How to speak the language of thought
By Tom Stafford
(Thinkstock)
We are now beginning to crack the brain’s
code, which allows us to answer such
bizarre questions as “what is the speed of
thought?”
When he was asked, as a joke, to explain how
the mind works in five words, cognitive
scientist Steven Pinker didn't hesitate. "Brain
cells fire in patterns", he replied. It's a good
effort, but all it really does is replace one
enigma with another mystery.
It’s long been known that brain cells
communicate by firing electrical signals to
each other, and we now have myriad
technologies for recording their patterns of
activity – from electrodes in the brain or on
the scalp, to functional magnetic resonance
scanners that can detect changes in blood
oxygenation. But, having gathered these data,
the meaning of these patterns is still an
enduring mystery. They seem to dance to a
tune we can't hear, led by rules we don't
know.
Neuroscientists speak of the neural code, and
have made some progress in cracking that
code. They are figuring out some basic rules,
such as when cells in specific parts of the
brain are likely to light up depending on the
task at hand. Progress has been slow, but in
the last decade various research teams
around the world have been pursuing a far
more ambitious project. We may never be
able to see the complete code book, they
realised, but by trying to write our own
entries, we can begin to pick apart the ways
that different patterns correspond to
different actions.
Albert Lee and Matthew Wilson, at the
Massachusetts Institute of Technology (MIT)
first helped to set out the principles in 2002.
It progresses like this. First, we record from
the brain of a rat – one of our closer
relatives, in the grand tree of life – as it runs
a maze. Studying the whole brain would be
too ambitious, so we can focus our recording
on an area known as the hippocampus,
known to be important for navigation and
memory. If you've heard of this area before
it is probably because of a famous result
which showed that London taxi drivers
developed larger hippocampi the longer
they had spent navigating the streets of
England's sprawling capital.
View image of The longer London taxi
drivers navigate the city's streets, the
bigger certain parts of their brain tasked
with memory and navigation grow
(Thinkstock)
While the rat runs the maze we record where
it is, and simultaneously how the cells in the
hippocampus are firing. The cell firing
patterns are thrown into a mathematical
algorithm which finds the pattern that best
matches each bit of the maze. The language
of the cells is no less complex, but now we
have a Rosetta Stone against which we can
decode it. We then test the algorithm by
feeding it freshly recorded patterns, to see if
it correctly predicts where the rat was at the
point that pattern was recorded.
It doesn’t allow us to completely crack the
code, because we still don't know all the
rules, and it can’t help us read the patterns
which aren't from this bit of the brain or
which aren't about maze running, but it is
still a powerful tool. For instance, using this
technique, the team was able to show that
the specific sequence of cell firing repeated
in the brain of the rat when it slept after
running the maze (and, as a crucial
comparison, not in the sleep it had enjoyed
before it had run the maze).
Fascinatingly, the sequence repeated faster
during sleep – around 20 times faster. This
meant that the rat could run the maze in
their sleeping minds in a fraction of the time
it took them in real life. This could be related
to the mnemonic function of sleep; by
replaying the memory, it might have helped
the rat to consolidate its learning. And the
fact that the replay was accelerated might
give us a glimpse of the activity that lies
behind sudden insights, or experiences where
our life “flashes before our eyes”; when not
restrained, our thoughts really can retrace
familiar paths in “fast forward”. Subsequent
work has shown that these maze patterns
can run backwards as well as forwards -
suggesting that the rats can imagine a goal,
like the end of the maze, and work their way
back from that to the point where they
are.
View image of A rat can mentally replay a
route around a maze 20 times faster
when sleeping than it can when it’s
awake (Thinkstock)
One application of techniques like these,
which are equal parts highly specialised
measurement systems and fiercely
complicated algorithms, has been to decode
the brain activity in patients who are
locked in or in a vegetative state. These
patients can’t move any of their muscles, and
yet they may still be mentally aware and able
to hear people talking to them in the same
room. First, the doctors ask the patients to
imagine activities which are known to active
specific brain regions – such as the
hippocampus. The data is then decoded so
that you know which brain activity
corresponds to certain ideas. During future
brain scans, the patients can then re-imagine
the same activities to answer basic questions.
For instance, they might be told to imagine
playing tennis to answer yes and walking
around their house to answer no – the first
form of communication since their injury.
There are other applications, both
theoretical science , to probe the inner
workings of our minds, and practical domains
such as brain-computer interfaces. If, in the
future, a paraplegic wants to control a robot
arm, or even another person, via a brain
interface, then it will rely on the same
techniques to decode information and
translate it into action. Now the principles
have been shown to work, the potential is
staggering.