Neuroscience and Biobehavioral Reviews 26 (2002) 761–767 www.elsevier.com/locate/neubiorev
Sensorimotor adaptation in young and elderly humans Otmar Bocka,*, Stefan Schneiderb b
a Department of Physiology, German Sport University, D-50927 Ko¨ln, Germany Department of Cardiovascular Research, German Sport University, D-50927 Ko¨ln, Germany
Received 26 February 2002; accepted 16 August 2002
Abstract Our brain’s capacity for adaptation allows us to interact meaningfully with an ever-changing environment. Experimental evidence suggests that the time course of sensorimotor adaptation is preserved or only moderately degraded in old age, and that seniors benefit from a previous adaptive experience even more than younger subjects. However, experimental evidence suggests that sensorimotor adaptation seems to be associated with a higher computational load in the elderly. We discuss two possible explanations for this pattern of findings: Older adults may take longer to consolidate newly gained information into long-term motor memory, or they may have problems to utilize supplementary (e.g. cognitive) strategies. In any case, the age-related deficits were relatively mild. If these deficits are related to an increased computational load, it should be possible to reduce them by extended practice on adaptation tasks. q 2002 Elsevier Science Ltd. All rights reserved. Keywords: Aging; Sensorimotor adaptation; Internal models; Strategic processing; Consolidation; Sequential adaptation; Workload; Manual tracking
1. Introduction The main task of our sensorimotor system is to integrate information from multiple sensory channels about the world around us, and to use this information to plan and execute meaningful motor responses. This task cannot be accomplished by rigid, non-adaptable processing algorithms. Since the neurobiological conditions and the physics of our body change throughout life, sensorimotor processing rules (or mechanisms) must change accordingly. For example, our sensorimotor system must adapt to visual adjustments produced by new prescription glasses, to the ‘feel’ of a new tool, the driving characteristics of a new car, and to age-related changes in the size, proportions, and biomechanical properties of our body [38]. The brain’s adaptive capacity is therefore a prerequisite for a continuing competence in everyday functioning. There is cumulating evidence suggesting that various aspects of brain functions gradually decline even during normal aging [6,27]. Thus, cognitive functions such as verbal memory, reasoning, and spatial abilities were found to decline (review in Ref. [32]), as were sensorimotor * Corresponding author. Tel.: þ49-221-498-2370; fax: þ 49-221-4973531. E-mail address:
[email protected] (O. Bock).
functions such as stimulus discrimination, simple reaction time, and the utilization of pre-cues (review in Refs. [7,15]). Given these age-related declines in elderly people’s cognitive and sensorimotor performance, it would not be surprising if the capacity for sensorimotor adaptation also decline in old age. Ever since the pioneering work of Stratton [35], sensorimotor adaptation has been studied by exposing subjects to visual distortions that are produced by lenses and prisms, or to mechanical distortions that are produced by external force fields. When naı¨ve subjects first experience such a manipulation, their motor behavior becomes grossly inadequate, but it gradually normalizes during prolonged practice under conditions of the distortions. This indicates that the processing rules of the sensorimotor system are progressively adjusted, in order to take into account the changes in sensory inputs due to the manipulations. Put differently, the brain’s ‘internal model’ of the body and environment is progressively modified such as to reflect the distortion [37]. It should be pointed out that the term ‘internal model’ (or internal representation) does not imply the existence of a specific area in the brain, where the mechanical properties of the body and its surround are coded in a static, topographical fashion. Rather, recent neuroimaging studies showed that adaptation-related neural activity is found in a
0149-7634/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 9 - 7 6 3 4 ( 0 2 ) 0 0 0 6 3 - 5
762
O. Bock, S. Schneider / Neuroscience and Biobehavioral Reviews 26 (2002) 761–767
widely distributed cortico-thalamo-cerebellar network [14, 18,21], and the exact contributions of this network and its subdivisions towards adaptive success still need to be explored. Given the present stage of knowledge, therefore, the term ‘internal model’ should be considered as a metaphor, which is useful to communicate experimental findings, but has no explanatory value beyond the alternative term ‘sensorimotor processing rule’: Models as well as processing rules can be modified in the presence of visual or mechanical distortions, and they both can yield performance characteristics such as those described in Section 2.
2. Characteristics of sensorimotor adaptation The adaptive capacity of the sensorimotor system has been investigated in sequential-adaptation studies, where subjects are repeatedly exposed to visual and/or mechanical distortions of sensory inputs. As one example, Fig. 1 illustrates the performance in a manual tracking task, when visual feedback about one’s own hand is distorted in various ways [4]. The leftmost part of Fig. 1 shows that when the visual feedback is veridical, the tracking error is low. The middle part shows that when the visual feedback is left – right reversed (i.e. targets moving towards the left required a hand movement to the right, and vice versa) while the up – down component of movement remains veridical, the tracking error increases abruptly at the onset of visual
Fig. 1. Examples of tracking performance during sensorimotor adaptation in three subjects. Inset: experimental setup, showing that visual targets were viewed through a tilted mirror, such that they appeared to move in a horizontal plane; the hand was not visible, but its position was registered and displayed along with the target, such that the feedback was either veridical or distorted. Main figure: root mean square tracking error (RMSE) for each tracking episode of 50 s duration. With veridical feedback (left part), the tracking error is small. During left –right reversal (middle part), the error is initially high, and then gradually decays. In a second experimental session (right part), the error is low when the left-right reversal is maintained (open squares), increases somewhat when an up – down reversal is added (filled squares), and increases substantially when an up– down reversal replaces the left–right reversal (crosses). Note that RMSE represents a global performance score, which can’t distinguish between various possible sources of tracking errors (e.g. feedforward, feedback, attention).
distortion, but gradually returns towards baseline. Thus, an initial performance decrement is followed by a gradual adaptive improvement. The right part represents data from separate experimental sessions, capturing performance at minutes, days, or weeks after the first exposure to the manipulation. Depending on the type of distortion, the tracking error follows distinctly different curves. Taken together, available evidence indicates that second-session performance falls into one of four categories: Retention. When the distortion in the second session is exactly the same as in the first, subjects typically continue to perform at the level they have achieved at the end of the first session (e.g. open squares in Fig. 1), even if the inter-session interval lasts as long as several weeks. This indicates that an adaptively modified internal model can remain for a long time in motor memory, which can be reactivated later when needed. Findings of this type are ubiquitous in literature. Facilitation. When the distortions are different across sessions but are still of the same type (e.g. both visual or both mechanical), second-session performance depends crucially on whether the distortions are compatible or conflicting. In literature, a compatible second distortion is commonly achieved by scaling the first [1,22] or by adding an orthogonal component to it (filled squares in Fig. 1). In either case, the second distortion led to an initial performance decrement with subsequent adaptive improvement. Most interestingly, the initial decrement was substantially smaller than in a control group that participated in the second session only. This phenomenon holds even when the two sessions are separated by several weeks [4]. In conclusion, it seems that adaptation is facilitated when it is based on an already available, adaptively modified internal model. Such a facilitation effect does not necessarily generalize to all task types, however. When adapting to visual rotation, subjects benefit from a previous exposure to a different angle of rotation only if the two angles do not stride a critical value of 90 – 120 degrees [2]. This suggests that at least under some task conditions it is more efficient to start ‘from scratch’ rather than to build upon a stored model. Interference. Two distortions can be of the same type but in mutual conflict with each other, e.g., when the second is an inverted version of the first [20] or is orthogonal to it (crosses in Fig. 1). In these cases, second-session performance is substantially worse than in naive subjects. This suggests that an adaptively modified internal model resists further change if it would revoke the existent modification. Such a property could be useful in preventing the internal model from tracking relatively short-lived fluctuations in the external world, for instance. Interestingly, this resistance effect lasts up to about 5 h in the case of mechanical manipulations [33], but days or weeks with visual distortions [4,20]. This seems to suggest that visual and mechanical distortions address different components of the internal model, which exhibit distinct temporal characteristics.
O. Bock, S. Schneider / Neuroscience and Biobehavioral Reviews 26 (2002) 761–767
Neutrality. If the distortions are of a different type across sessions (i.e. one visual and one mechanical), secondsession performance is similar to that of naive subjects [13, 20]. Thus, no interference is observed, even though the first modification must be revoked in order to implement the second one. This suggests again that visual and mechanical distortions are treated by different components of the internal model. A recent study [36] indicated that neutrality is not related to the visual versus mechanical nature of the distortion per se; rather, it is related to the fact that visual distortions typically depend on hand position, and mechanical distortions on hand velocity. Neutrality no longer applies if both distortions are coded in terms of hand position. Taken together, research on sequential-adaptation suggests that internal models can be modified and stored for a prolonged period. Subsequent modifications after the initial adaptation phase come with savings when they are complementary, but with extra cost when they are incompatible, and remain cost-neutral when they address different model components. In other studies, the computational demand of sensorimotor adaptation was gauged using the dual-task paradigm [26]. This approach is based on the view that two concurrently executed tasks compete for a limited pool of computational resources in the brain: When the computational demand of one task increases, between-task resource competition increases, and subsequently the performance on one or both tasks suffers. It was found that subjects adapt less when they have to perform a concurrent task, such as mental arithmetic [19,28], and that the decrement is more pronounced with the more difficult concurrent tasks [29]. These findings confirm that adaptation is computationally demanding, and therefore becomes less effective when computational resources must be shared with another task. Computational resources are thought to consist of several distinct categories [26]. Some of them may be required earlier, and others later during the adaptive process, since it is thought that sensorimotor modifications proceed in stages, each with different computational requirements.1 It has therefore been suggested that the resource demand of sensorimotor adaptation should be probed by administering repeatedly a whole battery of different concurrent tasks, each aimed at a different resource [16]. The first study using this approach dealt with the acquisition of a complex human-operator skill, which is often considered as a process similar to sensorimotor adaptation [23]. As predicted, some concurrent tasks affected the skill earlier and others somewhat later in training. However, the interpretation of 1
According to the stage model of Fitts [12], subjects gradually achieve an understanding of the new task requirements during an initial verbalcognitive stage, establish the necessary motor programs with the help of sensory feedback during the subsequent sensorimotor stage, and fine-tune their performance during a final automation stage.
763
Fig. 2. Example of tracking performance of one subject during sensorimotor adaptation under single-task (filled symbols), and dual-task (open symbols) conditions. Each symbol represents the RMSE during a 30 s tracking episode. The difference between single- and dual-task performance is small for tracking under veridical feedback (left part), but increases substantially under reversed visual feedback (right part). The solid line is an exponential fit to single-task data, which helps to visualize the dual-task decrement.
these results is encumbered by the fact that acquisition was probed only on two occasions, both scheduled very early in the acquisition process. Fig. 2 shows data from a more recent dual-task study [10], where subjects tracked a moving target before and during adaptation to vision reversal. The tracking task was performed either alone (filled symbols), or concurrently with a battery of concurrent tasks (open symbols). The concurrent tasks were designed such as to address different resource categories.2 The results show that before adaptation the concurrent tasks had little effect on tracking, but they had a considerable impact during adaptation. A detailed analysis revealed that an attention-demanding task had its highest impact on tracking immediately at adaptation onset, a task requiring substantial visuo-spatial processing somewhat later, and a task with high load on movement preparation still later during the adaptation process. This outcome fully confirms the above view of multiple adaptation stages with different computational demands. Moreover, the observed order of task demand first attentional, then visuo-spatial, and then response-related resources agrees well with the nature of the three postulated stages proposed by the above stage model [12].
3. Aging and adaptive capacity Several studies compared the adaptive improvements of young and elderly persons during exposure to a visual distortion. In one study [3], eleven young (20 – 30 years of age), and eleven community-dwelling older (60 –70 years of age) adults were trained to track an unpredictably moving 2
One of the concurrent tasks was a conventional four-choice reaction time task, which served as a control. The other tasks were variants which required more attention (by using precues), more visuo-spatial processing (by using spatial stimulus-response transformations), or more motor preparation (by using complex responses). For details see Ref. [10].
764
O. Bock, S. Schneider / Neuroscience and Biobehavioral Reviews 26 (2002) 761–767
visual target first under the condition of left – right reversed vision. One to two days later, the subjects performed under the condition of left – right plus up –down reversed vision. Previously, this distortion sequence has yielded facilitation in Fig. 1. The resultant tracking errors of both age groups are plotted in Fig. 3. The data show that between-subject tracking variability was substantially higher in the older group, which is confirmed by the Levene’s test for the homogeneity of variances ( p , 0.05 or p , 0.01 in 40 out of 65 learning episodes, respectively). The increased variability can be partly attributed to the larger differences between individual subjects across all episodes, since the Levene’s test remained significant ( p , 0.01) when applied to cross-episode means. Indeed, an inspection of the individual data revealed that the tracking error scores of some elderly subjects remained within the range of the younger group throughout all episodes, while the error scores of the others were consistently higher. A similar agerelated increase of variability has been observed in many other studies, dealing not only with sensorimotor performance. This effect has been interpreted as an evidence for a differential rate for aging across individuals (see overview in Ref. [34]). However, the reasons for such a divergence between chronological and biological age are still largely unknown. The age-related increase of variability is not exclusively due to a larger spread between good and poor performers. When between-subject differences are numerically eliminated by subtracting each subject’s cross-episode mean from the data in Fig. 3, the Levene’s test remained largely significant ( p , 0.05 or p , 0.01 in 19 out of 65 episodes). Thus, some of the higher variability appears related to within-subject, episode-to-episode variability. Another observation in Fig. 3 is the substantial difference between mean tracking errors in the two age groups: Elderly individuals had substantially higher errors across all episodes. This result corresponds well with previous research, since the tracking task required accuracy as well
Fig. 3. Tracking performance of young (triangles) and elderly (circles) subjects during sensorimotor adaptation. Data points are across-subjects means of the RMSE in each 50 s tracking episode, and error bars are the pertinent standard deviations. In a first session, subject tracked under veridical visual feedback (left part), and then under left –right reversal (middle part). In a second session 1–2 days later, an up –down reversal was added (right part).
as speed, and elderly people are known to be impaired on speeded tasks much more so than on tasks that focus on accuracy only [9,25]. Importantly, however, higher tracking errors do not necessarily imply that the adaptive capacity declines with aging as well. Rather, it is evident from Fig. 3 that the adaptive improvement, i.e. the decrease of tracking error from the first to the last episode under a given distortion, is comparable for both age groups: the two adaptation curves seem to run nearly parallel. These observations are confirmed by repeated-measures analyses of variance, with Huynh –Feldt adjustments for unequal variances [17]. In either session, the effect of age was highly significant (F(1,20) ¼ 23.1 and 16.8, respectively; p , 0.001), and so was the effect of episode (F(29,580) ¼ 33.7 and 11.7, respectively; p , 0.001). However, the interaction term was not significant (F(29,580) ¼ 0.7 and 0.6, respectively; p . 0.05), which indicated that the time-course of adaptation did not differ reliably between age groups. This lack of age-related differences in adaptive improvement cannot be due to the large between-subject variability in old adults, which is already taken into account by the repeated-measures analysis design. One might still suspect that the old adults who are poor trackers are also poor adapters. If so, adaptive improvement should be negatively correlated with overall performance. However, this alternative interpretation is not supported statistically. Adaptive improvement, expressed as difference between the first and last episode under a given distortion, was not significantly correlated with subjects’ cross-episode means (r ¼ 2 0.19 to þ 0.45, depending on session and age group, all p . 0.05). The data in Fig. 3 therefore do not support the view that adaptive improvement and tracking performance are closely linked. Clearly, the statistical power of the above analysis is restricted by a small sample size; it should be noted, however, that a dissociation between impaired motor performance and preserved motor learning has also been observed by others [8]. Findings in literature on the adaptive capacity of older people are not unequivocal. Results from some studies are in agreement with Fig. 3, and report that sensorimotor adaptation is largely preserved in old age [9,31]. However, other studies found that the elderly adapt less [5,25], or at least more slowly [11], than younger subjects. An evaluation of the methods used in the various studies yields no support for the view that the discrepant findings are due to differences between the respective age ranges, types of visual distortion, or classes of arm movement executed during visual distortion. This is most obvious in two studies which both investigated the throwing of objects at a vertical wall 3 m ahead under the condition of lateral visual shift. In spite of their methodological similarities, one study found that adaptation is substantially slower in older people [11], but the other yielded no age-related deficits [31]. One possible explanation of the above discrepancies is based on the length of resting breaks between individual
O. Bock, S. Schneider / Neuroscience and Biobehavioral Reviews 26 (2002) 761–767
trials. Pause length was not a controlled variable in most of the studies, even though pauses may play an important role for adaptation, by allowing a consolidation of the acquired knowledge into long-term memory. Indeed, it has been shown that longer resting breaks accelerate the acquisition of motor skills, and that this effect is more pronounced in younger subjects [39]. The above discrepancies could therefore be due to pause length differences between studies and/or age groups. An alternative interpretation of the discrepant findings is related to the widely accepted view, that adaptation curves, such as those in Fig. 3, actually may reflect several distinct processes. On of them is ‘true’ adaptation, achieved by progressive adjustments to the internal model (also called ‘spatial alignment’ or ‘perceptual-motor recalibration’). However, improvements may also result from movement corrections based on sensory feedback, cognitive decisions to aim past the perceived target position, and other so-called ‘strategic processes’ [25,30]. It is thought that the contributions of model adjustments and strategic processes can be separated by investigating the aftereffects following removal of the distortion. Given that only the effects of model adjustments are stored in motor memory, only they will manifest as aftereffects [25,30]. Evaluating data on effects and aftereffects of visual distortions in different age groups in light of these considerations, an intriguing picture emerges. One study found that old people’s adaptation slowed down, but was not reduced in magnitude, while aftereffects were even more pronounced in this age group [11]. Another study reported that adaptation magnitude was smaller in elderly subjects, but aftereffects were as strong as in younger ones [25]. Both studies therefore agree that relative to their performance during exposure, older participants had stronger aftereffects than younger ones. Such an outcome suggests that the two age groups seem not to differ much in their capacity for model adjustment. Rather, the superior adaptive improvement of young participants seems mainly to result from more powerful strategic processes. A third study provides further support for this view [31], finding no age-related differences, neither for adaptive improvement nor for aftereffects, as expected under experimental conditions where strategic processes played a minor role. In short, one could hypothesize that studies encouraging strategic processes by explicit instructions or circumstantial factors will yield age-dependent adaptive improvement, while experimental conditions focusing more on the stage of model adjustments will not. These two interpretations are still tentative, future research is needed to disentangle the effects of strategic processing, inter-trial pauses, and other factors, on age-related differences in sensorimotor adaptation. Finally, the data shown in Fig. 3 indicate that exposure to a visual distortion led to an initial increase of tracking error in both sessions, but this increase was much more pronounced in the second one. Indeed, the difference
765
between sessions is highly significant (F(1,20) ¼ 141.8; p , 0.001), which confirms earlier findings of facilitation during sequential adaptation (Section 2). The effect of age on the initial increase in errors was not significant (F(1,20) ¼ 0.2; p . 0.05), but the age x session interaction reached significance level (F(1,20) ¼ 5.7; p , 0.05), which suggests that facilitation might have been even more pronounced in elderly subjects. At a first glance, it might seem perplexing that older people benefited more than young adults from previous exposure to a compatible distortion. However, the explanation could well be the same as that used to explain the stronger aftereffects in that age group (see above). Specifically, if older people are affected more by modifications of the internal model than by strategic adjustments, any long-term phenomena should be more pronounced in this age group, including aftereffects and facilitation. As an alternative interpretation, it is conceivable that older subjects filled the 1 – 2 day pause between sessions with different activities than the younger group: they may have spent more time thinking about the experiment and discussing it with their peers, thus achieving more mental practice, and a higher degree of memory consolidation. If this is the case, the data in Fig. 3 would not reflect a stronger facilitation in seniors, but rather their ability to ‘catch up’ with younger subjects during the break between sessions. This interpretation once again underlines the need to investigate the potential importance of breaks for sensorimotor adaptation.
4. Aging and the computational demand of sensorimotor adaptation The observation that adaptive capacity may be largely preserved in older people (Fig. 3) does not necessarily imply that adaptation is as effortless as it is in younger persons. Instead, older people may require a higher proportion of their available computational resources in order to achieve the same adaptive improvement. This possibility can be investigated by the dual-task approach. As an example, Fig. 4 shows the data of thirteen younger (22 –32 years of age) and thirteen older (60 – 70 years of age) subjects, who performed the tracking task either alone (i.e., Single-task condition, S), or concurrently with a four-choice manual reaction time task (i.e., Dual-task condition, D). Single- and dual-task episodes varied within each block in the order SDDS. The leftmost part of Fig. 4 illustrates that under normal visual feedback, detrimental effects of the concurrent task on tracking performance were stronger in older than in younger subjects. Indeed, the dual-task tracking decrement (i.e., dual-task cost as computed by D – S performance) under normal vision differed significantly between age groups (F(1,24) ¼ 7.5; p , 0.05). This outcome is in accordance with a large number of studies documenting higher dual-task decrements in the elderly.
766
O. Bock, S. Schneider / Neuroscience and Biobehavioral Reviews 26 (2002) 761–767
rather than a higher computational demand [26]. If so, the higher tracking cost should also be reflected by a correspondingly better performance on the concurrenttask. This, however, was not the case: in the concurrenttask data, the age x block interaction was not significant (F(7,168) ¼ 0.4; p . 0.05). These data, therefore, is suggestive of the notion that sensorimotor adaptation was indeed computationally more demanding for our older subjects. More future research is needed to confirm this notion. Fig. 4. Tracking performance of young (triangles) and elderly (circles) subjects during sensorimotor adaptation in a dual-task study. In each block, single- and dual-task tracking episodes of 35 s duration were administered in the order SDDS. Data from both single-task episodes were then averaged across subjects (open symbols), as were those from both dual-task episodes (filled symbols). The difference between open and filled symbols in each age group therefore represents the dual-task decrement. Subjects tracked first under veridical (left part), and then under reversed vision (right part).
Typically, this age-related impairment is attributed to generalized slowing of neural processing, difficulties to divide attention between multiple tasks, or problems to combine two motor responses into an integrative action (for overview see [24]). Whichever interpretation one takes, the crucial question in the present context concerns the effects of sensorimotor adaptation on the dual-task tracking cost. If that cost increases during sensorimotor adaptation more in elderly than in younger subjects, such evidence would indicate that old people may need to allocate more of their computational resources for sensorimotor adaptation task. The data in the right part of Fig. 4 illustrate that for both groups the dual-task cost became larger at the onset of adaptation, and dropped again later during exposure; this confirms earlier observations on the increase of computational demand during adaptation ([9,17,26,27]; see also Fig. 2). More importantly, the data presented in Fig. 4 also show that the increase of dual-task cost during adaptation was more pronounced in the older than in the younger subject group. Both observations are confirmed by repeated-measures analyses of variance, using the multivariate approach to account for unequal variances: The dual-task tracking cost depended significantly ( p , 0.01) not only on age (F(1,24) ¼ 17.3) and block (RaoR(7,18) ¼ 18.0), but also on the age x block interaction (RaoR(7,18) ¼ 4.6). The latter finding is in accordance with the view that during the time-course of adaptation, old people allocate more computational resources to the task than do young people. It should be noted, however, that dual-task data cannot be interpreted conclusively when only one of the two tasks is taken into account, due to potential performance trade-off between the two tasks. For example, one could argue that during adaptation, older subjects shifted their priority away from tracking and towards the concurrent task, and that the above interaction reflects this priority reassignment
5. Conclusions Available experimental data suggest that aging might have a differential influence on sensorimotor adaptation: The time course of adaptation is preserved or moderately degraded in old age, while the benefits from a previous adaptive experience are even more pronounced in elderly than in younger subjects. We propose two interpretations for these findings. First, older people may need a longer time to consolidate newly gained information into long-term motor memory. If so, their performance scores during exposure to a distortion may be impaired, but they catch up during prolonged resting breaks between successive exposures. Such a consolidation deficit could be due to generalized slowing [7,32], and/or to a specific age-related problem of consolidation. Further studies that systematically vary the duration of resting breaks are needed to investigate this issue. Second, old people may be able to modify their internal models at least as efficiently as young people, but their ability to supplement model adjustments with strategic processes may be degraded. The latter deficit has been linked to a higher computational burden of adaptation [25], and indeed, there is initial evidence that adaptation is more resource demanding in old age (Fig. 4). The question remains open, however, whether such an increase of the computational demand reflects the effort to maintain at least a low level of strategic processing, to preserve a high level of adaptive adjustment, or both. Future research should address these questions by quantifying, for a given adaptive process, the interrelation between model adjustments, strategic processes, computational load, and age.
Acknowledgements This review was supported by DLR-grant 50WB9547 and DFG-grant BO 649/8 to the first author. Responsibility for the content rests with the authors. We would like to thank three anonymous reviewers for their insightful comments on previous versions of this paper.
O. Bock, S. Schneider / Neuroscience and Biobehavioral Reviews 26 (2002) 761–767
References [1] Abeele S, Bock O. Mechanisms for sensorimotor adaptation to rotated visual input. Exp Brain Res 2001;139:248 –53. [2] Abeele S, Bock O. Sensorimotor adaptation to rotated visual input: different mechanisms for small versus large rotations. Exp Brain Res 2001;140:407 –10. [3] Bock O, Schneider S. Acquisition of a sensorimotor skill in younger and older adults. Acta Physiol Pharmacol Bulg 2001;26:89 –92. [4] Bock O, Schneider S, Bloomberg J. Conditions for interference versus facilitation during sequential sensorimotor adaptation. Exp Brain Res 2001;138:359 –65. [5] Brown S. Control of simple arm movements in the Elderly. In: Ferrandez A-M, Teasdale N, editors. Changes in sensory motor behavior in aging, Amsterdam: Elsevier; 1996. p. 27–52. [6] Cabeza R. Functional neuroimaging of cognitive aging. In: Cabezza R, Kingstone A, editors. Handbook of functional neuroimaging of cognition, Cambridge: MIT Press; 2001. [7] Cerella J. Aging and information-processing rate. In: Birren JE, Schaie KW, editors. Handbook of the psychology of aging, San Diego: Hartcourt; 1990. p. 210–21. [8] Durkin M, Prescott L, Furchtgott E, Cantor J, Powell D. Performance but not acquisition of skill learning is severely impaired in the elderly. Arch Gerontol Geriatr 1995;20:167–83. [9] Etnier JL, Landers DM. Motor Performance and motor learning as a function of age and fitness. R Q Exercise Sports 1998;69:136–46. [10] Eversheim U, Bock O. Evidence for processing stages in skill acquisition: a dual-task study. Learn Mem 2001;8:183– 9. [11] Ferna´ndez-Ruiz J, Hall C, Vergara P, Dı´az R. Prism adaptation in normal aging: slower adaptation rate and larger aftereffect. Cogn Brain Res 2000;9:223– 6. [12] Fitts PM. Perceptual-motor skill learning. In: Melton AW, editor. Categories of human learning, New York: Academic Press; 1964. p. 243–85. [13] Flanagan JR, Nakano E, Imamizu H, Osu R, Yoshioka T, Kawato M. Composition and decomposition of internal models in motor learning under altered kinematic and dynamic environments. J Neurosci 1999; RC34:1– 5. [14] Ghilardi M-F, Ghez C, Dhawan V, Moeller J, Mentis M, Nakamura T, Antonini A, Eidelberg D. Patterns of regional brain activation associated with different forms of motor learning. Brain Res 2000; 871:127–45. [15] Hale S, Myerson J, Wagstaff D. General slowing of non-verbal information processing: evidence for a power law. J Gerontol 1987; 42:131–6. [16] Heuer H. Motor learning as a process of structural constriction and displacement. In: Prinz W, Sanders AF, editors. Cognition and motor processes, Berlin: Springer; 1984. p. 295–305. [17] Huynh H, Feldt L. Conditions under which mean square ratios in repeated measures designs have exact F-distributions. J Am Stat Assoc 1970;65:1582– 9. [18] Imamizu H, Miyauchi S, Tamada T, Sasaki Y, Takino R, Pu¨tz B, Yoshioka T, Kawato M. Human cerebellar activity reflecting an acquired internal model of a new tool. Nature 2000;403:192–5.
767
[19] Ingram HA, van Donkelaar P, Cole J, Vercher J-L, Gauthier GM, Miall RC. The role of proprioception and attention in a visuomotor adaptation task. Exp Brain Res 2000;132:114–26. [20] Krakauer JW, Ghilardi M-F, Ghez C. Independent learning of internal models for kinematic and dynamic control of reaching. Nat Neurosci 1999;2:1026 –31. [21] Krebs HI, Brashers-Krug T, Rauch SL, Savage CR, Hogan N, Rubin RH, Fischman AJ, Alpert NM. Robot-aided functional imaging: application to a motor learning study. Hum Brain Map 1998;6:59– 72. [22] Lazar G, van Laer J. Adaptation to displaced vision after experience with lesser displacements. Percept Motor Skills 1968;26:579–82. [23] Logie R, Baddelley A, Mane´ A, Donchin E, Sheptak R. Working memory in the acquisition of complex cognitive skills. Acta Psychol 1989;71:53 –87. [24] McDowd J, Vercruyssen M, Birren J. Aging, divided attention, and dual-task performance. In: Damos D, editor. Multiple-task performance, London: Taylor and Francis; 1991. p. 387 –414. [25] McNay EC, Willingham DB. Deficit in learning of a motor skill requiring strategy, but not of perceptualmotor recalibration, with aging. Learn Mem 1998;4:411 –20. [26] Navon D, Gopher D. On the economy of the human-processing system. Psychol Rev 1979;86:214– 55. [27] Raz N, editor. The handbook of aging and cognition. Aging of the brain and its impact on cognitive performance: integration of structural and functional findings, Hillsdale, NJ: Erlbaum; 2000. p. 1– 90. [28] Redding G, Wallace B. Perceptual-motor coordination and adaptation during locomotion: determinants of prism adaptation in hall exposure. Percept Psychophys 1985;38:320–30. [29] Redding GM, Wallace B. Cognitive interference in prism adaptation. Percept Psychophys 1985;37:225–30. [30] Redding GM, Wallace B. Prism adaption during target pointing from visible and non-visible starting locations. J Motor Behav 1997;29: 119–30. [31] Roller C, Cohen H, Kimball K, Bloomberg J. Effects of normal aging on visuo-motor plasticity. Neurobiol Aging 2002;23:117–23. [32] Salthouse TA. The processing-speed theory of adult age differences in cognition. Psychol Rev 1996;103:403–28. [33] Shadmehr R, Holcomb HH. Inhibitory control of competing motor memories. Exp Brain Res 1999;126:235 –51. [34] Spirduso W. Physical dimensions of aging, Human Kinetics, Champaign, 1995. [35] Stratton GM. Vision without inversion of the retinal image. Psychol Review 1897;4:341–481. [36] Tong C, Wolpert D, Flanagan J. Kinematics and dynamics are not represented independently in motor working memory: evidence from an interference study. J Neurosci 2002;22:1106–13. [37] Wolpert DM, Ghahramani Z, Jordan MI. An internal model for sensorimotor integration. Science 1995;269:1880–2. [38] Woollacott M. Systems contributing to balance disorders in older adults. J Gerontol Series A—Biol Sci Med Sci 2000;55:M424–8. [39] Wright B, Payne R. Effects of aging on sex differences in psychomotor reminiscence and tracking proficiency. J Gerontol 1985;40:179 –84.