Cognitive Processes 409imitation. The idea of learning by imitation would seem to re-
quire three things to be entirely convincing: The behavior to
be imitated should be stored as a representation by the ob-
server; the observer’s behavior should be a more or less faith-
ful copy of the demonstrator’s behavior; and the behavior
imitated should be reasonably complex and not a species-
typical behavior such as pecking by a chicken. A useful ex-
perimental design to study imitation is called the two-action
test. In this design, a given result may be accomplished
in two different ways. Observers should be shown to engage
in the particular behavior that they were allowed to observe
rather than the one not observed (Heyes, 1996).
Rats prefer foods their mothers ingested during pregnancy
and after birth. Nursing rats come to prefer foods ingested by
their mothers. Rats also come to prefer foods eaten by their
conspecific; one way they determine what this is food is is by
smelling the breath of their conspecific. It has been shown
that a group of rats that has come to prefer a specific flavor
will pass on that preference to new members of the group.
Roof rats have come to occupy the pine forests of Israel,
where they subsist on pine seeds, which are rendered difficult
to extract because of the tough scales that must be removed.
To remove the scales, the rat must begin chewing on them
at the base of the cone, removing them scale by scale by fol-
lowing the spiral pattern that goes to the top of the cone. Rats
have failed to learn this if left to their own devices. If raised
with a mother who is an efficient stripper they do acquire the
knack of getting to the seeds—but not, apparently, through
imitation. They acquire the behavior by getting access to
cones that have already been partially stripped, even if they
have been partially stripped by the experimenter.
In a noteworthy case of observational conditioning, mon-
keys have acquired fear of snakes by observing a demonstrator
monkey exhibiting such fear (Mineka & Cook, 1988). Fear is
acquired even by merely observing a demonstrator on video
showing fear to a toy snake. Monkeys have observed demon-
strators on video showing an apparent fear of flowers that had
been spliced into the film in place of the snake. Fear of snakes
is acquired more readily by naive monkeys than fear of flowers.
In addition to observational conditioning, three others
categories of learning have been distinguished: stimulus en-
hancement, emulation, and imitation. Instimulus enhance-
ment,the demonstrator’s behavior simply directs the observer
to stimuli that makes copying the demonstrated behavior
more likely.Emulationoccurs when behavior is copied in a
more or less general way by employing techniques different
from that of the demonstrator.Imitationinvolves faithful
copying of the demonstrator’s behavior. Children appear to
imitate behavior more faithfully than chimpanzees (see, e.g.,
Whiten & Boesch, 1999).
Interest in imitation or learning from others has a long his-
tory (e.g., Romanes, 1882; Thorndike, 1911), and over that
long period of time scores of observational and experimental
reports have been published. Nevertheless, we know little
about imitation. One problem is the lack of a useful theory of
imitation that can direct our efforts into useful channels.
Another is that only in relatively recent times have we devel-
oped useful experimental procedures for investigating learn-
ing by observation. Two examples here would include the
video techniques mentioned earlier for examining fear acqui-
sition in monkeys, and the two-action experimental design. It
may not be too optimistic to conclude that our understanding
of imitation may undergo rapid and significant development
over the next few years.Interval TimingThe ability of animals to time arbitrary events has been inten-
sively investigated in recent years through a variety of proce-
dures. How animals time intervals is seen as important for
understanding learning and cognition generally. For example,
according to one view, animals employ the same mechanisms
to time events as to count them. Even more generally, timing
events has been seen as fundamental to understanding all va-
rieties of learning and cognition (Gallistel & Gibbon, 2000).
That is, both Pavlovian and instrumental procedures are seen
as involving the learning of temporal intervals. For example,
in the Pavlovian preparation the animal responds most vigor-
ously at the termination of the conditioned stimulus in de-
layed conditioning. This finding suggests that the animals are
sensitive to the time elapsed since the stimulus was pre-
sented. In instrumental conditioning, it has been suggested
that the animals compared the time to reinforcement in the
trial to the overall time between reinforcements (but see
Capaldi, Alptekin, & Birmingham, 1996).
Most studies explicitly concerned with timing have used
instrumental conditioning. In the peak procedure,animals re-
ceive many daily trials in which reinforcement occurs after a
fixed time following the onset of a signal (Roberts, 1981). A
major finding is that animals respond most at approximately
the time that reinforcement is due. For example, if the inter-
val is 20 s, most responding occurs at about 20 s. Another
major finding using the peak procedure is that maximum
response rate is reached at a certain proportion of the way
into this interval regardless of the interval’s length.
In tests of temporal generalization,an animal is rein-
forced for responding to one signal duration but not others. A
major finding using this procedure is that a typical general-
ization gradient is obtained with maximum responding con-
fined to the reinforced duration (Church & Gibbon, 1982).