Hi, my name is Liang Zhou Koh. I am a graduate student in philosophy
at the University of Toronto and today I want to talk to
you about the false belief task. In the previous videos we were
introduced to our capacity for mindreading that is our capacity to
attribute mental states to others and to reason about those states. In this video we will discuss
an important experiment designed to test for this capacity. The false belief task is a test to see
who can attribute false beliefs to others. Why the emphasis on false beliefs? Let us first consider some limitations
of a test that involves only true beliefs. In a pioneering study on mindreading
in animals published in 1978, psychologists and primatologists
David Premack and Guy Woodruff conducted an experiment with
an adult chimpanzee named Sarah. Sarah was shown a series of videos showing
a trainer trying to reach some fruits that were placed in
inaccessible locations: hanging from the ceiling,
blocked by a crate, etc. At the end of each video Sarah was
shown two photos and told to pick one. The photos depicted the
trainer engaged in some action, for instance, stepping onto a crate. Only one of these actions
would allow him to reach the fruits in the video, just shown. Sarah picked the correct photo
in a significant number of trials. The experimenters concluded that Sarah was
able to predict the actions of the trainer by attributing to him
true beliefs such as “there are fruits hanging
from the ceiling” and "I can reach those fruits
by stepping onto a crate.” However, this conclusion
was soon met with objections. Sarah, many argued, could have completed
the task without engaging in mindreading. For example she could have
based her prediction on past observations about how
people act in similar situations. Or, she could have
picked out the correct action by thinking about what she herself
would do in order to reach those fruits. In order to rule out these
alternative hypotheses, philosophers Daniel Dennett
and Gilbert Harman suggested that a proper test
for mindreading would check whether you
can attribute to other agents beliefs different from those
you yourself hold - beliefs that you think are false. This would allow you to predict
that someone would act in ways that would be incongruent
with their own goals, or in ways different from how
you yourself would act. These suggestions were taken
up by developmental psychologists Heinz Wimmer and Josef Perner who designed a standard
version of the false belief task, commonly known
as the Sally-Anne task. Participants are shown a
story with two characters. Sally is the first shown hiding her
toy in a red box before leaving the room While she is away Anne comes into the
room and moves the toy to the blue box Participants are then
asked to predict where Sally will look for her
toy when she returns. The correct answer, as you would
have guessed, is: in the red box. This is also the answer,
Wimmer and Perner found, that most children
aged 5 and above gave. But they also found that this is not
the case when it comes to 3-year-olds. Most of these younger children
consistently made the incorrect prediction that Sally would look
for her toy in the blue box where they (but not Sally)
knew that it actually was. To date, hundreds of studies have
been conducted on the false belief task, with different variations
in task structure. And the results of Wimmer and Perner’s
original study were generally replicated. What are we to make of this? According to the
standard interpretation, older children can make the right
prediction because they can represent Sally as having a false belief
about the location of her toy. This requires them to recognise that
Sally has a perspective on the world just different from their own. They also have to understand that Sally’s
beliefs are tied to her own perspective, which means that it is possible for
her beliefs to be different from their own. They understand, for
example, that it makes sense for Sally to continue
believing that her toy is in the red box even after it has
been shifted because she, unlike them, didn’t see the transfer that
took place while she was gone. 3-year-olds, on the other hand,
seem to lack this understanding and give answers from
their own perspective. Failing the Sally-Anne test
suggests that they can’t appreciate that others can have beliefs
which do not match their own. On this interpretation,
passing the false belief task is a major developmental
milestone in human mindreading, which takes place
around the age of 4. It is around this age that children
are thought to acquire the ability to distinguish between mind and reality,
to correctly attribute beliefs to others and to make predictions and
explanations about others' actions on the basis of
those attributions. This standard interpretation
faces some challenges. One objection argues that
it grossly underestimates the mindreading capacities
of younger children. There are two main factors that performance
on any cognitive task reflects: the conceptual understanding
required to solve the problem (in this case, an understanding
of beliefs as mental states), and the cognitive skills
needed for completing the task. For the standard version of the
false belief task, those skills include the ability to follow the
narrative, remember its key details and understand the
questions being asked. Perhaps younger children fail to do
well in the standard false-belief task not because they can’t
track false beliefs in others but because of the demands
that such a task places on their developing executive functions
such as working memory. Recent studies using non-verbal
versions of the false belief task seem to provide some
support for this objection. Young infants, some of these studies
suggest, show signs of being sensitive to the false beliefs of others. For example, their looking
patterns seem to anticipate how someone with
false beliefs will act. How robust are these results
and how should we interpret them? These questions have
sparked major debates and are likely to continue to capture the
attention of psychologists and philosophers. In the next video, we will take a closer
look at at the development of mindreading.
