Impaired Implicit Sequence Learning in ... - Semantic Scholar

Download PDF
2 downloads 0 Views 439KB Size Report
Comment
are possibly key structures in understanding both the motor and cognitive components of FOG (Lewis & Barker, 2009). Recent studies have already indicated ...
Neuropsychology 2013, Vol. 27, No. 1, 28 –36

© 2013 American Psychological Association 0894-4105/13/$12.00 DOI: 10.1037/a0031278

Impaired Implicit Sequence Learning in Parkinson’s Disease Patients With Freezing of Gait Jochen Vandenbossche, Natacha Deroost, Eric Soetens, and Daphné Coomans

Joke Spildooren, Sarah Vercruysse, and Alice Nieuwboer

Vrije Universiteit Brussel

Katholieke Universiteit Leuven

Eric Kerckhofs Vrije Universiteit Brussel Objective: Freezing of gait (FOG) in Parkinson’s disease (PD) may involve specific impairments in acquiring automaticity under working memory load. This study examined whether implicit sequence learning, with or without a secondary task, is impaired in patients with FOG. Method: Fourteen freezers (FRs), 14 nonfreezers (nFRs), and 14 matched healthy controls (HCs) performed a serial reaction time (SRT) task with a deterministic stimulus sequence under single-task (ST) and dual-task (DT) conditions. The increase in reaction times (RTs) for random compared with sequenced blocks was used as a measure of implicit sequence learning. Neuropsychological tests assessing global cognitive functioning and executive dysfunction were administered in order to investigate their relation to sequence learning. Results: nFRs and HCs showed significant implicit sequence learning effects (p ⬍ 0.001). FRs demonstrated a tendency to learn sequence-specific information in the SRT-ST task (p ⫽ 0.07) but not in the SRT-DT task (p ⫽ 0.69). Severity of FOG, however, correlated positively with SRT-DT task performance (r ⫽ ⫺0.56; p ⬍ 0.05). Conclusions: The present results suggest that PD patients suffering from FOG pathology exhibit a specific impairment in the acquisition of automaticity. When working memory capacity is supplementarily loaded by adding a DT, sequence learning in FRs becomes increasingly impaired. These findings indicate that therapies should focus on extensive training in acquiring novel motor activities and reducing working memory load to improve learning in FOG. Keywords: freezing of gait, Parkinson’s disease, sequence learning, dual tasking, executive function

cognitive components is freezing of gait (FOG). Although not present in all PD patients, FOG is defined as “a brief, episodic absence or marked reduction of forward progression of the feet despite the intention to walk” (Nutt et al., 2011). FOG contributes to the development of major disability and frequent falls, and can be considered as an independent cardinal sign of PD (Giladi et al., 2001; Giladi & Nieuwboer, 2008). At present, the underlying mechanisms of FOG remain unclear. Affective disturbances, advanced PD motor symptoms, and cognitive challenges can increase the occurrence of FOG episodes (Giladi & Hausdorff, 2006). The involvement of the pedunculopontine nucleus (PPN) and thalamus are possibly key structures in understanding both the motor and cognitive components of FOG (Lewis & Barker, 2009). Recent studies have already indicated that patients with FOG suffer from a specific impairment in controlled processing, that is, processes demanding working memory capacity. Amboni, Cozzolino, Longo, Picillo, and Barone (2008), for example, found that freezers (FRs) exhibited a generalized executive dysfunction (Frontal Assessment Battery), cognitive inflexibility (Controlled Oral Word Association Test), as well as an impaired performance on the manual Stroop task. In recent studies, set-shifting difficulties under temporal pressure (Naismith, Shine, & Lewis, 2010), measured by the Trail Making Test, and a conflict resolution deficit (Vandenbossche et al., 2011, 2012), exposed in the Attention Network Test, were associated with FOG. These findings

Parkinson’s disease (PD) is characterized by both motor and nonmotor features (Jankovic, 2008). These symptoms result from a loss of dopaminergic neurons in the basal ganglia, which is crucial for voluntary motor control, procedural learning, as well as cognitive and emotional functions (Stocco, Lebiere, & Anderson, 2010). A common movement problem affecting both motor and

Jochen Vandenbossche, Research Unit for Cognitive Psychology, Research Unit for Neurological Rehabilitation, and Center for Neurosciences, Vrije Universiteit Brussel, Brussels, Belgium; Natacha Deroost and Eric Soetens, Research Unit for Cognitive Psychology and Center for Neurosciences, Vrije Universiteit Brussel; Daphné Coomans, Research Unit for Cognitive Psychology, Vrije Universiteit Brussel; Joke Spildooren, Sarah Vercruysse, and Alice Nieuwboer, Department of Rehabilitation Sciences, Katholieke Universiteit Leuven, Leuven, Belgium; Eric Kerckhofs, Research Unit for Cognitive Psychology, Research Unit for Neurological Rehabilitation, and Center for Neurosciences, Vrije Universiteit Brussel. This research was funded by the Flanders Research Funds (G.0691.08) and the Research Council of the Vrije Universiteit Brussel (Grant OZR1933 BOF). We thank Wim Vandenberghe for his assistance in recruiting PD patients. Correspondence concerning this article should be addressed to Jochen Vandenbossche, Research Unit for Cognitive Psychology, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels, Belgium. E-mail: Jochen [email protected] 28

SEQUENCE LEARNING IN FREEZING OF GAIT

indicate that controlled processes are significantly impaired in FOG. Giladi and Hausdorff (2006; see also Lewis & Barker, 2009) go one step further by suggesting that episodes of FOG can be evoked by increasing working memory load, for example, by adding a dual task (DT). Indeed, several studies indicated that walking under DT conditions compromised gait in PD patients (e.g., Hausdorff, Balash, & Giladi, 2003; O’Shea, Morris, & Iansec, 2002). A recent study, conducted by Spildooren and colleagues (2010), showed that turning in combination with a DT is the most important trigger in inducing freezing. This indicates that limited cognitive capacity plays a key role in the occurrence of FOG: The execution of automatic behavior in FOG is severely compromised by increased cognitive demands, even to the extent that it can evoke episodes of FOG. In the present study, we determined automatic processing, with or without cognitive load, in FOG. To our knowledge, this issue had not yet been investigated in patients expressing FOG. Yet this topic is very important for rehabilitation: When the acquisition of motor behavior itself is impaired and not only the execution, one should focus on (re)learning acquired automatic behavior in a different way (e.g., by defragmenting or dividing well-known motor programs into smaller segments; Kamsma, Brouwer, & Lakke, 1995). This is especially important in FOG because automaticity allows saving cognitive resources for handling complex situations, like turning during walking. To study the influence of increased cognitive demands on learning of automatic behavior, we examined the effect of working memory load on implicit sequence learning in FRs, nonfreezers (nFRs), and healthy controls (HCs). To determine automaticity, we used the serial reaction time (SRT) task (Nissen & Bullemer, 1987), a computerized reaction time (RT) experiment where subjects incidentally learn a repeating sequence of stimuli. The SRT task has been successfully applied in previous studies investigating automaticity (Deroost & Soetens, 2006a, 2006b; Willingham, 1999). Because variations in global cognitive functioning have a strong impact on implicit sequence learning abilities in PD (Deroost, Kerckhofs, Coene, Wijnants, & Soetens, 2006; Vandenbossche, Deroost, Soetens, & Kerckhofs, 2009), only participants scoring normal to high on global cognitive functioning, as measured by the Scales for Outcomes in Parkinson’s Disease-Cognition (SCOPACOG; Marinus et al., 2003), were recruited to participate in the current study. Consequently, we anticipated that FRs, nFRs, and HCs would show preserved learning under single-task (ST) conditions. To increase cognitive demands, we additionally implemented a secondary task (Jiménez & Vazquez, 2005; Shanks, Rowland, & Ranger, 2005). Previous research indicated that PD patients (without FOG) are still able to learn under DT conditions as long as cognitive demands do not exceed attention capacity limits (Kelly, Jahanshahi, & Dirnberger, 2004). However, given that controlled processes are hampered in FOG (Amboni et al., 2008; Naismith et al., 2010; Vandenbossche et al., 2011, 2012), we expected additional working memory load to completely eliminate learning in FRs.

Method Participants Twenty-eight PD patients (14 with FOG [FRs]; 14 without FOG, [nFRs]) and 14 HCs participated in the study. All PD patients were diagnosed by a neurologist specialized in movement

29

disorder. Diagnosed patients were subjected to both neurological and neuropsychological examinations. Patients who scored above zero on the new freezing of gait questionnaire (NFOGQ; Nieuwboer et al., 2009), defined by at least one episode of FOG during the preceding month, were assigned to the FR group. The NFOGQ score was constructed in three parts: (a) distinguishing FRs from nFRs, (b) rating the severity of freezing, and (c) rating the impact of FOG on daily life. A video showing several examples of FOG was presented to increase the recognition of this phenomenon in patients. All participants had normal to corrected-to-normal vision and had no orthopedic or additional neurological disorders. Participation in the experiment was voluntary, with informed consent in accordance with the Ethics Committee of the Vrije Universiteit Brussel (VUB). All three groups were matched for age, gender, and education (see Table 1). All participants scored above the standard cutoff score of 24 on the Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975). The MMSE was used as a general screening instrument for intellectual functioning. In addition, the SCOPA-COG (Marinus et al., 2003) was used to select patients with average or relatively high global cognitive functions in PD (Verbaan et al., 2011). The SCOPA-COG is comprised of 10 items, including subscales as “memory” (replicating the order in which cubes were pointed out, digit span backward, immediate and delayed word recall), “attention” (counting down by threes and months backward), “executive functioning” (successive repetitions of fist-edge-palm movements, set shifting with dice and fluency animals), and “visuospatial functioning” (mental reconstruction of figures). A measure of affective disturbance was obtained by administering the Hospital and Anxiety Depression Scale (HADS; Zigmond & Snaith, 1983). Scores higher than 7 points on HADS subscales indicated increased complaints associated with anxiety and depression. Measures of task switching (Brixton Spatial Anticipation Test; Burgess & Shallice, 1997) and cognitive flexibility (Controlled Oral Word Association Test [COWAT]; Benton, Hamsher, & Sivan, 1994) were also administered to assess executive functions and their relation with sequence learning under ST and DT conditions. Low scores on these neuropsychological tests are indicative of a deficit (the reverse is true for the Brixton Spatial Anticipation Test).

Design and Procedure The study took place in the participants’ home environment and the experiment was conducted on an Intel Core 2 Duo portable computer with 15.6-in. screen, using E-prime Version 1.1 software (Schneider, Eschman, & Zuccolotto, 2002). SRT testing took place individually under the supervision of the examiner. All participants were asked to complete the SRT task under both ST and DT conditions. To diminish fluctuations in cognitive and motor disease profiles, we implemented 1-week intervals between both sessions. Because the SRT task contains an important motor component (Deroost & Soetens, 2006a), testing occurred in the ONphase, about 60 to 90 min after patients took their morning dose of antiparkinson medication. SRT task order was counterbalanced: Half of the participants of each group started with the SRT-ST task, whereas the other half started with the SRT-DT task. SRT-ST. To investigate implicit learning under ST conditions, we used a standard SRT task (Nissen & Bullemer, 1987).

VANDENBOSSCHE ET AL.

30

Table 1 Clinical, Motor, and Neuropsychological Measurements Across PD Groups (FRs and NFRs) and Healthy Controls (HCs)

Measure Gender (male/female) H&Y (ON) (2; 2.5; 3) Levodopa therapy (n) Adjunct therapy (n) Disease duration (years) UPDRS–III (ON) (1) Rest tremor (Item 20) (2) Action and postural tremor (Item 21) (3) Rigidity (Item 22) (4) Repetitive movements (Items 23–26) (5) Bradykinesia (Item 31) Age (years) Education (years in school) Mini-Mental State Examination (ON) HADS–Anxiety (ON) HADS–Depression (ON) SCOPA–COG (ON) (1) Memory and learning (2) Attention (3) Executive functions (4) Visuospatial functions Brixton spatial anticipation test (ON) COWAT (ON)

FRs (n ⫽ 14)

nFRs (n ⫽ 14)

HCs (n ⫽ 14)

M

M

M

SD

11/3 (79% male) (42%; 29%; 29%) 14 (100%) 12 (86%) 10.21 2.97 37.21 15.99 0.93 2.20 1.14 1.16 9.21 3.66 15.36 6.72 3.71 2.55 65.72 7.91 19.36 2.84 27.93 1.07 6.57 3.13 5.93 4.53 27.64 5.00 10.28 3.43 3.21 0.80 9.71 1.90 4.43 0.51 11.43 6.60 23.79 9.40

SD

11/3 (79% male) (29%; 57%; 14%) 13 (93%) 13 (93%) 8.21 3.40 35.64 9.10 1.14 1.70 1.50 1.70 8.00 2.94 13.93 4.12 4.57 2.17 68.03 5.11 20.21 3.70 28.79 1.37 5.64 3.56 6.07 2.79 30.43 4.29 11.71 3.15 3.71 0.61 10.64 1.15 4.36 0.63 12.71 6.71 24.57 10.93

SD

11/3 (79% male) NA NA NA NA NA NA NA NA NA NA 67.07 6.64 19.14 3.08 29.00 1.30 5.29 5.24 4.50 4.01 32.43 5.06 13.29 4.03 4.00 0.00 10.64 1.74 4.50 0.65 8.64 2.73 32.79 9.92

p value

␩2

.109 .752 .776 .522 .342 .504 .348 .658 .653 .069 .693 .496 .040a .095 .004b .232 .822 .164 .043c

.096 .004 .003 .016 .035 .017 .034 .021 .022 .128 .019 .035 .153 .113 .251 .072 .010 .089 .149

Note. COWAT ⫽ Controlled Oral Word Association Test; FRs ⫽ freezers; H&Y ⫽ Hoehn and Yahr rating scale; HADS ⫽ Hospital and Anxiety Depression Scales; M ⫽ mean; NA ⫽ not applicable; nFRs ⫽ nonfreezers; ON ⫽ on phase; PD ⫽ Parkinson’s disease; SCOPA-COG ⫽ Scales for Outcomes in Parkinson’s disease–Cognition; SE ⫽ standard error of mean; UPDRS–III ⫽ Unified Parkinson’s Disease Rating Scale. a Post hoc testing: HCs vs FRs significant (p ⫽ .036, d ⫽ 0.95). b Post hoc testing: HCs vs FRs significant (p ⫽ .003, d ⫽ 1.40). c Post hoc testing: HCs vs FRs significant (p ⫽ .071, d ⫽ 0.93).

Before starting the experimental SRT blocks, participants completed a practice block of 50 trials in random order to train the stimulus–response mapping. After practice, they executed 12 experimental blocks of 72 trials. Implicit learning of a deterministic first-order conditional (FOC) sequence was tested: 132342134142 (ST) and 241431243231 (DT). The numbers 1 to 4 denoted the leftmost, left, right, and rightmost target position, respectively. Both sequence structures were identically (1 ⫽ 2; 3 ⫽ 4) so that possible differences in sequence learning could not be attributed to differences in sequence structure. The FOC sequence was continuously repeated over the experimental Blocks 1 to 12. During Block 11, the sequence was presented in a random order to assess sequence learning. For details, we refer to our previous work (Vandenbossche et al., 2009). SRT-DT. The procedure for the SRT-DT task was similar to the one used in the SRT-ST task. The only difference between both tasks was the presence of a secondary counting task during training phase (practice block ⫹ experimental Blocks 1 to 9). Participants were instructed to count the number of “gunshots” heard among irrelevant low-pitched tones (1000 Hz). Both sounds had a duration of 50 ms and were provided through a headphone. The shots were presented in the time interval between a participant’s response to the target dot and the presentation of the next target in the SRT task (50 ms response–stimulus interval), and varied between 24 and 30 times per block. After each block, they were asked to report the perceived number of gunshots they counted during the previous block. We used gunshots instead of common high-pitched tones in an effort to increase stimulus contrast. Par-

ticipants were encouraged to keep their average gunshot-counting accuracy to less than 10% in errors. The counting task was deleted during the testing phase (Blocks 10, 11, and 12) to determine the effect of the secondary task on learning assessed under similar conditions as the SRT-ST task (Frensch, Lin, & Buchner, 1998). Sequence awareness questionnaire. At the end of the second session, we administered a uniform questionnaire to assess awareness of the sequence in all participants. This was important because learning effects could be larger when sequence awareness is more developed. Measures to categorize (able to reproduce a part of, minimally, two successive elements of the sequence) and to quantify awareness (percentage of the entire sequence correctly reproduced) were tested with sequence-related questions (“Did you notice something particular?”; “Did the target dots appear randomly on the screen or was there a regularity involved?”; “Was this regularity present in all of the blocks?”; “When you noticed a regularity, try to describe it as precise and accurate as possible?”).

Results Independent t tests and a correlation analysis (Pearson statistics) were used to analyze group differences for clinical, motor, and demographical measures, and their relationship with severity of freezing. Mixed factorial ANOVAs (with Huyhn–Feldt corrections for violations of sphericity) were implemented to analyze training and sequence learning effects for SRT-ST and SRT-DT performance. Bonferroni post hoc tests were performed in case of significant group differences. Data analysis was performed using

SEQUENCE LEARNING IN FREEZING OF GAIT

SPSS Version 17.0 and all analyses were two-tailed, using a significance level of 0.05. Differences showing a significance level of 0.10 were reported as a tendency.

Clinical, Motor, and Neuropsychological Differences At the time of testing, all patients were in the ON-phase of antiparkinson medication. Medications used by patients consisted mostly of levodopa products (Prolopa, Stalevo, and Sinemet Control) and adjunct therapies (Mirapexin, ReQuip, Amantan, Eldepryl, Comtan, and Azilect). One nFR and one FR were on anticholinergic treatment (Artane) and two patients (one in each group) were taking a selective serotonin reuptake inhibitor (Sipralexa). The Hoehn and Yahr (1967) scale, providing a gross assessment of disease progression through several stages (0 to 5), from no signs of the disease to complete dependency, and Section III of the Unified Parkinson’s Disease Rating Scale (UPDRS; Fahn, Elton, & Members of the UPDRS Development Committee, 1987), administering global motor evaluation and specific subscales (e.g., rest tremor, action and postural tremor, rigidity, repetitive movements, bradykinesia), were both assessed in the ON phase. Despite normal cognition scores on the MMSE for all groups, there were significant differences between groups for both the global SCOPA-COG score (HCs ⬎ FRs, p ⬍ 0.05, d ⫽ 0.95) and performance on the attention subscale (HCs ⬎ FRs, p ⬍ 0.05, d ⫽ 1.40; see Table 1). Although a one-way ANOVA pointed to a main significant difference for cognitive flexibility (COWAT), post hoc analysis only showed a tendency between FRs and HCs (FRs ⬍ HCs; p ⫽ 0.07, d ⫽ 0.93). Differences between groups for the remaining general descriptive cognitive, affective, and motor measures (including medication profiles) all failed to reach significance.

SRT Task RT analysis was performed on participants’ median RTs of correct trials, with the exclusion of practice trials. Erroneous responses and responses following an error were discarded from the analysis. Results primarily focused on the RT data: Because of a higher number of observations, RT data provide a more informative and reliable estimate of learning abilities than error rates. Error rates generally also have the tendency to fluctuate in a nonspecific direction over experimental blocks, often complicating the interpretation of their exact significance. For completeness, however, error analyses are reported for sequence-specific learning.1

SRT-ST Training effects. We first investigated general training effects for all groups, indicated by the decrease in RTs across all experimental blocks, except Random Block 11. To estimate training effects, we carried out a repeated measures ANOVA with group (FRs, nFRs, and HCs) as a between-subjects factor and block (all experimental blocks, with the exclusion of Random Block 11) as a within-subjects factor. Due to the z-score transformation, there was no main effect of group, F(2, 39) ⫽ 0.02, p ⫽ 0.99, ␩p2 ⬍ 0.01. As predicted, the main effect of block was significant, implying a decrease in RTs over training, F(2.18,84.94) ⫽ 18.14, p ⬍ 0.001,

31

␩p2 ⫽ 0.32. No significant Group ⫻ Block interaction, however, could be found, meaning that there were no differences in training effect between groups, F(4.36,84.94) ⫽ 1.00, p ⫽ 0.42, ␩p2 ⫽ 0.05 (see Figure 1a).

Sequence Learning Sequence-specific learning was determined for each group by comparing RTs/error rates in Random Block 11 with the mean of the adjacent sequenced Blocks 10 and 12. Slower performance in Random Block 11 compared with the surrounding sequenced blocks implied that sequence-specific learning took place. RTs. A repeated measures ANOVA was carried out with group (FRs, nFRs, and HCs) as a between-subjects factor and sequence learning (Random Block 11 vs. the mean of Blocks 10 and 12) as a within-subjects factor. Due to z-score transformation, there was no main effect of group, F(2, 39) ⫽ 0.05, p ⫽ 0.94, ␩p2 ⬍ 0.01. Importantly, both sequence learning, F(1, 39) ⫽ 94.77, p ⬍ 0.001, ␩p2 ⫽ 0.71, and the Sequence Learning ⫻ Group interaction was significant, F(2, 39) ⫽ 16.09, p ⬍ 0.001, ␩p2 ⫽ 0.45. A Post hoc Bonferroni test showed that FRs (transformed, M ⫽ 0.09, SD ⫽ 0.17; untransformed, M ⫽ 26.82 ms, SD ⫽ 51.03 ms) were more impaired in sequence learning compared with nFRs (transformed, M ⫽ 0.33, SD ⫽ 0.24; untransformed, M ⫽ 41.73 ms, SD ⫽ 30.81 ms), who, in turn, experienced a lower learning effect than HCs (transformed, M ⫽ 0.54, SD ⫽ 0.22; untransformed, M ⫽ 47.13 ms, SD ⫽ 19.28 ms), all p ⬍ 0.05 (except for FRs vs. HCs, p ⬍ 0.001). A paired-samples t test showed that sequence learning effects were significant for the HCs and nFRs, t(13) ⫽ 9.15, p ⬍ 0.001, d ⫽ 5.08, and t(13) ⫽ 5.07, p ⬍ 0.001, d ⫽ 2.81, respectively. Freezers, however, only showed a tendency toward significance in acquiring sequence-specific knowledge, t(13) ⫽ 1.97, p ⫽ 0.07, d ⫽ 1.09 (see Figure 2a). These results indicate impaired implicit learning under ST conditions in FRs. Error rates. A repeated measures ANOVA was carried out with group (FRs, nFRs, and HCs) as a between-subjects factor and sequence learning (Random Block 11 vs. the mean of Blocks 10 and 12) as a within-subjects factor. There was no main effect of group, F(2, 39) ⫽ 0.44, p ⫽ 0.65, ␩p2 ⫽ 0.02, nor a main effect of sequence learning, F(1, 39) ⫽ 1.18, p ⫽ 0.29, ␩p2 ⫽ 0.03, nor a Sequence Learning ⫻ Group interaction, F(2, 39) ⫽ 1.31, p ⫽ 0.28, ␩p2 ⫽ 0.06. Additional paired-samples t tests showed that no significant sequence learning could be derived from the error rates in either group (FRs, t[13] ⫽ 1.34, p ⫽ 0.20, d ⫽ 0.74; nFRs, t[13] ⫽ 1.37, p ⫽ 0.20, d ⫽ 0.76; HCs, t[13] ⫽ 0.74, p ⫽ 0.47, d ⫽ 0.41).

SRT-DT Counting performance. A one-way ANOVA with group (FRs, nFRs, and HCs) as a between-subjects factor and counting 1 RT and error rate analyses were performed on z scores in all conditions to control for baseline differences (Faust, Balota, Spieler, & Ferraro, 1999) under both ST conditions (FRs, M ⫽ 580 ms, SD ⫽ 268 ms; nFRs, M ⫽ 546 ms, SD ⫽ 115 ms; HCs, M ⫽ 451 ms, SD ⫽ 69 ms), F(1,41) ⫽ 3.97, p ⬍ 0.05, ␩2p ⫽ 0.09, and DT conditions (FRs, M ⫽ 627 ms, SD ⫽ 149 ms; nFRs, M ⫽ 696 ms, SD ⫽ 166 ms; HCs, M ⫽ 563 ms, SD ⫽ 111 ms), F(1,41) ⫽ 4.24, p ⬍ 0.05, ␩2p ⫽ 0.10.

32

VANDENBOSSCHE ET AL.

Figure 1. (a): Z-score transformations of reaction times (RTs [ms]) per block in the serial reaction time single task (SRT-ST) for the Parkinson’s disease (PD) groups (freezers [FRs] and nonfreezers [nFRs]) and the healthy controls (HCs). All blocks are structured, except for Random Block 11. Vertical bars denote standard errors. (b) Z-score transformations of RTs (ms) per block in the serial reaction time dual task (SRT-DT) for the PD groups (FRs and nFRs) and the HCs. Blocks 1 through 9 are with a secondary shot-counting task, Blocks 10 through 12 are under ST conditions. All blocks are structured, except for Random Block 11. Vertical bars denote standard errors.

performance as a dependent factor, revealed no group differences in counting performance in the DT (FRs, M ⫽ 12.91%, SD ⫽ 7.76%; nFRs, M ⫽ 9.74%, SD ⫽ 5.86%; HCs, M ⫽ 9.06%, SD ⫽ 6.79%), F(2, 41) ⫽ 1.26, p ⫽ 0.30, ␩p2 ⫽ 0.06. Training effects. We investigated a general training effect for all groups for experimental Blocks 1 through 9 (only blocks under

DT conditions were included in this analysis). A repeated measures ANOVA was run with group (FRs, nFRs, and HCs) as a between-subjects factor and block (all experimental blocks, with the exclusion of Blocks 10, 11, and 12) as a within-subjects factor. There was no main effect of group, F(2, 39) ⫽ 0.01, ns. The main effect of block, however, was significant, implying a decrease in

SEQUENCE LEARNING IN FREEZING OF GAIT

33

Figure 2. Sequence learning under (a) single-task (SRT-ST) conditions, and (b) dual-task (SRT-DT) conditions is expressed here as a difference between Random Block 11 (Random) and the mean of the adjacent structural Blocks 10 and 12 (Structured) for the Parkinson’s disease (PD) groups (freezers [FRs] and nonfreezers [nFRs]) and the healthy controls (HCs). Vertical bars denote standard errors. ⴱ (p ⬍ .10). ⴱ p ⬍ .05. ⴱⴱ p ⬍ .001.

RTs over training, F(3.25,126.72) ⫽ 21.99, p ⬍ 0.001, ␩p2 ⫽ 0.36. No significant Group ⫻ Block interaction was found, F(6.5,126.72) ⫽ 0.47, p ⫽ 0.84, ␩p2 ⫽ 0.02 (see Figure 1b).

Sequence Learning RTs. A repeated measures ANOVA was carried out with group (FRs, nFRs, and HCs) as between-subjects factor and sequence learning (Random Block 11 vs. the mean of Blocks 10 and 12) as within-subjects factor. No main effect of group was revealed, F(2, 39) ⫽ 0.23, p ⫽ 0.79, ␩p2 ⫽ 0.01. Higher scores in Random Block 11, compared with the mean of adjacent Blocks 10 and 12, indicated significant sequence learning under DT conditions, F(1, 39) ⫽ 8.13, p ⬍ 0.01, ␩p2 ⫽ 0.17. However, the Sequence Learning ⫻ Group interaction failed to reach significance, F(2, 39) ⫽ 1.02, p ⫽ 0.37, ␩p2 ⫽ 0.05, suggesting that sequence learning did not differ between groups. To determine sequence-specific learning effects in each group, we used paired-samples t tests. Similar to our findings in the SRT-ST task, we found significant sequence learning effects for the HCs (transformed, M ⫽ 0.11, SD ⫽ 0.14; untransformed, M ⫽ 15.50 ms, SD ⫽ 19.17 ms) and nFRs (transformed, M ⫽ 0.11, SD ⫽ 0.19; untransformed, M ⫽ 22.07 ms, SD ⫽ 38.52 ms), respectively, t(13) ⫽ 3.03, p ⫽ 0.01, d ⫽ 1.68 and t(13) ⫽ 2.14, p ⫽ 0.05, d ⫽ 1.19. However, FRs did not acquire sequencespecific learning (transformed, M ⫽ 0.02, SD ⫽ 0.22; untransformed, M ⫽ 4.23 ms, SD ⫽ 38.77 ms), t(13) ⫽ 0.41, p ⫽ 0.69, d ⫽ 0.23 (see Figure 2b). Thus, sequence learning under DT conditions proved to be absent in FRs. Error rates. A repeated measures ANOVA was carried out with group (FRs, nFRs, and HCs) as a between-subjects factor and sequence learning (Random Block 11 vs. the mean of Blocks 10

and 12) as a within-subjects factor. The main effect of group was not significant, F(2, 39) ⫽ 0.45, p ⫽ 0.64, ␩p2 ⫽ 0.02. Higher scores in Random Block 11 compared with the mean of adjacent Blocks 10 and 12 revealed sequence learning under DT conditions, F(1, 39) ⫽ 19.18, p ⬍ 0.001, ␩p2 ⫽ 0.33. Similar to the RT analyses, the Sequence Learning ⫻ Group interaction failed to reach significance, F(2, 39) ⫽ 0.40, p ⫽ 0.68, ␩p2 ⫽ 0.02. Additional paired-samples t tests showed that FRs, nFRs, and HCs all were able to acquire sequence specific knowledge, respectively, t(13) ⫽ 3.78, p ⬍ 0.01, d ⫽ 2.10, t(13) ⫽ 2.15, p ⫽ 0.05, d ⫽ 1.19, and t(13) ⫽ 2.15, p ⫽ 0.05, d ⫽ 1.19 (FRs transformed, M ⫽ ⫺15.60, SD ⫽ 0.51, and untransformed, M ⫽ 3.23%, SD ⫽ 3.19%; nFRs transformed, M ⫽ ⫺17.49, SD ⫽ 0.54, and untransformed, M ⫽ 1.74%, SD ⫽ 3.02%; HCs transformed, M ⫽ ⫺36.36, SD ⫽ 0.82, and untransformed, M ⫽ 1.29%, SD ⫽ 2.24%).

Sequence Awareness Questionnaire Although no participant could reproduce the sequence entirely, up to six FRs, five nFRs, and five HCs were able to reproduce at least two consecutive elements of the sequence correctly (derived from the question, “When you noticed a regularity, try to describe it as precise and accurate as possible?”), and were therefore categorized as “aware.” A repeated measures ANOVA with awareness (aware, not aware) as a between-subjects factor and sequence learning (Random Block 11 vs. the mean of Blocks 10 and 12) as a within-subjects factor was conducted for ST and DT conditions. Significant main sequence learning effects (ST, F[1, 40] ⫽ 45.84, p ⬍ 0.001, ␩p2 ⫽ 0.53; DT, F[1, 40] ⫽ 6.35, p ⫽ 0.02, ␩p2 ⫽ 0.14), but not the Significant Learning ⫻ Awareness interactions (ST, F[1, 40] ⫽ 0.36, p ⫽ 0.55, ␩p2 ⬍ 0.01; DT, F[1, 40] ⫽ 0.09, p ⫽

VANDENBOSSCHE ET AL.

34

0.77, ␩p2 ⬍ 0.01), showed that sequence learning did not depend on the participants’ awareness of any sequence regularity in the SRT task. Subsequently, we determined how well subjects could reproduce the entire sequence correctly by calculating scores on the post-test questionnaire, reflected in percentages (a score of 100% was assigned when all 12 positions were predicted correctly). A high percentage indicates a high chance that subjects were aware of the sequence. The FR group (M ⫽ 11.29%, SD ⫽ 14.03%), nFR group (M ⫽ 14.86%, SD ⫽ 21.93%), and HC group (M ⫽ 9.57%, SD ⫽ 15.62%) did not differ significantly on the explicit awareness task, F(2, 41) ⫽ 0.33, p ⫽ 0.72, ␩p2 ⬍ 0.02. Correlations between the questionnaire scores and RT sequence learning (Random Block 11 vs. the mean of Blocks 10 and 12) on the SRT-ST (r ⫽ 0.10, p ⫽ 0.55) and SRT-DT task (r ⫽ 0.10, p ⫽ 0.53) were not significant, confirming that participants’ performance was not determined by sequence awareness. Important to note is also that cognitive measures (SCOPA-COG, MMSE, Brixton, and COWAT) showed poor correlation with awareness in FRs (all ps ⬎ 0.10).

Correlation Analysis Neuropsychological Measures and SRT Task Table 2 shows correlations between clinical and neuropsychological measures and sequence learning effects (difference between Random Block 11 and the mean of Blocks 10 and 12) under ST and DT conditions. Interestingly, global cognitive functioning

(MMSE) correlated positively with both ST and DT learning. Furthermore significant positive correlations were found between DT sequence learning and (a) global SCOPA-COG score, (b) learning and memory, and (c) executive functioning subscales. Importantly, a significant negative correlation was observed between the SRT-DT sequence learning and severity of freezing (FRs, M ⫽ 13.2, SD ⫽ 7.03). None of the other clinical or neuropsychological tests related significantly to implicit learning.

General Discussion The goal of the current study was to investigate learning of automatic behavior with or without cognitive load in FOG. Our results show that sequence learning under ST conditions was present in FRs, nFRs, and HCs. However, FRs only showed a tendency toward a learning effect. This is important to consider, as general motor characteristics cannot account for differences in learning between FRs and nFRs. More precisely, disease duration and the UPDRS-III scores (including subscores for rest tremor, action and postural tremor, rigidity, repetitive movements, and bradykinesia) were similar. Thus, although FRs and nFRs had comparable general motor profiles, only FRs showed deficits in acquiring novel motor activities. This indicates that FOG patients suffer from a specific impairment in learning automatic behavior. When observing the correlations between the MMSE score and SRT-ST learning, one could argue that these differences in learning could be explained by differences in global cognitive function-

Table 2 Correlations Between Clinical and Neuropsychological Measures and Implicit Sequence Learning Effects Under Single Task (ST) and Dual Task (DT) Conditions Implicit learning ST

Implicit learning DT

FRs, nFRs, HCs (n ⫽ 42)

r (Pearson)

p value

r (Pearson)

p value

Age (years) Disease duration (years) Education H&Y (ON) UPDRS–III (ON) (1) Rest tremor (Item 20) (2) Action and postural tremor (Item 21) (3) Rigidity (Item 22) (4) Repetitive movements (Items 23–26) (5) Bradykinesia (Item 31) NFOGQ HADS–Anxiety HADS–Depression MMSE SCOPA⫺COG (ON) (1) Memory and learning (2) Attention (3) Executive functions (4) Visuospatial functions COWAT (ON) Brixton spatial anticipation test (ON) Implicit learning ST Implicit learning DT

⫺.11 ⫺.17 ⫺.08 ⫺.13 .05 ⫺.03 ⫺.29 .05 ⫺.08 .14 ⫺.35 ⫺.04 ⫺.01 .36 .24 .25 .12 .20 ⫺.18 .24 .08 1.00 .43

.50 .40 .60 .50 .79 .89 .13 .80 .70 .48 .24 .78 .99 .02ⴱ .13 .12 .45 .20 .26 .13 .63 NA .004ⴱⴱ

⫺.01 ⫺.18 .06 ⫺.12 ⫺.10 .14 ⫺.33 ⫺.23 ⫺.25 .22 ⫺.56 ⫺.07 ⫺.05 .37 .34 .31 ⫺.10 .32 .22 .27 ⫺.16 .43 1.00

.95 .36 .70 .55 .60 .48 .09 .24 .20 .27 .04ⴱ .65 .76 .02ⴱ .03ⴱ .05ⴱ .55 .04ⴱ .16 .08 .31 .004ⴱⴱ NA

Note. COWAT ⫽ Controlled Oral Word Association Test; FRs ⫽ freezers; H&Y ⫽ Hoehn & Yahr rating scale; HADS ⫽ Hospital and Anxiety Depression Scales; HCs ⫽ healthy controls; MMSE ⫽ Mini-Mental State Examination; NA ⫽ not applicable; NFOGQ ⫽ Revised Freezing of Gait Questionnaire; nFRs ⫽ nonfreezers; ON ⫽ on phase; SCOPA-COG ⫽ Scales for Outcomes in Parkinson’s Disease–Cognition; UPDRS–III ⫽ Unified Parkinson’s Disease Rating Scale. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.

SEQUENCE LEARNING IN FREEZING OF GAIT

ing between FRs, nFRs, and HCs (Vandenbossche et al., 2009). Although dissimilarities on several cognitive measures did not reach significance between FRs and nFRs, subtle contrasts could possibly influence the acquisition of automaticity. Differences in cognitive capacity between FRs and nFRs, however, can never be completely ruled out because FOG is essentially characterized by a genuine cognitive deficit. Nevertheless, we aimed to mitigate this factor by only selecting patients in the FR group (M ⫽ 27.64, SD ⫽ 5.00; see Table 1) who scored within the boundaries of normal cognition as assessed in the SCOPA-COG (see PROPARK study, Verbaan et al., 2007; M ⫽ 28.0, SD ⫽ 4.6). We also expected less sequence learning in FRs under DT conditions compared with nFRs and HCs. However, a nonsignificant interaction effect indicated that FRs, nFRs, and HCs acquired sequence-specific knowledge under additional working memory load equally. Likely, the rather small sample sizes, a limitation of the current study, as well as reduced learning under DT conditions in all groups,2 probably challenged our ability to find significant group differences. Nevertheless, it is important to note that no sequence learning in the RTs emerged in patients with FOG under DT conditions when learning was assessed for each group separately (albeit some in errors), whereas HCs and nFRs did manage to learn the sequence to some extent. However, we should be prudent not to draw too strong of a conclusion, as error rate analyses revealed that FRs potentially acquired sequential knowledge under DT conditions. Thus, overall, our results suggest that implicit sequence learning is not only impaired in FOG but also tends to be more susceptible to cognitive load in FRs. These findings confirm with previous research investigating gait under dual-task conditions in patients with FOG (Hausdorff, Schaafsma, et al., 2003; Spildooren et al., 2010). Automatic motor activities are apparently affected by the manipulation of cognitive load in FOG. An interesting question for future research would therefore be whether the acquisition of novel motor actions is also able to evoke FOG episodes. Furthermore, correlation analyses revealed an interesting association between the SRT-DT task and severity of FOG, measured by the NFOGQ. As FOG pathology worsened, sequence learning under DT conditions declined. However, correlations with executive functioning were not significant. Yet, as a limitation of the study, diagnostic screening measures—instead of more sensitive computerized tasks—were used to assess executive functioning. Future research is necessary to address the relation between executive functions, dual tasking and FOG. Important to note is that although FRs showed reduced learning compared with nFRs and HCs, they still managed to acquire some knowledge under ST conditions. This could lead to an important direction for rehabilitation, namely, that (re)learning of automatic behavior remains possible in FOG as long as actions are simple and extensive training is provided. For example, a strategy to ameliorate automaticity could be lowering cognitive demands by defragmenting actions into smaller parts. This would also have a positive influence on the availability of remaining cognitive resources in FOG.

Conclusions The present study is the first to show that PD patients suffering from FOG pathology exhibit a specific impairment in learning

35

automatic behavior. When cognitive capacity is supplementarily loaded by adding a secondary task, automaticity in freezers becomes increasingly compromised. Cognitive load therefore seems to interrupt both learning and execution of automatic behavior, hereby possibly triggering FOG.

2 We carried out a repeated measures ANOVA on transformed sequence learning effects with group (FRs, nFRs, and HCs) as a between-subjects factor and task (ST and DT) as a within-subjects factor. A main effect of group showed that sequence learning levels differed between groups, F(2,39) ⫽ 10.18, p ⬍ 0.001, ␩2p ⫽ 0.34. Both the main effect of task, F(1,39) ⫽ 43.67, p ⬍ 0.001, ␩2p ⫽ 0.53, as well as the Task ⫻ Group interaction were significant, F(2,39) ⫽ 8.74, p ⫽ 0.001, ␩2p ⫽ 0.31. A post hoc Bonferroni test showed that sequence learning under DT conditions, when compared with an ST, declined significantly more in HCs (M ⫽ 0.43, SD ⫽ 0.26) than in FRs (M ⫽ 0.06, SD ⫽ 0.19; p ⬍ 0.05, d ⫽ 1.62) and tended to decline more in HCs than in nFRs (M ⫽ 0.22, SD ⫽ 0.25; p ⫽ 0.06, d ⫽ 0.82).

References Amboni, M., Cozzolino, A., Longo, K., Picillo, M., & Barone, P. (2008). Freezing of gait and executive functions in patients with Parkinson’s disease. Movement Disorders, 23, 395– 400. doi:10.1002/mds.21850 Benton, L. A., Hamsher, K., & Sivan, A. B. (1994). Controlled oral word association test. Multilingual aphasia examination. (3rd ed.). Iowa City, IA: AJA Associates. Burgess, P., & Shallice, T. (1997). The Hayling and Brixton Tests. Test manual. Bury St Edmunds, UK: Thames Valley Test Company. Deroost, N., Kerckhofs, E., Coene, M., Wijnants, G., & Soetens, E. (2006). Learning sequence movements in a homogenous sample of patients with Parkinson’s disease. Neuropsychologia, 44, 1653–1662. doi:10.1016/j .neuropsychologia.2006.03.021 Deroost, N., & Soetens, E. (2006a). Perceptual or motor learning in SRT tasks with complex sequence structures. Psychological Research, 70, 88 –102. doi:10.1007/s00426-004-0196-3 Deroost, N., & Soetens, E. (2006b). Spatial processing and perceptual sequence learning in SRT Tasks. Experimental Psychology, 53, 16 –30. doi:10.1027/1618-3169.53.1.16 Fahn, S., Elton, R., & Members of the UPDRS Development Committee. (1987). Unified Parkinson’s Disease Rating Scale. In M. C. D. Fahn, D. B. Calne, & M. Goldstein (Eds.), Recent developments in Parkinson’s disease (pp. 153–163, 293–304). Florham Park, NJ: MacMillan. Faust, M. E., Balota, D. A., Spieler, D. H., & Ferraro, F. R., (1999). Individual differences in information processing rate and amount: Implications for group differences in response latency. Psychological Bulletin, 125, 777–799. doi:10.1037/0033-2909.125.6.777 Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). “Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12, 189 –198. doi: 10.1016/0022-3956(75)90026-6 Frensch, P. A., Lin, J., & Buchner, A. (1998). Learning versus behavioral expression of the learned: The effects of a secondary tone-counting task on implicit learning in the serial reaction task. Psychological Research, 61, 83–98. doi:10.1007/s004260050015 Giladi, N., & Hausdorff, J. M. (2006). The role of mental function in the pathogenesis of freezing of gait in Parkinson’s disease. Journal of the Neurological Sciences, 248, 173–176. doi:10.1016/j.jns.2006.05.015 Giladi, N., McDermott, M. P., Fahn, S., Przedborski, S., Jankovic, J., Stern, M., . . . Tanner, C. (2001). Freezing of gait in PD: Prospective assessment in the DATATOP cohort. Neurology, 56, 1712–1721. doi:10.1212/ WNL.56.12.1712

36

VANDENBOSSCHE ET AL.

Giladi, N., & Nieuwboer, A. (2008). Understanding and treating freezing of gait in parkinsonism. Proposed working definition and setting the stage. Movement Disorders, 23, S423–S425. doi:10.1002/mds.21927 Hausdorff, J. M., Balash, Y., & Giladi, N. (2003). Effects of cognitive challenge on gait variability in patients with Parkinson’s disease. Journal of Geriatric Psychiatry and Neurology, 16, 53–58. Hausdorff, J. M., Schaafsma, J. D., Balash, Y., Bartels, A. L., Gurevich, T., & Giladi, N. (2003). Impaired regulation of stride variability in Parkinson’s disease subjects with freezing of gait. Experimental Brain Research, 149, 187–194. Hoehn, M. M., & Yahr, M. D. (1967). Parkinsonism: Onset, progression, and mortality. Neurology, 17, 427– 442. doi:10.1212/WNL.17.5.427 Jankovic, J. (2008). Parkinson’s disease: Clinical features and diagnosis. Journal of Neurology, Neurosurgery, & Psychiatry, 79, 368 –376. doi: 10.1136/jnnp.2007.131045 Jiménez, L., & Vazquez, G. A. (2005). Sequence learning under dual-task conditions: Alternatives to a resource-based account. Psychological Research, 69, 352–368. doi:10.1007/s00426-004-0210-9 Kamsma, Y. P. T., Brouwer, W. H., & Lakke, J. P. W. (1995). Training of compensational strategies for impaired gross motor skills in Parkinson’s disease. Physiotherapy: Theory and Practice, 11, 209 –229. doi:10.3109/ 09593989509036407 Kelly, S. W., Jahanshahi, M., & Dirnberger, G. (2004). Learning of ambiguous versus hybrid sequences by patients with Parkinson’s disease. Neuropsychologia, 42, 1350 –1357. doi:10.1016/j.neuropsychologia .2004.02.013 Lewis, S. J., & Barker, R. A. (2009). A pathophysiological model of freezing of gait in Parkinson’s disease. Parkinsonism & Related Disorders, 15, 333–338. doi:10.1016/j.parkreldis.2008.08.006 Marinus, J., Visser, M., Verwey, N. A., Verhey, F. R. J., Middelkoop, H. A. M., Stiggelbout, A., & van Hilten, J. J. (2003). Assessment of cognition in Parkinson’s disease. Neurology, 61, 1222–1228. doi: 10.1212/01.WNL.0000091864.39702.1C Naismith, S. L., Shine, J. M., & Lewis, S. J. (2010). The specific contributions of set-shifting to freezing of gait in Parkinson’s disease. Movement Disorders, 25, 1000 –1004. doi:10.1002/mds.23005 Nieuwboer, A., Rochester, L., Herman, T., Vandenberghe, W., Emil, G. E., Thomaes, T., & Giladi, N. (2009). Reliability of the new freezing of gait questionnaire: Agreement between patients with Parkinson’s disease and their carers. Gait & Posture, 30, 459 – 463. doi:10.1016/j.gaitpost.2009 .07.108 Nissen, M. J., & Bullemer, P. T. (1987). Attentional requirements for learning: Evidence from performance measures. Cognitive Psychology, 19, 1–32. doi:10.1016/0010-0285(87)90002-8 Nutt, J. G., Bloem, B. R., Giladi, N., Hallett, M., Horak, F. B., & Nieuwboer, A. (2011). Freezing of gait: Moving forward on a mysterious clinical phenomenon. The Lancet Neurology, 10, 734 –744. doi:10.1016/S1474-4422(11)70143-0

O’Shea, S., Morris, M. E., & Iansec, R. (2002). Dual task interference during gait in patients with Parkinson’s disease: Effects of motor versus cognitive secondary tasks. Physical Therapy, 82, 888 – 897. Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-prime version 1.1. Pittsburgh, PA: Psychology Software Tools. Shanks, D. R., Rowland, L. A., & Ranger, M. S. (2005). Attentional load and implicit sequence learning. Psychological Research, 69, 369 –382. doi:10.1007/s00426-004-0211-8 Spildooren, J., Vercruysse, S., Desloovere, K., Vandenberghe, W., Kerckhofs, E., & Nieuwboer, A. (2010). Freezing of gait in Parkinson’s disease: The impact of dual-tasking and turning. Movement Disorders, 25, 2563–2570. doi:10.1002/mds.23327 Stocco, A., Lebiere, C., & Anderson, J. R. (2010). Conditional routing of information to the cortex: A model of the basal ganglia’s role in cognitive coordination. Psychological Review, 117, 541–574. doi:10.1037/ a0019077 Vandenbossche, J., Deroost, N., Soetens, E., & Kerckhofs, E. (2009). Does implicit learning in nondemented Parkinson’s disease depend on the level of cognitive functioning? Brain and Cognition, 69, 194 –199. doi:10.1016/j.bandc.2008.07.005 Vandenbossche, J., Deroost, N., Soetens, E., Spildooren, J., Vercruysse, S., Nieuwboer, A., & Kerckhofs, E. (2011). Freezing of gait in Parkinson disease is associated with impaired conflict resolution. Neurorehabilitation and Neural Repair, 25, 765–773. doi:10.1177/1545968311403493 Vandenbossche, J., Deroost, N., Soetens, E., Zeischka, P., Spildooren, J., Vercruysse, S., . . . Kerckhofs, E. (2012). Conflict and freezing of gait in Parkinson disease: Support for a response control deficit. Neuroscience, 206, 144 –154. doi:10.1016/j.neuroscience.2011.12.048 Verbaan, D., Jeukens-Visser, M., Van Laar, T., van Rooden, S. M., Van Zwet, E. W., Marinus, J., & van Hilten, J. J. (2011). SCOPA-cognition cutoff value for detection of Parkinson’s disease dementia. Movement Disorders, 26, 1881–1886. doi:10.1002/mds.23750 Verbaan, D., Marinus, J., Visser, M., van Rooden, S. M., Stiggelbout, A. M., Middelkoop, H. A. M., & van Hilten, J. J. (2007). Cognitive impairment in Parkinson’s disease. Journal of Neurology, Neurosurgery and Psychiatry, 78, 1182–1187. doi:10.1136/jnnp.2006.112367 Willingham, D. B. (1999). Implicit motor sequence learning is not purely perceptual. Memory & Cognition, 27, 561–572. doi:10.3758/ BF03211549 Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta Psychiatrica Scandinavica, 67, 361–370. doi:10.1111/j .1600-0447.1983.tb09716.x

Received September 9, 2011 Revision received November 6, 2012 Accepted November 13, 2012 䡲