| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Motor Control Laboratory, Arizona State University, Tempe
b Department of Neuroscience and Brain Sciences Center, University of Minnesota, Veterans Affairs Medical Center, Minneapolis
George E. Stelmach, Motor Control Laboratory, P.O. Box 870404, Arizona State University, Tempe, AZ 85287-0404 E-mail: stelmach{at}asu.edu.
Decision Editor: Margie E. Lachman, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Birren 1974
and Salthouse 1985 hypothesized that all fundamental neural events become slower with advanced age for cognitive and motor functions, resulting in overall movement slowing in older adults. A substantial portion of research has shown older adults to be 30% to 70% slower than young adults on a variety of motor tasks, with pronounced slowing observed as task difficulty increases (Amrhein, Goggin, and Stelmach 1991; Bellgrove, Phillips, Bradshaw, and Gallucci 1998; Cooke, Brown, and Cunningham 1989; Goggin and Meeuwsen 1992; Pohl, Winstein, and Fisher 1996; Stelmach, Goggin and Amrhein 1988; Walker, Philbin, and Fisk 1997; Welford 1984). Older adults tend to move more slowly, but do not necessarily make more errors than young adults (Goggin and Meeuwsen 1992; Salthouse 1988). Goggin and Meeuwsen 1992 used a speed–accuracy task to assess spatial control in an aiming task by manipulating movement amplitude and target size. They found that older adults emphasized the later portion of the movement to maintain accuracy.
Research has described movement slowing in terms of kinematic parameters and has identified differences between young and older adults on features such as longer deceleration profiles (Bellgrove et al. 1998; Brown 1996; Cooke et al. 1989; Darling, Cooke, and Brown 1989; Goggin and Meeuwsen 1992; Pratt, Chasteen, and Abrams 1994) and lower peak velocity amplitudes (Bellgrove et al. 1998; Brown 1996; Cooke et al. 1989; Goggin and Meeuwsen 1992; Pratt et al. 1994). Cooke and colleagues 1989 studied kinematics of older adults' arm movements in which participants performed tracking movements to varying amplitudes. Both young and older adults increased movement durations and velocity amplitude as movement amplitude was increased. Young adults produced symmetric velocity profiles, whereas older adults showed lengthened deceleration curves. These researchers also reported that maximum velocities were significantly lower in older adults across all amplitudes and were more variable at the short amplitudes. Further, they observed that older adults showed hypometric movements in which they made "discrete submovements" to reach the target.
Velocity profiles can further be parsed using the movement optimization model proposed by Meyer, Abrams, Kornblum, Wright, and Smith 1988 to assess the underlying microstructure of the movement. This method demonstrated that adjustments to the microstructure occur when accuracy constraints are imposed and under heightened task difficulty. Research has also shown that older adults have a reduced ability to propel the limb near the target with the initial ballistic portion of their movement, consequently making more secondary, corrective submovements to reach the target (Bellgrove et al. 1998; Darling et al. 1989; Pratt et al. 1994; Seidler-Dobrin and Stelmach 1998; Walker et al. 1997). However, a systematic manipulation of target size and movement amplitude in the context of Fitts's law has not been well documented using the movement optimization model.
The present study was designed to assess movement slowing and variability in older adults to provide insight into how movement kinematics and microstructure are adjusted with a systematic manipulation of a speed–accuracy task (Fitts 1954). We sought to determine whether the impairments observed in the older adults are similar for changes in task difficulty as predicted by Fitts or whether they are related specifically to manipulations in target size or movement amplitude. Further, we sought to determine which kinematic parameter or parameters are most related to movement slowing in older adults and if they are influenced by task difficulty, target size, and movement amplitude. If kinematic and movement-parsing analyses reveal similar changes for target-size and movement-amplitude manipulations, this would suggest that a single deficit may cause movement slowing and increased variability in the older adults. However, if analyses reveal separate changes as target size and movement amplitude are manipulated, it would suggest that movement slowing and variability are not necessarily a result of a unitary deficit but are rather task-feature specific.
Thus, three specific hypotheses were assessed:
The present study furthers the understanding of contributions to movement slowness in older adults by the systematic manipulation of target size and movement amplitude in conjunction with kinematic and movement-parsing analyses techniques.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
) to exclude those with other neurological impairments such as dementia. We set the minimum score for inclusion at 28 out of 30. Furthermore, all participants filled out a health-history questionnaire to exclude those who had a history of stroke, arthritis, or other neurological or movement impairments. To assess functionality of movement, we gave all participants a 20-s tapping task. Older adults produced significantly fewer taps in 20 s (M = 27.47, SD = 1.92) compared with young adults (M = 33.47, SD = 5.93). Furthermore, older adults (M = 3.60 years, SD = 2.69) and young adults (M = 3.67 years, SD = 0.62) did not differ on years of education past the high school level.
Procedures
Participants made point-to-point aiming movements to different targets, which were presented on a computer screen. Target size and movement amplitude were manipulated to yield seven different indices-of-difficulty (ID) conditions (Fig. 1). Two IDs of 4 (4a, 4b) and two IDs of 5 (5a, 5b) were used to make comparisons as individual parameters changed (i.e. target size and movement amplitude were manipulated independently). The conditions were IDs 2, 3, 4, 5, and 6 (two each of IDs 4 and 5). Movement amplitudes and target widths were: ID 2 = 9.6 cm, 4.8 cm; ID 3 = 9.6 cm, 2.4 cm; ID 4a = 19.2 cm, 2.4 cm; ID 4b = 14.4 cm, 1.8 cm; ID 5a = 9.6 cm, 0.6 cm; ID 5b = 14.4 cm, 0.9 cm; and ID 6 = 19.2 cm, 0.6 cm. There were 22 blocks of 12 trials. The first contained two of each possible ID to allow the participant to become oriented with the task. Following this were 7 familiarization blocks of 12 trials, which were not analyzed. The last 14 blocks constituted the test phase, in which the first 7 blocks were randomly presented to the participants and the last 7 blocks were counterbalanced. Participants were instructed to move as fast and accurately as possible to the target after a computer-generated go stimulus (tone). The trial ended when participants stopped in the target.
|
) was used to control the experiment and collect data.
Data Analysis
The pen-tip data were filtered using a second-order dual pass (no phase lag) Butterworth filter (Winter 1990). A residual analysis was conducted to determine the appropriate cutoff frequency for the data (7 Hz). Velocity and acceleration were computed with a three-point finite difference derivative with endpoint padding to eliminate endpoint problems (Winter 1990). Normalized jerk score was computed as 
(1/2 
dt j2 (t) x duration5 / length2) to evaluate the smoothness of the movement. This variable has no units, because it was normalized for both amplitude and movement duration (Teulings, Contreras-Vidal, Stelmach, and Adler 1997). The optimal algorithm of Teasdale, Bard, Fleury, Young, and Proteau 1993 was used to determine movement onset from velocity profiles. The algorithm worked by locating the sample at which the velocity time series first exceeded 10% of its maximum value (Vmax). It then worked backward from this point and stopped at the first sample (S) less than or equal to Vmax/10-Vmax/100. The standard deviation of the series between Sample 1 and Sample S (SD) was then determined. The onset sample was from S stop, the first sample less than or equal to S-SD. The same algorithm in reverse was used for movement offset.
The end of the primary submovement was determined by the second zero crossing of acceleration profile. Each of the subsequent acceleration and deceleration pairs (two zero crossings) were recorded as a secondary submovement (Fig. 2).
|
Statistical Analyses
A multivariate analysis of variance with repeated measures was used to analyze the relevant subsets of data. The Geisser-Greenhouse corrected degrees of freedom were used when violations to sphericity occurred. The observed power was reported if it was less than 1.0 for each effect as was the effect size (
2 = ES), which is an estimation of the total variance explained by the treatment variation (Keppel 1991). Values ranged between 0 to 1.0 (.03 is a small effect size, .06 is a medium effect size, .15 is a large effect size; Cohen 1977). Age main effects as well as ID main effects and Age x ID interactions were reported. For target-size manipulations, ID changes were reported as target-size effects, and for movement-amplitude manipulations, ID effects were reported as movement-amplitude effects. Age main-effect means were pooled across ID, and ID main-effect means were pooled across age groups. For simplicity of data presentation, text and figures refer to a representative comparison for manipulations of target size and movement amplitude, but do not include all pairings. All data pairings are reported in Table 2 and Table 3 in their entirety for target-size and movement-amplitude pairings respectively.
|
|
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
Influence of Target Size (Table 2 )
Three combinations of IDs (4a-6, 4b-5b, 2-3-5a) were used to evaluate differences between young and older adults when movement amplitude was held constant and target size was manipulated. Thus, increases in ID were elicited by decreases in target size. The comparison between 2, 3, and 5a is highlighted (Fig. 3, Fig. 3, Fig. 3). Movement amplitude was fixed at 9.6 cm, and target sizes were 4.8, 2.4, and 0.6 cm respectively. The multivariate test indicated a significant effect of target size, F(12,17) = 30.23, p < .001, a significant effect of age, F(6,23) = 3.94, p < .01, P = .91, and a significant Age x Target Size interaction, F(12,17) = 2.97, p < .05, P = .89. Therefore, we analyzed univariate tests for each dependent variable to analyze the effects of age and decreases in target size on observed kinematics. Geisser-Greenhouse corrected degrees of freedom rounded to the nearest whole number were used when sphericity violations occurred in these comparisons. Both young and older adults produced slower movement times as target size decreased, (significant target size main effect), F(2,56) = 98.4, p < .001, ES = .68. Older adults were significantly slower than young adults at all target sizes (significant age main effect), F(1,28) = 15.1, p < .005, P = .96, ES = .14, and were differentially slower as target size decreased (significant Age x Target Size interaction), F(2,56) = 7.2, p < .005, P = .92, ES = .12 (Fig. 5). Furthermore, as target size decreased, amplitude of peak velocity decreased for both groups, (significant target size effect), F(2,56) = 37.8, p < .001, ES = .45). Older adults had significantly lower peak velocities across all target sizes (significant age main effect), F(1,28) = 15.8, p < .001, ES = .14, and were differentially affected at the larger target size (significant Age x Target Size interaction, F(2,56) = 3.3, p < .05, P = .61, ES = .05 (Fig. 5). Relative time to peak velocity decreased as target size decreased for both groups, (significant target size effect), F(2,56) = 128, p < .001, ES = .74). There was neither a significant age main effect (ES = 0) nor an Age x Target Size interaction (ES = 0) suggesting that the change in relative time to peak velocity was a function of change in target size solely (Fig. 5).
|
Influence of Movement Amplitude
Two combinations of IDs (3-4a, 5a-6) were designed to evaluate differences between young and older adults when target size was held constant and movement amplitude was changed. Thus, ID increases were elicited by increases in movement amplitude. The comparison between 3 and 4a is highlighted (Fig. 3, Fig. 3). Target size was held constant at 2.4 cm, and amplitude was increased from 9.6 to 19.2 cm. The multivariate test indicated a significant effect of movement amplitude, F(6,23) = 196.7, p < .001, a significant effect of age, F(6,23) = 3.9, p < .05, P = .76, and a significant Age x Movement Amplitude interaction, F(6,23) = 3, p < .05, P = .89. Therefore, we analyzed univariate tests for each dependent variable to analyze the effects of age and increases in movement amplitude on observed kinematics. Geisser-Greenhouse corrected degrees of freedom rounded to the nearest whole number were used when sphericity violations occurred in these comparisons. Both young and older adults were significantly slower as amplitude increased (significant movement amplitude main effect), F(1,28) = 34.5, p < .001, ES = .36. Older adults were significantly slower than young adults at both movement amplitude (significant age main effect), F(1,28) = 13.1, p < .005, P = .94, ES = .17), and were differentially slower as movement amplitude increased (significant Age x Movement Amplitude interaction), F(1,28) = 7.3, p < .05, P = .74, ES = .09 (Fig. 6). Amplitude of peak velocity increased as movement amplitude increased for both groups (significant movement amplitude main effect), F(1,28) = 139.8, p < .001, ES = .70. Older adults produced significantly lower peak velocities across movement amplitude compared with young adults (significant age main effect), F(1,28) = 14.8, p < .005, P = .96, ES = .19, and did not increase amplitude of peak velocity at the same rate as young adults as movement amplitude increased (significant Age x Movement Amplitude interaction), F(1,28) = 7.9, p < .01, P = .77, ES = .10 (Fig. 6). Furthermore, relative time to peak velocity decreased as movement amplitude increased for both groups (significant movement amplitude main effect), F(1,28) = 213.7, p < .001, ES = .78. There was not a significant age effect (ES = 0), however there was a significant Age x Movement Amplitude, F(1,28) = 6.2, p < .05, P = .67, ES = .08 (Fig. 6).
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Cooke et al. 1989; Goggin and Meeuwsen 1992; Pohl et al. 1996; Stelmach et al. 1988; Walker et al. 1997; Welford 1984).
Movement characteristics that differentiated older adults from young adults were peak velocity, relative time to peak velocity, number of secondary submovements, and normalized jerk scores. Peak velocity did not change systematically for either group, suggesting that it was not a major contributor to lengthened movement durations; however, relative time to peak velocity did change systematically for young and older adults. Although there were no age group differences, as ID increased, time to peak velocity decreased such that the majority of the time was spent in the deceleration phase as task difficulty increased (Billon, Bootsma, and Mottet 2000). Distance traveled in the primary submovement was shortened as a function of ID, but it did not interact with age. The number of secondary submovements and normalized jerk scores increased differentially between young and older adults as a function of ID. However, both these variables are a function of distance traveled in the primary submovement, which did not systematically differentiate young and older adults. For the collapsed data comparison, combining target-size and movement-amplitude manipulations created ID changes. Kinematic changes were not systematic in relation to task difficulty alone as would be expected if there is a global affect of information-processing slowing (Fitts 1954; Haaland, Harrington, and Grice 1993; Welford 1984). To get a better understanding of how each of these variables influenced performance, we partitioned data into subsets in which increases in ID were achieved by systematically decreasing target size or increasing movement amplitude.
Influence of Target Size
To determine the influence of target size, we held movement amplitude constant while target-size was manipulated. Older adults showed greater increases in movement time compared with young adults as target size decreased (Fig. 5). Peak velocity, relative distance traveled in the primary submovement, number of secondary submovements, and normalized jerk scores all exhibited significant Age x Target Size interactions, suggesting that older adults behave differently than young adults for decreasing target sizes (Fig. 5). The movement durations of older adults were differentially slower at the smallest target size (highest ID) at which the lowest peak velocities were observed for both young and older adults. Pearson r correlations of the differences across changes in target size between movement time and particular kinematic variables were squared to interpret as variance accounted for in slower movement times of older adults. Peak velocity accounted for 22% (r = -.47) of variance in movement time of young adults but only 1% (r = .10) of variance in older adults.
Relative distance traveled in the primary submovement showed differential effects between groups as target size was decreased such that older adults covered substantially less distance in the primary submovement across the three target sizes, specifically between IDs 3 and 5a, the older adults shortened the relative primary submovement distance to a greater extent than the young adults. Relative distance traveled in primary submovement accounted for 23% (r = -.48) of variance in movement time for older adults, but only 2% (r = -.13) for young adults. Thus, relative distance traveled in primary submovement determined a large portion of slower movement times in older adults when target size was decreased.
Furthermore, increasing accuracy constraints caused the older adults to alter the microstructure of their movements such that more secondary, corrective movements were needed to achieve a target, thus contributing to differentially slower movement times observed as target size was decreased (Bellgrove et al. 1998; Darling et al. 1989; Pratt et al. 1994; Seidler-Dobrin and Stelmach 1998; Walker et al. 1997). In the present study, secondary submovements accounted for 76% (r = .87) of the variance in movement times for older adults, whereas for young adults they only accounted for 23% (r = .48) of the variance. These results indicate that accuracy constraints influence movement substructure in older adults and lead to increased variability in older adults. Pratt and colleagues 1994 observed shortened primary submovements in older adults compared with young adults. They reported that older adults were unable to lengthen the distance traveled in the primary submovement with extensive practice. The present data furthers these findings and establishes that shortened primary submovements were a substantial contributor to movement slowing in older adults when target size was decreased, placing an emphasis on accuracy.
Influence of Movement Amplitude
To determine the influence of movement amplitude, target size was held constant while movement amplitude was manipulated. Although in both groups movement time was increased by movement amplitude increases, older adults again were differentially slower than young adults at the longer movement amplitudes (Fig. 6). Several characteristics exhibited Age x Movement Amplitude interactions including peak velocity, relative time to peak velocity, secondary submovements, and normalized jerk scores (Fig. 6). Peak velocity increased for both groups as movement amplitude increased, however older adults did not increase their peak velocity similarly to young controls. Pearson r correlations of the differences across changes in movement amplitude between movement time and particular kinematic variables were squared to interpret as variance accounted for in slower movement times of older adults. Peak velocity accounted for 20% (r = -.45) of variance in movement time for young adults, but only 1% (r = -.10) for older adults. This suggests that older adults did not scale their velocity as movement amplitude was increased in the same manner as young adults did. However, differential increases in peak-velocity amplitude as a function of movement amplitude increases led to differentially slower movement times, and thus the slower movement times observed in older adults in this comparison.
There was an overall slowing of older adults compared with young adults that may be explained by the shorter relative distance traveled in the primary submovement (Pratt et al. 1994; Seidler-Dobrin and Stelmach 1998). However, of interest in this experiment was the specific contribution to the increased rate of slowing observed at the longer movement amplitudes. This difference was explained by the inability to ramp force as observed by differentially lower peak velocities across movement amplitudes. The number of secondary submovements and normalized jerk scores showed significant interactions and may be a consequence of lower peak velocities resulting in slow movements with multiple inflections and less smooth movements. From this comparison it was determined that the inability of older adults to increase peak velocity at the same rate as young adults primarily contributed to slower movement times observed in older adults as movement amplitude increased and higher movement variability appeared. The inability to ramp velocity may be a consequence of neuromuscular changes (Galganski, Fuglevand, and Enoka 1993; Welford 1984).
Summary
Older adults respond to changes in task difficulty by making different adjustments in response to accuracy and amplitude constraints. When the data were separated into influences of target size and of movement amplitude, the kinematic characteristics that differentiated young and older adults were dependent on the task constraints imposed. Although other parameters changed, relative distance traveled in the primary submovement was the primary source of movement slowing when target size was decreased, whereas differential scaling of peak velocity was the primary contributor when movement amplitude was increased.
These findings suggest that older adults are unable to effectively propel their limb to the target in a single step, which results in multiple secondary, corrective submovements, causing their limb movement to be less smooth and slower. The exact cause for having a reduced ability to propel the limb to a target in a single step is not understood. Some have thought it may be related to a central planning deficit but this seems unlikely (Amrhein et al. 1991; Goggin and Meeuwsen 1992; Haaland, Harrington, and Grice 1993; Seidler-Dobrin and Stelmach 1998; Stelmach et al. 1988; Welford 1984). If differential effects were a result of central mechanisms it might be expected that all task conditions would show similar impairments, which was not the case in this study.
Another possibility may be that older adults could not produce the necessary forces to complete the desired movement resulting in lower peak velocities and shorter relative submovement distances. Galganski and colleagues 1993 reported that age affected force variability at low target forces. This consequently led to the inability to ramp forces accurately and/or efficiently. The present data for manipulations in movement amplitude could be explained by this interpretation. However, Walker and colleagues 1997 have previously shown that when accuracy was not a factor in an aiming movement, older adults produced movement velocities that were very similar to those of the young adults. The combination of the findings from Galganski and colleagues and Walker and colleagues suggests that older adults are able to produce forces needed to propel the limb to the target, however the ability to modulate forces is constrained by the presence of terminal accuracy requirements. This decrement in force modulation may result from observed problems in the timing and phasing of muscle activation (Darling et al. 1989). A force modulation deficit may lead to the inability to propel the limb to the target effectively; however, similar kinematic outcomes would be expected in the task manipulations if this were the only mechanism.
Older adults have been shown to produce normal agonist muscle bursts, but abnormal phasic antagonist muscle bursts during the deceleration phase of the movement resulting in increased cocontraction (Darling et al. 1989). Seidler-Dobrin, He, and Stelmach 1998 have shown that older adults have considerably more cocontraction during point-to-point movements compared with young adults. Cocontraction may help reduce the variability and speed of the movement allowing participants to have more control of the terminal phase of the movement, thus leading to lower velocities and shorter distances traveled in the initial phase of their movement. Furthermore, opposing muscles may act to reduce the torque about the joints, thus restricting the ability to propel the limb accurately towards the target. Again, this may lead to the outcomes observed in this study, but further research is necessary to confirm.
The present study has established that older adults are affected differently than young adults during a Fitts's law task, suggesting slowing in older adults' aiming movements is not necessarily a consequence of a unitary factor. Components of movement, measured by kinematic analyses, contributed differently to movement slowing when target size and movement amplitude were manipulated separately.
|
Acknowledgments |
|---|
Portions of these data were presented at the 1999 Neuroscience Conference, Miami, Florida.
Received for publication January 10, 2001. Accepted for publication May 31, 2001.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. J. Goble and S. H. Brown Upper Limb Asymmetries in the Matching of Proprioceptive Versus Visual Targets J Neurophysiol, June 1, 2008; 99(6): 3063 - 3074. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. O. Weiss, L. P. Fried, and K. Bandeen-Roche Exploring the Hierarchy of Mobility Performance in High-Functioning Older Women J. Gerontol. A Biol. Sci. Med. Sci., February 1, 2007; 62(2): 167 - 173. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
