can see in this example how an innate grouping rule could help a learning
process to get started. (I am not suggesting that the establishing of acoustic
boundaries at discontinuities is the only method that infants use to discover
units, but I would be very surprised if it were not one of them.)
Another example of innate segregation that was given earlier concerned an
infant trying to imitate an utterance by her mother. It was argued that the fact
that the infant did not insert into her imitation the cradle’s squeak that had
occurred during her mother’s speech displayed her capacity for auditory scene
analysis. I am also proposing that this particular capacity is based on innately
given constraints on organization.
There is much experimental evidence drawn from experiments on the vision
of infants that supports the existence of innate constraints on perceptual orga-
nization. Corresponding experiments on auditory organization, however, are
still in short supply.
One such study was carried out by Laurent Demany in Paris.^15 Young infants
from 1^12 to 3^12 months of age were tested with sequences of tones. The method
of habituation and dishabituation was used. This is a method that can be used
with infants to discover whether they consider two types of auditory signals
the same or different. At the beginning, a sound is played to the babies every
time they look at a white spot on a screen in front of them. The sound acts as a
reward and the babies repeatedly look at the white spot to get the interesting
sound. After a number of repetitions of this ‘‘look and get rewarded’’ sequence,
the novelty of the sound wears off and it loses its potency as a reward (the
infants are said to have habituated to the sound). At this point the experimenter
replaces the sound by a different one. If the newness of the sound restores its
ability to act as a reward, we can conclude that the infants must consider it to
be a different sound (in the language of the laboratory, they have become dis-
habituated), but if they continue ignoring it, they must consider it to be the
same as the old one.
Using this method, Demany tried to discover whether infants would percep-
tually segregate high tones from low ones. The proof that they did so was indi-
rect. The reasoning went as follows: Suppose that four tones, all with different
pitches, are presented in a repeating cycle. Two are higher in pitch (H1 and H2)
andtwoarelower(L1andL2),andtheyarepresentedintheorderH1,L1,H2,
L2,.... If the high and low tones are segregated into different perceptual
streams, the high stream will be heard as
H1–H2–H1–H2–H1–H2–...and the low stream will be perceived as
L1–L2–L1–L2–L1–L2–...(where the dashes represent brief within-stream silences). In each stream all
that is heard is a pair of alternating tones.
Now consider what happens when the reverse order of tones is played,
namely L2, H2, L1, H1,.... If the high tones segregate from the low ones, the
high stream is heard as
H2–H1–H2–H1–H2–H1–...244 Albert S. Bregman