INTRODUCTION
Dual-task methodology is often employed to determine the cognitive
processes that memory tasks involve. Within this field of research, the
focus has mainly been on the structure of verbal or visuospatial
memory; however, recently there have been investigations of memory for
movement sequences and whether “movement memory” is
distinct from verbal and visuospatial memory (e.g., Feyereisen & Van der Linden, 1997; Helstrup, 1999; Smyth &
Pendleton, 1989). While these recent investigations have
attempted to ascertain the cognitive processes involved in memory for
movement, tasks used clinically, such as the Kaufman Hand Movements
Test (KHMT; Kaufman & Kaufman, 1983),
have not been investigated (Frencham et al.,
2003), and as such, the processes that they involve remain
unclear.
The KHMT was developed from Luria's fist-edge-palm test of motor
function, a task in which the individual must mimic a sequence of hand
postures (fist, edge, and palm) demonstrated by the examiner. It has
been found to be sensitive to cognitive deficits demonstrated in
various populations such as individuals with mild traumatic brain
injury or alcohol-related brain damage (Fox &
Fox, 2001; Fox et al., 1993).
The development and structure of the KHMT may partially explain why
it has not been investigated in the working memory literature. The task
design does not allow for traditional span analysis, and the overall
format is not the same as other commonly used verbal and visuospatial
span tasks, such as the Digit Span and Spatial Span subtests from the
Wechsler Memory Scale (e.g., Wechsler, 1987).
To address this problem and investigate the processes involved in
memory for sequential hand movements, Frencham et al. (2003) compared letter and Corsi span tasks with a
hand movement task of comparable span structure (span-HMT). The
span-HMT was developed from both the KHMT and the movement memory task
used by Smyth and Pendleton (1989) and
involves recall of sequences of two to nine items selected from an item
pool of seven distinct hand postures. Based on an observed detrimental
effect of articulatory suppression (repeatedly articulating a digit
sequence) on span-HMT performance when compared to visuospatial or
movement interference, it was concluded that memory for hand movements
involves the phonological loop to an extent. It was proposed that this
might occur due to participants spontaneously using a verbal labelling
strategy when encoding and recalling hand movement sequences. While
Kaufman and Kaufman (1983) presented their
task as one that did not depend on language, they also commented that
performance could benefit from adopting a mediating strategy, such as
using verbal labelling, and that it also relied upon visual short-term
memory to an extent.
Other researchers have recognized the role of verbal strategies in
the recall of movement sequences. Helstrup (2000) investigated the manner in which strategies
of various modalities affected memory for patterned movement, using a
primary task that required serial recall of movement sequences within a
matrix rather than sequences of different postures. Instructions to
participants were quite lenient, directing them to adopt a verbal,
spatial, or movement strategy of their choice during stimulus
presentation, with general examples provided. In the baseline
condition, participants were free to employ strategies according to
personal preference. Results showed that, while recall was superior
with a verbal than with a visuospatial or movement strategy, there was
no significant difference between performance in the verbal strategy
condition and baseline. This equivalence between baseline and verbal
strategy conditions may indicate that many participants employed a
verbal strategy during baseline. Given the lenience of the strategy
conditions, it is possible that had participants been instructed more
explicitly to use a specific verbal strategy, their performance would
have surpassed that obtained in the baseline condition. Under
Helstrup's methodology it is plausible that strategies employed
under the baseline condition, and perhaps in the verbal strategy
condition to a lesser extent, would be less beneficial to recall than
predetermined and imposed strategies, due to the added demands with the
former of generating and modifying strategies during task performance.
Regardless, Helstrup's findings support the hypotheses that (1)
movement sequences are coded as verbal strings rather than motorically
or visuospatially and (2) when the verbal information attended to
during movement task performance is context specific and presented as a
strategy that should be consciously employed, it is not detrimental to
performance.
Others have acknowledged the supportive role of verbal processes in
movement recall more indirectly. In a dual-task study, Woodin and Heil
(1996) acknowledged the potential for verbal
involvement in their movement task by instructing participants to
perform articulatory suppression in all interference
conditions, including baseline, “to prevent the use of verbally
mediated strategies” (p. 360). Overall, research such as this
supports the notion that memory for movements involves a significant
verbal component, and that verbal strategies can support movement
memory performance.
Despite the reported involvement of verbal processing in memory for
movement (e.g., Feyereisen & Van der Linden,
1997; Helstrup, 1999; Remoundou & Humphreys, 2001), evidence for
verbal labelling during encoding has not been directly addressed using
a primary and secondary task format. The aim of the present study was
to further delineate the role of verbal labelling in recall of
movements by presenting verbal labels concurrent with hand postures and
having the participant shadow them during a computerized version of the
KHMT (Experiment 1) and during the span-HMT (Experiment 2). Shadowing
is a technique that has been widely used in selective attention
research (Banich, 1997), but has also been
incorporated as a secondary task in memory research (Orwig, 1979). Shadowing tasks typically involve
repeating auditorily presented material aloud, and have been argued as
being suitable for determining the degree to which a task involves a
verbal component (Klauer & Stegmaier,
1997). Using this design, the intention was to identify the
differential effects of verbal labelling on performance, depending on
whether the labels did or did not correspond to the hand postures
presented.
The KHMT was modified for computerized presentation to make it more
comparable with the span-HMT, and so that the overall dual-task
structure would be constant across the memory tasks. An additional
reason for this modification was to regulate the rate of stimulus
presentation by using a dynamic video format, and to determine the
comparability of this format relative to the in vivo
presentation of the KHMT outlined by Barry and Riley (1987) in their administration of the task in an
adult population.
The hypotheses of this study were as follows. A specific hypothesis
for Experiment 1 was that performance during the baseline condition of
the computerized-KHMT would be comparable to published norms for the
KHMT. It was also predicted that instructing participants to shadow
verbal labels that were congruent with the hand postures being
displayed would improve performance on both the computerized-KHMT and
the span-HMT when compared to a baseline condition for at least three
reasons; firstly, as participants would not have to generate verbal
labels whilst performing the task; secondly, as the relevant
information was presented in two modalities rather than one; and
thirdly, as the labels provided may be more efficient than those
participants would generate spontaneously. Finally, in keeping with the
detrimental effect of articulatory suppression on span-HMT performance
reported by Frencham et al. (2003), the
incongruent verbal labelling condition was hypothesized to result in
poorer recall for both memory tasks when compared to a baseline
condition.
General Methods
The computerized-KHMT and span-HMT (the latter is described in Frencham et al., 2003) were used in Experiments 1
and 2, respectively. Both tasks involved the participant viewing
dynamic video footage of sequences of hand postures (made on a flat
surface with the right hand), and then reproducing them with his or her
own right hand. The main difference between tasks was that the span-HMT
item pool was seven hand postures, while the computerized-KHMT had only
three. For both tasks, stimuli and instructions were presented on a
computer monitor, and following presentation of the stimuli, a visual
cue (the word “recall”) prompted the participant to
reproduce the sequence.
For each task, prior to practice and test items, individuals were
familiarized with hand posture stimuli. During this familiarization
stage the hand postures for each task were presented in random order on
the screen as 10 cm × 12 cm still images, positioned so they
resembled the orientation of the examiner's right hand if sitting
opposite the participant. Stimuli were presented in random order, and
the individual was instructed to reproduce each one. The examiner
corrected any incorrect imitations before subsequent stimuli were
presented. Prior to the presentation of practice and test sequences,
the participant was instructed to remember the hand postures presented
and then repeat them in the same order using his or her right hand when
told. The examiner manually recorded responses and classified each
sequence as correct if the items were recalled in their correct ordinal
positions.
Computerized-KHMT
Three hand postures made up the stimulus set (see Figure 1), and the sequences presented were
reproduced from the original KHMT such that there was one practice item
of sequence length two, and then 21 test trials, with sequences ranging
in length from two to six. The rate of presentation was one hand
posture per second, as in the KHMT. All 21 items were administered to
every participant.
Still images of hand postures used in the computerized-KHMT for
Experiment 1. Verbal labels assigned to the hand postures are as
follows: 1—palm; 2—fist; 3—side.
Span-HMT
Seven hand postures made on a flat surface formed the stimulus set
for this task (see Figure 2). After the
familiarization phase, four practice trials were given, two each at
sequence lengths 2 and 3. For the test proper, sequences involving
three to nine hand movements were presented. Four trials were presented
at each sequence length, and the task was discontinued when all four
sequences at a particular length were not correctly recalled. The
stimuli for each test sequence were selected randomly, with
replacement, from the pool of seven items. Stimuli were presented with
approximately equal frequencies in the test sequences. The rate of
stimulus presentation was one hand posture every 1.5 s as in Frencham
et al. (2003).
Still images of hand postures used in the span-HMT for Experiment 2.
Verbal labels assigned to the hand postures are as follows:
1—curve; 2—fist; 3—side; 4—claw; 5—palm;
6—star; 7—pinch.
Secondary tasks
For both experiments, each participant completed either the
computerized-KHMT or the span-HMT under three secondary task
conditions: congruent verbal labels, incongruent verbal
labels, and baseline. Secondary tasks were performed
during the stimulus presentation phase such that under the congruent
and incongruent verbal labels conditions, immediately after hearing
each verbal label, the participant had to shadow it (i.e., repeat it
out aloud). The examiner monitored secondary task performance to ensure
that the participant correctly shadowed each verbal label. The verbal
labels were presented via computer speakers by a female voice
at the same rate and with onsets corresponding to the onset of visually
presented stimuli in the memory task (i.e., one per second for the
computerized-KHMT and one per 1.5 s for the span-HMT). Participants
were instructed not to say the labels aloud while they were recalling
the hand position sequences.
Verbal labels were generated to accompany the hand postures used in
both tasks, such that there were three for the computerized-KHMT and
seven for the span-HMT, as indicated in Figures 1 and 2, respectively. In
the congruent labels condition, the word was presented concurrent with
the corresponding hand posture, and for the incongruent condition,
labels and hand postures were mismatched. For the baseline condition,
there was no additional task during the stimulus presentation phase.
The participant was familiarized with each secondary task during the
practice trials. Prior to both the congruent and incongruent verbal
labels conditions for each task, the participant was told, “the
most important thing is to watch and remember the hand movements. Just
echo the words that are presented”. The participant was then
informed that after echoing the labels, he/she would not be asked
to recall them. During the familiarization phase under the baseline
condition, participants were instructed to watch and remember the hand
movements, and were not specifically encouraged to or prevented from
generating their own labels.
EXPERIMENT 1
Methods
Research Participants
Eighteen participants were recruited (6 male, 12 female), being
undergraduate students in the School of Psychology at the University of
Western Australia. All participants were naïve to the nature and
hypotheses of the experiment. The mean age was 21.3 years (SD
= 3.8), and mean education level was 14.3 years (SD = 0.96).
All except one reported being right handed. All participants reported
being able to speak English fluently.
Equipment
An IBM compatible computer with a 53-cm color touch-sensitive screen
was programmed using MetaCard 2.3 to present all stimuli and control
timing of the auditory and visual display presentations.
Procedure
Each participant completed the computerized-KHMT under three
conditions; baseline, congruent labels, and incongruent labels. The
order of these secondary task conditions was counterbalanced across
participants.
Results
The total score for each participant during each condition was the
number of correctly recalled test sequences, with a maximum of 21.
Baseline condition performance was compared to published KHMT adult
normative data (Barry & Riley, 1987) to
assess comparability of the tasks and subsequent generalizability of
the results to the wider KHMT literature, given the modifications used
in this study. Using an alpha level of .05, an independent samples
t test demonstrated that there was no significant difference
between the mean computerized-KHMT score (M = 17.0,
SD = 3.09) and the normative mean published for individuals of
comparable age (M = 16.7, SD = 2.62; t (36)
= .32, p >.05). The mean number of items recalled for each
condition revealed that congruent labels led to increased
computerized-KHMT recall when compared to baseline, while incongruent
labels led to decreased recall as illustrated in Figure 3.
Mean performance on the computerized-KHMT during the congruent
labels, baseline, and incongruent labels conditions (within-subjects
95% confidence intervals shown as error bars).
A significant effect of secondary task condition was observed,
F(2,34) = 46.05, p < .001. Paired-samples
t tests demonstrated that computerized-KHMT recall during the
congruent verbal labels condition was higher than during the baseline
condition, t (17) = 2.13, p < .05, while shadowing
incongruent labels significantly reduced recall, t (17) =
6.96, p < .001.
Discussion
These results demonstrate that instructing participants to shadow
congruent verbal labels enhanced performance, whereas incongruent
verbal labels decreased computerized-KHMT performance. It was also
shown that the average computerized-KHMT performance was comparable
with performance of adults in their twenties on the standard KHMT, thus
indicating that the difficulty of the task was not significantly
altered by modifying it for computer administration.
Using similar methodology, Experiment 2 investigated whether the
previously described interference pattern would be found with the
span-HMT. Thus it assessed whether the two target tasks were recruiting
similar cognitive processes, and by inference, whether the span-HMT
would be clinically useful when assessing similar cognitive domains as
the KHMT.
EXPERIMENT 2
Methods
Research Participants
Eighteen participants were recruited (7 male, 11 female), being
undergraduate students in the School of Psychology at the University of
Western Australia. All participants were naïve to the nature and
hypotheses of the experiment. The mean age was 20.4 years (SD
= 3.2), and mean education level was 14.0 years (SD = 0.97).
Three participants reported being left handed while the rest were right
handed. All reported speaking English fluently.
Equipment
This was as for Experiment 1.
Procedure
Each participant completed the span-HMT under three conditions,
baseline, congruent labels, and incongruent labels. Again, the order of
secondary task presentation was counterbalanced across participants.
Results
The span score for each participant during each condition was
calculated using a fractional scoring method similar to that used by
Chuah and Maybery (1999); 2 + (number of
correctly recalled sequences)/4 (where 4 is the number of trials
per sequence length, with two units added as the testing began at
sequence length three). The resultant mean spans for each condition
revealed that congruent verbal labels led to increased span-HMT
performance when compared to baseline, while incongruent verbal labels
led to decreased span, as illustrated in Figure
4.
Mean performance on the span-HMT during the congruent labels,
baseline, and incongruent labels conditions (within-subjects 95%
confidence intervals shown as error bars).
A significant effect of secondary task was observed, F(2,34)
= 71.69, p < .001. Paired-samples t tests
demonstrated that recall during the congruent labels condition was
significantly higher than recall in the baseline condition, t
(17) = 3.38, p < .001, and that recall during the baseline
condition exceeded that during the incongruent condition, t
(17) = 7.65, p < .001.
Discussion
Verbalizing congruent labels while encoding hand postures was found
to enhance subsequent recall, whereas concurrent verbalization of
incongruent labels decreased span-HMT recall when compared to baseline.
These results are in keeping with the self-report of pilot study
participants in Frencham et al. (2003) who
stated that they used verbal labels to help them remember the hand
postures, and with respect to the detrimental effect of incongruent
labels on recall, are in keeping with findings in the wider verbal
short-term memory literature.
GENERAL DISCUSSION
The results demonstrated that baseline performance on the
computerized-KHMT was comparable with published normative data for the
KHMT under standardized administration. Experiments 1 and 2
investigated the effects of secondary tasks involving the shadowing of
verbal labels upon hand movement memory. For both primary tasks, the
computerized-KHMT and span-HMT, recall was enhanced when the verbal
labels were congruent and reduced when they were incongruent with the
hand postures presented. Thus, the meaning or content of the secondary
task, and the relation that this meaning had to the primary task (i.e.,
match or mismatch), was found to affect recall.
To date, dual-task memory experiments addressing the effects of
secondary tasks that are directly congruent or incongruent with the
primary task are scarce; however, two studies deserve mention. Buchner
et al. (1996) conducted a study of recall of
two-digit numbers in which one secondary task condition also involved
auditory presentation of two-digit numbers. However, the primary and
secondary tasks were not completely congruent, as the order of numbers
presented was random for both. In keeping with the wider literature,
this interference condition had a detrimental effect on recall. These
findings are consistent with the current result that incongruent labels
reduced primary task performance. In a study of the role of movement in
spatial memory, Quinn and Ralston (1986)
incorporated secondary spatial tapping tasks that were either
compatible or incompatible with the primary task, the Brooks matrix
task. They reported that recall was detrimentally affected by the
incompatible task, whereas recall was not significantly reduced in the
compatible condition.
It is possible that the learning literature may be more amenable to
the approach of the current study. For example, selected results of a
larger study of memory for sign language support the current findings.
English-speaking individuals with no knowledge of sign language were
tested on their recall of cherological (sign language) hand postures.
Results demonstrated that when signing stimuli were presented with
spoken verbal labels, recall was better than when the signs were
presented without verbal labels (Hamilton &
Holzman, 1989). It was concluded that this demonstrated that
recall of less familiar material was supported when it was associated
with more recognizable information.
The role of familiarity with task material has also been discussed in
the movement memory literature. Citing the wider problem-solving
literature, Helstrup (1996) proposed that
verbal codes were easier to rehearse than actions because verbal codes
are more familiar. Also, movement memory is a less familiar task for
most people than verbal or visuospatial memory. Accordingly, he argued
that more verbal than motor support cues would arguably be available at
the retrieval stage. Cumulatively, our research has generated evidence
for the use of verbal codes in movement memory. For example, in pilot
studies, when presented with hand movement stimuli, the majority of
participants reported spontaneously using a verbal strategy instead of
a kinesthetic or visuospatial one (see Frencham et
al., 2003).
This study, by demonstrating that imposing a verbal strategy can
improve recall of hand movement sequences, extends upon Helstrup's
(2000) findings. Admittedly, the memory tasks
used currently differed to that used by Helstrup: in this study, the
target tasks involve memory for “configured” movement
according to Smyth and Pendleton's (1989) terminology, that is, recalling sequences of
different hand postures, whilst Helstrup's memory task involved
what he termed “motor-spatial” memory for sequences of
movements within a matrix. Yet despite these differences, one possible
interpretation of the current results is that strategies that are both
imposed and performed (at least via shadowing) are more
beneficial to movement recall than the range of individually initiated
verbal strategies that may have occurred during baseline performance.
It is likely that if labels were spontaneously adopted during baseline,
they would have been less specific and less consistent than those
imposed in the congruent verbal labels condition. Future research could
explore this issue by comparing recall for hand movement sequences in a
control group and individuals who are taught labels prior to task
performance. Another explanation for superior recall in the congruent
labels condition as opposed to baseline is that information was
presented via both visual and auditory modalities.
Furthermore, participants also generated articulatory codes through
shadowing the congruent labels. Useful further research would be to
disentangle contributions to recall of presentation modality (visual,
auditory) and modality of shadowing (oral, manual) by manipulating
these two factors independently.
Turning to the detrimental effect of incongruent labels on primary
task performance, one interpretation is that recall of hand movement
sequences involved at least some verbal recoding of the hand postures.
Shadowing incongruent labels may have reduced the use of spontaneous
verbal rehearsal strategies. It is also possible that the detrimental
effect of shadowing incongruent labels is partly attributable to the
overlap in content with the labels that are applied to the
to-be-remembered hand postures. Incompatible serial ordering of these
two sets of verbal labels may have contributed to poorer performance
(see Jones, 1999). Nevertheless, Frencham et
al. (2003) found that repeatedly articulating
digit sequences that were unrelated to hand posture labels also had a
detrimental effect on span-HMT performance, suggesting that at least a
component of the detrimental effect observed in the present study is
due to verbal interference.
However shadowing incongruent verbal labels may exert interference on
recall of hand movement sequences over and above that of articulating
unrelated verbal items, as with vocalizing 1,2,3,4 repeatedly, employed
in Frencham et al. (2003). Direct comparison
of results across our two studies supports this postulation, indicating
that while baseline span-HMT performance was comparable in the two
studies, t (32) = .77, p = .45, performance during
the incongruent verbal labelling condition was poorer than that under
digit articulation, t (32) = 2.36, p < .05. The
additional interference incurred in the incongruent labels condition
may stem from competition for mental resources or “confusion of
like content” between the shadowed labels and verbal labels
generated for the to-be-remembered hand postures. Alternatively, the
conflict could be between processing the visual movement stimuli and
incongruent visuo-motor movement representations deriving from the
verbal labels being shadowed. Thus, while interference due to digit
articulation demonstrates the involvement of the verbal coding in
recall for hand movement sequences, the additional interference under
the incongruent labels condition may represent conflict between either
verbal or nonverbal codes for hand postures.
The likelihood of an additional executive load stemming from
shadowing incongruent labels also warrants discussion. However,
shadowing auditory verbal stimuli has been shown not to adversely
affect memory for serial spatial material (Klauer
& Stegmaier, 1997). Thus, shadowing of words does not appear
to involve an executive component that could impact on retention of
serial memory for nonverbal material.
The current findings can be interpreted as indicating that (1)
application of a verbal labelling strategy through congruent shadowing
significantly improves performance on both tasks, and (2) the recall of
hand movement sequences involves at least some spontaneous use of
verbal recoding, evidenced by the deleterious effect of incongruent
labelling. With regard to movement memory in general, this study
expands upon the findings of our previous research and demonstrates a
differential effect of verbal material of different content on memory
for movement sequences.
Importantly, these results expand our understanding of the cognitive
processes involved in KHMT and span-HMT performance. In conjunction
with the results of our previous study, and taking into account the
opposite effects of shadowing incongruent and congruent labels when
compared to baseline performance, it is inferred that at least some
verbal labelling is taking place during task performance. It is
possible that the utility of these verbal labels in aiding movement
sequence recall depends upon how well the participant employs them as a
recall strategy, and how well he/she can rehearse and subsequently
remember the labels and their order.
In the context of a full neuropsychological assessment, it may be
possible to obtain a more comprehensive explanation of the
individual's performance by administration of these tasks.
Individual differences in KHMT performance may reflect not only
individual differences in verbal memory capacities, if verbal labels
are recalled in order to correctly reproduce movement sequences, but
also the extent to which there is spontaneous use of verbal strategies
to aid movement sequence recall. Richardson and Barry (1985) reported that memory problems following mild
traumatic brain injury might reflect a reduced ability to adopt
effective strategies to support memory. Results of the current study
are consistent with the possibility that impairments in KHMT
performance reflect failure to adopt an effective verbal labelling
strategy for hand postures. Additionally, given the reported
equivalence of performance on the KHMT and the current
computerized-KHMT tasks, the computerized-KHMT could potentially be
used as an additional or alternative task. Currently, research is
underway into the sensitivity of the span-HMT to the effects of mild
traumatic brain injury during the acute and chronic phases
postinjury.
ACKNOWLEDGMENTS
We would like to thank Motohide Miyahara for his assistance and
helpful comments with regard to the rationale and design of this
experiment. We are also grateful for the thoughtful comments provided
by Fabrice Parmentier and two anonymous reviewers. Thanks go to Matt
Huitson and Herb Jurkiewicz for their assistance with task development.
This study was supported by an Australian Postgraduate Award provided
to K. Frencham.
