

as they intend their own limb to move.<br />

It is unclear how many individual neurons are needed to accurately decode full<br />

arm and hand movements over a continuous range <strong>of</strong> possible joint configurations<br />

and muscle forces. Most studies so far try to predict some subset <strong>of</strong> movement<br />

characteristics, such as predicting two or three dimensional hand position in space<br />

and ignoring other details <strong>of</strong> the hand and arm configuration. Figure 4 compares<br />

actual hand trajectories and brain-controlled movement trajectories from a study by<br />

Taylor et al. where a monkey used intracortical signals to move a virtual cursor to<br />

targets in 3D space [15]. This figure illustrates the importance <strong>of</strong> visual feedback<br />

and the brain’s ability to adapt to the BCI. Figure 4 also illustrates the benefit <strong>of</strong><br />

concurrently adapting the decoding function to track and encourage learning and<br />

beneficial changes in the brain signals.<br />

Part(a)<strong>of</strong>Fig.4 shows the monkey’s actual hand trajectories as it made centerout<br />

movements in 3D space (split into two 2D plots for easier viewing on the printed<br />

page). Part (b) plots trajectories decoded from the intracortical signals recorded<br />

while the animal made the hand movements in part (a). These trajectories were<br />

generated after the animal completed the experiment, so the animal did not have any<br />

real-time visual feedback <strong>of</strong> where the neural trajectories were going. Part (c) shows<br />

what happened when the animal did have visual feedback <strong>of</strong> the brain-controlled trajectories<br />

in real time. Here the animal used its intracortical signals to directly control<br />

the 3D movements <strong>of</strong> the cursor to the targets. In this case, the monkey modified<br />

its brain signals as needed to correct for errors and steer the cursor to the targets.<br />

Part (d) shows what happened when the experimenters also adapted the decoding<br />

function to track and make use <strong>of</strong> learning induced changes in the animal’s brain<br />

signals. Adapting the BCI decoder to the brain as the brain adapts to the decoder<br />

resulted in substantial improvements in movement control. Many neural signals that<br />

did not convey much movement information during normal arm movements became<br />

very good at conveying intended movement with practice. Regularly updating the<br />

decoding function enabled the decoder to make use <strong>of</strong> these improvements in the<br />

animal’s firing patterns.<br />

Dr. Schwartz and colleagues have moved further toward generating whole arm<br />

and hand control with intracortical signals. They trained a monkey to use a braincontrolled<br />

robotic arm to retrieve food and bring it back to its mouth. This<br />

self-feeding included 3D control <strong>of</strong> the robot’s hand position as well as opening<br />

and closing <strong>of</strong> the hand itself for control <strong>of</strong> four independent dimensions or<br />

actions [16].<br />

The United States government’s Department <strong>of</strong> Defense is currently sponsoring<br />

research investigating use <strong>of</strong> intracortical signals to control very sophisticated prosthetic<br />

limbs to benefit soldiers who have lost limbs in the war efforts in the Middle<br />

East. The goal <strong>of</strong> this funding is to develop realistic whole-arm-and-hand prosthetic<br />

limbs with brain signals controlling 22 independent actions or dimensions including<br />

control <strong>of</strong> individual finger movements. Researchers involved in this effort so far<br />

have been able to decode individual finger movements from the intracortical signals<br />

recorded in monkeys [17]. This government funding effort is intended to fast-track

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!