2
C. Mayor-Duboisa,∗, P. Maederb, P. Zesigerc, E. Roulet-Pereza
3
aPediatric Neurology and Neurorehabilitation Unit, University Hospital, Lausanne, Switzerland 4
bRadiology, University Hospital, Lausanne, Switzerland 5
cFaculty of Psychology and Educational Sciences, University of Geneva, Switzerland 6
Received in revised form 24 March 2010 12
We investigated procedural learning in 18 children with basal ganglia (BG) lesions or dysfunctions of various aetiologies, using a visuo-motor learning test, the Serial Reaction Time (SRT) task, and a cognitive learning test, the Probabilistic Classification Learning (PCL) task. We compared patients with early (<1 year old,n= 9), later onset (>6 years old,n= 7) or progressive disorder (idiopathic dystonia,n= 2). All patients showed deficits in both visuo-motor and cognitive domains, except those with idiopathic dystonia, who displayed preserved classification learning skills. Impairments seem to be independent from the age of onset of pathology. As far as we know, this study is the first to investigate motor and cognitive procedural learning in children with BG damage. Procedural impairments were documented whatever the aetiology of the BG damage/dysfunction and time of pathology onset, thus supporting the claim of very early skill learning development and lack of plasticity in case of damage.
© 2010 Published by Elsevier Ltd.
1. Introduction
20
According to the conception of multiple memory systems
21
(Squire, 1992; Squire, Knowlton, & Musen, 1993; Squire &
22
Zola-Morgan, 1988), procedural learning is generally defined as
23
“knowing how” to do things (for review,Knowlton & Moody, 2008).
24
It is part of non-declarative memory and refers to a progressive
25
improvement of motor or cognitive performance through practice
26
or repetition. The procedural learning system includes perceptual
27
and motor skill learning as well as habit learning, which refers to
28
the gradual acquisition of stimulus-response associations. It
con-29
trasts with declarative memory (“knowing that”), a flexible system
30
responsible for the conscious retrieval of information (Cohen &
31
Squire, 1980). These different memory systems are subserved by
32
distinct neuroanatomical networks (Squire, 1992; Squire et al.,
33
1993). While declarative memory is clearly assumed by the medial
34
temporal lobes and the diencephalon, procedural learning is mainly
35
supported by the striatum (putamen and caudate nuclei), with
36
the participation of the cerebellum, the frontal cortex and other
37
cortical areas, depending on the nature of the task (Doyon et al.,
38
1997; Packard & Knowlton, 2002; Poldrack, Prabhakaran, Seger, &
39
Gabrieli, 1999).
40
Dissociations between procedural learning and declarative
41
learning systems were reported, with normal motor procedural
42
∗Corresponding author. Tel.: +41 21 31 435 63; fax: +41 31 435 72.
E-mail address:Claire.Mayor@chuv.ch(C. Mayor-Dubois).
learning being observed despite severe amnesia due to medial 43
temporal lobe lesions (Brooks & Baddeley, 1976; Corkin, 1968; 44
Damasio, Eslinger, Damasio, Van Hoesen, & Cornell, 1985; Gabrieli, 45
Corkin, Mickel, & Growdon, 1993; Milner, 1962; Tranel, Damasio, 46
Damasio, & Brandt, 1994) and impaired procedural learning 47
together with preserved declarative memory in patients with basal 48
ganglia damage like Parkinson’s (Haaland, Harrington, O’Brian, & 49
Hermanovicz, 1997; Harrington, Haaland, Yeo, & Marder, 1990; 50
Sarazin et al., 2002) and Huntington diseases (Gabrieli, Stebbin, 51
Singh, Willingham, & Goetz, 1997; Schmidtke, Manner, Kaufmann, 52
& Schmolck, 2002; Willingham, Koroshetz, & Peterson, 1996). Even 53
if the participation of declarative memory or executive function 54
in optimal skill learning is likely (Curran, 1997; Foerde, Poldrack, 55
& Knowlon, 2007; Knowlton, Squire, & Gluck, 1994; Packard & 56
Knowlton, 2002; Poldrack et al., 2001; Smith & McDowall, 2004), it 57
appears that conscious recollection is not a necessary component 58
of the process. 59
The basal ganglia, which include the striatum, globus pallidus, 60
subthalamic nucleus and substancia nigra, are involved in move- 61
ment control and planning but also in the regulation between 62
motivational, cognitive and motor components of behavior (Bédard 63
et al., 2003; Saint-Cyr, Taylor, & Lang, 1988). Lesions or dysfunc- 64
tion of these structures typically lead to either loss of postural 65
adjustment, rigidity and bradykinesia (as seen in parkinsonism), 66
or to abnormal movements like chorea, athetosis, dystonia and tics 67
(Adams & Victor, 2001). 68
Procedural learning has mostly been investigated in its motor 69
facet with the Serial Reaction Time test (SRT,Nissen & Bullemer, 70 0028-3932/$ – see front matter© 2010 Published by Elsevier Ltd.
doi:10.1016/j.neuropsychologia.2010.03.022
Please cite this article in press as: Mayor-Dubois, C., et al. Visuo-motor and cognitive procedural learning in children with basal ganglia pathology.
Neuropsychologia(2010), doi:10.1016/j.neuropsychologia.2010.03.022
UNCORRECTED PROOF
sequences of stimuli, the participant being unaware of the
repeti-74
tive structure of the task. This task does not depend on declarative
75
memory since patients with amnesia show preserved learning
76
skills (Knopman & Nissen, 1987;Nissen & Bullemer, 1987).
Func-77Q1
tional neuroimaging studies with healthy subjects show striatal
78
activations in SRT learning in association with premotor cortex
79
and supplementary motor area (Grafton, Hazeltine, & Ivry, 1995;
80
Rauch et al., 1997). Performance in SRT tasks is impaired in patients
81
with Huntington disease (Knopman & Nissen, 1991; Willingham
82
& Koroshetz, 1993) and in Parkinson disease, although the nature
83
and extent of the deficit are found to vary across studies (Doyon et
84
al., 1997; Jackson, Jackson, Harrison, Henderson, & Kennard, 1995;
85
Muslimovic, Post, Speelman, & Schmand, 2007; Pascual-Leone et al.,
86
1993; Smith & McDowall, 2004, 2006; Stefanova, Kostic, Ziropadja,
87
Markovic, & Ocic, 2000).
88
Procedural learning was initially thought to support only motor
89
learning but progressively, its role in cognitive learning, such as
90
associative learning has been recognized (Packard & Knowlton,
91
2002). This cognitive procedural component has been
investi-92
gated with a probabilistic classification task (Shohamy et al., 2004)
93
adapted from the original«Weather Prediction»Task (Knowlton,
94
Mangels, & Squire, 1996), in which the participant has to learn
95
the association between a combination of cues and a given result
96
(=outcome). Healthy subjects show increased activation in the right
97
caudate nucleus in this associative learning compared to baseline
98
(Poldrack et al., 1999). Patients with focal (Keri et al., 2002) or more
99
diffuse degenerative lesions of basal ganglia (Knowlton, Mangels, et
100
al., 1996; Knowlton, Squire, Paulsen, Swerdlow, & Swenson, 1996)
101
perform poorly in this task, unlike patients with medial temporal
102
lobe or diencephalic pathology (Eldridge, Masterman, & Knowlton,
103
2002; Knowlton, Mangels, et al., 1996; Knowlton et al., 1994),
cere-104
bellar disease (Witt, Nuhsman, & Deuschl, 2002) or frontal lobe
105
lesions (Knowlton, Mangels, et al., 1996).
106
Studies of procedural learning in children are sparse and
con-107
cern either infants in their first year of life or school-aged children.
108
It has been suggested that during development, children move from
109
a largely procedural to a more declarative knowledge
(Karmiloff-110
Smith, 1994). Automatic, non-conscious learning, acquired through
111
repetition, is thought to be very important for the acquisition of
112
motor and language skills and is viewed as a more primitive
sys-113
tem from the phylogenetic and ontogenetic point of view (Nelson,
114
1987; Reber, 1993). A simplified SRT task – the visual expectancy
115
task – has indeed shown reliable anticipatory responses in infants
116
as young as 3 months old for simple sequence patterns (Haith
117
& McCarty, 1990) and from 5 months old for longer sequences
118
(Smith, Loboschefski, Davidson, & Dixon, 1997). This indicates that
119
sequence learning is already present very early in development and
120
possibly increases during the first year of life. Studies with SRT tasks
121
show no evidence of age-related improvements from 4 to 10 years
122
old (Meulemans, Van Der Linden, & Perruchet, 1998; Thomas &
123
Nelson, 2001), as well as no difference from childhood to adulthood
124
(Meulemans et al., 1998), suggesting age invariance of procedural
125
learning skills beyond infancy. If learning performance is similar
126
across life-span and if the neural structures underlying these
learn-127
ing skills are the same in adults and children, then it would imply
128
that basal ganglia lesions would affect procedural learning in
chil-129
dren just like in adults. However, procedural learning has never
130
been studied in children with basal ganglia pathology in order to
131
test this hypothesis.
132
The few existing paediatric studies on procedural learning have
133
focused on children with diffuse brain pathology like traumatic
134
brain injury (Ward, Shum, Wallace, & Boon, 2002) or various kinds
135
of brain development disorders: in genetic syndromes, such as
136
Japikse, & Eden, 2006; Stoodley, Ray, Jack, & Stein, 2008; Vicari et al., 140
2005). Procedural learning impairments – but preserved declara- 141
tive memory – has been reported in Williams syndrome and in most 142
studies on children with autism or dyslexia, whereas the reversed 143
pattern of performance has been found in children with traumatic 144
brain injury. These results indicate that both memory systems can 145
also be dissociated in children. 146
The only studies addressing procedural learning in a develop- 147
mental disorder involving the basal ganglia, were carried out on 148
patients with Gilles de la Tourette syndrome. This syndrome is 149
thought to arise from a fronto-striatal dysfunction and consists of 150
a severe and chronic tic disorder starting during childhood, and 151
frequently associates cognitive and psychiatric disturbances (for 152
review:Leckman, Bloch, Scahill, & King, 2006). Results obtained in 153
children and adults showed preserved perceptivo-motor learning 154
(Marsh, Alexander, Packard, Zhu, & Peterson, 2005), but impaired 155
cognitive procedural learning (Kéri, Szlobodnyik, Benedek, Janka, & 156
Gadoros, 2002; Marsh et al., 2004). Children with chorea due to con- 157
genital or acquired conditions or children with focal basal ganglia 158
damage have never been studied. 159
We investigated 18 children aged 8–15 years with lesions or a 160
dysfunction of the basal ganglia, by giving them a classical SRT task 161
and a probabilistic classification task. 162
The first aim of our study was to investigate procedural visuo- 163
motor sequence learning and cognitive (classification) learning 164
skills in children with basal ganglia pathology by comparing their 165
learning skills with those of healthy, control children. We expected 166
learning deficits in the clinical group compared to the control group, 167
and similar procedural learning deficits in children as in adults with 168
basal ganglia damage, since their procedural memory system is 169
considered to be already differentiated early in development. The 170
second aim was to investigate a possible plasticity of the proce- 171
dural learning system, by comparing congenital or early acquired 172
pathologies with later acquired ones in terms of their impact on 173
the SRT and classification tasks. Finally, we wondered if dissocia- 174
tions between visuo-motor and cognitive skills could be observed, 175
as shown in some adults (Foerde et al., 2008). 176
2. Methods 177
2.1. Participants 178
Eighteen patients (9 girls, 9 boys) aged 8–15 years (mean = 11.5) were recruited 179 via our paediatric neurology outpatient clinic. Criteria for study participation was 180 a minimum age of 8 years, normal cognitive development (see below) and either 181 combined neurological and radiological, or only clinical signs of uni- or bilateral 182 basal ganglia pathology of various aetiology (Table 1). Neurological signs included 183 choreic movements, dystonia, hemiparesis and tics. Five patients with clear neu- 184 rological signs at initial consultation were included, although abnormal movement 185 had disappeared at the time of neuropsychological assessment, because of sponta- 186 neous evolution or medication. These 18 patients constituted the basal ganglia (BG) 187
group. 188
Age of pathology onset ranged from prenatal to 13 years old. Four patients had 189 early acquired damage (before 1 year old) and five had a neurodevelopmental dys- 190 function. These 9 patients were defined as the early acquired lesion/dysfunction 191 group. Seven patients had a lesion/dysfunction acquired after the age of 6 years 192 and constituted the late acquired lesion group. Two children had progressive dis- 193 ease (clinical onset at respectively 7 and 9 years old) with a diagnosis of idiopathic 194
dystonia of unknown etiology (DYT1 negative). 195
Two patients had unilateral lesions of basal ganglia; the 16 others had clinically 196 or MRI documented bilateral involvement. Only four patients had additional lesion 197 outside basal ganglia. All MRI were reviewed by the radiologist (PM). Six patients 198
had no MRI (no clinical indication or refusal). 199
Fifteen patients had comprehensive neuropsychological testing for clinical pur- 200 poses (Table 2). All patients had normal intelligence, defined by a Verbal Reasoning Q2 201 Index superior to 85 at the Wechsler Intelligence Scale, 4th edition (Wechsler, 2007), 202 or based on a single subtest of logico-deductive reasoning for a few children (3), who 203 had no academic difficulties and were recruited for research purposes only (Table 1). 204
Please cite this article in press as: Mayor-Dubois, C., et al. Visuo-motor and cognitive procedural learning in children with basal ganglia pathology.
Neuropsychologia(2010), doi:10.1016/j.neuropsychologia.2010.03.022
UNCORRECTED PROOF
testing/sex presentation b) at testing (if different)
T2/spectrography Scale (WISC IV) and behavior
1 9/M Choreic movements,
dysarthria
Not done Meningitis (neonatal) VCI = 118, PRI = 104 Normal
2 9/F Choreic movements,
mild dysarthria
Not done Hyperbilirubinemia
(neonatal)
PRI = 92 ADD, dyslexia, dyspraxia
3 15/Ma Choreic movements, (b) None
Normal Broncodysplasia, severe
neonatal pulmonar hypertension (neonatal)
VCI = 110, PRI = 88 ADD, executive dysfunction, poor visuo-spatial skills,
VCI = 99, PRI = 77 ADD, poor arithmetic skills
5 15/M Choreic movements,
(b) None
Normal Unknown (Adopted child) VCI = 99, PRI = 65 ADD, executive dysfunction, dyspraxia, conduct disorder
6 10/F Choreic movements Not done Neurodevelopmental VCI = 94, PRI = 77 ADD, dyscalculia, dyspraxia
7 12/Ma Tics Not done Neurodevelopmental VCI = 116, PRI = 107 ADD, OCD
8 8/M Tics Not done Neurodevelopmental Matrices: SS = 12 Normal, OCD
9 8/M Tics (b) none Not done Neurodevelopmental VCI = 116, PRI = 102 Dyspraxia, autistic features
10 9/Ma Choreic movements,
VCI = 155, PRI = 132 Normal, limited social skills
11 11/Fa Choreic movements (b) none
Normalb Sydenham chorea (7 years) VCI = 116, PRI = 104 Dyslexia, episodic dyscontrol syndrome, OCD
VCI = 86, PRI = 92 ADD, dyslexia, dyscalulia
13 13/F Dystonia Lesion of L caudate, L
putamen (head), and R thalamus
Lyme disease (8 years) VCI = 99, PRI = 99 Dyscalculia
14 15/Fa Dystonia, dysarthria Bilateral atrophy of caudate nuclei (heads) and putamen
Wilson disease (13 years) VCI = 99, PRI = 92 ADD, poor arithmetic skills
15 14/M Right motor paresis Lesions of L pallidum and L capsula interna
Infarct L choroidal artery (13 years)
Matrices: SS = 13 Poor verbal declarative memory
16 12/F Right motor paresis Lesion of L caudate, L lenticular nucleus, L thalamic atrophy, L frontal lobectomy
Infarct L middle cerebral artery (10 years)
PRI = 88 Mild conduction aphasia, deep alexia-agraphia, dysexecutive signs
17 15/Ma Dystonia Normal Idiopathic progressive, DYT1
negative (9 years)
VCI = 135, PRI = 96 Normal
18 13/F Dystonia Mild bilateral atrophy
of ant.
F: female, M: male, L: left, R: right, NAA:N-acetylaspartate, VCI: verbal comprehension index, PRI: perceptive reasoning index, SS: standard score, C: percentile, ADD: attention deficit disorder, OCD: obsessive compulsive syndrome. * Standard Raven Progressive Matrices (PMS 47).
aPatients with medication.
bPET showed hypermetabolism of caudate (left > right) and of bilateral frontal cortex.
Results were compared to a control group of 72 healthy school-aged children 205
(39 girls, 33 boys) aged 8–12 years (mean age = 10) (Mayor-Dubois et al., in prepa-206
ration), recruited from the same mainstream school. They had no developmental 207
problems, no learning difficulties and no medication, according to their teacher and 208
to a questionnaire filled out by their parents.
209
This research project was accepted by the medical ethical committee of our 210
institution and informed consent was obtained by the patient’s parents.
211
2.2. Measures 212
The E-Prime software version 1.1 was used to run both procedural tasks.
213
2.2.1. Serial Reaction Time (SRT) 214
The experimental paradigm developed for children and adults byMeulemans 215
et al. (1998)was chosen. The child sat at a comfortable distance from the computer, 216
and placed his middle finger and forefinger of each hand on four keys (X, C, N, M 217
on the Swiss French QWERTZ keyboard). He was instructed to respond as quickly as 218
possible to an asterisk appearing in one out of four possible positions displayed on 219
the screen by pushing the corresponding key. Four arrows, indicating the possible 220
locations, remained displayed during the whole session. Five blocks of 84 trials were 221
administered (total of 240 trials). In each block, a sequence composed of random 222
trials alternated with a repeated sequence (C-M-X-N-M-C-X-M-N-X) of 10 positions.
223
In this task, we expected a decrease of reaction times for the repeated sequences 224
but not for the random stimuli as the trials proceed.
225
Learning was measured by computing the median reaction times for correct 226 responses for both the repeated and the random sequences in each block. Accuracy 227 was also taken into account, by computing the number of errors in each block for 228
each type of sequence. 229
A recognition test was administered at the end of the learning phase in order 230 to check whether some declarative knowledge of the sequences had been acquired. 231 The child had to judge if a given sequence was displayed or not in the learning test 232 by giving a yes/no answer. Sixteen sequences of four positions were proposed, eight 233 sequences belonging to the repeating sequence («known»), and eight sequences 234 not presented in the learning phase («new»). The known and new sequences were 235
presented in a random order. 236
2.2.2. Probabilistic Classification Learning (PCL) 237
We used the paradigm described inShohamy et al. (2004), designed for adults. 238 Participants were instructed to predict the flavour of ice cream (vanilla or choco- 239 late) that the presented figure was going to choose. The figure consisted of a basic 240 shape (head with black eyes, red nose, mouth, arms and feet) on which one to three 241 out of four different attributes (=cues) were added (hat, sunglasses, moustache, tie), 242 for a total of 14 different figures. Each cue was associated with a flavour (=outcome) 243 with a distinct level of probability. Over the whole test, each outcome occurred 244
equally often (50% vanilla, 50% chocolate). 245
The figures were presented one at a time (total of 200 trials) on a computer 246 screen, in a randomized order (but fixed for all subjects). Once the prediction was 247 done, feedback was given (the correct ice cream appears together with the figure 248
Please cite this article in press as: Mayor-Dubois, C., et al. Visuo-motor and cognitive procedural learning in children with basal ganglia pathology.
Neuropsychologia(2010), doi:10.1016/j.neuropsychologia.2010.03.022
UNCORRECTED PROOF
The number of correct responses was computed in 4 blocks of 50 trials over the 252
whole test (200 trials). Following previous PCL studies, the participant’s response 253
was judged correct if it corresponded to the most probable outcome for that figure 254
(if the probability for this outcome is superior to 0.5).
255
At the end of the 200 trials learning phase, a questionnaire was adminis-256
tered in order to evaluate the declarative knowledge developed by the participants 257
(for instance: what cue was associated with vanilla?) with a maximum score of 258
10 points.
259
2.3. Statistical analyses 260
Analyses were carried out using the SPSS statistical package, version 15.0. The 261
level of significance was set atP< 0.05, unless otherwise specified.
262
Descriptive data instead of statistical analyses were provided for the basal gan-263
glia patient subgroups (early versus late versus progressive pathology; bilateral 264
versus unilateral) because of the small number of participants.
265
2.3.1. SRT 266
2.3.1.1. Reaction times (RT).An analysis of variance (ANOVA) with sequence type 267
(repeated versus random) and blocks (5) as repeated measures was performed on 268
the median reaction times of the BG group, in order to see whether a learning effect 269
for the repeated sequence could be observed. A comparison with the control (C) 270
group was then performed, with an analysis of variance (ANOVA) with group (BG 271
versus C) as the between-subject variable, Sequence (random versus repeated) and 272
blocks (5) as repeated measures.
273
2.3.1.2. Errors.Analysis of accuracy was measured through the numbers of errors 274
committed in random versus repeated sequences. The number of errors in each 275
block was too low to perform an analysis of variance. Therefore the total number of 276
errors was computed and the differences were evaluated through paired-samples 277
t-tests.
278
2.3.1.3. Declarative knowledge.We tested whether participants developed explicit 279
knowledge of the repeating sequence with a singlet-test for each group, by com-280
paring the number of correct Reponses to the chance level performance (8 out of 281
16 = 50%).
282
2.3.2. Probabilistic Classification Learning (PCL) 283
2.3.2.1. Learning phase.In order to judge whether the scores were different from 284
chance level, we used single samplet-test in each group. Then, the number of correct 285
responses were compared with an analysis of variance (ANOVA) with blocks (4) as 286
repeated measures and groups (C versus BG) as a between subject variable.
287
2.3.2.2. Declarative knowledge.Scores of declarative knowledge on the question-288
naire were compared with independent samplet-tests in each group. Bivariate 289
correlations (Pearson) between explicit knowledge of the cue-outcome associations 290
and learning scores in the different blocks of the test were performed.
291
3. Results
292
3.1. Serial Reaction Time (SRT)
293
Fig. 1 shows the evolution of reaction times (RT) for both
294
repeated and random sequences across the 5 blocks, for the BG
295
group and the C group.
296
Fig. 2indicates the mean RT gain (in ms) for repeated sequences
297
compared to random stimuli at the end of the test (block 5) for the
298
BG, and C groups.
299
3.1.1. Basal ganglia group (BG)
300
3.1.1.1. Reaction times (RT). Results are given for 14 children
301Q3
only since four patients (cases 4, 13, 15, 16) with severe
302
motor dysfunction could not perform this task. These patients
303
had unilateral lesions and/or additional lesions which thus
304
allowed no comparison for subgroups according to extent of the
305
lesion/dysfunction.
306
Reaction times were obviously slower in the BG group
com-307
pared to the C group (Fig. 1). Individual analysis showed that the
308
patients with choreic movements at the moment of the testing had
309
the slowest reaction times together with participants with
symp-310
tomatic dystonia (case numbers 1, 2, 6, 10, 14). Patients with tics,
311
Fig. 1. Mean of the median reaction times (RT) in SRT task of basal ganglia (BG) and the control (C) groups. Unlike the C group, the BG group displays no decrease of
Fig. 1. Mean of the median reaction times (RT) in SRT task of basal ganglia (BG) and the control (C) groups. Unlike the C group, the BG group displays no decrease of