Home > News > Intracranial Ventriloquism

Intracranial Ventriloquism

Futuristic-sounding story out of University of Utah today: electrodes implanted in the skull allow patients to speak without making a sound. While this is certainly a cool new trick, the investigator was quick to add: “It doesn’t mean the problem is completely solved and we can all go home. It means it works, and we now need to refine it so that people with locked-in syndrome could really communicate.”

University of Utah Department of Neurosurgery

When he says “it works,” what he means is that they get output that can be interpreted as individual words at a rate that is significantly better than chance. What they’ve done is implant small grids of electrodes over two areas of cortex and ask the subjects to think of individual words (the grids in the picture at left are artist renderings since the actual grids are too small to see, and they are next to “standard” size EEG leads). The researchers then try to characterize the electrical output to see if they detect any patterns. They then ask the subject to think of individual words (from a pre-determined list of 10 words) at random to see if they can recognize the electrical pattern and guess the right word. By honing the technique, they got up to about 50% accuracy which isn’t bad. You can’t really have a conversation at this level, and it doesn’t approach the near-100% accuracy you get by having someone blink when you read the correct word/letter (think of The Diving Bell and the Butterfly), but as a first attempt it’s pretty cool.

One confusing thing, however: in the press release they describe the “surprising” result that Wernicke’s area was less active then facial motor cortex during speech production, and that Wernicke’s area was most active when the subjects were thanked by the researchers. I don’t know why this is surprising, since Wernicke’s area is mostly involved with language recognition, interpretation, and meaning. Anyone who’s seen a patient with a stroke in that area who develops a Wernicke’s Aphasia is well aware that they have no problem speaking. They just have trouble making sense. Putting a grid over facial motor cortex is clever, though, as this are is likely active when movements are contemplated as well as when they are executed, and distinct words would be expected to have a different pattern of activation (since by definition a different series of muscles would be activated).

So we’re a long, long way away from anybody talking without moving a muscle, but in order to get there, we have to start somewhere. And apparently that somewhere is in Utah.

Advertisements
Tags: , ,
  1. September 19, 2010 at 11:29 am

    Interesting study. I wonder if you could do this without cutting people’s skulls open. With fMRI, using similar mathematical “decoding” approaches, you can infer what pictures people are thinking about and also what words.

    However I suspect that the trick will be being able to distinguish between two very similar words. That might not be possible with fMRI.

    • September 19, 2010 at 11:02 pm

      You probably could do something similar with fMRI. There was a report not long ago about patients in minimally conscious states being able to answer yes/no questions by thinking about different movements, and the fMRI could distinguish them. So a code like “thinking of moving left leg = yes,” “right leg=no,” “left arm=leave me alone now” could be worked out. The main obstacle would be rolling around with the MRI scanner strapped to one’s head!

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: