However, the experimenter had actually put a pencil inside the tube. These
non-standard contents are now revealed to the child. The original point of
the Smarties test had been toWnd out what children would say about
another child who was going to be presented with the Smarties tube, with
the pencil still inside – ‘What will your friend say the tube contains?’. As in
the Maxi-type false-belief test younger children, under the age of about
four, erroneously attribute the knowledge they now have to other children
- that is, they say the other child wouldsay there was a pencil in the tube.
But from about four years on, children correctly predict that the other
child will get the contents wrong and say that the tube containssweets.So
results of the Smarties test corroborate the developmental watershed
indicated by the original false-belief tests (Hogrefeet al., 1986; Perneret
al., 1987).
But the really interestingWnding on the Smarties test emerges from
responses to a supplementary question: ‘What didyouthink was in the
tube?’ It turns out that younger children – at an age when they are still
failing the false-belief task – also fail to acknowledgetheir own past false
belief. They answer the question by saying ‘a pencil’, even though only a
few moments before they had said that there were sweets in the tube
(Astington and Gopnik, 1988). Yet their failure is not merely one of
memory – they can recall what they hadsaidwas in the tube, for example.
Very similar results to those for the Smarties test have also been obtained
onHollywood Rock– sponges tricked out to look like rocks. When they see
them, children will atWrst say they are rocks. They are then invited to
handle them and discover, to their surprise, that they are not really rocks.
Once again the younger children, not yet possessing a developed concept
of (false) belief, will say that other children looking at them will think they
are sponges; and they will also say thatthey themselves thought they were
sponges before too! (See Gopnik, 1993, for a survey of test results and
theoretical discussion.)
From the perspective of mature mind-reading, these may appear the
strangest and most surprising results. In fact, they count heavily in favour
of a nativist version of theory-theory and against the developmental
theorisingview. (So here is a second point to add to the argument from
developmental rigidity advanced in section 1.1 above.) The theorising
account must be that children learn to revise a more primitive theory,
which uses conceptions ofdesireandknowledgeand/orperception,to
include possiblefalse beliefas well. Since children will sometimes have
beliefs which are false, the suggestion could be that they will learn from
their own case that there are such things as false beliefs. But natural as that
thought might be, it is covertly Cartesian: it just assumes that introspection
is a specially privileged form of observation, which is not theory-laden.
Developmental studies 93