Lateralized implicit sequence learning in uni- and bi-manual conditions

Brain and Cognition 81 (2013) 1–9

Contents lists available at SciVerse ScienceDirect

Brain and Cognition journal homepage: www.elsevier.com/locate/b&c

Lateralized implicit sequence learning in uni- and bi-manual conditions Rémy Schmitz a, Antoine Pasquali b,c, Axel Cleeremans b, Philippe Peigneux a,⇑ a UR2NF (Unité de Recherches en Neuropsychologie et Neuroimagerie Fonctionnelle) at CRCN (Centre de Recherches en Cognition et Neurosciences), Université Libre de Bruxelles, Campus du Solbosch CP191, Avenue F.D. Roosevelt 50, B-1050 Brussels, Belgium b Co3 (Consciousness, Cognition, and Computation Group) at CRCN, Université Libre de Bruxelles, Campus du Solbosch CP191, Avenue F.D. Roosevelt 50, B-1050 Brussels, Belgium c Neurogenics Research Unit, Adam Neurogenics, 133 Marine de Solaro, 20240 Solaro, France

a r t i c l e

i n f o

Article history: Accepted 25 September 2012 Available online 9 November 2012 Keywords: Sequence learning Implicit learning Cerebral dominance Bimanual learning Unimanual learning

a b s t r a c t It has been proposed that the right hemisphere (RH) is better suited to acquire novel material whereas the left hemisphere (LH) is more able to process well-routinized information. Here, we ask whether this potential dissociation also manifests itself in an implicit learning task. Using a lateralized version of the serial reaction time task (SRT), we tested whether participants trained in a divided visual field condition primarily stimulating the RH would learn the implicit regularities embedded in sequential material faster than participants in a condition favoring LH processing. In the first study, half of participants were presented sequences in the left (vs. right) visual field, and had to respond using their ipsilateral hand (unimanual condition), hence making visuo-motor processing possible within the same hemisphere. Results showed successful implicit sequence learning, as indicated by increased reaction time for a transfer sequence in both hemispheric conditions and lack of conscious knowledge in a generation task. There was, however, no evidence of interhemispheric differences. In the second study, we hypothesized that a bimanual response version of the lateralized SRT, which requires interhemispheric communication and increases computational and cognitive processing loads, would favor RH-dependent visuospatial/attentional processes. In this bimanual condition, our results revealed a much higher transfer effect in the RH than in the LH condition, suggesting higher RH sensitivity to the processing of novel sequential material. This LH/RH difference was interpreted within the framework of the Novelty-Routinization model [Goldberg, E., & Costa, L. D. (1981). Hemisphere differences in the acquisition and use of descriptive systems. Brain and Language, 14(1), 144–173] and interhemispheric interactions in attentional processing [Banich, M. T. (1998). The missing link: the role of interhemispheric interaction in attentional processing. Brain and Cognition, 36(2), 128–157]. Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction Implicit learning can be defined as the incidental learning of complex information without awareness about what it has been learned (Shanks, 2005). This phenomenon has been investigated using three main paradigms (for a review see Cleeremans, Destrebecqz, & Boyer, 1998): artificial grammar learning (Reber, 1967), dynamic systems control (Berry & Broadbent, 1984) and sequence learning (Nissen & Bullemer, 1987). Because sequencing of actions and information is a fundamental human skill that can be considered as a complex form of implicit learning, and because sequence learning experiments are easy to conduct in controlled settings, this paradigm has become increasingly popular (Clegg, Digirolamo, & Keele, 1998). Sequence learning has been mostly studied using the serial reaction time task (SRT). In the original SRT study by Nissen and ⇑ Corresponding author. Fax: +32 (0)2 650 22 09. E-mail address: [email protected] (P. Peigneux). URL: http://dev.ulb.ac.be/ur2nf (P. Peigneux). 0278-2626/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.bandc.2012.09.002

Bullemer (1987), participants were sequentially presented visual stimuli, displayed horizontally at one of several fixed locations on a computer screen. They were instructed to press as fast and as accurately as possible on the spatially compatible response key upon each appearance of the stimulus, after what the next location was displayed. Unknown to them however, stimuli were not randomly distributed but followed a repeated sequence of ten positions (e.g. 1-4-6-3-2-1-5-4-3-2, each number representing a location on the screen). Results showed that participants exposed to this structured material became gradually faster and more accurate, as compared to participants exposed to a purely random pattern of stimuli, suggesting that they had learned the sequential regularities allowing them to anticipate on the next location. Notwithstanding, they were unable to verbally describe the repeated sequence at the end of the learning session (Nissen & Bullemer, 1987). Such dissociation between a gradual increase in participants’ performance and the lack of ability to consciously describe the underlying structured material was interpreted as implicit learning. In further studies, sequence learning was tested at the within-subject level by introducing a block of random trials

2

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9

or a different sequence at the end of the learning session after repeated exposure to the material, with the effect that participants’ reaction times (RTs) increased when confronted to random or novel material, i.e. a transfer effect. Response slowing in this context was interpreted as reflecting participant’s sensitivity to the violation of learned regularities in the unstructured or different material (e.g. Cohen, Ivry, & Keele, 1990; Reed & Johnson, 1994). Transfer effects reflecting successful sequence learning have been investigated in many different conditions including aging (e.g. Cherry & Stadler, 1995; Curran, 1997; Howard & Howard, 1989), childhood (Meulemans, Van der Linden, & Perruchet, 1998; Thomas & Nelson, 2001; Thomas et al., 2004), brain damage (e.g. Smith, Siegert, McDowall, & Abernethy, 2001; Vandenberghe, Schmidt, Fery, & Cleeremans, 2006) and psychopathology (Brown, Aczel, Jimenez, Kaufman, & Grant, 2010). Whether and to what extent sequence learning is implicit can be disputed if the assessment of awareness is based solely on verbal reports (Shanks & St. John, 1994). The lack of sensitivity of such measures has prompted the use of more sophisticated methods to evaluate participants’ awareness (for a review see Destrebecqz & Peigneux, 2005). For instance, recognition tasks in which participants must decide whether sequential fragments (e.g. chunks of three successive elements) belong or not to the learned sequence have been used as a better estimate of conscious sequential knowledge (e.g. Perruchet & Amorim, 1992). Alternatively, some authors have also advocated using generation tasks (e.g. Jimenez, Mendez, & Cleeremans, 1996; Shanks & Johnstone, 1999), in which participants are asked to generate the learned sequence, or what they think it was, instead of merely reacting to the displayed stimuli. In this case, generation of learned sequence chunks above chance level can be taken as an index of conscious knowledge. The results of such sensitive tests have been generally suggestive that sequence learning involves largely conscious knowledge. However, such generation task cannot be taken to constitute exclusive tests of conscious knowledge. Rather, they can also involve familiarity, and hence implicit knowledge. Thus, a participant may successfully generate the successor of a sequence fragment but claim that he was merely guessing. Generation tasks, therefore, involve a mixture of implicit and explicit knowledge, and probably overestimate the extent of conscious knowledge. This contamination problem can be solved by adapting the Process Dissociation Procedure (PDP, see Jacoby, 1991) to sequence generation tasks (Destrebecqz & Cleeremans, 2001). The rationale behind the PDP is that during a classical generation task in which participants must reproduce the learned sequence (an ‘‘inclusion’’ condition), both explicit and implicit knowledge of the sequence contribute to performance and generation of learned chunks. However, if participants are requested to generate a different or the inverse sequence than the learned one (i.e. an ‘‘exclusion’’ condition), then conscious knowledge should allow them to avoid producing any learned element. If, on the contrary, they continue to generate learned chunk elements above chance level, one can then conclude that these responses depend on the influence of implicit processes that cannot be controlled by conscious knowledge, thus demonstrating implicit sequence learning (Destrebecqz & Cleeremans, 2001; Destrebecqz & Peigneux, 2005). To the best of our knowledge, cerebral hemispheric specialization has never been investigated in the context of implicit sequence learning. This is all the more surprising since specific, testable assumptions concerning the role of the left (LH) and right (RH) hemispheres during learning of novel material can be formulated in this context. Indeed, according to the Novelty-Routinization model (Goldberg & Costa, 1981; see also Goldberg & Podell, 1995; Goldberg, Podell, & Lovell, 1994), the RH should be crucial in the initial acquisition stage during learning, when exploratory processing of new cognitive situations is necessary and preexisting

cognitive strategies and representations are not yet firmly established. On the other hand, LH processing would favor preexisting representations and well-routinized cognitive strategies. Nevertheless, the model has so far not been thoroughly tested (for a short review see Dien, 2008, p. 296). In line with this prediction however, RH dominance in a commissurotomized patient performing an implicit visual statistical learning task has been recently reported (Roser, Fiser, Aslin, & Gazzaniga, 2011). In this study, the patient was unwittingly exposed to scenes composed of random combinations of fixed pairs of shapes. After a phase of incidental exposition, the patient was asked to make a two-alternative forced-choice and to decide which pair of shapes had appeared together during the familiarization phase. The two shapes were briefly presented either within the left (LVF) or right (RVF) visual fields, thus primarily targeting the patient’s RH or LH, respectively. Results in this split-brain patient showed above-chance shape discrimination only when the RH was stimulated. Additionally, he was unable to explicitly describe the relations between the different pairs of shapes. Given that callosotomy impairs interhemispheric cortical transfer, this study suggests that visual implicit statistical learning may be established at first within the RH. RH predominance in hippocampus and caudate nucleus activation was also observed using fMRI in healthy participants performing the same statistical learning task (Turk-Browne, Scholl, Chun, & Johnson, 2009). It should be noted that these specific regions also participate in sequence learning (e.g. Albouy et al., 2008; Destrebecqz et al., 2005; Peigneux et al., 2000). In addition, the Novelty-Routinization model (Goldberg & Costa, 1981) makes the prediction of an initial RH dominance during the acquisition of a novel material, shifting toward a LH dominance when the material is sufficiently integrated. Accordingly, preferential RH activation was found in a functional neuroimaging study (Seger et al., 2000) when healthy participants learned to distinguish between exemplars of two categories made of variations of different unseen prototype stimuli, and LH activations were found at the end of the learning session in participants showing the best performance. In this context, the aim of the present study was to study the influence of hemispheric specialization on implicit sequence learning. In line with the Novelty-Routinization model (Goldberg & Costa, 1981), we hypothesized that transfer effects reflecting sequence learning would be observed at first in the RH after a relatively short, unique training session on the SRT task. To probe this hypothesis, participants were tested using a divided visual field, lateralized version of the SRT to ensure preferential unihemispheric processing, either in the LVF (i.e. RH stimulation) or in the RVF (i.e. LH stimulation). Moreover, to maximize interhemispheric differences, LVF and RVF groups performed the lateralized SRT task with the hand ipislateral to the stimulated visual field, thus promoting both visual perception and motor response related to sequence learning prioritized in the same hemisphere (Bourne, 2006; Gazzaniga, 2000). Finally, after the learning session, participants performed a generation task in both inclusion and exclusion conditions using reversible sequences (Pasquali, 2009) to evaluate their degree of conscious knowledge of the incidentally learned sequence.

2. Study 1 2.1. Method 2.1.1. Participants Twenty-four young healthy right-handed volunteers participated in this experiment. Twelve participants were assigned to the left visual field (LVF) group (1 male) and the remaining to the right visual field (RVF) group (2 males). Mean age did not differ

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9

between groups (LVF: 19.8 ± 1.1 years vs. RVF: 20.8 ± 1.7 years; t(22) = 1.76, p > .05), as well as handedness laterality quotient evaluated using the Edinburgh Handedness Inventory (Oldfield, 1971; LVF: 98.3 ± 3.9 vs. RVF: 95.2 ± 9.3; t(22) = 1.07, p > .05). 2.1.2. Tasks and materials The experiment was run on a PC system with a 1700 width screen, using Cogent 2000 software (http://www.vislab.ucl.ac.uk) implemented on Matlab 6.1. The display on the screen consisted of five permanent grey circles positioned on a black background in two oblique lines left- or right-sided relative to a permanent central fixation cross (see Fig. 1). To ensure extrafoveal presentation, the inner and outer circles were located at 4.6° and 10.3° of visual angle, respectively. Each circle was matched with a key (S, E, R, G and spacebar vs. spacebar, H, U, I and L for the left and right displays, respectively) on the PC keyboard. The spatial configuration of the keys and fingers corresponded to the screen positions (left little, ring and middle fingers, index and thumb vs. right thumb, index, middle, ring and little fingers, for the left and right displays, respectively). The target stimulus was a white circle that substituted one of the five grey circles for 100 ms only to avoid ocular saccade during the lateralized display (Bourne, 2006). 2.1.2.1. Letter detection task. To ensure participants’ central ocular fixation, a letter detection task was concurrently administered during the SRT task. At each trial, a letter substituted the central cross and appeared simultaneously with the target SRT stimulus, for the same 100 ms duration. Letters were randomly selected, and participants were asked to keep track of the displayed letters following the alphabetical order within the random succession, and to report the last identified letter in increasing alphabetical order at the end of the block. For instance, if participants are presented with the

3

letters suite letters ‘‘X-N-K-A-R-F-G-L-B-S-D-E-T-Y-B-P-O-R-C-KS’’ during a block, then they should report the letter ‘‘C’’ because ‘‘A’’ then ‘‘B’’ letters have been previously presented, thus constituting an alphabetical suite). The next block restarted with the latest correct letter to be detected (i.e. the letter ‘‘C’’ in the present example) and participants had to continue keeping track of the letters arising in ascending alphabetical order (i.e., C - . . . - D - . . . - E - . . . - F - . . .). Hence participants had to keep track simultaneously of the centrally displayed letter (ensuring foveal central fixation) and of the displayed location in the SRT task using peripheral vision (see below). As instruction emphasized the importance of accurate performance in the letter detection task also, it favored central gaze fixation while performing to the task. 2.1.2.2. SRT task. During the SRT task, participants had to constantly fixate the central fixation cross (and perform the concurrent letter detection task) while pressing as fast and as accurately as possible on the key corresponding to the target stimulus flashed for 100 ms in the trained hemifield. Stimuli appeared in the LVF for the first half of participants and in the RVF for the other half, ensuring primary processing by the right (RH) or the left (LH) hemisphere respectively. Participants had 1500 ms to respond to the target stimulus, and response-stimulus interval was 850 ms. Unknown to participants, each of the 13 training blocks contained 8 repetitions of a 10-elements sequence governing the apparition of the target stimuli. Whereas blocks 1–11 and block 13 were ruled by one of the two possible sequences of ten elements (4-2-1-4-5-32-5-1-3 or 3-1-5-2-3-5-4-1-2-4), the 12th block was made with the reverse sequence to assess transfer effects (i.e. 3-1-5-2-3-54-1-2-4 or 4-2-1-4-5-3-2-5-1-3). These two sequences consisted of so-called ‘‘second order conditional’’ transitions (SOC; Reed & Johnson, 1994). In other words, with SOC sequences, two elements

Fig. 1. Upper right panel: Time-course and sample display of the unimanual response lateralized SRT task in the LVF condition. Left panels: percentage of correct responses (top) and associated reaction times (bottom) for the learned (blocks 1–11 and 13) and the transfer, reversed (block 12, Re) sequences in the LVF and RVF groups. Lower right panel: Magnitude of the transfer effect in LVF and RVF groups. t-Test against the value 0: p < .01.

4

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9

of temporal context are always necessary to predict the location of the next stimulus. Moreover, the sequence contained no common transition with its reverse, and all the transitions in the sequence appeared with the same frequency-of-occurrence (Pasquali, 2009). 2.1.2.3. Generation task. At the end of the SRT session, participants were informed that the succession of locations was not random but that it instead followed a recurrent pattern. They were then presented with the same display, this time centrally located on the screen. In the inclusion condition, participants were asked to generate a series of 144 trials that matched the previously learned sequence as best as possible with their trained hand (left vs. right hand for LVF vs. RVF group, respectively). They were told to rely on intuition whenever they felt unable to recollect the location of the next stimulus. After the inclusion session, participants performed an exclusion task, in which they were asked to generate the reverse pattern of the previously learned sequence for another 144 trials. In both sessions, the first circle was displayed randomly at one of the five locations and participants had to press one of the five keys to generate the next stimulus. Each stimulus then appeared at the corresponding location on the screen immediately after the key press. 2.1.3. Procedure Each participant was tested individually in a quiet room. They were asked to remain still and maintain a constant distance from the computer screen during the entire experiment to ensure correct visual angle. They first started with the SRT task, either in the LVF-RH or the RVF-LH condition. They were asked to fixate the fixation cross all along the SRT session, and to respond as fast as possible to targets while keeping their accuracy score above 85% and successfully performing the concurrent letter detection task (see above). At the end of each block, a feedback related to participants’ speed and accuracy on the SRT task was automatically displayed. After having written down the last letter detected in alphabetical order (see above), the correct letter was shown and participants were prompted to begin with the next block of trials when ready. At the end of the last block, participants were informed that the sequence of successive stimuli was not random but followed a specific rule. The generation task was then explained. The exclusion condition always followed the inclusion condition. Testing ended with filling out a handedness questionnaire. The entire experimental session lasted approximately 1 h depending on participants’ speed. 2.2. Results 2.2.1. Letter detection task The proportion of letters correctly reported on the entire SRT session was calculated for each participant. Participants in the LVF and RVF groups correctly reported 76.8 ± 9.7% and 78.1 ± 10.9% of the letters, respectively. t-Test for independent samples did not reach significance (t(22) = 0.31, p > .05). These results show that both groups satisfactorily performed the letter detection task, hence reflecting correct fixation on the center of the screen during the concurrent apparition of target stimuli. 2.2.2. SRT task A mixed ANOVA with Block (1–13) as within-subjects factor and Hemifield (LVF vs. RVF) as between subjects factor computed on responses accuracy revealed a significant main effect of Block (F(12, 264) = 2.42, p < .01). Tukey’s post hoc comparisons failed to show between-blocks differences (all ps > .05). Fig. 1 shows the time course of participants’ accuracy within the two experimental groups. Both the main effect of Hemifield (F(1, 22) = 1.57) and the Block by Hemifield interaction (F(12, 264) = .87) failed to reach

significance (both ps > .05), indicating that participants in the LVF vs. RVF conditions exhibited similar accuracy all along the experiment (mean accuracy across all blocks: LVF 86.42 ± 10.59% vs. RVF 90.83 ± 8.62%, respectively; see Fig. 1). The same analysis computed on RTs associated to correct responses revealed a significant main effect of Block (F(12, 264) = 17.17, p < .0001), reflecting a gradual decrease in RTs throughout the session (see Fig. 1). Both the main effect of Hemifield (F(1, 22) = .55) and the Block by Hemifield interaction (F(12, 264) = .38) failed to reach significance (both ps > .05). To evaluate RTs increase during the transfer block (i.e., the block containing the reversed sequence), we computed the difference between the RTs for the 12th block (i.e. the transfer block) and the average of the RTs over the 11th and 13th blocks within each experimental group (see Fig. 1). t-Tests against 0 value disclosed a significant transfer effect both in LVF (35.8 ± 36.5 ms; t(12) = 3.40, p < .01; Cohen’s d = .98) and RVF (27.6 ± 29.9 ms; t(12) = 3.19, p < .01; d = .92) conditions. Additionally, a one tailed t-test for independent samples failed to disclose between-group differences in delta reflecting the size of the transfer effect (t(22) = 0.59, p > .05; d = .75). Similar transformation and analyses computed on accuracy also failed to reach significance (all ps > .05).

2.2.3. Generation task Generation performance was computed as the number of generated chunks of three elements over the 144 trials belonging either to the training sequence in the inclusion condition or to its reverse (transfer sequence) in the exclusion condition (see Fig. 2). Chance level was computed to be 16.67% of generated triplets: each sequence contains 10 possible triplets and there are 60 (5  4  3) possible triplets within a sequence of 5 possible locations (i.e. random level = 10/60 = 16.67%). Two-sided t-tests against the random value (i.e. 16.67%) only showed a significant effect in the RVF group for the proportion of chunks belonging to the learned sequence in the exclusion condition (11.4 ± 5.6%; t(12) = 3.24, p < .01; d = .93). In other words, when asked to generate parts of the reversed sequence, participants fail to perform above random level but were able to exclude some chunks belonging to the trained sequence. However, a mixed ANOVA with Instructions (Inclusion vs. Exclusion) and Sequence (Trained vs. Reversed) as within-subjects factors and Hemifield (LVF vs. RVF) as between subjects factor computed on the proportion of generated triplets failed to reveal any statistical significant results (all ps > .05). Overall, performance in inclusion and exclusion conditions of the generation task indicates that participants in both groups are equivalent to generate chunks belonging to the trained and reversed sequences, and do not differentiate from chance level.

Fig. 2. Generation scores for the unimanual response lateralized SRT task (proportion of generated triplets belonging to the learned or the reverse sequences) in the inclusion and exclusion conditions in LVF and RVF groups. Chance level was 16.67%.

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9

2.3. Discussion study 1 The aim of our first study was to investigate whether implicit learning would be present in conditions primarily targeting the RH. According to the Novelty-Routinization model (Goldberg & Costa, 1981), the RH acquires more easily novel information whereas the LH has more facilities to recruit routines already well established. Using a lateralized and unimanual version of the SRT, a paradigm used to assess implicit sequence learning, we predicted that the RH would acquire first sequential knowledge. In other words, we predicted that participants’ RTs would increase during the block of transfer composed to random trials (i.e. a transfer effect), more in the LVF condition as compared to the RVF condition targeting in priority the RH vs. LH, respectively. Our results show that participants were able to perform both the lateralized SRT and the letter detection task, the latter ensuring central ocular fixation during lateral displays and therefore peripheral visual processing of the targets. In both groups, performance on the letter detection task was satisfactory (about 77%) suggesting that participants correctly fixated the center of the screen during lateral presentation of the visual cues of the SRT. Therefore, we assume that the RH and LH were primarily targeted within the LVF and RVF groups, respectively. During SRT practice, groups did not differ in accuracy and stayed above 85% correct responses, suggesting that they were focused and performed accurately on the task. Additionally, the time course of RTs associated to correct responses was similar, suggesting that both groups progressively develop sensitivity to the structured material to the same extent. Corroborating this interpretation, and conversely to our RH hypothesis, the LVF and RVF groups were slowed down to the same extent during the block of transfer (Cohen’s d > .90). Finally, the generation task indicates that sequence learning was implicit as suggested by the lack of differences between the inclusion and exclusion conditions. Indeed, our participants were close to chance level in both conditions. Several factors may explain the fact that the transfer effects between the LH and RH were equivalent. First, our material was perhaps not sensitive enough: a longer training session might be necessary to reveal hemispheric asymmetries. However, it should be noted that participants’ responses were already slowed down on the transfer sequence after a training session of only eleven blocks. This suggests that the sequence regulating the succession of stimuli was already learned at that time, at least to some extent. Nevertheless, it should also be noted that a longer training session might be associated with larger transfer effects (Pasquali, 2009). Therefore, in our experiment, the LH may have already become sensitive enough to the sequential material after 11 blocks, thus reducing the gap with the RH. An alternative explanation is that the letter detection task in itself could interfere with sequence learning. Indeed the letter detection task may induce a dual-task condition in which working memory becomes saturated. However, implicit learning in such a condition has been shown possible (for a review see Shanks, 2003). Of note, since a dual-task condition is thought to reduce sequential knowledge (Frensch & Miner, 1994; Shanks & Channon, 2002; Stadler, 1995), this may explain why the transfer effect was weaker in our study than in previously published ones. Additionally, the verbal nature of the letter detection task may be a supplementary confound as a verbal task might artificially enhance LH processing (Kinsbourne, 1970). Nevertheless, if it was the case it could have been expected that the LH would be more affected than the RH by the block of transfer, which was not the case, ruling out this potential confound. Finally, it can be hypothesized that the unimanual condition was globally too easy to disclose a RH dominance in implicit learning in young healthy participants. Indeed, it has been proposed that when attentional and/or computational load is high, the

5

segregation of information processing across the two cerebral hemispheres may overcome the performance of an isolated hemisphere despite the costs of interhemispheric transfer (Banich, 1998, 2003). Conversely, a unique hemisphere may better process the whole information under low cognitive costs. Therefore, compared to its unimanual version, a lateralized bimanual response SRT might enhance such computational costs because it requires an additional interhemispheric processing. Indeed, transcallosal relay has been showed to be crucial to express behaviorally a transfer effect after a training session of a bimanual SRT both in commisurotomized patients (de Guise et al., 1999) and children in which callosal maturation is not yet fully achieved (de Guise & Lassonde, 2001). Accordingly, functional synchronization between the two hemispheres has been also revealed using electroencephalography in healthy participants performing bimanual sequence learning (Andres et al., 1999; Gerloff & Andres, 2002). These elements indicate that an increase of computational and cognitive loads may be expected in a bimanual lateralized version of the SRT. In addition, even if the LH is generally considered to be dominant in motor control (Goldenberg, 2009; Haaland & Harrington, 1996), RH specializations are also thought to be important in this context (Richards & Chiarello, 1997; Serrien, Ivry, & Swinnen, 2006). Within this framework, we hypothesized that sequential material acquisition would be better under an LVF stimulation condition assuming that the RH is more specialized both in visuospatial (Corballis, 2003; Vogel, Bowers, & Vogel, 2003) and attentional (Posner & Petersen, 1990; Sturm & Willmes, 2001) processes. Also, a transfer effect might be more pronounced in the LVF condition because motor and visuospatial/attentional computations may be processed in parallel within the LH and RH, respectively. In contrast, in the RVF condition, both motor and visuospatial/attentional inputs will primarily reach the LH and a transcallosal transfer will be needed to complete these two processes. Hence, this potential LH overloading might impair the acquisition of the sequential material. Our second study applied a bimanual version of the lateralized SRT to test these specific assumptions. 3. Study 2 3.1. Method 3.1.1. Participants Twenty-four young healthy right-handed volunteers participated in this experiment. Twelve participants were assigned to the left visual field (LVF) group (5 males) and the remaining in the right visual field (RVF) group (5 males). Mean age did not differed between group (LVF: 23.5 ± 3.7 years vs. RVF: 23.7 ± 3.2 years; t(22) = 0.18, p > .05). Handedness laterality quotient based on the Edinburgh Handedness Inventory (Oldfield, 1971) did not differed neither (LVF: 82.4 ± 17.9 vs. RVF: 76.4 ± 21.2; t(22) = 0.74, p > .05). 3.1.2. Task and materials The experiment was run on a PC system with a 1700 width screen, using Cogent 2000 software (www.vislab.ucl.ac.uk) implemented on Matlab 6.1. The screen display consisted of six permanent grey circles arranged on a black background in two oblique lines left- or right-sided relative to a permanent central fixation cross. To ensure extrafoveal presentation, the inner and outer circles were located at 4.6° and 10.3° of visual angle from the central fixation cross, respectively (see Fig. 3). Each target circle was matched with a key (S, D, F, J, K and L) on the PC keyboard. The spatial configuration of the keys and fingers corresponded to the screen positions: left vs. right ring, middle and index fingers for

6

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9

Fig. 3. Upper right panel: Time-course and sample display of the bimanual response lateralized SRT task in the LVF condition. Left panels: percentage of correct responses (top) and associated reaction times (bottom), for the learned (blocks 1–11 and 13) and the transfer, reversed (block 12, Re) sequences in the LVF and RVF groups. Lower right panel: Magnitude of the transfer effect in LVF and RVF groups. t-Test against the value 0: p < .00001. One-tailed t-test for independent samples: p < .05.

the upper-, middle- and lower-left vs. right circles, respectively. The target stimulus was a white circle that substituted one of the six grey circles for 100 ms to avoid ocular saccade during the lateralized display. 3.1.2.1. Letter detection task. The letter detection task was the same as the study 1. 3.1.2.2. SRT task. The parameters of the unimanual SRT task were the same as the experiment 1 except that the sequence and its reversed were composed of six elements (5-3-1-6-2-4-1-5-2-3-6-4 vs. 4-6-3-2-5-1-4-2-6-1-3-5). 3.1.2.3. Generation task. Instructions provided to the participants were the same as in Study 1. The only difference was that participants were confronted to six circles displayed centrally and had to perform the generation task bimanually. 3.1.3. Procedure The procedure was the same as study 1. 3.2. Results 3.2.1. Letter detection task Participants in the LVF and RVF groups correctly reported 62.17 ± 23.52% and 62.82 ± 20.70% of the letters, respectively. tTest for independent samples was not significant (t(22) = 0.07, p > .05), indicating correct performance and central screen fixation in both groups during lateralized presentation of target stimuli.

3.2.2. SRT task A mixed ANOVA with within-subjects Block (1–13) factor and between subjects Hemifield (LVF vs. RVF) factor computed on accuracy revealed a significant main effect of Block (F(12, 264) = 9.00, p < .0001) and a Block by Hemifield interaction (F(12, 264) = 2.15, p < .05). However, Tukey’s post hoc comparisons failed to disclose any significant between-group difference at each successive block (all ps > .05). The main effect of Hemifield (F(1, 22) = 1.56) failed to reach significance (p > .05). Overall, these results suggest that participants in the LVF and RVF conditions were similarly accurate all along the experiment (respective mean accuracy across all blocks: LVF 87.91 ± 12.96% vs. RVF 91.33 ± 6.67%, respectively; see Fig. 3). A similar ANOVA computed on RTs for correct responses showed a significant main effect of Block (F(12, 264) = 49.24, p < .0001) and a Block by Hemifield interaction (F(12, 264) = 1.95, p < .05). Again, Tukey’s post hoc comparisons failed to disclose any significant between-group difference at each successive block (all ps > .05). The main effect of Hemifield (F(1, 22) = .38) failed to reach significance (p > .05). Overall, these results illustrate the progressive improvement in response speed over practice on the SRT task (see Fig. 3). To evaluate RTs increase during the block of transfer, we computed the difference between RTs of the 12th block (i.e. transfer block) and average RTs over blocks 1 and 13 (see Fig. 3). t-Tests against 0 value disclosed a highly significant transfer effect in the LVF (54.5 ± 24.2 ms; t(12) = 7.79, p < .00001; Cohen’s d = 2.25) but only a trend in the RVF group (26.9 ± 48.1 ms; t(12) = 1.94, p = .078; d = .55). Additionally, a one-tailed t-test for independent samples reached significance (t(22) = 1.77, p < .05; d = .72), suggesting a larger transfer effect in the LVF group (54.5 ± 24.2 ms) compared to the RVF group (26.9 ± 48.1 ms). Similar transforma-

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9

tion and analyses computed on responses accuracy scores failed to reach significance (all ps > .05). 3.2.3. Generation task Generation performance was computed as the number of generated chunks of three elements all along the 144 trials belonging to the training sequence or its reverse. Chance level was set at 10% of generated triplets: 12 possible triplets in the second order sequence and 120 (6  5  4) possible triplets within a sequence with 6 possible locations (i.e. random level = 12/120 = 10%; see Fig. 4). Two-sided t-tests against chance value (i.e. 10%) only disclosed a significant effect in the LVF group for the proportion of chunks belonging to the reversed sequence in the exclusion condition (6.82 ± 4.1%; t(12) = 2.61, p < .05; d = .75). In other words, when asked to generate the reverse sequence, participants failed to perform above chance level. Moreover, a mixed ANOVA with Instruction (Inclusion vs. Exclusion) and Sequence (Trained vs. Reverse) as within-subjects factors and Hemifield (LVF vs. RVF) as between subjects factor computed on the proportion of generated triplets failed to reveal any significant effect (all ps > .05). Overall, performance in inclusion and exclusion conditions of the generation task indicates that participants in both groups failed to generate chunks belonging to the trained and reverse sequences, thus a lack of conscious knowledge about the learned sequential material. 3.3. Discussion study 2 Study 2 aimed at evaluating whether interhemispheric differences would be evidenced in the implicit acquisition of a structured sequential material, under a bimanual response condition enhancing computational load. We hypothesized (1) that a bimanual SRT would require interhemispheric coordination (Andres et al., 1999; de Guise & Lassonde, 2001; de Guise et al., 1999; Gerloff & Andres, 2002) in addition to a LH dominance in motor processing (Haaland & Harrington, 1996; Serrien et al., 2006), (2) that the LVF condition would facilitate the visuospatial/attentional processing according to the RH dominance for these functions (Posner & Petersen, 1990; Vogel et al., 2003) and (3) that a segregation of processing across the two hemispheres would be beneficial under higher computational and cognitive loads (Banich, 1998, 2003). Likewise Study 1, both LVF and RVF groups did not differ regarding to their accuracy in the concurrent letter detection task and the time course of RTs in the lateralized SRT. Additionally, performance was satisfactory in both tasks. These results indicate that participants succeeded in fixating the cross at the center of the screen while target stimuli were flashed in the lateral visual field, and were able to successfully perform on the lateralized SRT under a dual-task condition. In line with our hypotheses, participants in

Fig. 4. Generation scores for the bimanual response lateralized SRT task (proportion of generated triplets belonging to the learned or the reverse sequences) in the Inclusion and Exclusion conditions in LVF and RVF groups. Chance level was 10%.

7

the LVF condition exhibited a highly significant transfer effect (p < .00001; Cohen’s d = 2.25) whereas the effect was much weaker and non-significant in the RVF condition (p = .078; d = .55). In addition, between-group differences on the magnitude of the transfer effect was significant, with effect size ranged medium to high (d = .72) according to Cohen’s classification (Cohen, 1988). Finally, generation in inclusion and exclusion conditions was at chance level in both groups. Overall, these results suggest that participants implicitly learned the regularities embedded in the sequential material to a greater extent when the LVF/RH was primarily targeted. Results of study 2 are in line with the Novelty-Routinization model (Goldberg & Costa, 1981; see also Dien, 2008) predicting a RH superiority during the acquisition of novel material. The transfer effect was indeed more pronounced in the LVF condition favoring this hemisphere, suggesting a greater sensitivity of the RH in the acquisition of sequential regularities. At variance, transfer effects in the LH might be less pronounced due to its feebler capacity at acquiring novel material not yet sufficiently integrated (Goldberg & Costa, 1981). According to our hypothesis, introducing a bimanual response condition in the lateralized SRT had potentiated interhemispheric differences. Computational and cognitive costs are in line with studies showing that bimanual learning requires interhemispheric coordination (Andres et al., 1999; de Guise & Lassonde, 2001; de Guise et al., 1999; Gerloff & Andres, 2002). Our results are also consistent with the hypothesis of a specific RH propensity to visual statistical learning as evidenced in a split-brain patient (Roser et al., 2011) and healthy participants (Turk-Browne et al., 2009). Since implicit sequence learning and statistical learning may be considered as two different approaches of the same phenomenon (see Perruchet & Pacton, 2006), our results extend the RH specificity hypothesis to implicit sequence learning in healthy young adults. It should be noted the existence of a potential confound with response set-up and sequence length. In study 1, the unimanual condition involved a sequence of 10 elements with five alternatives whereas, in study 2, the bimanual condition used a sequence of 12 elements with six alternatives. Although this may be seen as a confounding effect impeding the interpretation of the results, it must be considered also that it is rather difficult if not impossible to control at the same time both for the complexity of the sequence and for the motor demands in unimanual and bimanual versions of the SRT. Indeed, the six alternatives inherent to the sequence of 12 elements are technically impossible to apply in the unimanual response condition, i.e. within a condition anatomically constraining to five alternatives (i.e., fingers) maximum, unless using twice the same finger for two locations. Conversely, the use of five alternatives, like in the unimanual condition, is also problematic to apply to the bimanual condition. Indeed, in this specific condition, should the participants respond with three fingers on one hand and two fingers on the other, or vice versa? Moreover, even if using a sequence of 8 elements (i.e. equating for complexity), the unimanual condition would require using four fingers with the same hand whereas only two fingers of each hand would be used in the bimanual condition. To summarize on this point, administering such a ‘‘better-controlled’’ condition would actually lead to other confounds related to the motor demands of the task. In this respect, an interesting follow-up for this experiment would be to manipulate the sequence length within the bimanual condition. Indeed, if our computational hypothesis is correct, a sequence of 8 elements should accentuate both the transfer effect in the LVFRH group, and the differences between the two groups, whereas a sequence of 4 or 6 elements should lead to reduced hemispheric effects due to lower cognitive and computational loads. Although a within-subjects design is generally preferable, especially as it increases statistical power and controls better for

8

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9

between-subjects variability, this strategy is actually difficult to apply in the present experiments without introducing at least two potential major confounds: the effect of prior practice and transfer effects. Indeed, studies have shown that learning a first sequence may impact the acquisition of a second sequence (e.g. Koedijker et al., 2010) even if performed at separate days (e.g. Goedert & Willingham, 2002; Panzer & Shea, 2008; Panzer, Wilde, & Shea, 2006). In addition, data have shown that sequential knowledge learnt using one particular hand may be transferred to the other hand in a unimanual SRT paradigm (e.g. Chase & Seidler, 2008; Perez, Tanaka, et al., 2007; Perez, Wise, Willingham, & Cohen, 2007; Perez, Tanaka, Wise, Willingham, & Cohen, 2008). A within-subject design might therefore induce interference effects both in the uni- and bimanual version of the SRT and/or intermanual transfer effects in the unimanual version. Each of these specific issues should be taken into account in future studies. It is also important to mention that the verbal nature of the letter detection task might have played a significant role since cognitive load was increased. Indeed, if functional hemispheric segregation favors performance in the bimanual SRT, the attentional resources within the LH may have been overloaded when participants had at the same time to process each letter and update it in verbal working memory (Kinsbourne, 1970). Articulatory suppression, a widespread method used to disturb memory updating, may indeed impair the acquisition of sequential knowledge (Gaillard, Destrebecqz, & Cleeremans, 2012; Shanks, 2003 for a review). Therefore, this specific condition may have disturbed the implicit acquisition of the sequential knowledge more in the RVF/LH group. Further studies are needed to test these specific assumptions, as well as to probe whether transfer effects in the RVF/LH group would normalize at the same level than in the LVF/RH group over more extended practice sessions eventually leading to increased routinization of the learned material. Within this framework, it should be investigated whether a concurrent visuospatial task would differentially impact the acquisition of the sequential material as compared to the verbal dual-task used in the present study. If our assumption is correct, an advantage of the RVF/LH group might therefore be predicted when participants simultaneously perform a central visuospatial task and the lateralized version of the SRT. Gender influence cannot be excluded in this study. Indeed, it has been proposed that women exhibit reduced structural and functional cerebral asymmetries (e.g. Hausmann, 2010 for a review). Therefore, a lack of differences between LVF/RH and RVF/LH conditions observed in the unimanual group might have been related to a higher proportion of women in Study 1. Besides, women are also known to perform better on verbal skills such as verbal memory and fluency tasks, whereas men perform better on visuospatial tasks (for a review see Kimura, 2002). Thus, participants’ gender might be an interesting variable to manipulate according to the nature of the lateralized SRT and the letter detection task. If the nature of the concurrent task ensuring central fixation may indeed specifically impact more one hemisphere, women as compared to men should acquire the sequential material more easily (i.e. exhibit a higher transfer effect) if the letter detection task is used. On the contrary, men should be shown more sensitive to the presentation of a reversed sequence after practicing a SRT session with a concurrent visuospatial task. Further research is needed to delineate the respective contributions of these potential factors. Finally, an additional interesting issue would be to investigate whether explicit sequence learning produces similar hemispheric effects than its implicit version. A simple way might be to instruct the participants about the existence the sequence (i.e., an intentional vs. incidental learning) and/or to modulate the response to stimulus interval (RSI) in the lateralized SRT. Previous studies

using the classical SRT have indeed showed that higher RSI made the participants more conscious of the sequence as compared to a null RSI (Destrebecqz & Cleeremans, 2001; Destrebecqz et al., 2005). A more explicit and/or an intentional version of the lateralized SRT might attenuate the difference between the two hemispheres or give an advantage to the LH during the acquisition of a new sequential material. Indeed, the transparency of the sequence to be learnt might require verbalizations requiring cognitive routines that are better established in the LH (Kinsbourne, 1970). 4. Conclusion We conducted two studies aimed at investigating cerebral hemispheric asymmetries in implicit learning using an original, lateralized version of the SRT task. Cerebral asymmetries in learning processes (Goldberg & Costa, 1981; Roser et al., 2011; Turk-Browne et al., 2009), suggested that the RH would learn faster than the LH a recurrent sequential pattern practiced over the training session. We therefore predicted that the RH primacy in learning would be expressed in a higher transfer effect, i.e. an increase in RTs upon introduction of a novel, unlearned material in left visual field conditions favoring primary processing in the RH. Although future replications and extensions will be needed, our results demonstrate that implicit sequence learning is possible in divided visual field conditions both using uni- and bi-manual response SRT. In addition, implicit sequence learning was impaired in the bimanual response version, in which cognitive and computational loads were increased. A RH specialization for the acquisition of novel material (Goldberg & Costa, 1981) coupled to the advantage of parallel hemispheric processing during higher computational and cognitive demands (Banich, 1998, 2003) may explain the weaker transfer effect in the RVF/LH condition. Because the RH is better suited to process visuospatial and attentional information (Posner & Petersen, 1990; Vogel et al., 2003), we propose that sequential knowledge acquisition may be facilitated in the LVF condition, in which stimuli and the recurrent pattern are initially processed in the RH. Further studies are needed to confirm and extend these results but also to test these potential explanatory hypotheses. Acknowledgments The authors thank Anicée Duterte and Sendybel Pagliara for help in data acquisition, and Prof. Iring Koch and an anonymous reviewer for their careful reading and constructive comments on a previous version of this manuscript. R.S. and A.P. are supported by the Belgian Fonds National de la Recherche Scientifique (FNRS). A.C. is a Research Director with the F.R.S.-FNRS. This study has been conducted with support of FRSM Grants 3.4.594.08.F and 3.4.582.09.F. References Albouy, G., Sterpenich, V., Balteau, E., Vandewalle, G., Desseilles, M., Dang-Vu, T., et al. (2008). Both the hippocampus and striatum are involved in consolidation of motor sequence memory. Neuron, 58(2), 261–272. Andres, F. G., Mima, T., Schulman, A. E., Dichgans, J., Hallett, M., & Gerloff, C. (1999). Functional coupling of human cortical sensorimotor areas during bimanual skill acquisition. Brain, 122(Pt 5), 855–870. Banich, M. T. (1998). The missing link: The role of interhemispheric interaction in attentional processing. Brain and Cognition, 36(2), 128–157. Banich, M. T. (2003). Interaction between the hemispheres and its implications for the processing capacity of the brain. In K. Hugdahl & R. J. Davidson (Eds.), The asymmetrical brain (pp. 261–302). Cambridge, MA: The MIT Press. Berry, D. C., & Broadbent, D. E. (1984). On the relationship between task performance and associated verbalizable knowledge. Quarterly Journal of Experimental Psychology, 39, 585–609. Bourne, V. J. (2006). The divided visual field paradigm: Methodological considerations. Laterality, 11(4), 373–393.

R. Schmitz et al. / Brain and Cognition 81 (2013) 1–9 Brown, J., Aczel, B., Jimenez, L., Kaufman, S. B., & Grant, K. P. (2010). Intact implicit learning in autism spectrum conditions. Quarterly Journal of Experimental Psychology (Colchester), 63(9), 1789–1812. Chase, C., & Seidler, R. (2008). Degree of handedness affects intermanual transfer of skill learning. Experimental Brain Research, 190(3), 317–328. Cherry, K. E., & Stadler, M. A. (1995). Implicit learning of a nonverbal sequence in younger and older adults. Psychology and Aging, 10(3), 379–394. Cleeremans, A., Destrebecqz, A., & Boyer, M. (1998). Implicit learning: News from the front. Trends in Cognitive Sciences, 2(10), 406–416. Clegg, B. A., Digirolamo, G. J., & Keele, S. W. (1998). Sequence learning. Trends in Cognitive Sciences, 2(8), 275–281. Cohen, A., Ivry, R. I., & Keele, S. W. (1990). Attention and structure in sequence learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16(1), 17–30. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. Corballis, P. M. (2003). Visuospatial processing and the right-hemisphere interpreter. Brain and Cognition, 53(2), 171–176. Curran, T. (1997). Effects of aging on implicit sequence learning: Accounting for sequence structure and explicit knowledge. Psychological Research, 60(1–2), 24–41. de Guise, E., del Pesce, M., Foschi, N., Quattrini, A., Papo, I., & Lassonde, M. (1999). Callosal and cortical contribution to procedural learning. Brain, 122(Pt 6), 1049–1062. de Guise, E., & Lassonde, M. (2001). Callosal contribution to procedural learning in children. Developmental Neuropsychology, 19(3), 253–272. Destrebecqz, A., & Cleeremans, A. (2001). Can sequence learning be implicit? New evidence with the process dissociation procedure. Psychonomic Bulletin and Review, 8(2), 343–350. Destrebecqz, A., & Peigneux, P. (2005). Methods for studying unconscious learning. Progress in Brain Research, 150, 69–80. Destrebecqz, A., Peigneux, P., Laureys, S., Degueldre, C., Del Fiore, G., Aerts, J., et al. (2005). The neural correlates of implicit and explicit sequence learning: Interacting networks revealed by the process dissociation procedure. Learning and Memory, 12(5), 480–490. Dien, J. (2008). Looking both ways through time: The Janus model of lateralized cognition. Brain and Cognition, 67(3), 292–323. Frensch, P. A., & Miner, C. S. (1994). Effects of presentation rate and individual differences in short-term memory capacity on an indirect measure of serial learning. Memory and Cognition, 22(1), 95–110. Gaillard, V., Destrebecqz, A., & Cleeremans, A. (2012). The influence of articulatory suppression on the control of implicit sequence knowledge. Frontiers in Human Neuroscience, 6, 208. Gazzaniga, M. S. (2000). Cerebral specialization and interhemispheric communication: Does the corpus callosum enable the human condition? Brain, 123(Pt 7), 1293–1326. Gerloff, C., & Andres, F. G. (2002). Bimanual coordination and interhemispheric interaction. Acta Psychologica (Amsterdam), 110(2–3), 161–186. Goedert, K. M., & Willingham, D. B. (2002). Patterns of interference in sequence learning and prism adaptation inconsistent with the consolidation hypothesis. Learning and Memory, 9(5), 279–292. Goldberg, E., & Costa, L. D. (1981). Hemisphere differences in the acquisition and use of descriptive systems. Brain and Language, 14(1), 144–173. Goldberg, E., & Podell, K. (1995). Lateralization in the frontal lobes: Searching the right (and left) way. Biological Psychiatry, 38(9), 569–571. Goldberg, E., Podell, K., & Lovell, M. (1994). Lateralization of frontal lobe functions and cognitive novelty. Journal of Neuropsychiatry and Clinical Neuroscience, 6(4), 371–378. Goldenberg, G. (2009). Apraxia and the parietal lobes. Neuropsychologia, 47(6), 1449–1459. Haaland, K. Y., & Harrington, D. L. (1996). Hemispheric asymmetry of movement. Current Opinion in Neurobiology, 6(6), 796–800. Hausmann, M. (2010). Hormonal effects on the plasticity of cognitive brain functions. Wiley Interdisciplinary Reviews Cognitive Sciences, 1, 607–612. Howard, D. V., & Howard, J. H. Jr., (1989). Age differences in learning serial patterns: Direct versus indirect measures. Psychology and Aging, 4(3), 357–364. Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30(5), 513–541. Jimenez, L., Mendez, C., & Cleeremans, A. (1996). Comparing direct and indirect measures of sequence learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22(4), 948–969. Kimura, D. (2002). Sex hormones influence human cognitive pattern. Neuroendocrinology Letters, 23(Suppl 4), 67–77. Kinsbourne, M. (1970). The cerebral basis of lateral asymmetries in attention. Acta Psychologica (Amsterdam), 33, 193–201. Koedijker, J. M., Oudejans, R. R., & Beek, P. J. (2010). Interference effects in learning similar sequences of discrete movements. Journal of Motor Behavior, 42(4), 209–222. Meulemans, T., Van der Linden, M., & Perruchet, P. (1998). Implicit sequence learning in children. Journal of Experimental Child Psychology, 69(3), 199–221. Nissen, M. J., & Bullemer, P. (1987). Attentional requirement of learning: Evidence from performance measures. Cognitive Psychology, 19, 1–32.

9

Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9(1), 97–113. Panzer, S., & Shea, C. H. (2008). The learning of two similar complex movement sequences: Does practice insulate a sequence from interference? Human Movement Science, 27(6), 873–887. Panzer, S., Wilde, H., & Shea, C. H. (2006). Learning of similar complex movement sequences: Proactive and retroactive effects on learning. Journal of Motor Behavior, 38(1), 60–70. Pasquali, A. (2009). Learning with and without consciousness: Empirical and computational explorations. Unpublished PhD thesis, Université Libre de Bruxelles, Brussels. Peigneux, P., Maquet, P., Meulemans, T., Destrebecqz, A., Laureys, S., Degueldre, C., et al. (2000). Striatum forever, despite sequence learning variability: A random effect analysis of PET data. Human Brain Mapping, 10(4), 179–194. Perez, M. A., Tanaka, S., Wise, S. P., Sadato, N., Tanabe, H. C., Willingham, D. T., et al. (2007). Neural substrates of intermanual transfer of a newly acquired motor skill. Current Biology, 17(21), 1896–1902. Perez, M. A., Tanaka, S., Wise, S. P., Willingham, D. T., & Cohen, L. G. (2008). Timespecific contribution of the supplementary motor area to intermanual transfer of procedural knowledge. Journal of Neuroscience, 28(39), 9664–9669. Perez, M. A., Wise, S. P., Willingham, D. T., & Cohen, L. G. (2007). Neurophysiological mechanisms involved in transfer of procedural knowledge. Journal of Neuroscience, 27(5), 1045–1053. Perruchet, P., & Amorim, M. A. (1992). Conscious knowledge and changes in performance in sequence learning: Evidence against dissociation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18(4), 785–800. Perruchet, P., & Pacton, S. (2006). Implicit learning and statistical learning: One phenomenon, two approaches. Trends in Cognitive Sciences, 10(5), 233–238. Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain. Annual Review of Neuroscience, 13, 25–42. Reber, A. S. (1967). Implicit learning of artificial grammars. Journal of Verbal Learning and Verbal Behavior, 6, 855–863. Reed, J., & Johnson, P. (1994). Assessing implicit learning with indirect tests: Determining what is learned about sequence structure. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20(3), 585–594. Richards, L., & Chiarello, C. (1997). Activation without selection: Parallel right hemisphere roles in language and intentional movement? Brain and Language, 57(1), 151–178. Roser, M. E., Fiser, J., Aslin, R. N., & Gazzaniga, M. S. (2011). Right hemisphere dominance in visual statistical learning. Journal of Cognitive Neuroscience, 23(5), 1088–1099. Seger, C. A., Poldrack, R. A., Prabhakaran, V., Zhao, M., Glover, G. H., & Gabrieli, J. D. (2000). Hemispheric asymmetries and individual differences in visual concept learning as measured by functional MRI. Neuropsychologia, 38(9), 1316–1324. Serrien, D. J., Ivry, R. B., & Swinnen, S. P. (2006). Dynamics of hemispheric specialization and integration in the context of motor control. Nature Reviews Neuroscience, 7(2), 160–166. Shanks, D. R. (2003). Attention and awareness in ‘‘implicit’’ sequence learning. In L. Jimenez (Ed.), Attention and implicit learning (pp. 11–42). Amsterdam, The Netherlands: John Benjamins Publishing Co.. Shanks, D. R. (2005). Implicit learning. In K. Lamberts & R. Goldstone (Eds.), Handbook of cognition (pp. 202–220). London, UK: Sage. Shanks, D. R., & Channon, S. (2002). Effects of a secondary task on ‘‘implicit’’ sequence learning: Learning or performance? Psychological Research, 66(2), 99–109. Shanks, D. R., & Johnstone, T. (1999). Evaluating the relationship between explicit and implicit knowledge in a sequential reaction time task. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25(6), 1435–1451. Shanks, D. R., & St. John, M. F. (1994). Characteristics of dissociable human learning systems. Behavioral and Brain Sciences, 17, 367–395. Smith, J., Siegert, R. J., McDowall, J., & Abernethy, D. (2001). Preserved implicit learning on both the serial reaction time task and artificial grammar in patients with Parkinson’s disease. Brain and Cognition, 45(3), 378–391. Stadler, M. A. (1995). Role of attention in sequence learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21(3), 674–685. Sturm, W., & Willmes, K. (2001). On the functional neuroanatomy of intrinsic and phasic alertness. Neuroimage, 14(1 Pt 2), S76–84. Thomas, K. M., Hunt, R. H., Vizueta, N., Sommer, T., Durston, S., Yang, Y., et al. (2004). Evidence of developmental differences in implicit sequence learning: An fMRI study of children and adults. Journal of Cognitive Neuroscience, 16(8), 1339–1351. Thomas, K. M., & Nelson, C. A. (2001). Serial reaction time learning in preschooland school-age children. Journal of Experimental Child Psychology, 79(4), 364–387. Turk-Browne, N. B., Scholl, B. J., Chun, M. M., & Johnson, M. K. (2009). Neural evidence of statistical learning: Efficient detection of visual regularities without awareness. Journal of Cognitive Neuroscience, 21(10), 1934–1945. Vandenberghe, M., Schmidt, N., Fery, P., & Cleeremans, A. (2006). Can amnesic patients learn without awareness? New evidence comparing deterministic and probabilistic sequence learning. Neuropsychologia, 44(10), 1629–1641. Vogel, J. J., Bowers, C. A., & Vogel, D. S. (2003). Cerebral lateralization of spatial abilities: A meta-analysis. Brain and Cognition, 52(2), 197–204.