Introduction

Motor imagery (MI) is an active mental process in which a subject internally simulates a movement without any corresponding motor output. Accumulating evidence indicates functional similarities between actually and mentally executed actions, notably regarding temporal characteristics, neural correlates and autonomic responses1. A plethora of studies shows that actual and mental movements are subject to common motor rules and principles (e.g., Fitts’ law)2,3,4. Indeed, results on healthy adults usually indicate a temporal coupling between the duration of actual and mental movements3, 5,6, the so-called “isochrony principle”2. Congruently, fMRI studies have shown that the neural networks involved in MI overlap considerably with those recruited for actual motor execution7,8,9.

On the contrary, anisochrony, i.e., a temporal discrepancy between actual and mental movements, has been frequently reported in populations of neurological patients, as people after stroke10, with Parkinson disease (PD)11,12 and with multiple sclerosis (MS)13,14,15, gaining increasing attention as a promising additional clinical investigation tool16. Recently, Tacchino et al.17 showed that behavioural performance and brain activations during MI are correlated to disease severity in people with MS (PwMS) and proposed anisochrony as a surrogate behavioural marker of MS evolution, especially in the early stages of the disease.

As for healthy subjects (HS), most investigations on MI of people with neurological diseases have mainly examined upper limb movements with different motor tasks such as sequential pressing of buttons18, finger-to-thumb opposition19, pointing towards targets15,20, hand manipulation17,21, hand grasping and arm-lifting22, and bimanual circle-line coordination23. However, examples of MI involving lower limb are also present24,25,26,27,28,29,30.

Walking is particularly interesting for a broader understanding of MI. In fact, this represents a unique window into the study of MI because subjects are requested to simulate full-body movements and simultaneously update environmental spatial information5. In general, findings indicate that healthy good imagers usually preserve temporal congruence31,32,33 as well as recruiting very similar cerebral networks (e.g. fronto-parietal areas, basal ganglia, brainstem, cerebellum)34,35 when they are asked to actually and mentally execute tasks involving locomotion.

Since impaired locomotion is a frequent and major source of disability in patients with neurological diseases and new techniques are continuously required to improve gait rehabilitation, a better characterization of MI of locomotion in these populations is fundamental. People after stroke seem to retain locomotion MI abilities36 and mental practice appears to be a beneficial intervention for stroke rehabilitation37; nevertheless the existing evidence remains inconclusive because of significantly statistical heterogeneity and methodological flaws38. More contradictory results have been found in patients with PD, mainly due to differences in testing procedures and task instructions27,39,40.

The few studies investigating MI in PwMS are mainly limited to considering upper limb actions13,15,17, showing also how cognitive and mood disorders affect MI28,41,42. Only recently the focus has been shifted to lower limb movements43,44,45. Interestingly, evidence indicates that MI has been shown to improve not only walking, but also balance, fatigue, mood and quality of life in MS43,44,45. However, to date, it remains unclear whether PwMS show anisochrony between actual and mental locomotion as well and, if so, whether this temporal discrepancy is modulated by spatial constraints5,13,15,17 and/or affected by cognitive fatigue.

Investigating the relationship between fatigue and MI performances in PwMS could be key to shedding new light on the motor representation of PwMS and providing insights into new rehabilitative treatments. In fact, fatigue is one of the most debilitating symptoms in MS as it significantly impacts patients’ daily life activities and quality of life46. Given its multifactorial nature, frequently divided into various components such as physical vs. cognitive, fatigue is difficult to define or operationalize, although neurophysiological and fMRI studies have indicated that a cortico-subcortical disconnection could support the central origin of fatigue and explain its complex interaction with other clinical conditions (e.g. mood, cognitive functioning, disability)47. Interestingly, given the significant cognitive effort required to concentrate on MI tasks, loss of attention and declined arousal level due to cognitive fatigue have been found to significantly alter neural signals in the Brain Computer Interface (BCI) system48. As indicated by Deluca et al.49, abnormal cerebral activation in the basal ganglia and frontal lobes (that are also recruited in the MI task) may represent the extra “effort” (i.e., allocation of more neural resources) required to maintain the same level of performance, supporting the notion that increased cerebral activation may reflect the additional effort (i.e., cognitive fatigue) to adequately perform behavioral tasks in PwMS. Therefore, investigation of the complex inter-relationship between cognitive fatigue and MI is very relevant.

Although there is clear and consistent evidence that PwMS may have difficulties in the mental simulation of actions and that several aspects could affect performances, previous results are far from conclusive. Thus, to shed light on this aspect, we asked PwMS to actually and mentally walk along three paths with different widths and we compared their performances to those of HS.

Methods

Participants

Sample size was determined using values from the study by Tacchino et al.17 which suggested a 0.3 point difference [standard deviation (SD) = 0.3] in Index of Performance (IP), a measure of participants’ mental movement ability, of PwMS. Considering 80% power and 5% (two-sided) level of significance, the planned sample size was 15 PwMS and 15 HS. PwMS were recruited among outpatients at the Rehabilitation Service of Genoa of the Italian Multiple Sclerosis Society (AISM). Inclusion criteria for PwMS were clinically defined MS according to McDonald criteria50, a stable phase of the disease without relapses in the previous 3 months, all disease courses, an Expanded Disability Status Scale (EDSS) ≤ 6.5 (i.e. able to walk with at most one aid, e.g. cane or single crutch)51. The HS were age-matched, did not present any neurological or muscular disorders and had normal or corrected-to-normal vision. Moreover, the subjects belonging to the two groups were included if they were right-handed as determined by the Edinburgh Handedness Inventory52 and showed preserved cognitive functioning as measured by a Symbol Digit Modalities Test53 (SDMT) score > 38. Subjects with a history of severe psychiatric disorders as indicated by Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria54, blurred vision or cardiovascular and respiratory disorders were excluded. Participants from both groups were assessed with the following tests: the Timed 25-Foot Walk (T25-FW)55 to measure performances in walking speed, the short form of the International Physical Activity Questionnaire (IPAQ)56,57 to gather information about the usual time spent in walking, in vigorous and moderate intensity activity, and in sedentary behaviours, the Dual-tasking questionnaire (DTQ)59 which collects information about everyday difficulties with dual-tasking and the Kinesthetic and Visual Imagery Questionnaire (KVIQ)58 for MI vividness. PwMS were also assessed with the Modified Fatigue Impact Scale (MFIS)60 to gather information about fatigue perception.

All subjects gave written informed consent in accordance with the revised Declaration of Helsinki61. The study was approved by the Ethics Committee of San Martino Hospital, Genoa, Italy. All methods were carried out in accordance with relevant guidelines and regulations.

Procedure

The experimental setup was an adapted version from Personnier et al.5. Participants stood upright behind the starting line; their arms were hanging along the body, and their feet were parallel and slightly apart. They were asked to actually or mentally walk through three 5-m long paths, each with a different width: 20 cm, 35 cm and 50 cm. The different spatial constraints made it possible to test walking under several conditions, from normal to very accurate according to the path’s width. During actual movements, participants were invited to perform the task without walking on the lines limiting the paths at the self-paced speed. During mental movements, participants had to imagine themselves walking at a self-paced speed in the first-person perspective (i.e., as if one were the actor of the action). This has been considered to be closer to the real execution of movement14 since individuals elicit kinaesthetic sensation representations of actions as if they were actually performing them (see Fig. 1 for a representation of the experimental setup).

The duration of actual and mental movements was recorded by means of an electronic stopwatch (temporal resolution: 10 ms). When the experimenter gave the “Go!” signal, temporal acquisition started, and participants had to begin walking or imagining they were walking. During the actual movements, the temporal acquisition was stopped when the participants’ shoulders clearly crossed the finish line; during mental trials participants were required to verbally indicate “Stop!” when they mentally crossed the finish line. The time elapsed between the “Go!” signal and crossing the line corresponded to the movement duration.

Participants performed 12 walking movements for each path width (20 cm, 35 cm and 50 cm) in each experimental condition (actual and mental); for a total of 72 movements distributed over two different sessions separated by a time interval of 48 h. Within each session, actual and mental trials were presented to the participants in two different blocks. In the first session, the mental movements were performed before the actual ones, while in the second session the actual movements were performed before the mental ones. The three paths were randomly presented to the participants within each block. The two experimental conditions were separated by a 5-min break. In order to familiarize with the task, all participants performed one actual and one mental movement within each path. No information concerning their temporal performance was given to the participants during the practice or the experimental trials.

Statistical analysis

For each actually (A) and mentally (M) executed path, the mean movement duration of locomotion was calculated. In order to better examine participants’ mental movement ability, we computed an Index of Performance (IP) expressed in percentage:

$$IP=\frac{A-M}{A} x 100$$

The IP reflects whether participants overestimate (negative values) or underestimate (positive values) movement durations during MI with respect to actual performances, where an IP approaching zero would represent a good motor imagery performance. However, although it is considered a reliable and objective measure of MI accuracy and an indicator of motor system integrity7,62, this index could hinder differences between imagined and actual movement durations when, for instance, some subjects underestimate and others overestimate movement durations during MI. As suggested by5, in order to provide a clearer and more complete analysis of MI performance, we also calculated the absolute values of IP.

A Shapiro–Wilks test revealed that all variables were normally distributed (always p > 0.05) and, therefore, parametric tests were used. Independent t tests were performed on SDMT, T25FW, IPAQ, DTQ and KVIQ total scores to assess potential differences between groups in cognitive, locomotion, usual physical abilities, dual-task and MI vividness performance. A two-way analysis of variance (ANOVA), with group as a between-subjects factor (PwMS; HS) and path width as a within-subjects factor (20 cm; 35 cm; 50 cm) was performed on IP values. Furthermore, an ANOVA with group as a between-subjects factor (PwMS; HS), path width (20 cm; 35 cm; 50 cm) and type of movement (actual; mental) as within-subjects factors was performed on actual and mental movement durations. Post-hoc tests (Bonferroni’s correction; p < 0.05) were applied to explore significant effects and interactions. To test any potential association between anisochrony and fatigue perception among PwMS, a correlation analysis was performed between the absolute values of IP (IPabs)5 across different path widths and MFIS (total score and subscale scores). Statistical analyses were run using SPSS 23.

Results

In order to evaluate MI in PwMS, we asked fifteen PwMS (14 with relapsing–remitting course; 1 with secondary progressive course; 4 males; mean EDSS: 2.97 ± 1.86; mean MFIS: 38.53 ± 17.57) and fifteen HS (7 males) to take part in this study. The two groups were age matched (PwMS: 42.67 ± 11.67 years; HS: 41.60 ± 13.35 years; t28 = 0.23; p = 0.82). No significant differences were found between groups in walking speed as indicated by the T25-FW (PwMS: 7.14 ± 1.93 s; HS: 6.23 ± 1.03 s; t28 = 1.59; p = 0.122), in usual physical activity by IPAQ (walking in MET-min/week: PwMS: 1178.10 ± 519.19; HS: 1214.40 ± 208.78; t28 = -0.25; p = 0.803; moderate in MET-min/week: PwMS: 1420 ± 949.13; HS: 1536 ± 586.48; t28 = − 0.40; p = 0.690; vigorous in MET-min/week: PwMS: 1760 ± 1146.22; HS: 1872 ± 422.56; t28 = − 0.35; p = 0.725; total in MET-min/week: PwMS: 4358.10 ± 2366.04; HS: 4622.40 ± 820.83; t28 = − 0.41; p = 0.686; sedentary in h/day: PwMS: 4.4 ± 2.2; HS: 3.4 ± 1.40; t28 = 1.48; p = 0.149), in dual-task abilities as evaluated by DTQ (PwMS: 0.16 ± 0.07; HS 0.13 ± 0.07; t28 = 1.00; p = 0.326), in information processing speed measured by SDMT (PwMS: 48.27 ± 19.34; HS: 51.00 ± 11.56; t28 = − 0.47; p = 0.64) and in MI vividness as tested by KVIQ (PwMS: 126.20 ± 33.28; HS: 137.53 ± 21.47; t28 = − 1.11; p = 0.28). See Table 1 for a summary.

Overestimation of mental movement duration in PwMS

The results of an ANOVA on IP showed a main effect of group [F (1,28) = 12.79, p = 0.001, η2 = 0.314]. IP approached zero at all path widths in HS (− 3.5% ± 4.7%), while PwMS tended to mentally overestimate the duration of the required action (− 27.3% ± 4.7%), reflecting the presence of anisochrony (i.e. mentally executed locomotion lasted longer than actually executed locomotion). No significant differences in the path width factor (p = 0.456, η2 = 0.028) or in group*path width interaction (p = 0.103, η2 = 0.078) were found. However, we observed a trend towards anisochrony in PwMS with the decreasing of the path width (PwMS M ± SE: 20 cm: − 29.6% ± 4.6%, 35 cm: − 27.9% ± 5.1%, 50 cm: − 24.3% ± 5.0%; HS M ± SE: 20 cm: − 1.9% ± 4.6%, 35 cm: − 4.5% ± 5.1%, 50 cm: − 4.0% ± 5.0%), suggesting a progressive increment of both actual and mental durations along with the increase in spatial precision.

Influence of path spatial constraints on both actual and mental movement duration in PwMS

Although IP offered a single and informative measure to examine participants’ mental movement ability, it did not clarify whether and to what extent subjects tended to overestimate or underestimate mental movements across paths with different spatial widths. The ANOVA on actual and mental movement duration further qualified the results found in the IP statistical analysis. A significant group*path width*type of movement interaction was found [F (2,56) = 6.801, p = 0.040, η2 = 0.108] (Fig. 2): post-hoc analysis showed that actual locomotion across all path widths in PwMS was significantly slower compared to that in HS (PwMS M ± SE: 20 cm: 5.68 ± 0.37 s, 35 cm: 5.37 ± 0.32 s, 50 cm: 5.15 ± 0.25 s; HS M ± SE: 20 cm: 4.26 ± 0.37 s, 35 cm: 4.23 ± 0.32 s, 50 cm: 4.23 ± 0.25 s) (always p < 0.05). The same result was found for mental locomotion (PwMS M ± SE: 20 cm: 7.31 ± 0.50 s, 35 cm: 6.81 ± 0.49 s, 50 cm: 6.41 ± 0.41 s; HS M ± SE: 20 cm: 4.36 ± 0.50 s, 35 cm: 4.45 ± 0.49 s, 50 cm: 4.25 ± 0.41 s) (always p < 0.005). Moreover, mental locomotion of PwMS was significantly slower with respect to actual locomotion at all path widths (always p < 0.001). We did not observe the same result in HS (always p > 0.05). In addition, while mental locomotion in PwMS was significantly different across all path widths (always p < 0.005), considering actually executed locomotion, significant differences were only found between the narrower (20 cm) and the wider paths (35 cm and 50 cm) (p < 0.005 in both cases) (see Fig. 2 for a graphical representation of major results).

Role of cognitive fatigue on PwMS’ performances

In PwMS, the correlation between IPabs and the total score of MFIS was significant at all path widths (20 cm: r = 0.719, p = 0.002; 35 cm: r = 0.362, p = 0.011; 50 cm: r = 0.707, p = 0.003): the higher the reported fatigue perception, the higher the anisochrony. More interestingly, IPabs significantly correlated with the MFIS cognitive subscale at all path widths (20 cm: r = 0.752, p = 0.001; 35 cm: r = 0.584, p = 0.022; 50 cm: r = 0.673, p = 0.006), suggesting a role of cognitive fatigue on PwMS’ performances (see Fig. 3). No correlations with physical (20 cm: r = 0.439, p = 0.102; 35 cm: r = 0.479, p = 0.071; 50 cm: r = 0.512, p = 0.051) or psychosocial (20 cm: r = 0.472, p = 0.075; 35 cm: r = 0.354, p = 0.195; 50 cm: r = 0.408, p = 0.131) subscales were found.

Discussion

MI is a dynamic state during which the representation of an action is internally reproduced within working memory without any overt output2. The aim of the present study was to test MI in people with Multiple Sclerosis (PwMS). Although there is clear and consistent evidence that PwMS may have difficulties in the mental simulation of actions, previous results are far from being conclusive. Thus, a critical advance of these studies was necessary. Here, we investigate if PwMS show anisochrony in a task involving lower limbs and if the dependency on spatial constraints was present. In addition, due to the major role of fatigue in the life of PwMS, the relationship between cognitive fatigue and MI performances of patients was investigated. To this aim, fifteen PwMS with mild disability and 15 age-matched healthy subjects (HS) were required to either actually walk or imagine themselves walking on three paths of the same length, but different widths. When asked to mentally simulate a walking movement along a constrained pathway, PwMS tended to overestimate mental movement duration with respect to actual movement duration, thus confirming the presence of anisochrony. Moreover, the investigation of movement durations revealed that a progressive increment of both actual and mental durations occurred with the augmentation of spatial precision; this increment was more consistent for the mental task at the narrowest paths. Interestingly, in line with previous evidence, cognitive fatigue was found to play a role in the MI of PwMS.

The analysis of actual movement durations across different path widths showed that PwMS were slower than HS. Although no statistical differences were found between groups in T25-FW, this result was expected because MS typically influences postural control and body equilibrium as of the first stages of the disease; thus, ambulation was more challenging for PwMS with respect to HS, especially when walking on a path with spatial constraints was required. This effect is also present in mental locomotion; in fact, the mental simulation of movements reflects the updating of motor representation in line with disease evolution. Thus, the slower the actual movements, the slower mental simulation. Interestingly, PwMS and HS reported to spent almost the same time in walking or in other physical activities, as indicated by IPAQ scores. This further corroborated our results suggesting that PwMS could not have less experience of walking to refer when performing the MI tasks and, thus, anisochrony could not be attributed to differences in physical practice or, conversely, sedentary behaviours.