Your mind-reading skills are more impressive than you might think,

When I raise my index finger on
entering my local bar in the
Netherlands, by some mysterious
process a bottle of La Chouffe is
brought to the table. This is a trivial
example of mind-reading. That last
sentence might read like an
oxymoron, but mind-reading is
commonplace; and like many things
we are very good at, we are hardly
even aware that we are doing it.
A raised finger can have infinite
meanings - as a warning, a reproach,
a sign of agreement, or in this case, a
request for a drink. But we seem to
be able to negotiate these ambiguities
by sowing common ground, or a
mental space, where we can easily
share information with another, in
the confidence that our meaning is
understood.
A recent study published in the
journal Proceedings of the National
Academy of Sciences looked at mind-
reading in both directions, by
recording the brain activity of both a
sender and interpreter of non-verbal
signals. Within a two-player game
scenario, the sender communicates by
moving a circle on the grid of a
computer screen. The interpreter,
sitting in a separate room, can see
these movements on her own screen.
For each game, the sender has a
specific task - he must try to tell the
interpreter where she should place a
triangle. He also has to indicate to the
interpreter how the triangle should be
oriented on the target square.
A 'Sender' has to use the movements
of his token (a blue circle) to communicate
with an 'Interpreter'. The Interpreter,
located in a separate room, can see these
movements on her own screen. For each
game, the Sender must try to tell the
Interpreter where she should place her
token (an orange triangle). He also has to
indicate how the triangle should be
oriented on the target square. Illustration:
PNAS
It's a complicated little game, and
there's no right or wrong way to
communicate. What's important is
that the players build up their own
improvised dictionary, gradually
assigning meanings to specific sets of
movements (watch an example here).
A sender might hit on an ingenious
signal, but if the other player doesn't
understand his intention, then that
strategy has to be abandoned, no
matter how clever it is. In this way,
an intuitive flow evolves, and
meaning is refined.
It is a model of what happens in
everyday interactions, where we read
the minds of others more or less
continuously. But the depth to which
we do so, depends on how much time
and effort we expend in building
common ground with another
person.
What emerges from these experiments
is that the very same movements may
mean "orient south" for one pair,
while for another pair, they may
mean "orient north". It doesn't
matter that the same movements have
been assigned opposite meanings by
different teams. So long as the sender
and interpreter reach a consensus,
their communication has been
successful, no matter how strange or
illogical the signal may seem to be.
The potential ambiguity is turned on
its head, and allows for endless
inventiveness - much as the limited
repertoire of a pawn can have
echoing possibilities in a game of
chess.
While all of this mind-reading was
going on during the game, brain
activity was also recorded using
magnetoencephalography (MEG) . The
researchers found that the areas that
were activated when a sender was
trying to work out a signal were the
same areas that were activated in the
interpreter when she was decoding
the signal. Importantly, this overlap
points to brain areas that allow for
common ground to take hold.
Even before each game began, there
was a priming of these mind-reading
nodes, as both sides readied
themselves for the game. It is as
though both players were already
setting up a co-operative frame of
mind. Once the game began, the
sender used specific parts of the
brain to work out a communicative
strategy, while the interpreter in turn
decoded these signals using the same
brain areas which the sender used to
create them.
In engineering terms, we can think of
this as an efficient two-way learning
algorithm. Many learning algorithms
need to run a large number of trials
before they converge on an assigned
meaning, but this isn't very useful in
everyday life. If I am in a remote
village where I do not know the local
language and I have diarrhoea, I
don't want to be fooling about with
an algorithm that converges on the
meaning, "show me a toilet", after
three to the power of n loops.
Instead, I want to use our mind-
reading algorithm to convey my
meaning as rapidly as possible.
But it is also commonplace that we
come across people who are simply
unwilling to communicate or co-
operate. Once, on a train in China, I
needed to know if we were
approaching a certain town. As the
train began to slow, I found a
conductor and pointed out the name
of the town to him, written in Chinese
characters in my guidebook. Then I
pointed out the window to the
approaching station, as if to say, "Are
we now coming into this town that
I'm pointing to here in my book?" He
shrugged. I tried to improvise, but he
shrugged again, unwilling to play our
own little two-player game.
In such cases, there is no way to
communicate, in part because the
other person is unwilling to try.
From these experiments it seems
likely that this conductor was
refusing to activate the mind-reading
nodes that would allow us to share a
common space.
Earlier on the same trip, though, I
came across an old man by a bus
station in the suburb of a city. Not
quite sure how to get into town, I
made a few gestures, which he
understood immediately. As clear as
crystal, he told me in our own silent
language which bus to take and
roughly when it would come. All that
information turned out to be correct,
and his mind-reading skills took me
all the way to the centre of town.