508 Procedural Memory and Skill Acquisition
forces observers to attend to the distinctive and unique fea-
tures of stimuli; having attended to these features, observers
can use them to improve performance (e.g., Rabin, 1988). It
may be more than a matter of affectation that wine tasters
have developed such an elaborate vocabulary for classifying
wines. Training that directs the observer’s attention to unique
features has also been shown to result in better perceptual
learning (e.g., Biederman & Shiffrar, 1987).
Sowden, Davies, and Roling (2000) investigated whether
improved sensitivity in detecting basic features could be a
basis for improvement in reading X-ray images. Experts were
found to be more sensitive than novices in detecting dots in
X-ray images. Further, novices were found to improve over
4 days of training but to show no transfer to reversed contrast
images when these images were simple. When more com-
plicated images were used, transfer (although not perfect) did
occur. Sowden et al. interpreted these results as evidence that,
in addition to strategic components, stimulus-specific sensory
learning is important in learning to read X-ray images.
Perceptual learning leads to improved recognition and
classification of stimuli, but it may also reflect improved pro-
cessingof stimuli. Processing may become more efficient
because stimuli are unitized in a sort of visual chunking
process (LaBerge, 1973), or because observers become more
fluent in applying learned operations. Kolers and Roediger
(1984) developed the idea that stimuli are not remembered
independently of the operations performed on them. That is,
learning can be viewed as reflecting both experience with the
stimuli and experience processing them. Evidence for this
view comes from a series of studies in which observers read
geometrically inverted text (i.e., text presented upside down
and from right to left; Kolers, 1975a). After about two
months of practice, participants became quite proficient in
the task. Because different texts were used on different days,
the learning was not tied only to the particular stimuli used
in the task. In fact, when participants were tested in the same
task more than a year later, reading times were only 5% faster
when the same passages used in training were read than when
completely new passages were used (Kolers, 1976). The
advantage for the previously read pages was likely due to
specific practice with the analysis of the graphemic patterns
and not due to prior exposure to the content of the text. This
is suggested by a study from Kolers (1975b) in which prior
reading of the same text in a normal orientation did not facil-
itate reading of inverted text. A similar result was noted
by Thorndike and Woodworth (1901a, 1901b, 1901c), who
trained people on simple tasks such as estimating areas of
geometric shapes or crossing out specific letters in a text, and
then transferred them to related tasks. They found that the
benefits of practice were restricted in scope, suggesting that
the benefits were partly, or even primarily, due to perceptual
learning of the training stimuli.Cognitive SkillCognitive skills range from learning to make simple associa-
tions between stimuli and responses to solving complex prob-
lems or flying fighter planes. Complex skills usually have
perceptual or motor components or depend on background
knowledge, but much can be gained by examining what is
arguably the simplest form of cognitive skill, response-
selection skill. Response-selection processes are those
processes that are important in determining which response is
to be made to which stimulus. Increased facility in response
selection is often the most important determinant of improve-
ment in task performance (Teichner & Krebs, 1974; Welford,
1968, 1976), outweighing the importance of making percep-
tual discriminations or executing motor responses.
Developmental studies have shown that children’s im-
provement as a function of age in a selective-attention task in
which one stimulus dimension has to be attended and another
ignored is largely attributable to increases in the speed with
which stimulus-response translation can occur (Ridderinkhof,
van der Molen, Band, & Bashore, 1997). Numerous studies
have shown that stimulus-response translation is the locus of
performance improvements in choice-reaction tasks among
adults. For example, Pashler and Baylis (1991) used a number
of practice and transfer conditions to determine the locus of
performance improvements in choice-reaction tasks. Partici-
pants practiced pressing keys in response to stimulus category
(e.g., pressing a key with the index finger if the stimulus was
a letter, a middle key with the middle finger for a digit, and a
left key with the ring finger for a nonalphanumeric symbol).
During practice sessions, a small set of only two stimuli from
each category was used. After substantial improvement in
performance had occurred, two additional stimuli from each
category were added. Importantly, responses were just as fast
for new stimuli as for already practiced stimuli, suggesting
that the locus of the practice effect was in assigning stimuli to
categories and selecting the right category key. Changing the
hand used to make the key presses had no effect on perfor-
mance, ruling out a motor locus for improvements. However,
consistent with a response-selection account of performance
improvements, reassigning the categories to different keys
completely eliminated the benefits of practice.
Although practice effects in choice-reaction tasks are con-
centrated in response-selection or stimulus-response transla-
tion processes, it does not seem to be the case that response
selection becomes automatized such that stimuli automati-
cally activate their corresponding responses. Ehrenstein,