INTRODUCTION
Apraxia is commonly defined as a disorder of
learned skilled movement (Rothi and Heilman,
1997; Buxbaum, 2001), but a peculiarity of
research into ideomotor apraxia (IMA) is that there
has been relatively little detailed investigation of
dexterity. The reasons for this are largely historical.
In Liepmann’s pioneering studies at the start of the
twentieth century he used pantomime of object use
and imitation of gesture to elicit deficits and
explore the relationship between concept and action
(Goldenberg, 2003). It emerged that these methods
were sensitive to a selective impairment after left
hemisphere damage, and this impairment in
imitation or pantomime is now accepted as the
defining feature of IMA. Over the decades it has
been repeatedly noted, often as an aside, that
ideomotor apraxics do not seem to be significantly
impaired when handling real objects or in
naturalistic action. For example, De Renzi et al.
(1980) commented that “in the great majority of
cases [apraxia] only appears in the testing situation
and does not trouble the patient in everyday life”
(p. 9). This anecdotal evidence of normality in
everyday settings has kept the focus of research on
imitation and pantomime, where deficits are
strongly displayed and any impact on dexterity or
naturalistic action has not been investigated in
depth. This is unsatisfactory for both theoretical
and clinical reasons. In building a theory of IMA,
there is a need to fully explore the manifestations
of apraxia in all contexts, including naturalistic
action and manipulation of real objects. From a
rehabilitation perspective, the final concern is
seldom about deficits in pantomime or imitation
but on whether there are implications for dexterity
and everyday actions. This paper will review the
limited evidence on the impact of IMA on
functional ability, consider theoretical and clinical
implications, and outline some priorities for future
work.
A continuing source of confusion in this field
has been the absence of clear and consistent
definitions. The general definition of apraxia as a
disorder of skilled action not due to motor or
sensory loss, is too broad to be useful since
abnormalities in actions can arise for a multitude of
reasons after brain damage. For example, the
demented patient who is unable to use a knife and
fork may be suffering from any of a range of
attentional, perceptual, semantic or executive
deficits. In this paper the focus will be on IMA,
and we define this as an inability to pantomime
object-use or to imitate gesture which is normally
seen after focal left parietal or left frontal damage
(Haaland et al., 2000). It must be recognised that
even this simple operational definition is not free
from difficulties as there are important differences
between patients. For example, ability to imitate
gesture can dissociate from inability to pantomime
tool-use (Dumont et al., 1999). However, the
working hypothesis which underpins most of the
published work on IMA is that there is damage to
a system for the control of skilled action and that
differences in individual pattern of deficit reflect
Cortex, (2007) 43, 359-367
SPECIAL ISSUE: ORIGINAL ARTICLE
IDEOMOTOR APRAXIA AND FUNCTIONAL ABILITY
Alan Sunderland and Caroline Shinner
(School of Psychology, University of Nottingham, Nottingham, UK)
ABSTRACT
The impact of ideomotor apraxia (IMA) on functional ability has been a relatively neglected topic in research. This has
been due to the continued focus on performance on gesture imitation and pantomime of tool-use, together with widespread
acceptance of anecdotal evidence that IMA has no effect when directly manipulating objects. An increasing number of
studies have shown that IMA does in fact result in increased clumsiness when handling objects and may contribute to
disability in everyday life. However the effect seems relatively mild compared to the stark abnormalities on gesture
imitation and pantomime. The conventional explanation for this is that the cues provided by naturalistic contexts improve
retrieval of action representations, but an alternative account concerns task-specific cognitive demands. Performance on
simple prehensile tasks can be successfully guided by physical affordances whereas motor tasks may be failed if they
require the support of memory or problem solving ability. A central deficit in IMA may be impaired postural representation
causing inability to solve the problem of how to manipulate objects where neither affordance nor memory can dictate
action. However, this account still fails to explain fully the patterns of error seen on complex naturalistic tasks such as
dressing. Future research needs to further our understanding of how IMA maps on to disability, which will have implications
for theory building and for therapeutic intervention.
Key words: apraxia, motor skill, affordance, rehabilitation
variations in which parts of this system have been
impaired. This leads to the general prediction that a
diagnosis of IMA should have implications for the
wider control of action as seen in the functional
tasks of everyday life. This aim of this paper is to
review the evidence on these functional
consequences.
Another problem of definition has been whether
and how a distinction is made between IMA and
ideational apraxia (IA). This distinction has its
origins in Liepmann’s insight (Goldenberg, 2003)
that difficulties in performing actions might be due
either to the loss of the concept of an action (IA)
or an inability to execute the related motor
programme (IMA). There is general agreement (De
Renzi, 1985; Hanna-Pladdy et al., 2003) that such
a distinction should be made on the basis that the
ideational apraxic will show loss of action concepts
by making inappropriate movements (content
errors), whereas the ideomotor apraxic will show
movements that are appropriate but spatially or
temporally inaccurate (accuracy errors). Clear
evidence for the existence of these two forms of
apraxia comes from the study by De Renzi and
Lucchelli (1988). These authors investigated a
group of left hemisphere lesioned patients who
were diagnosed as showing IA because they
exhibited a high rate of content errors on multiplestage
tasks, such as lighting a candle with a match.
A dissociation from IMA was indicated by the
absence of a correlation between rate of these
content errors and accuracy on a gesture imitation
task, and the presence of individuals who were
selectively impaired on either one of these tests.
However, in practice, making this distinction is
difficult because there is often no clear boundary
between grossly inaccurate performance and a
content error. For example, if when asked to
pantomime slicing bread a patient makes a
chopping rather than a slicing motion (Poizner et
al., 1995), this could be due to inaccurate control
of co-ordination of arm joints or loss of knowledge
of how bread is sliced. In reality, it seems that both
types of deficit may be present in many cases – De
Renzi and Lucchelli (1988) found co-occurrence of
IMA in 75% of their cases of IA. For the purposes
of this paper, we will side-step this issue of the
IMA/IA distinction by focussing initially on
whether failure of any type on tests of imitation or
pantomime has predictive validity for function.
This will then lead to consideration of the nature of
errors.
In the complex confusion of the research
literature on apraxia it can be easy to lose sight of
the fact that IMA is a common consequence of left
hemisphere damage. De Renzi et al. (1980) tested a
large sample of stroke patients with neurological
signs of left hemisphere damage (100 cases) or
right hemisphere damage (80 cases). Abnormality
in imitation of gestures was detected in 20% of the
right hemisphere cases and 50% of the left
hemisphere cases. In a more selective sample of
consecutive stroke ward admissions with brain scan
evidence of parietal or posterior frontal damage,
Sunderland et al. (1999) found impaired gesture
imitation in 66% of left hemisphere cases while a
marginal impairment was seen in 13% of right
hemisphere cases. This solid evidence of a high
incidence of IMA after left hemisphere damage
gives impetus to theoretical and clinical research –
we are not dealing with some rare or non-specific
disorder but a common selective deficit whose
nature and clinical implications need to be fully
understood.
The first section of this paper reviews the
evidence of the impact of IMA on functional
ability. The aim here is to reveal what we know
about the ecology of IMA as a phenomenon
affecting functional ability. The second section of
the paper (Explaining Preserved Ability) places this
ecological evidence alongside evidence from
experimental studies to consider how well we can
explain the observed patterns of behaviour.
THE FUNCTIONAL IMPACT OF IMA
Studies of Dependence and Complaint
Studies to test the assertion that IMA “does not
trouble the patient in everyday life” can be divided
into those that have observed performance on
functional motor tasks and those that have looked
for correlations between functional dependence and
apraxia scores. An example of the latter is the
study of left hemisphere stroke patients 6 months
after discharge from hospital by Sundet et al.
(1988). They reported that gesture imitation scores
were a significant predictor of questionnaire ratings
of dependency on care-givers for help in activities
of daily living (ADL), and concluded that IMA had
a significant impact on functional outcome. The
weakness of such correlational studies is that IMA
is never seen as an isolated deficit. Multiple
regression analysis was used by Sundet et al.
(1988) to provide statistical control for some
confounding variables such as the impact of
hemiplegia on dependency scores, but there may be
other important confounding variables.
Consequently, this result cannot be taken as more
than circumstantial evidence that IMA may
increase dependence. In this context, it is
interesting to note some negative findings. Both
Walker and Lincoln (1991) and Hanna-Pladdy et al.
(2003) found that IMA scores were not a predictor
of dressing ability. The manipulation of clothing
when dressing requires the execution of complex
learned action sequences, and preservation of this
skill would support the assertion that IMA does not
translate into everyday life. However, as discussed
below, there is evidence from direct observation
that IMA can impact on dressing but interacts with
360 Alan Sunderland and Caroline Shinner
other factors to obscure any direct correlation. The
conclusion on correlational studies is therefore that
they are inadequate to assess the functional impact
of IMA – the presence of correlations could be due
to confounding variables, whilst their absence
could be due to interactions between the multitude
of variables which can affect dependence.
There is a consensus that few patients with
IMA complain about any impact on everyday life
(De Renzi et al., 1980; Foundas et al., 1995), but
the reasons for this may again be complex. In a
study of patients with cortical damage six months
after stroke (Sunderland, 2000), questionnaire
ratings of dexterity problems such as difficulty in
using eating implements were very high for some
apraxic patients, but others reported fewer
problems than those without apraxia. However,
there was evidence that people found it difficult to
find a meaningful point of comparison in deciding
whether things should be labelled as problematic.
Non-apraxic patients with hemiplegia or their
carers were likely to say there was “no problem”
with activities such as tying shoelaces which
require bimanual control for normal performance.
Presumably, they meant that the task could be
successfully completed in one-handed fashion,
which they saw as a significant achievement rather
than a problem! Interestingly, the most frequent
dexterity problems were reported by a patient with
severe IMA but no hemiparesis, suggesting that in
other cases the major deficits in motor skill due to
right-sided hemiparesis may have overshadowed
any impact of IMA on function with the
ipsilesional hand. Furthermore, any clumsiness that
is noted by the patient or carers is often ascribed to
use of the non-dominant hand for most everyday
tasks.
In summary, studies of dependency and
complaint make the obvious point that for the
typical stroke patient there are many deficiencies in
functional skill which “trouble them in everyday
life”; the difficulty is in finding solid evidence that
IMA contributes to these troubles. To do this we
probably have to turn to direct observation of
motor performance rather than relying on
dependency ratings or patients’ complaints.
Use of Tools or Objects under Test Conditions
It is a common clinical observation that the
apraxic patient who is unable to pantomime how a
tool would be used seems to be able to
demonstrate use when handed the actual tool.
However, where this has been studied in detail, it
appears that actual tool use is not entirely normal
(De Renzi and Lucchelli, 1988; Poizner et al.,
1995; McDonald et al., 1994; Goldenberg and
Hagmann, 1998a). As with the previously discussed
studies of dependency and complaint, a major issue
is again how much this can be ascribed to IMA as
opposed to other causes. The traditional view in
apraxia research has been that testing ability to
pantomime or imitate is a way of demonstrating
disruption of some of the same neural systems that
are involved when handling actual objects (De
Renzi, 1985). From a modern perspective this is a
bold assumption. It seems that action relies on
parallel pathways and multiple dynamic spatial
representations (Buxbaum, 2001) and it is not clear
which of these systems may be recruited when no
actual object is present (Westwood et al., 2000). It
is therefore important to critically evaluate
correlations between gesture or pantomime and
actual object use to determine whether there is
evidence of a shared underlying deficit. Direct
observation of object use allows us to ask not only
whether there is a correlation but also whether the
pattern of errors suggests a common cause. The
weight of evidence is in support of this and the
most convincing studies are reviewed below.
De Renzi and Lucchelli (1988) selected 20 left
hemisphere patients in whom clinical observation
suggested a problem in handling objects
appropriately. They were assessed on 5 tests of
manipulation of multiple objects, such as opening a
bottle with a bottle opener then pouring the drink
into a glass. They were also tested on pantomime
of tool use and ability to imitate gestures. A control
group of patients with right hemisphere damage
only made minor errors on the multiple object test
whereas the majority of the left hemisphere group
made both content errors (e.g., putting the bottle
opener into the glass) and accuracy errors (i.e.,
appropriate but inaccurate movements). A gesture
imitation test was also included and a significant
proportion of the sample were impaired, suggesting
IMA. There was however no correlation between
severity of IMA and frequency of content errors.
However, a re-analysis of the data presented by De
Renzi and Lucchelli (1988) indicates that there is a
significant correlation (r = 0.31) between severity
of IMA and accuracy errors. So their data is
entirely consistent with the view that IA and IMA
occur as independently operating deficits – the
former causing content errors, and the latter
accuracy errors.
Evidence of the predictive value of gesture
imitation for accuracy in object manipulation also
emerged in the study by Sunderland et al. (1999)
which included 15 patients with recent left parietal
or posterior frontal damage. There was a significant
impairment on simulations of everyday dexterity
such as ability to spoon-up beans and deposit them
into a container, or to unlatch and open a cupboard
door. Accuracy errors were the most common, such
as clumsiness when unlocking the latch, but in no
case were these errors so severe as to prevent
eventual success in completing the functional tasks.
In a multiple regression analysis controlling for
severity of stroke, a brief test of imitation and
pantomime (Kimura and Archibald, 1974) emerged
as a strong predictor of frequency of errors on the
Ideomotor apraxia and functional ability 361
dexterity tasks, whereas other variables such as
severity of aphasia or visuospatial ability did not.
Evidence of a shared pattern of errors in
pantomime and in actual use of objects emerged
from the study by McDonald et al. (1994), who
investigated cases of dominant hemisphere damage,
most of whom were aphasic. Demonstration of use
of 12 objects was assessed by pantomime to
command, imitation and when holding the actual
object. Accuracy and content errors were assessed
by two raters, with high inter-rater agreement.
Normal controls made few errors but a majority of
patients were impaired in at least one test condition
and only 11% of patients did not make errors when
handling actual objects. The pattern of errors was
similar in all conditions, with accuracy errors being
the most frequent. A more objective measure of
error type was obtained in the study by Clark et al.
(1994) who used motion analysis to investigate
performance during pantomime and when using
actual objects. In both conditions apraxics showed
difficulty in producing movement in the correct
plane and poor co-ordination between the joints of
the arm (Poizner et al., 1995), leading these
authors to propose a shared deficit of disruption of
spatiotemporal representations of action.
In summary, there is solid evidence that
impaired imitation or pantomime is associated with
clumsiness when handling actual objects.
Furthermore the quality of errors seems similar in
pantomime and actual object manipulation. This is
entirely in line with the classical theory of IMA,
that impairment on gesture or pantomime is
indicative of a more general disorder of skilled
action, and we can put aside worries that
correlations might not indicate a common cause
under these very different task conditions.
However, two questions arise. From a theoretical
standpoint we can ask why the impairment with
actual objects seems relatively mild, while from a
rehabilitation perspective we can ask whether this
mild deficit in a test situation translates into a
disability when using objects in their normal
functional context.
Activities of Daily Living
The basic ADL such as dressing, grooming and
eating are complex tasks involving a diverse range
of perceptual, cognitive and motor skills. The
picture may be further complicated after stroke by
the presence of hemiparesis, which prevents
completion of these tasks in a normal bimanual
fashion. It is therefore difficult to say if and how
IMA impacts on performance. However, evidence
that there is an impact comes from several studies.
Function within a wholly naturalistic context was
studied by Foundas et al. (1995), who videotaped
10 aphasic patients as they ate lunch. All
succeeded in eating the meal but compared to
neurologically normal controls, patients were more
disorganised in sequencing their mealtime
behaviour and clumsy in their use of eating
implements. There was a significant correlation
between the frequency of these action errors and
errors on a gesture-to-command test (Florida
Apraxia Battery). Furthermore, there was a trend
towards a higher incidence of errors involving use
of eating implements compared to errors in nontool
actions such as moving a bowl. Foundas et al.
(1995) interpreted this as evidence of a selective
effect of apraxia rather than general clumsiness due
to hemiparesis.
Goldenberg and Hagmann (1998a) studied
aphasic patients as they carried out 3 ADL tasks –
buttering bread, donning and doffing a T-shirt, and
cleaning teeth. Only 25% of the sample were able
to complete the tasks without error, and 17% were
unable to complete any task. The total frequency of
errors (content and accuracy combined) correlated
significantly with tests of imitation or pantomime
but was only weakly related to severity of aphasia,
suggesting a specific impact of IMA. T-shirt
dressing was studied in detail by Walker et al.
(2004) in a study of recovery of independent
dressing in a representative sample of stroke
patients. IMA was assessed using the Kimura Box
(Kimura, 1977) which requires imitation of 3 hand
postures in sequence to press a button, pull a
handle and press a bar. It emerged that patients
with IMA were able to dress independently if they
had sufficient power in the right arm to use a
normal bimanual strategy. However where there
was hemiplegia requiring the deployment of a new
unimanual strategy, no apraxic patient was able
initially to put on a T-shirt.
Goldenberg and Hagmann (1998a) and Walker
et al. (2004) came to similar conclusions regarding
the impact of IMA on basic ADL tasks – when a
motor task such as dressing can be completed
using an over-learned strategy then the apraxic
patient may be successful, whereas if hemiparesis
demands discovery of a new compensatory strategy
then IMA will present a barrier. This interaction
between hemiparesis and apraxia may explain why
IMA has not emerged as a simple correlate of
dressing ability in other studies (Walker and
Lincoln, 1991; Hanna-Pladdy et al., 2003).
It is not clear if specific effects of IMA can be
observed in the case of more complex naturalistic
tasks. Goldenberg et al. (2001) studied tasks such as
changing the batteries in a tape recorder. It was
found that apraxic patients exhibited more
difficulties than left hemisphere patients without
apraxia, who in turn were impaired compared to
controls. Goldenberg et al. (2001) suggested that
IMA may act as an additional factor, augmenting
the difficulties caused by left hemisphere damage
alone, but acknowledged the alternative possibility
that the greater deficit in the apraxic cases might
reflect a non-specific depletion of cognitive
resources where brain damage is severe. They
362 Alan Sunderland and Caroline Shinner
concluded that such complex naturalistic tasks may
display “cognitive opacity”. In other words, the
complexity of the interaction between actor and
environment may be such that we are unable to
detect the impact of specific neuropsychological
deficits. A similar conclusion was reached by
Schwartz et al. (2002) who studied tasks such as
making a packed lunch and found a similar pattern
of errors in normal controls and in those with right
or left hemisphere brain damage. While it is
plausible that such complex naturalistic tasks may
indeed be cognitively opaque, it is worthy of note
that both Goldenberg et al. (2001) and Schwartz et
al. (2002) used error rating schemes which focussed
on content errors such as the inappropriate use of
objects or omitting part of the required action
sequence. Accuracy errors (clumsiness when
handling objects appropriately) were not considered
in detail. As the evidence from studies regarding the
handling of objects discussed earlier is that IMA is
associated with accuracy rather than content errors,
then this might explain the failure to find evidence
of specific effects of IMA on these complex tasks.
To summarise, the evidence is fairly clear that
IMA does disrupt ADL tasks where there is a
requirement to deploy non-routine motor strategies.
This being so, it must have an impact on more
complex naturalistic tasks as well, but it may be
difficult to distinguish that impact of IMA from the
multitude of other factors which will affect
performance. However it may be that future studies
which focus on accuracy rather than content errors
will detect specific effects.
EXPLAINING PRESERVED ABILITY
This review has shown that we can discard the
traditional view that IMA has no functional impact,
but it has to be acknowledged that the functional
deficit is relatively mild. While patients with IMA
show clumsiness on naturalistic tasks, apraxia
seldom prevents eventual success in completing
tasks (Foundas et al., 1995; Sunderland et al.,
1999). The conventional account for why the
effects are mild in most cases has been that
contextual cues can support performance. For
example, De Renzi et al. (1980) suggested that
action representations that are inaccessible for
controlling imitation or gesture to a command can
become available if “elicited by a particularly
strong flow of stimulation” (p. 10) as found within
a natural functional context. So, the chain of events
leading to waving goodbye in a natural context will
prime the appropriate action even if there is a
weakened representation of that action so that it
cannot be accessed out of context on command or
by imitation. The less obvious deficit seen when
handling actual objects than when pantomiming
object use (Clark et al., 1994; McDonald et al.,
1994) might also be explained in these terms with
the increased stimulation received from
proprioceptive and visual inputs when handling
actual objects serving to access damaged action
representations. However, such contextual effects
have not been investigated in depth and alternative
explanations are possible. For example, the
presence of physical objects may constrain
movement so that even if the action representation
is not more accessible, movements might fall into a
more normal pattern. Also, naturalistic tasks tend to
be forgiving of slight errors in performance. This
was illustrated by the study of functional dexterity
tasks by Sunderland et al. (1999) where, although
apraxics showed multiple accuracy errors, these
could be self-corrected and a successful outcome
achieved. Close observation showed that an apraxic
patient might make multiple inaccurate attempts to
position their fingers around a card to turn it over
but these occurred in rapid succession and the
desired outcome was achieved fairly rapidly, albeit
with greater effort than for normal controls.
It would be interesting to compare actions
under conditions of subtle changes in context, such
as following a command to stir water with a spoon
versus stirring a cup of coffee prior to drinking it.
Only when such research has been done could we
decide if the improvements when performing
actions in context is due to action representations
being “elicited by a particularly strong flow of
stimulation” or simply reflects the different nature
of a task when physical props are provided.
The combined effects of contextual cueing and
the forgiving nature of functional tasks may
therefore explain much of preserved everyday
ability. However, it also appears that some
functional tasks are inherently less affected by
IMA than others. Sunderland and Sluman (2000)
compared performance across different dexterity
tasks. This study included 15 patients with left
parietal or frontoparietal damage, 11 of whom
showed IMA on imitation or pantomime. An initial
analysis (Sunderland et al., 1999) had shown that
severity of IMA was a selective correlate of overall
rate of dexterity errors but Figure 1 indicates that
certain dexterity tasks were more affected than
others. There were frequent accuracy errors on the
bean spooning task described earlier and there was
a milder deficit on turning over cards placed flat
on the table top. In contrast, there was no
significant accuracy deficit on picking up small
objects such as a coin or a paperclip, or in stacking
4 wooden discs (checkers) to form a tower – all
apraxic patients in the study were within the
normal range on checker stacking. Other studies
are consistent with this finding of no deficit related
to IMA on some dexterity tasks. Haaland (1984)
found no correlation between IMA and speed on a
pegboard task, and Kimura (1993) reported that
apraxics were as fast as non-apraxic left
hemisphere patients in rapidly screwing a nut up
and down a bolt.
Ideomotor apraxia and functional ability 363
Apraxia and Physical Affordance
We suggest that a partial explanation for this
pattern of abilities on dexterity tasks is that apraxics
retain control of reaching and grasping under
perceptual guidance. Hermsdorfer et al. (2003)
studied patients with left hemisphere damage, two
thirds of whom had mild or severe IMA, as they
reached with their left hand for cylindrical objects
placed vertically in front of them. They showed
normal increases in velocity of reach with more
distant cylinders, together with preserved ability to
scale the aperture of their grasp to different
diameters of cylinder. Compared to normal controls,
there was some increase in total movement time and
some abnormalities in kinematic profile of the
reach-to-grasp response, but there was no
correlation between these abnormalities and severity
of IMA. The overall picture of the effect of left
hemisphere damage on prehension is therefore that
the basic skills of reaching and grasping with the
left hand remain, albeit with some deficits in the
control of timing of movement trajectory (Haaland
et al., 2004), which are probably of separate origins
from IMA (Hermsdorfer et al., 2003). Reduced
accuracy related to IMA has been found for rapid
aiming movements to the smallest visual targets
(Haaland and Harrington, 1994) or reaching in
conditions of restricted visual feedback (Haaland et
al., 1999; Hermsdorfer et al., 2003) but not under
less demanding conditions.
To what extent could this preservation of simple
prehension explain the performance of different
dexterity tests shown in Figure 1? An important
characteristic of simple prehension is that the shape
of the emerging grasp may be dictated by the
structural properties of the target object (Jeannerod
et al., 1995). This will be the case when the object
offers few degrees of freedom in how it can be
grasped or manipulated; that is to say, it offers a
strong physical affordance (Norman, 1988). So the
apraxic patient may have an intact route to action
sufficient to perform functional tasks such as
picking up a small object where a precision grip of
a certain amplitude is afforded by the object. Other
dexterity tasks present more degrees of freedom or
require physical attributes to be secondary to other
factors. The card turning task in Figure 1 offered
many degrees of freedom in how the cards could
be picked up and turned over (control subjects
showed 7 different strategies), and the bean
spooning task would also be influenced by
knowledge and experience of using a spoon.
The Role of Memory for Action
Figure 2 suggests three major influences on
ability to manipulate objects. This is not an
information processing diagram, but a conceptual
framework which specifies three different
categories of explanatory factor which may be
important in understanding apraxic behaviour. The
first of these is physical affordance, and, as
discussed above, this may underpin good
performance of apraxic patients on some tasks. A
second broad category is the use of memory to
364 Alan Sunderland and Caroline Shinner
Fig. 1 – The frequency of accuracy errors on different dexterity tasks for stroke patients with left or right hemisphere damage, and
age-matched controls using the left or right hand (from Sunderland and Sluman, 2000).
guide action, within which we include both
memory for actions and semantic knowledge (e.g.,
knowledge of what specific tools are used for). We
have seen that the pattern of errors in object
manipulation supports the distinction between IA
and IMA, with the former giving rise to content
errors and the latter to errors of accuracy of
execution. This implies that if disturbed memory
for action is a key factor in explaining disturbed
object manipulation in pure IMA, then the deficit
is more procedural than semantic. This is in line
with the conventional theory of IMA which is that
it is not due to loss of knowledge about how
objects are used but to disruption or disturbed readout
of mental representations of movements
(visuokinesthetic engrams; Rothi and Heilman,
1997).
A corollary of this engram theory would seem
to be that habitual actions should be more affected
by IMA than novel ones, because habitual actions
rely on memory whereas novel actions do not. The
limited available evidence on functional ability
points in the opposite direction. This was noted for
complex tasks such as dressing (Goldenberg and
Hagmann, 1998a; Walker et al., 2004) where
difficulties emerged when patients had to learn
novel dressing strategies to compensate for
hemiparesis but not when previously learned
routine procedures could be used. The vast
majority of research related to action engram
theory has stayed within the Liepmann tradition of
investigation of imitation or pantomime, and
implications for object manipulation and functional
ability are therefore not entirely clear. Recent
theorizing within this tradition has focussed on
single case dissociations in ability to imitate
familiar meaningful versus novel meaningless
gestures, and contrasts between ability to produce
or recognise gestures. This has lead to the
suggestion that there are sub-types of IMA, with
impaired memory for action being crucial in only
one of these. Buxbaum (2001) termed this sub-type
“representational IMA”, in which damage to action
engrams is demonstrated by deficiency in gesture
recognition (Heilman et al., 1982). Buxbaum
(2001) contrasts this with “dynamic IMA” in which
intact gesture recognition combined with impaired
imitation of meaningless gestures indicates a
problem with spatiomotor processing but not with
action memory. The implications of this model for
functional ability have yet to be explored. For
example, might it be that the high rate of errors on
the bean spooning task in Figure 1 is due to the
inclusion of cases of representational IMA where
memory for action is impaired whereas dynamic
IMA accounts for errors on novel tasks such as
card turning? Future research on functional ability
will need to distinguish these sub-types.
Postural Problem Solving
There are many naturalistic tasks where action
can neither be entirely affordance led nor dictated
by memory. For example, when faced with opening
a door with an unfamiliar latch, the structure of the
door handle will afford grasping it in a particular
way and memories of the actions required by
familiar latches may help in attempting to open it.
However, a degree of problem solving ability will
be required to discover how this particular latch
operates and what orientation of grasp is necessary
to open it efficiently. The third process indicated in
Figure 2 is therefore the use of problem solving to
tackle novel tasks. If a problem solving deficit is
central to IMA then this could explain why the
apraxic deficit tends to emerge when hemiparesis
Ideomotor apraxia and functional ability 365
Fig. 2 – Three major influences on ability to manipulate objects. Memory and problem solving may be disrupted in IMA but the
intact affordance-led route allows manipulation of objects where the structural characteristics of objects and/or the biomechanical
constraints of the context guide appropriate action.
Environmental
Affordance
ACTIONMemory
Knowledge of use and
action representations
Problem solving
Mechanical or postural
Intact route
in IMA
forces the learning of new strategies for tackling
functional motor tasks (Goldenberg and Hagmann,
1998a; Walker et al., 2004). However the evidence
on the nature of any problem solving deficit in
IMA is not yet clear. Goldenberg and Hagmann
(1998b) studied the relationship between errors in
use of actual tools and pantomime of tool use in a
group of left hemisphere damaged aphasic patients.
They interpreted deficits in pantomime as
indicating impaired semantic memory of
“instructions of use” for the tools but found that it
was a poor predictor of errors in actual tool use.
Only when patients were also impaired on a
mechanical problem solving task was there a
consistent deficit in actual tool use. Sunderland and
Sluman (2000) suggested that the problem solving
deficit which was specific to IMA related not to
mechanical reasoning but to mental representation
of body posture. That is to say, that when neither
physical affordance nor memory dictated how an
object should be handled then there was a
requirement to mentally model what configurations
of grasp could be used to manipulate it. So, the
interpretation of the profile of performance shown
in Figure 1 was that the large number of degrees of
freedom in how objects could be grasped for card
turning required that a mental simulation of
grasping be used to determine the most efficient
strategy. Similarly bean spooning, because it
involved use of the non-dominant hand, could not
be performed by relying on action memory but
required postural problem solving.
Evidence consistent with there being a deficit in
a high level representation of body posture in IMA
can be drawn from several sources. A general
finding is that ideomotor apraxics make spatial
errors in imitation of gestures, leading Buxbaum
(2001) to define her sub-type of dynamic IMA as
due to deficient calculation of information about
spatial relationships between body parts. More
specifically, Goldenberg (1995) found that
ideomotor apraxics were not only unable to imitate
gesture but also were unable to model the gesture
on a life-size mannikin, suggesting a conceptual
deficit in appreciation of body configuration.
Finally, consistent with a spatial deficit is the
evidence that ideomotor apraxics show slightly
reduced spatial accuracy when high demands are
placed on awareness of limb position, for example
when visual feedback is restricted (Hermsdorfer et
al., 2003) or rapid aiming movements are made to
very small visual targets (Haaland et al., 1999).
However, these latter studies also seem to raise
a difficulty with the postural theory in that
problems with egocentric spatial coding would
seem to predict not minor but major inaccuracies in
visuomotor control, including impairment on the
simple prehensile tasks which we have seen are
spared in IMA. In answer to this we would argue
that the postural representation which is deficient
in IMA is at a higher level than primary
sensorimotor representation and is only called upon
as a specialist left hemisphere system to solve
problems where there is high demand on postural
modelling. We can speculate that this is a function
of the left inferior parietal lobe which is known to
be involved in awareness of the body scheme
(Haggard and Wolpert, 2005) and in the mental
simulation of actions (Sirigu et al., 1996). Also, it
must be remembered that all studies of objectrelated
actions in IMA concern use of the
ipsilesional left hand, which has unimpeded access
to the motor control systems of the intact right
hemisphere. It is interesting to note that affordance
led action has been equated with the functioning of
the dorsal route for action control (Norman, 2002)
and that the right hemisphere dorsal route is intact
in IMA.
The conceptual framework of Figure 2 therefore
appears to offer a useful way of viewing the profile
of functional disability in IMA, but further
elaboration and expansion is needed. There is a
need to explore the extent to which demands on
problem solving abilities of the type outlined can
successfully predict disability on a given task, and
to compare this with accounts of IMA which place
the focus of the deficit in disrupted action engrams
for at least some patients (Buxbaum, 2001). An
example of the challenge facing both accounts
comes from the study of dressing difficulties in
apraxia (Walker et al., 2004). The postural
hypothesis might predict that IMA would lead to
problems in learning to orient the garment and
limbs appropriately, whereas an engram account
might predict reduced accuracy of habitual action
sub-components such as threading a limb through a
sleeve. In fact, apraxics appeared to have no
particular difficulty on either count when putting
on a shirt but instead showed a persistent error of
dressing the non-paretic arm first, leaving them
entangled with the shirt when attempting to dress
the paretic side. This perhaps suggests that theories
of IMA which stress learning of motor sequences
(Harrington and Haaland, 1992) might be more
successful in predicting functional ability in this
instance. Overall it is clear that we are still some
way from having an understanding of IMA
sufficient to predict its functional impact on
different naturalistic tasks.
CONCLUSIONS
One firm conclusion is that anecdotal accounts
of intact functional ability in IMA can be
dismissed. There is solid evidence of an apraxic
impairment when handling actual objects and
circumstantial evidence of abnormalities in
naturalistic action. The extent to which these
translate into clinically significant disability
remains unclear but it is plausible that IMA may be
a major hidden obstacle in rehabilitation.
366 Alan Sunderland and Caroline Shinner
The conceptual framework we have offered
suggests several directions for future clinical and
theoretical work. Major theoretical concerns are to
seek further evidence on the nature of the proposed
postural impairment and also to extend the studies
of the role of action engrams from traditional work
on gesture imitation into work on object
manipulation. Foremost among the clinical
concerns are to clarify the relative benefits of
supporting action through maximum use of
physical affordance or through minimum reliance
on postural problem solving, and to clarify if
impaired memory for habitual actions is the critical
deficit in a sub-group of apraxic patients with
representational IMA.
REFERENCES
BUXBAUM LJ. Ideomotor apraxia: A call to action. Neurocase, 7:
445-458, 2001.
CLARK M, MERIANS A, KOTHARI A, POIZNER H, MACAULEY B,
ROTHI LJ and HEILMAN KM. Spatial planning deficits in limb
apraxia. Brain, 117: 1093-1106, 1994.
DE RENZI E. Methods of limb apraxia examination and their
bearing on the interpretation of the disorder. In Roy EA (Ed),
Neuropsychological Studies of Apraxia. Amsterdam: NorthHolland,
1985.
DE RENZI E and LUCCHELLI F. Ideational apraxia. Brain, 111:
1173-1185, 1988.
DE RENZI E, MOTTI F and NICHELLI P. Imitating gestures: A
quantitative approach to ideomotor apraxia. Archives of
Neurology, 37: 6-10, 1980.
DUMONT C, SKA B and SCHIAVETTO A. Selective impairment of
transitive gestures: An unusual case of apraxia. Neurocase, 5:
447-458, 1999.
FOUNDAS AL, MACAULEY BL, RAYMER AM, MAHER LM, HEILMAN
KM and ROTHI LJ. Ecological implications of limb apraxia:
Evidence from mealtime behaviour. Journal of the
International Neuropsychological Society, 1: 62-66, 1995.
GOLDENBERG G. Imitating gestures and manipulating a mannikin:
The representation of the human-body in ideomotor apraxia.
Neuropsychologia, 33: 63-72, 1995.
GOLDENBERG G. Apraxia and beyond: Life and work of Hugo
Liepmann. Cortex, 39: 509-524, 2003.
GOLDENBERG G, DAUMULLER M and HAGMANN S. Assessment and
therapy of complex activities of daily living in apraxia.
Neuropsychological Rehabilitation, 11: 147-169, 2001.
GOLDENBERG G and HAGMANN S. Therapy of activities of daily
living in patients with apraxia. Neuropsychological
Rehabilitation, 8: 123-141, 1998a.
GOLDENBERG G and HAGMANN S. Tool use and mechanical
problem solving in apraxia. Neuropsychologia, 36: 581-589,
1998b.
HAALAND KY. The relationship of limb apraxia severity to motor
and language deficits. Brain and Cognition, 3: 307-316, 1984.
HAALAND KY and HARRINGTON DL. Limb-sequencing deficits
after left but not right hemisphere damage. Brain and
Cognition, 24: 104-122, 1994.
HAALAND KY, HARRINGTON DL and KNIGHT RT. Spatial deficits in
ideomotor limb apraxia: A kinematic analysis of aiming
movements. Brain, 122: 1169-1182, 1999.
HAALAND KY, HARRINGTON DL and KNIGHT RT. Neural
representations of skilled movement. Brain, 123: 2306-2313,
2000.
HAALAND KY, HARRINGTON DL, KNIGHT RT and LEE RR.
Hemispheric asymmetries for kinematic and positional aspects
of reaching. Brain, 127: 1145-1158, 2004.
HAGGARD P and WOLPERT DM. Disorders of body scheme. In
Freund HJ, Jeannerod M, Hallett M and Leiguarda R (Eds),
Higher-Order Motor Disorders. Oxford: Oxford University
Press, 2005.
HANNA-PLADDY B, HEILMAN KM and FOUNDAS AL. Ecological
implications of ideomotor apraxia: Evidence from physical
activities of daily living. Neurology, 60: 487-490, 2003.
HARRINGTON DL and HAALAND KY. Motor sequencing with left
hemisphere damage: Are some cognitive deficits specific to
limb apraxia? Brain, 115: 857-874, 1992.
HEILMAN KM, ROTHI LJ and VALENSTEIN E. Two forms of
ideomotor apraxia. Neurology, 32: 342-346, 1982.
HERMSDORFER J, BLANKENFELD H and GOLDENBERG G. The
dependence of ipsilesional aiming deficits on task demands,
lesioned hemisphere, and apraxia. Neuropsychologia, 41:
1628-1643, 2003.
JEANNEROD M, ARBIB MA, RIZZOLATTI G and SAKATA H. Grasping
objects: The cortical mechanisms of visuomotor
transformation. Trends in Neurosciences, 18: 314-320, 1995.
KIMURA A and ARCHIBALD Y. Motor functions of the left
hemisphere. Brain, 97: 337-350, 1974.
KIMURA D. Neuromotor Mechanisms in Human Communication.
Oxford: Oxford University Press, 1993.
KIMURA K. Acquisition of a motor skill after left-hemisphere
damage. Brain, 100: 527-542, 1977.
MCDONALD S, TATE RL and RIGBY J. Error types in ideomotor
apraxia: A qualitative-analysis. Brain and Cognition, 25: 250-
270, 1994.
NORMAN DA. The Psychology of Everyday Things. New York:
Basic Books, 1988.
NORMAN J. Two visual systems and two theories of perception: An
attempt to reconcile the constructivist and ecological
approaches. Behavioral and Brain Sciences, 25: 73-144, 2002.
POIZNER H, CLARK M, MERIANS A, MACAULEY B, ROTHI LJ and
HEILMAN KM. Joint coordination deficits in limb apraxia.
Brain, 118: 227-242, 1995.
ROTHI LJ and HEILMAN KM. Introduction to limb apraxia. In Rothi
LJ and Heilman KM (Eds), Apraxia: The Neuropsychology of
Action. Hove: Psychology Press, 1997.
SCHWARTZ MF, SEGAL M, VERAMONTI T, FERRARO M and
BUXBAUM LJ. The naturalistic action test: A standardised
assessment for everyday action impairment.
Neuropsychological Rehabilitation, 12: 311-339, 2002.
SIRIGU A, DUHAMEL JR, COHEN L, PILLON B, DUBOIS B and AGID
Y. The mental representation of hand movements after
parietal cortex damage. Science, 273: 1564-1568, 1996.
SUNDERLAND A. Recovery of ipsilateral dexterity after stroke.
Stroke, 31: 430-433, 2000.
SUNDERLAND A, BOWERS MP, SLUMAN S-M, WILCOCK DJ and
ARDRON ME. Impaired dexterity of the ipsilateral hand after
stroke and the relationship to cognitive deficit. Stroke, 30:
949-955, 1999.
SUNDERLAND A and SLUMAN S-M. Ideomotor apraxia, visuomotor
control and the explicit representation of posture.
Neuropsychologia, 38: 923-934, 2000.
SUNDET K, FINSET A and REINVANG I. Neuropsychological
predictors in stroke rehabilitation. Journal of Clinical and
Experimental Neuropsychology, 10: 363-379, 1988.
WALKER CM, SUNDERLAND A, SHARMA J and WALKER MF. The
impact of cognitive impairment on upper body dressing
difficulties after stroke: A video analysis of patterns of
recovery. Journal of Neurology, Neurosurgery and Psychiatry,
75: 43-48, 2004.
WALKER MF and LINCOLN NB. Factors influencing dressing
performance after stroke. Journal of Neurology, Neurosurgery
and Psychiatry, 54: 699-701, 1991.
WESTWOOD DA, CHAPMAN CD and ROY EA. Pantomimed actions
may be controlled by the ventral visual stream. Experimental
Brain Research, 130: 545-548, 2000.
Alan Sunderland, Reader in Clinical Neuropsychology, School of Psychology,
University of Nottingham, Nottingham NG7 2RD, UK.
e-mail: alan.sunderland@nottingham.ac.uk
Ideomotor apraxia and functional ability 367
(Received 13 October 2004; revised 11 January 2005; accepted 12 March 2005)