8 June 2013 | NewScientist | 35keith negleyphone will be on the table. If such a construct
represents “second-order” theory of mind,
most of us can understand a fourth-order
sentence like: “John said that Michael thinks
that Anne knows that Sally thinks her phone
will be on the table.”
In fact Dunbar’s team has shown that such
a concept would be the limit of about 20 per
cent of the general population (British Journal
of Psychology, vol 89, p 191). Sixty per cent of
us can manage fifth-order theory of mind and
the top 20 per cent can reach the heights of
sixth order.
As well as letting us keep track of our
complex social lives, this kind of mentalising
is crucial for our appreciation of works of
fiction. Shakespeare’s genius, according to
Dunbar, was to make his audience work at the
edge of their ability, tracking multiple mind
states. In Othello, for instance, the audience
has to understand that Iago wants jealous
Othello to mistakenly think that his wifeDesdemona loves Cassio. “He’s able to lift
the audience to his limits,” says Dunbar.
So why do some of us operate at the
Bard’s level while others are less socially
capable? Dunbar argues it’s all down to the
size of our brains.
According to one theory, during human
evolution the prime driver of our expanding
brains was the growing size of our social
groups, with the resulting need to keep track
of all those relatives, rivals and allies. Dunbar’s
team has shown that among monkeys and
apes, those living in bigger groups have a
larger prefrontal cortex. This is the outermost
section of the brain covering roughly the
front third of our heads, where a lot of higher
thought processes go on.
Last year, Dunbar applied that theory to
a single primate species: us. His team got
40 people to fill in a questionnaire about the
number of friends they had, and then imaged
their brains in an MRI scanner. Those with thebiggest social networks had a larger region of
the prefrontal cortex tucked behind the eye
sockets. They also scored better on theory of
mind tests (Proceedings of the Royal Society B,
vol 279, p 2157). “The size of the bits of
prefrontal cortex involved in mentalising
determine your mentalising competencies,”
says Dunbar. “And your mentalising
competencies then determine the number of
friends you have.” It’s a bold claim, and one
that has not convinced everyone in the field.
After all, correlation does not prove causation.
Perhaps having lots of friends makes this part
of the brain grow bigger, rather than the other
way round, or perhaps a large social network is
a sign of more general intelligence.Lying robots
What’s more, there seem to be several parts of
the brain involved in mentalising – perhaps
unsurprisingly for such a complex ability. In
fact, so many brain areas have been implicated
that scientists now talk about the theory of
mind “network” rather than a single region.
A type of imaging called fMRI scanning,
which can reveal which parts of the brain
“light up” for specific mental functions,
strongly implicates a region called the right
temporoparietal junction, located towards the
rear of the brain, as being crucial for theory
of mind. In addition, people with damage to
this region tend to fail the Sally-Anne test.
Other evidence has emerged for the
involvement of the right temporoparietal
junction. When Rebecca Saxe temporarily
disabled that part of the brain in healthy
volunteers, by holding a magnet above the
skull, they did worse at tests that involved
considering others’ beliefs while making
moral judgments (PNAS, vol 107, p 6753).
Despite the explosion of research in this
area in recent years, there is still lots to learn
about this nifty piece of mental machinery.
As our understanding grows, it is not just our
own skills that stand to improve. If we can
figure out how to give mentalising powers to
computers and robots, they could become a
lot more sophisticated. “Part of the process
of socialising robots might draw upon things
we’re learning from how people think about
people,” Apperly says.
For instance, programmers at the Georgia
Institute of Technology in Atlanta have
developed robots that can deceive each other
and leave behind false clues in a high-tech
game of hide-and-seek. Such projects may
ultimately lead to robots that can figure out
the thoughts and intentions of people.
For now, though, the remarkable ability to
thoroughly worm our way into someone else’s
head exists only in the greatest computer of
all – the human brain. nKirsten Weir is a science writer based in Minneapolis130608_F_Mindreading.indd 35 31/5/13 14:24:15