I Know Kung Fu
New technology may soon enable us download knowledge directly into our brains, says a new study.
By decoding activation patterns from fMRI scans and then reproducing them as direct input to a precise area of the brain, the new system may be able to “teach” neural networks by example – priming them to fire in a certain way until they learn to do it on their own.
This has led everyone from io9 to the National Science Foundation to make Matrix references – and it’s hard to blame them. After all, immersive virtual reality isn’t too hard to imagine – but learning kung-fu via download, like grabbing an mp3? Sounds like pure sci-fi – especially since we know that the way muscle pathways form memories is pretty different from how we remember facts and images.
The basic idea is this: when you learn to perform a physical action – say, riding a bike or shooting a basketball – your muscles learn to coordinate through repetition. This is called procedural memory, because your muscles (and parts of your brain) are learning by repeating a procedure – in other words, a sequence of electrochemical actions – and (hopefully) improving the precision of that procedure with each run-through.
In contrast to this, we have declarative memories – memories of, say, the color of your favorite shirt, or where you had lunch today. Though declarative memories can certainly improve with practice – think of the last time you studied for an exam – there’s typically not an “awkward” stage as your brain struggles to learn how to recreate these memories. In short, once a bit of information is “downloaded” into your conscious awareness, it’s pretty much instantly available (until you forget it, of course).
Now, I could give some examples that blur the lines between these two types of memory – reciting a long list of words, for instance, seems to involve both procedural and declarative memory – but my point here is that procedural memories tend to require practice.
So it’s pretty surprising to read, in the journal Science, that a team led by Kazuhisa Shibata at Boston’s Visual Science Laboratory may have found a way to bridge the gap between these two types of memory.
The team began by taking fMRI scans of the visual cortex as volunteers looked at particular visual images – objects rotated at various angles. Once the team had isolated a precise activation pattern corresponding to a particular angle of orientation, they turned around and directly induced that same activation pattern in the volunteers’ brains:
We induced activity patterns only in early visual cortex corresponding to an orientation without stimulus presentation or participants’ awareness of what was to be learned. The induced activation caused VPL specific to the orientation.
In other words, the researchers triggered brain activity patterns corresponding to specific angles of visual orientation without telling the volunteers what the stimulus was going to be.
Then, when the scientists asked the volunteers what they’d “learned,” the volunteers had no idea. But when the researchers asked them to pick a “best guess” orientation that seemed “right” to them, a significant percentage chose the orientation their brains had been trained to remember.
This isn’t exactly downloadable kung-fu – but it provides some of the first conclusive evidence that not only do repeated visual stimuli help sculpt brain networks – direct stimulation can sculpt the way those networks learn about visual stimuli.
Could we use technology like this to teach people to feel less anxious or depressed? Or to understand concepts they have trouble grasping? And what might happen if we could harness this technology to our own mental output? What if we could literally dream our way to new skills, or more helpful beliefs?
We’re not quite there yet, but it may very well happen in our lifetime.