| | |
Summary: Human sensory systems handle enormous amounts of
complex information during every waking moment. An
increasing number of studies supports the idea that the
processing of this information is implemented by an un-
supervised form of continuous observational learning,
which organizes both old and new sensory information
into maximally efficient codes (Fiser &Aslin, 2005).This
statistical learning mechanism has been demonstrated
using speech streams in adults (Saffran, Newport, &Aslin,
1996), infants (Saffran, Aslin, & Newport, 1996), and
nonhuman primates (Hauser, Newport, & Aslin, 2001),
and using visual sequences in adults (Fiser &Aslin, 2002)
and infants (Kirkham, Slemmer, & Johnson, 2002).
A key question is how such a powerful learning mech-
anism is sufficiently constrained to enable associative
learning without suffering from intractable computa-
tional demands, given the infinite number of potentially
relevant statistical structures in the input. We chose to
investigate this question within the framework of visual
statistical learning that occurs during brief periods of ob-
|
