The study examined the effects of localized muscle fatigue on gait velocity, stride length, and stride length variability under single and dual-task conditions in young and older adults. The main findings can be summarized as follows. First, significantly lower maximal torque of the knee extensors and flexors were observed at baseline in older compared to young adults. Second, older adults needed significantly more repetitions to reach 50% of Mmax of the knee extensors/flexors than the young adults. Third, stride length variability under dual-task conditions was significantly greater at baseline in older compared to young adults. Fourth, localized muscle fatigue resulted in significant decreases in single-task gait velocity and stride length in young adults. Fifth, muscle fatigue produced significant increases in dual-task gait velocity and stride length in older adults which were accompanied by significant decreases in stride length variability under dual-task conditions. Sixth, muscle fatigue did not significantly affect dual-task costs in all analyzed gait parameters in both, young and older adults. Finally, muscle fatigue resulted in significant improvements in cognitive performance during walking in young and older adults. These findings indicate that our initially formulated hypothesis (i.e., localized muscle fatigue affects gait characteristics in young and older adults under single and dual-task conditions, and increases dual-task costs while walking, particularly in the elderly) is only partially supported.
Differences in maximal torque between young and older adults
In this study, significant baseline differences between young and older adults were found for maximal torque of the knee extensors and flexors. This is consistent with the literature because Vandervoort et al.  examined a 53% lower concentric peak torque of the knee extensors in old as compared to young adults. The observed difference in maximal torque in this study could be caused by a reduced excitability of efferent corticospinal pathways resulting in lower levels of central muscle activation, a gradual loss of spinal motoneurons (particularly large alpha-motoneurons) due to apoptosis, a subsequent decline in muscle fibre number and size (sarcopenia) of especially type-II fibres, changes in muscle architecture, and decreases in tendon stiffness. For a review see Granacher et al. .However, due to the methodological approach applied in this study, we cannot directly infer on the underlying neuromuscular mechanisms responsible for the reduced level of maximal torque in old compared to young adults.
Differences in fatigue-resistance between young and older adults
Findings of this study indicated that the older adults needed significantly more repetitions to reach 50% of Mmax of the knee extensors/flexors than the young adults. This may suggest that the older adults were more fatigue resistant than the young adults. In fact, there is evidence showing that older adults fatigue less than young during isometric contractions . This can be explained by the occurrence of motor unit remodeling in old age, i.e., type II muscle fibres are denervated due to degenerative processes and subsequently re-innervated by an adjacent slow-twitch motor neuron resulting in a muscle fibre shift from type II to fatigue resistant type I fibres . Thus, proportionally more fatigue-resistant type I fibres contribute to force generation in the aged compared to the young muscle. This may explain why the older adults needed significantly more repetitions to reach 50% of Mmax of the knee extensors/flexors than the young adults.
Differences in stride length variability between young and older adults
The observed baseline differences in stride length variability under dual-task conditions between the two experimental groups are in accordance with a study conducted by Granacher et al. . These authors reported increased stride length variability in old compared to young subjects when walking while concurrently performing a cognitive (i.e., performing an arithmetic task) or a motor interference task (i.e., holding two interlocked sticks steady in front of the body). Gunning-Dixon and Raz  attributed age-related dual-task deficits to the shrinkage of prefrontal brain areas in old age, since those areas are related to executive functions (e.g., processing of multi-tasking). Of note, it was recently shown in healthy older adults that individuals with poorer executive function are more prone to falls . Other authors ascribe increased gait instability in old age to the age-related loss of visual, proprioceptive, and vestibular sensitivity . Notably, baseline differences between the experimental groups were only present for stride length variability under dual-task conditions, but not for the parameters stride length and gait velocity under single and dual-task conditions. This contradicts findings reported by Hollman et al.  and Hausdorff et al.  who observed differences between young and older adults in gait velocity as well as in variability of stride time, stance time, and swing time. The lack of age-related effects on gait velocity and stride length observed in this study can quite likely be explained by the high physical activity level of our older subjects which was not significantly different from that of the young adults. In addition, subjects in the studies of Hollman et al.  and Hausdorff et al.  were older than our subjects with a mean age of 81 and 82 years respectively, which might explain why their gait pattern was characterized by greater instability.
Effects of muscle fatigue on gait characteristics in young and older adults
The present results regarding the effects of muscle fatigue on gait characteristics in young adults are consistent with findings reported by Parijat et al. . These authors examined the impact of bilateral fatigue induced by repetitive isokinetic knee extension movements of the quadriceps on kinematic and kinetic gait characteristics in healthy young adults. After fatigue exertions, participants showed a tendency towards a decrease in gait speed. It was argued that reduced push-off force during the stance phase of the gait cycle reduces the transitional acceleration of the whole body centre of mass and may thus be responsible for the decrease in gait velocity in young adults . Reduced gait speed may represent a compensatory strategy to enhance dynamic stability during walking in order to keep from falling .
In our older adults, muscle fatigue produced an increase in gait speed and stride length coming along with a decrease in stride length variability particularly under dual-task conditions. This is in accordance with findings from two studies [42, 43]. Morris et al.  investigated changes in gait characteristics (tested on a walkway) and fatigue from morning to afternoon in people with multiple sclerosis. Although self rated fatigue significantly increased from the morning to the afternoon, increases in walking speed and stride length were observed over the course of the day. The authors suggested that practice effects could be responsible for the observed increases over the course of the trials. Yoshino et al.  examined how long-term free walking (3 hours) at a self-determined preferred pace on level ground affected the gait pattern of healthy subjects. Based on their level of performance during the 3 h walk, subjects were assigned to two groups. Group A showed longer gait cycle time during the second half of the walk and group B showed shorter gait cycle time during the same period. Variability of the parameter gait cycle time increased significantly in group A from 120 min on, whereas it tended to decrease gradually with time in Group B. For both groups, the mean subjective levels of fatigue increased monotonically with time. The mean heart rate during the walking task was almost constant until 120 min from the beginning, and it tended to increase gradually during the last 60 min in both groups. Unfortunately, the authors did not provide a reasonable explanation for this phenomenon. It was suggested that subjects in group B could have been more fatigue resistant than those in group A because of higher levels of stamina .
Four reasons may account for the observed fatigue-induced increase in gait speed and stride length in the older adults. First, walking faster with longer strides could represent a strategy of the older adults to overcome the short walking distance (10 m) and thus the feeling of physical discomfort due to muscle fatigue as quickly as possible. Therefore, it is suggested that future studies investigate the effects of muscle fatigue on gait characteristics by incorporating longer walking distances. In fact, longer distances may prevent older subjects from initially increasing their walking speed to levels higher than their preferred non-fatigued walking speed because they might not be able to keep up this walking speed for the entire distance. Second, it was reported that the age-related loss of ankle plantar flexor strength resulted in lower ankle plantar flexor power during the late stance phase of the gait. Interestingly, older adults learned to compensate for this muscular deficit by increasing hip flexor power . It is proposed that our older adults may have compensated fatigue-induced decreases in knee extensors/flexors by increasing hip flexor power during walking. In contrast, young adults probably never learned this compensatory strategy due to a lack of need. Third, it was reported that muscle fatigue has an impact on muscle spindle function in terms of an increase in sensitivity of this mechanoreceptor . Increased muscle spindle sensitivity may represent a fatigue-induced compensatory mechanism to maintain function and force output . Given that muscle spindle sensitivity decreases in seniors due to increased spindle capsule thickness and a loss of intrafusal- and nuclear chain fibers , it is speculated that particularly older adults could benefit from this compensatory mechanism in terms of enhanced leg extensor muscle activation and thus improved forward propulsion of the body. This hypothesis needs to be proven in future studies. Fourth, a practice and/or learning effect from pre to post tests could have resulted in an increase in gait speed and stride length in the older adults. However, due to the standardized testing procedures and because improvements were only present from pre to post but not from post to T5 testing, it is postulated that practice/learning may only play a minor role.
The observed increase in gait velocity and stride length post-fatigue was accompanied by a decrease in stride length variability indicating improved gait stability. Yet, it was recently reported that stride-to-stride variability appears to be speed dependent . Jordan et al.  observed that gait cycle variability was lowest at 100% and 110% of the preferred walking speed. Post-fatigue, our older adults showed a 2.8% and a 9.7% increase in gait speed under single-task and dual-task conditions as compared to the respective non-fatigued preferred walking speed. Both percentage rates are within the range of lowest gait cycle variability stated by Jordan et al. .
Age-related effects of muscle fatigue on gait characteristics are task dependent
Recently, Granacher et al.  investigated the effects of ankle fatigue on the ability to compensate for decelerating gait perturbations during walking on a treadmill in healthy young and older adults. The authors reported that muscle fatigue affected the compensatory mechanisms of young and older adults in terms of significant decreases in reflex activity and increases in antagonist co-activity of lower extremity muscles. Since young and elderly subjects were affected to a similar extent by muscle fatigue, the authors proposed that age-related deteriorations in the postural control system did not specifically affect the ability to compensate for gait perturbations under fatigued condition . However, the fatigue-induced changes in reflex activity may put young and older adults at high risk of sustaining a fall when encountering a balance threatening situation in a fatigued state. The finding that young and older adults showed similar fatigue-induced responses when compensating for gait perturbations contradicts the present results. In this study, muscle fatigue produced different gait characteristics in young and older adults. More specifically, young adults decreased their gait velocity and stride length particularly under single-task condition, whereas older adults increased their walking speed and stride length predominantly under dual-task conditions. In addition, young adults slightly increased their stride length variability under dual-task conditions, whereas older adults significantly decreased theirs. The observed discrepancy between the study of Granacher et al.  and ours can most likely be explained by different test conditions. Whereas Granacher et al.  investigated the impact of muscle fatigue on postural reflexes in young and older adults, we studied the effects of muscle fatigue on the gait pattern which is regulated by a complex interaction of reflexive and voluntary contributions to muscle activation . Notably, it was reported that neural control of volitional limb movements differs in some fundamental ways in comparison to reactions that are evoked by postural perturbation .
Effects of muscle fatigue on dual-task costs while walking in young and older adults
In the present study, muscle fatigue did not have a significant impact on dual-task costs in all analyzed gait parameters in both, young and older adults. Bock et al.  found that the occurrence of dual-task costs while walking in healthy young and elderly persons is task dependent with complex secondary tasks affording higher dual-task costs. Thus, the choice of our secondary task (reciting out loud serial subtractions by three) may have influenced the outcome of this study. Further, Simoneau et al.  investigated how moderate fatigue induced by fast walking on a treadmill challenged dynamic balance control in young healthy adults and whether the attentional demands for the performance of the balance task varied with fatigue. Fatigue induced an initial negative impact on balance control followed by a subsequent improvement in the performance of the balance task. Subjects achieved this performance enhancement by allocating a greater portion of the cognitive resources to the balance control task. In general, this finding seems to be in accordance with the results of the present study regarding the young adults. More specifically, our young participants chose a different strategy of allocating central resources than those in the study of Simoneau et al.  because we detected impaired performance in balance control following fatigue accompanied by improved performance in the cognitive interference task while walking. In other words, the young adults achieved better cognitive performance post-fatigue at the cost of impaired balance control.
Improvements in cognitive performance following fatigue were also observed in the older adults participating in this study. Emery et al.  evaluated the acute effects of exercise (i.e., 20 min bicycle ergometry stress test) on cognitive performance in a community-based sample of patients (mean age 68 ± 7 years) with chronic obstructive pulmonary disease (COPD). Acute exercise was associated with improved performance on the verbal fluency test, a measure of verbal processing. The fatigue protocol applied in the present study represents some kind of an acute bout of exercise and our results may thus be comparable to those investigated by Emery et al. . It was suggested that improved neurotransmitter functioning in the brain could be responsible for the enhanced cognitive function following acute bouts of exercise or fatigue .