TRENDS in Cognitive Sciences Vol.5 No.7 July 2001
Graded representations in behavioral dissociations Yuko Munakata Why do people sometimes seem to know things when they are tested in one
a weak representation, but fail another task
way, while seeming unaware of this information when tested in a different
designed to measure the same knowledge because it
way? Such task-dependent behaviors, or dissociations, often occur in infants
requires a stronger representation. At a conceptual
and children, and in adults following brain damage. To explain these
level, representations can be graded in terms of how
dissociations, researchers have posited separable knowledge systems that are
‘clean’ they are for signaling the appropriate
differentially tapped by various tasks, develop at different rates and can be
information, as opposed to being corrupted by noise
selectively impaired. There is an alternative account in which knowledge is
or damage. At the neural level, representations can
viewed as graded in nature. Certain tasks tap weaker representations, while
be graded in terms of the number of relevant
other tasks require stronger representations, leading to dissociations in
neurons firing, their firing rates and the coherence
behavior. The graded representations approach addresses dissociations observed in perception, attention, memory, executive functioning and
representations depends upon various factors, such
language, and has implications for the organization, development and
as the amount of environmental support for them,
impairment of our cognitive systems.
the state of development of an individual and theextent of neurological insult.
Knowledge can be highly task dependent. We can
On the one hand, the graded representations
simultaneously fail one test of knowledge while
alternative might seem obvious and not particularly
passing another test with flying colors. For example,
worthy of review. After all, researchers have long
people (with and without brain damage) can fail
been aware of the role that task difficulty might play
tests of memory for faces by being unable to
in dissociations; graded representations could be
viewed simply as the mechanism that underlies the
nonetheless relearning the correct pairings of
effects of difficulty, with stronger representations
names and faces more quickly than incorrect
needed for more difficult tasks. And, researchers
pairings1,2. Adults with prefrontal damage and
have thought carefully about how even double
children might mistakenly follow old rules for how to
dissociations (e.g. one group succeeding on task A
behave rather than new rules, while nonetheless
but not task B; a second group succeeding on task B
demonstrating verbally that they have learned the
but not task A) need not imply the existence of
new rules3,4. And, people with dyslexia might seem
separable underlying systems6–9. Thus, the
to know more or less about words depending on the
separable systems approach might seem like a
straw opponent. On the other hand, prevalent
approaches to dissociations across a range of
behaviors, or dissociations, might be crucial to
domains adopt the notion of separable systems and
answering fundamental questions about the
ignore the potential role of graded representations.
organization, development and impairment of our
Therefore, the graded representations approach
cognitive systems. This article explores these issues
warrants review. Furthermore, by providing a
in the context of dissociations observed in
mechanistic basis for behavioral dissociations, the
graded representations approach can inform
thinking about cognitive systems beyond simply
In each domain, the prevalent approach explains
caching out notions of task difficulty.
dissociations in terms of separable knowledgesystems that develop at different rates, can be
Graded representations in perception
selectively impaired, and contribute differentially to
Individuals with optic aphasia have difficulty
various tasks. A person might pass one task using
naming objects that are presented visually, despite
one functioning system, but fail another task
being able to gesture the use of visually presented
designed to measure the same knowledge because
objects and name objects presented auditorally. Most
the separate system it taps is underdeveloped or
accounts of these dissociations (reviewed in Ref. 10)
Yuko Munakata
impaired. An alternative approach explains
assume that optic aphasia cannot result from
dissociations in terms of knowledge representations
damage to a basic system that converts visual
that are graded, rather than simply being present or
absent, with certain tasks requiring stronger
representations, and then semantic representations
representations. A person might pass one task using
into representations for naming (Fig. 1). Based on
http://tics.trends.com 1364-6613/01/$ – see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S1364-6613(00)01682-X
TRENDS in Cognitive Sciences Vol.5 No.7 July 2001
propagated through the system, having a relatively
large effect on performance, whereas the auditorynaming task involves weak representations only at
the end stage of processing (semantics–naming),resulting in relatively less impairment. Thisprediction is supported by several studies (reviewedin Ref. 10). The model also suggested a method forimproving the performance of individuals by
preventing them from responding to a visual objectimmediately. Such a delay improved theperformance of the model dramatically, becauseweak representations from the visual–semanticspathway were strengthened before the
semantics–naming pathway operated on them. This prediction has also been supported11. Fig. 1. Theories of dissociations observed in optic aphasia. According to prevalent accounts (a), the Graded representations in attention
failure of individuals to name visually presented objects could not arise simply from damage to a
Individuals with damage to parietal cortex tend to
basic system (indicated by straight arrows). Damage to the visual–semantics pathway is inconsistent
neglect the contralateral side of space, instead
with the ability of individuals to gesture the use of visually presented objects (which presumably
occurs via the visual–semantic–gesture pathway); damage to the semantics–naming pathway is
focusing attention on the ipsilateral side12. However,
inconsistent with the ability of individuals to name objects presented auditorally (which presumably
the evaluation of attention again depends crucially on
occurs via the auditory–semantics–naming pathway). Thus, additional systems must be posited
how it is measured. Although individuals might be
(some of which are shown by curved arrows). According to a graded representations account (b), the
poor at identifying contralateral stimuli, they are able
observed dissociations could arise owing to partial damage to the visual–semantics and
semantics–naming pathways in the basic system. The resulting weak representations most affect
to make accurate same–different judgments about
tasks that tap both damaged pathways, such as naming visually presented objects (which presumably
contralateral and ipsilateral stimuli13,14. The
occurs via the visual–semantics–naming pathway). Adapted, with permission, from Ref. 10.
prevalent account of this dissociation was that theperception by the individual of the stimuli was
preserved, but a separable system for conscious
incorporate additional systems (in some cases
awareness of perception was selectively impaired.
That is, individuals could attend to and perceive
visual–semantic–naming system to explain the
contralateral stimuli, but they could not achieve the
dissociations observed in optic aphasia.
conscious awareness of the stimuli required for
representations account, these dissociations result
In contrast, according to a graded representations
from partial damage to both pathways in the basic
account, the dissociation following parietal damage
visual–semantics–naming system10. Specifically,
might result because less visual information
damage to the visual–semantics pathway could
(a weaker representation) is required for a
result in weak semantic representations for visually
same–different judgment than for identifying a
stimulus14. This graded representations account
representations might not be strong enough to drive
predicted that: (1) neurologically intact participants
a damaged semantics–naming pathway, leading to
should show the same dissociation with degraded
optic aphasics’ difficulty in naming visually
stimuli (e.g. when the stimuli on one side of space are
presented objects. However, these weak semantic
viewed through a diffusing mask); and (2) the
representations might be sufficient to drive an
dissociation should disappear in individuals with
intact semantics–gesture pathway, which allows
parietal damage and neurologically intact
optic aphasics to gesture the use of visually
participants when the amount of visual information
presented objects. Similarly, strong semantic
required for the two tasks is equated (e.g. with a
representations (e.g. via an intact auditory–
forced choice procedure for the identification task).
semantics pathway) might be sufficient to drive the
Both of these predictions were confirmed14.
damaged semantics–naming pathway, allowing
optic aphasics to name auditorally presented
demonstrated how graded activity patterns can
govern attention, and how damage to such systems
instantiates this graded representations account
can lead to deficits observed in neglect15,16.
simulated the dissociations in optic aphasia10. Furthermore, the model predicted that individuals
Graded representations in memory
with optic aphasia should have more difficulty
Some of the most compelling dissociations in
gesturing the use of visually presented objects than
memory occur early in development. For example,
naming objects presented auditorally. The gesturing
infants demonstrate an apparent memory for hidden
task leads to weak representations in the first stage
objects within the first few months of life in
of processing (visual–semantics) that are then
violation-of-expectation studies17,18, while
TRENDS in Cognitive Sciences Vol.5 No.7 July 2001
Box 1. Dissociations in memory development
Violation-of-expectation studies build on the assumption that
In one of the most well-known studies using this methodf,
infants look longer at events that they find novel or unnaturala,b.
infants viewed a drawbridge-like stimulus rotating back and
When the novelty or unnaturalness of such events arises based
forth, and then saw possible and impossible events involving
on objects that have been occluded, the longer looking times of
the drawbridge and a block that became occluded by the
infants at these events are taken as evidence of their memory for
drawbridge (Fig. I). Infants as young as 3.5 months looked
the hidden objects. Although these methods are not without
longer at the impossible event, indicating some apparent
controversy (e.g. see special issues covering debates on those
memory for the occluded block. The dissociation between the
methods in Refs c–e), they have provided important constraints
sensitivity of infants in such violation-of-expectation studies
and their failure to search for hidden objects is one of the most
Such dissociations are observed even within search
tasks. For example, infants appear to reach for toys that are
hidden by completely darkening a roomg,h months earlier than
they will reach for toys that are hidden by visible occluders in
the light. From a graded representations approach, a weak
memory of an occluded object might be strong enough to
guide a reach in the dark, when there is no direct visual
information conflicting with the memory of the object. This
same memory, however, might not survive the interference
from the visual stimulus of an occluder where the object used
to bei. This graded representations account predicts that
infants might search more for an object that has been hidden
behind a visible occluder if the room lights are then turned off
(reducing the interference from the occluder on the weak
representation of the hidden object) compared with if the room
lights are left on. Preliminary results from our laboratory
References
a Fantz, R.L. (1964) Visual experience in infants: decreased attention to
familiar patterns relative to novel ones. Science 146, 668–670
b Spelke, E.S. (1985) Preferential looking methods as tools for the study
of cognition in infancy. In Measurement of Audition and Vision in the First Year of Postnatal Life (Gottlieb, G. and Krasnegor, N., eds),pp. 323–363, Ablex
c Special Issue (2000) Infancy Vol. 1 (4)
d Special Issue (1999) Dev. Sci. Vol. 2 (2)
e Special Issue (1998) Infant Behav. Dev. 21 (2)
f Baillargeon, R. (1987) Object permanence in 3.5- and 4.5-month-old infants.
g Goubet, N. and Clifton, R.K. (1998) Object and event representation in
6-and-a-half-month-old infants. Dev. Psychol. 34, 63–76
Fig. I. An experiment using the violation-of-expectation method to test infants’
h Hood, B. and Willatts, P. (1986) Reaching in the dark to an object’s
memory for hidden objectsc. After infants viewed a drawbridge-like stimulus
remembered position: evidence for object permanence in 5-month-old
rotating back and forth (a), a block was placed in the path of the drawbridge. In the
‘possible’ event (b), the drawbridge rotated to the point where it would touch the
infants. Br. J. Dev. Psychol. 4, 57–65
block and then rotated back to its starting point. In the ‘impossible’ event (c), the
i Munakata, Y. et al. (1997) Rethinking infant knowledge: toward an adaptive
drawbridge appeared to rotate through the space occupied by the block before
process account of successes and failures in object permanence tasks.
rotating back to its starting point.
nonetheless failing for several more months to
infants successfully retrieve visible objects while
search for desired objects that are presented and
showing little memory for hidden objects,
then hidden (see Box 1). A prevalent account of this
demonstrating that their failures with hidden
dissociation is that the memory of young infants
objects cannot be due solely to problem-solving
for hidden objects is intact, but separable systems
for acting on this memory are underdeveloped
According to a graded representations account,
(e.g. a problem-solving system for lifting an
the abilities of infants to represent hidden objects
occluding cover to retrieve a toy underneath)19–21.
become stronger with development; dissociations in
Although problem-solving demands are likely to be
memory for hidden objects result because weak
part of the problem22, they do not fully explain the
representations of hidden objects suffice for some
dissociation. When problem-solving demands are
tasks but not others23. For example, a weak
equated for retrieving visible and hidden toys,
representation of a hidden object might allow infants
TRENDS in Cognitive Sciences Vol.5 No.7 July 2001
to recognize a strange event with that object in the
resolving a conflict (e.g. between the new rule of
violation-of-expectation paradigm. Such weak
color and the previous rule of shape). In contrast,
representations might not suffice for reaching for
answering a standard verbal question about the new
hidden objects, owing to the greater complexity and
rule (e.g. ‘Where do red things go in the color game?’)
effort of reaching. However, search failures would
does not involve any conflict. This graded
not simply reflect deficits in the action system for
representations account predicted that the
reaching; strong representations of objects (e.g. for
dissociation between sorting and verbal measures
visible objects) might be sufficient to drive the
would disappear when the amount of conflict in the
reaching system. This graded representations
two tasks was equated (e.g. by adding conflict to the
account of dissociations in memory development
verbal query by asking, ‘Where do red trucks go in
parallels the account of dissociations in optic
the color game?’). This prediction has been
aphasia. In both cases, particularly poor
confirmed in children25. Similarly, a weak prefrontal
performance arises when tasks tap the interaction
representation of a hidden object might allow
between two pathways with weak representations
infants to gaze correctly at the location where the
(the naming of visually presented objects or reaching
object is hidden, whereas a stronger representation
for hidden objects), with relatively spared
is needed to overcome the prepotent response to
performance when tasks tap only one pathway with
reach back to a previous hiding location. This graded
weak representations (gesturing the use of a visually
representations account was instantiated in a
presented object, naming an auditorally presented
neural network model that (like infants) gazed more
object, recognizing an impossible event with an
frequently than it reached, allowing the gazing
occluded object and reaching for a visible object). A
system to make better use of weak representations
neural network model has demonstrated how graded
representations can support the dissociationsobserved in memory development23. The model also
Graded representations in language
predicted that infants would search more for
People with phonological dyslexia have difficulty
familiar toys than novel toys after they were hidden
reading non-words (e.g. ‘nust’), despite being able to
read regular words (e.g. ‘mint’) and exception words
representations), whereas infants typically showed
(e.g. ‘pint’), and to repeat non-words after hearing
novelty preferences with visible objects. Preliminary
them. Again, this dissociation leads to the
work supports this prediction (Y. Munakata,
assumption that phonological dyslexia cannot result
from damage to a general phonological system;otherwise, people with phonological dyslexia would
Graded representations in executive functioning
not be able to read words or repeat non-words after
Adults with damage to prefrontal cortex, infants and
hearing them. One account of these dissociations
children show dissociations in executive functioning,
appearing to know things but failing to act
information from print (orthography) to sound
appropriately on the basis of this knowledge. For
(phonology): a word-specific system that is preserved
example, adults with prefrontal damage and
in phonological dyslexia and a grapheme-to-
children can verbally report new rules they have
phoneme conversion system that is impaired in
learned for sorting cards (e.g. to sort according to the
color of the objects on the cards), but incorrectly sort
the cards according to previously learned rules
representations account, the dissociations observed
(e.g. based upon the shape of the objects on the
in phonological dyslexia result from the weakening of
cards)3,4. Similarly, infants can look to a new location
general phonological representations, and the effects
where they have watched a toy being hidden, but
of these weak representations on an interactive
incorrectly reach back to an old location where the
language system including orthography and
toy was previously hidden26–28. Such compelling
semantics5,34,35. Specifically, weak phonological
dissociations have suggested that action systems
representations might suffice for people to respond
might be selectively impaired or underdeveloped
appropriately given other (non-phonological)
while separable knowledge systems are fully
support, as is available for reading words (semantic
support) and for repeating non-words after hearing
them (auditory support). These weak phonological
representations account, prefrontal representations
representations might not suffice without such other
might be weak because of neurological insult or
support, as in the case of reading non-words (no
state of development; dissociations in executive
semantic or auditory support). This graded
functioning result because weak prefrontal
representations account predicted that people with
representations suffice for some tasks but not
phonological dyslexia would show decreasing
others25,30–32. For example, stronger representations
abilities to repeat words and non-words as the
might be required to resolve conflict. Sorting a card
semantic and auditory support for them decreased.
(e.g. a red truck) according to a new rule requires
This prediction has been confirmed5. Furthermore,
TRENDS in Cognitive Sciences Vol.5 No.7 July 2001
Box 2. Explaining dissociations in terms of separable systems
Dissociations in recognition memory (between
hippocampus make it well suited for this task
measures of recollection and familiarity) might be
(e.g. with relatively sparse representations that
best understood in terms of the operation of two
minimize interference). In contrast, representing the
separate systems (the hippocampus and cortex).
underlying structure of the environment requires
Consider the case of meeting a new person, and
slow, interleaved learning using distributed,
overlapping representations. For example,
recognizing her. Recollection refers to the retrieval
abstracting a general schema for meeting people
of specific information about her (such as where
(e.g. what kinds of conversation are appropriate, how
you were when you met her, what her name is, etc.)
closely people stand) requires slow learning over
in this recognition process. Familiarity refers to the
multiple episodes, collapsing across the differences
assessment that she is known (‘I know I’ve seen her
of meeting individual people. The cortex appears to
before!’), without the recollection of specific
be well suited for this task, and can also support
familiarity based recognition. Thus, a fundamental
Recollection, but not familiarity, is associated with
computational trade-off in memory suggests the
hippocampal activitya. When people recognize words
need for two specialized systems that the
that they have studied, hippocampal activity
hippocampus and cortex appear to satisfy.
increases if they report recollecting the words, but not
Finally, recollection and familiarity are associated
if they report that they know that they have seen the
with distinct event-related potential components,
words without recollecting them (i.e. based on
with different timing and spatial topographyd. A
familiarity). In fact, hippocampal activity is similarly
relatively early frontal FN400 component varies with
low for such recognized, non-recollected words as for
the familiarity of words, whereas a relatively late
new (also non-recollected) words. Furthermore,
parietal component varies with the recollection of
hippocampal damage impairs recollection but not
familiarity, even when difficulty across tasks is
References
a Eldridge, L.L. et al. (2000) Remembering episodes: a
selective role for the hippocampus during retrieval. Nat.
supported the notion of distinct roles for the
hippocampus and cortex in memoryc. Recollecting
b Holdstock, J.S. et al. Memory dissociations following human
particular episodes requires rapid learning with
hippocampal damage. Hippocampus (in press)
non-overlapping representations. For example,
c McClelland, J.L. et al. (1995) Why there are complementary
recollecting meeting that particular person requires
learning systems in the hippocampus and neocortex: insightsfrom the successes and failures of connectionist models of
learning from a single episode and keeping that
learning and memory. Psychol. Rev. 102, 419–457
memory distinct from memories of meeting other
d Curran, T. (2000) Brain potentials of recollection and
people. The anatomy and physiology of the
familiarity. Mem. Cognit. 28, 923–938
neural network models have demonstrated how an
instantiating graded representations have simulated
interactive system of phonology, orthography and
a range of dissociations and led to numerous
semantics can support typical language processing
predictions that have been confirmed empirically.
(without specialized systems for regular words,
This approach has shown promise to advance
non-words and exception words), and how damage to
theorizing about how developmental transitions
the system can lead to the dissociations observed in
occur, and to inform intervention and rehabilitation.
different types of dyslexia16,34,36–38.
These notions surrounding the nature of our
cognitive systems might seem intuitive, or even
Conclusions
obvious; they have been part of the cognitive
Behavioral dissociations cut across the domains of
literature for some time41 and have been emphasized
perception, attention, memory, executive functioning
in both neural network and dynamic systems
and language. These dissociations might be most
approaches42. However, numerous theories ignore the
salient during development and following brain
potential role of graded representations, instead
damage, but are also observed in typical, adult
treating knowledge as all or nothing and attributing
populations (see, for example, Refs 2,14,39,40).
dissociations to deficits in separable systems. It can
The graded representations approach provides a
be difficult within this kind of framework to think
unified framework for understanding behavioral
about how developmental transitions occur43 or how
dissociations. Behaviors are viewed in terms of an
to intervene or rehabilitate. How could children go
interactive system that is sensitive to the strength of
from nothing to all, and how can individuals improve
representations, which can vary based on factors
if critical separable pathways are simply missing?
such as environmental support, state of development
Furthermore, theories within this framework can
become unwieldy in the number of separate systems
TRENDS in Cognitive Sciences Vol.5 No.7 July 2001
Box 3. Explaining associations as well as dissociations
Theories that posit multiple separable pathways to
exception words show some systematicity in the
explain dissociations in behavior might have more
othography-to-phonology mapping (e.g. in ‘pint,’ the
difficulty accounting for related associations than
pronunciation of the ‘p’, ‘n’ and ‘t’, and even the
graded representations approaches. Consider the
pronunciation of the ‘i’ as in ‘pie’ and ‘mind’), despite
case of reading, specifically reading regular words,
not following the most typical rules of pronunciation
(as in ‘mint’). Such ‘neighborhoods’ of similar and
These tasks might seem like natural candidates
discrepant pronunciations influence the reading of
for separable specialized systems. Regular words
regular words, exception words and non-words in
and non-words can be read according to rules about
both typical and brain-damaged populationsc–f.
orthography-to-phonology mappings: ‘mint’ is
A single system for reading regular and exception
pronounced according to the same rules that govern
words and non-words can naturally handle the
the pronunciation of ‘hint’ and ‘lint,’ and ‘nust’
similarities across these tasks, and (as described in
follows the same rules that govern ‘must’ and ‘rust.’
the main text) can account for the dissociations taken
In contrast, exceptions such as ‘pint’ violate such
to suggest the existence of separable systems. In
rules. And, regular and exception words might be
contrast, theories that posit distinct systems to
‘looked up’, based on prior experience, whereas
explain the dissociations require special adjustments
non-words cannot. Further, behavioral patterns
to account for the associations observed across
during development and following brain damage
suggest separate systems specialized for different
reading tasks. People with phonological dyslexia
References
a Coltheart, M. et al. (1993) Models of reading aloud: dual-route
can read words but have difficulty reading non-
and parallel-distributed-processing approaches. Psychol. Rev.
words, whereas people with surface dyslexia can
read non-words but have difficulty reading
b Plaut, D.C. et al. (1996) Understanding normal and impaired
exception words. Thus, some theories of word
word reading: computational principles in quasi-regular
reading posit two separate systems: a rule-based
domains. Psychol. Rev. 103, 56–115
c Glushko, R.J. (1979) The organization and activation of
system for mapping orthography to phonology (for
orthographic knowledge in reading aloud. J. Exp. Psychol.
reading regular words and non-words, impaired in
phonological dyslexia) and a look-up table for
d Jared, D. et al. (1990) The basis of consistency effects in word
lexical items (for exception and regular words,
naming. J. Mem. Lang. 29, 687–715
Acknowledgements
e Shallice, T. et al. (1983) Reading without semantics. Q. J. Exp.
f Taraban, R. and McClelland, J.L. (1987) Conspiracy effects in
similarities, across the tasks of reading regular
word pronunciation. J. Mem. Lang. 26, 608–631
words, exception words, and non-wordsb. Many
g Pinker, S. (1991) Rules of language. Science 253, 530–535
that must be posited to explain the array of
Finally, two important caveats to behavioral
dissociations observed during development, after
dissociations should be noted. First, although
damage and in the intact mature system.
graded representations provide a powerful tool forunderstanding dissociations across a range ofdomains, in some cases separable systems might
Questions for future research
provide the best explanation (see Box 2). This
• Are different causes and consequences associated with different
article emphasizes the importance of considering
instantiations of gradedness (e.g. firing rates of neurons, coherence of
the potential role of graded representations,
because in numerous other cases this approach
• In other dissociations, such as speech versus gesture44, why might one
appears to provide the best explanation, despite
behavior require stronger representations than another behavior?
being ignored by prevalent accounts. Second,
• If a common mechanism (the strengthening of representations)
although dissociations are compelling and reliable,
supports the ability to carry out new behaviors rather than old behaviors
behaviors do often group together. That is,
across a range of tasks (e.g. reaching to a new hiding location for a toy
performance on one measure of knowledge often
rather than an old one, sorting cards according to a new rule rather than
correlates with performance on a different
an old one), what accounts for the discrepancy in the age when children
measure designed to tap the same knowledge.
succeed at these different tasks (one year of age for the hiding task26,
These associations might fall out more naturally
four years of age for the cardsort task4)?
from a graded representations approach with
• Could weak representations be strengthened by using them for tasks that
interactive systems than from a separable systems
approach designed to address dissociations (see
• What are further implications for drawing parallels across developing,
Box 3). Theoretical frameworks must accommodate
brain-damaged, and mature neurologically intact populations?
both the associations and the dissociations observedin behavior.
TRENDS in Cognitive Sciences Vol.5 No.7 July 2001
References
16 O’Reilly, R.C. and Munakata, Y. (2000)
30 Cohen, J.D et al. (1990) On the control of
1 Farah, M.J. et al. (1993) Dissociated overt and
Computational Explorations in Cogntive
automatic processes: a parallel distributed
covert recognition as an emergent property of a
Neuroscience: Understanding the Mind by
processing model of the stroop effect. Psychol. Rev.
lesioned neural network. Psychol. Rev.
17 Baillargeon, R. (1993) The object concept
31 Cohen, J.D. and Servan-Schreiber, D. (1992)
2 Wallace, M.A. and Farah, M.J. (1992) Savings in
revisited: new directions in the investigation of
Context, cortex, and dopamine: a connectionist
relearning face-name associations as evidence for
infants’ physical knowledge. In Visual Perception
‘covert recognition’ in prosopagnosia. J. Cogn.and Cognition in Infancy. Carnegie Mellon
schizophrenia. Psychol. Rev. 99, 45–77
Symposia on Cognition (Granrud, C., ed.),
32 Munakata, Y. (1998) Infant preservation and
3 Milner, B. (1963) Effects of different brain lesions
implications for object permanence theories: a PDP
on card sorting. Arch. Neurol. 9, 90–100
18 Spelke, E. et al. (1992) Origins of knowledge.
model of the AB task. Dev. Sci. 1, 161–184
4 Zelazo, P.D. et al. (1996) An age-related
33 Coltheart, M. et al. (1993) Models of reading
dissociation between knowing rules and using
19 Baillargeon, R. et al. (1990) Why do young
aloud: dual-route and parallel-distributed-
infants fail to search for hidden objects?
processing approaches. Psychol. Rev.
5 Farah, M.J. et al. (1996) Phonological dyslexia:
loss of a reading-specific component of the
20 Diamond, A. (1991) Neuropsychological insights
cognitive architecture? Cognit. Neuropsychol.
into the meaning of object concept development.
Phonology, reading acquisition, and dyslexia:
In The Epigenesis of Mind (Carey, S. and Gelman,
insights from connectionist models. Psychol. Rev.
6 Farah, M.J. (1994) Neuropsychological inference
with an interactive brain: a critique of the
21 Willatts, P. (1990) Development of problem-
35 Patterson, K. et al. (1996) Interpreting a case of
‘locality’ assumption. Behav. Brain Sci.
solving strategies in infancy. In Children’s
Japanese phonological alexia: the key is in
Strategies: Contemporary Views of Cognitive
phonology. Cognit. Neuropsychol. 13, 803–822
7 Nosofsky, R.M. and Zaki, S.R. (1998) Dissociations
Development (Bjorklund, D.F., ed.), pp. 23–66,
36 Harm, M.W. and Seidenberg, M.S. Are there
between categorization and recognition in amnesic
and normal individuals: an exemplar-based
22 Diamond, A. et al. (1999) Early developments in
dyslexia? Cognit. Neuropsychol. (in press)
interpretation. Psychol. Sci. 9, 247
the ability to understand the relation between
37 Plaut, D.C. et al. (1996) Understanding normal
8 Plaut, D.C. (1995) Double dissociation without
stimulus and reward. Dev. Psychol.
principles in quasi-regular domains. Psychol. Rev.
neuropsychology. J. Clin. Exp. Neuropsychol.
23 Munakata, Y. et al. (1997) Rethinking infant
knowledge: toward an adaptive process account
38 Plaut, D.C. and Shallice, T. (1993) Deep dyslexia:
9 van Orden, C.G. et al. (2001) What do double
of successes and failures in object permanence
a case study of connectionist neuropsychology.
dissociations prove? Cognit. Sci. 25, 111–172
tasks. Psychol. Rev. 104, 686–713
Cognit. Neuropsychol. 10, 377–500
10 Sitton, M. et al. (2000) Superadditive effects of
24 Shinskey, J.L. and Munakata, Y. Detecting
39 Goodale, M.A. and Hy, Y. (2000) Grasping after a
multiple lesions in a connectionist architecture:
transparent barriers: clear evidence against the
delay shifts size-scaling from absolute to relative
implications for the neuropsychology of optic
means-end deficit account of search failures.
metrics. J. Cogn. Neurosci. 12, 856–868
aphasia. Psychol. Rev. 107, 709–734
40 Yonelinas, A.P. (1997) Recognition memory ROCs
11 Schnider, A. et al. (1994) Visual agnosia and optic
25 Munakata, Y. and Yerys, B.E. All together now:
for item and associative information: the
aphasia: are they anatomically distinct? Cortex
when dissociations between knowledge and action
contribution of recollection and familiarity. Mem.
disappear. Psychol. Sci. (in press)
12 Robertson, I.H. and Marshall, J.C. (eds) (1993)
26 Piaget, J. (1954) The Construction of Reality in the
41 McClelland, J.L. and Rumelhart, D.E. (1981) An
Unilateral Neglect: Clinical and Experimental
interactive activation model of context effects in
27 Diamond, A. (1985) Development of the ability
letter perception: Part 1. An account of basic
13 Volpe, B.T. (1979) Information processing of
to use recall to guide action, as indicated by
findings. Psychol. Rev. 88, 375–407
visual stimuli in an ‘extinguished’ field. Nature
infants’ performance on AB. Child Dev.
42 Thelen, E. and Smith, L.B. (1994) A DynamicSystems Approach to the Development of
14 Farah, M.J. (1991) Unconscious perception of
28 Hofstadter, M.C. and Reznick, J.S. (1996)
‘extinguished’ visual stimuli: reassessing the
43 Siegler, R.S. (2000) The rebirth of children’s
evidence. Neuropsychologia 29, 949–958
delayed-response performance. Child Dev.
15 Cohen, J.D. et al. (1994) Mechanisms of spatial
44 Alibali, M.W. and Goldin-Meadow, S. (1993)
attention: the relation of macrostructure to
29 Goodale, M.A. and Milner, A.D. (1992) Separate
Gesture-speech mismatch and mechanisms of
microstructure in parietal neglect. J. Cogn.
visual pathways for perception and action. Trends
learning: what the hands reveal about a child’s
state of mind. Cognit. Psychol. 25, 468–523
Articles of interest in other Elsevier journals
Short-term synaptic plasticity as a temporal filter, by E.S. Fortune and G.J. Rose Trends in
Multiple brain-memory systems: the whole does not equal the sum of its parts, by J.J. Kim and
M.G. Baxter Trends in Neurosciences 24, 324–330
Depression: a major, unrecognized risk factor for osteoporosis?, by G. Cizza, P. Ravn, G.P. Chrousos
and P.W. Gold Trends in Endocrinology & Metabolism 12, 198–203
St John’s Wort: Prozac from the plant kingdom, by G. Di Carlo, F. Borrelli, E. Ernst and A.A. Izzo
Trends in Pharmacological Sciences 11, 292–297
EMAIL YOUR QUESTIONS TO Taking care of your mouth and teeth is a very important, yet often overlooked, part of maintaining your general health. Good oral health can help you prevent or catch infections early. It can also give you clues as to the state of your overall health and the health of your immune system. BRUSHING motion, keeping bristles at a 45° angle to the gum line. Pay speci
Dynamic response of breast tumor oxygenation to hyperoxic respiratory challenge monitored with three oxygen-sensitive parameters Yueqing Gu, Vincent A. Bourke, Jae G. Kim, Anca Constantinescu, Ralph P. Mason,and Hanli LiuThe simultaneous measurement of three oxygen-sensitive parameters ͓arterial hemoglobin oxygen sat-uration ͑SaO ͒, tumor vascular-oxygenated hemoglobin concentration ͓͑HbO