Vidoni, Eric D. MS, PT; Boyd, Lara A. PT, PhD
Successful rehabilitation interventions require that physical therapists structure practice to maximize both motor skill acquisition and function. It is becoming clear that learning implicit motor skills may be influenced by explicit instructions as well as the loci of brain damage.1–5 Clinically, it is important to consider these findings during the development of individualized rehabilitation plans. The main purpose of this review is to discuss implicit and explicit learning, with a focus on how these two memory systems interact during motor skill acquisition. In the first part of this paper, we provide an overview of the learning and memory systems that subserve motor skill acquisition. To illustrate how focal lesions may disrupt the formation of motor skill memories, we also briefly consider which neuroanatomical substrates are associated with explicit and implicit learning. In the second part of this article, we present new data that illustrate how the autogenous development of explicit knowledge for a novel skill affects implicit motor performance in healthy individuals. The impact of acquiring explicit knowledge via physical practice has largely been neglected in past work, yet it represents a common circumstance during motor skill acquisition.
TWO MAJOR MEMORY AND LEARNING SYSTEMS
Learning and memory are not singular processes but are composed of many separate abilities. The broad category of long-term memory can be subdivided into two main types: declarative and nondeclarative or procedural.6 For the purpose of this article, the terms declarative and procedural are used to discuss hypothetical memory and learning systems, while the terms explicit and implicit refer to a distinction between tasks. In general, it is assumed that explicit tasks draw on the declarative, whereas implicit tasks engage the procedural memory system.7 Explicit learning may be assessed directly via memory tests that tap subjective, factual knowledge of the task as well as recognition and recall. Implicit learning is measured indirectly, as a subject's responses are altered or facilitated by the acquisition of knowledge about the structural properties of the task itself. At this level, task becomes critical, as it must guide the subject's responses even if he or she has had no prior experience or exposure.7–9
Another set of constructs critical to any discussion of declarative and procedural systems concerns the distinctions between learning and memory. Learning refers to an acquisition of knowledge about a task with practice or exposure. During explicit learning, factual knowledge is typically acquired. The knowledge that is gained during implicit learning concerns regularities for a task or relationships within a sequence of objects or events that are not intuitive or obvious. In contrast, memory refers to the accumulated effects of past experiences with a set of facts, objects, or events. Explicit memories are assessed in much the same fashion as is explicit learning, via evaluation of subjects' recognition and/or recall of factual information. Specific to implicit learning, however, the effect of prior experience or practice may be observed for a task via some performance measure (eg, decreased reaction times, fewer errors) despite the fact that subjects are not specifically asked to relate their current performance with a previous episode.8,9
Declarative and procedural memories differ fundamentally in the type of the information each stores and uses. Declarative knowledge is the conscious memory of facts, events, and episodes. It may be formed very quickly (even in one exposure) and is directly accessible to conscious recollection6 (Fig. 1). Declarative memory guides high-level cognition where decisions are based on complex rules and information. Multiple studies have demonstrated that declarative memory is severely impaired by damage to the medial temporal lobe (hippocampus and adjacent cortex).1,10
Procedural learning, on the other hand, is the capacity to acquire skill through physical practice; it is not directly accessible to conscious recollection as facts or data. The development of procedural knowledge occurs incrementally, with practice, over a period of time and exposure. Most rehabilitative tasks require clients to implicitly learn a movement pattern, with or without explicit knowledge of task characteristics. Thus, facilitating procedural learning is an important part of the therapeutic process. As this review attempts to make clear, physical therapists can alter procedural learning via their interaction with the client, both by how they structure practice sessions and also by the content of the explicit information they provide.
One of the most interesting and clinically relevant features of the declarative and procedural memory systems is their neuroanatomical isolation from one another. The declarative memory system is distinctly supported by the medial temporal lobe, and damage to this region causes profound declarative memory deficits (ie, amnesia) while leaving the procedural system intact. This dissociation was first described by Scoville and Milner1 who studied the declarative memory impairments of the patient H.M. In an attempt to ameliorate intractable seizures, H.M. underwent bilateral surgical removal of the medial temporal lobes. Following this procedure, H.M. lost all declarative memory function despite retaining normal IQ, short-term memory, and past memories for early life events. Overt explicit memory deficits notwithstanding, with practice H.M. was able to learn and retain skill for various implicit motor tasks. For example, over the course of several days of practice, H.M. was able to significantly decrease the number of errors made in a mirror-tracing task, despite a total lack of conscious awareness of previously having performed this implicit motor task.1,11 Further, H.M.'s ability to reduce errors over practice was similar to that seen for healthy control subjects. This finding of preserved procedural learning and memory in the absence of declarative memory function has been confirmed by numerous reports of procedural learning in individuals with amnesia (for examples, see references 2,12,13).
Procedural learning may be further subdivided into motor skills and habits (eg, sequence learning, mirror tracing), priming (eg, word completion), associative learning (eg, classic and operant conditioning) and nonassociative learning (eg, habituation, sensitization6,14 (Fig. 1). Each of these types of procedural learning and memory are preserved following amnesia.6 Additional criteria defining procedural learning include that (1) knowledge gained is not accessible to conscious awareness, (2) learning is an incidental consequence of information processing during practice and does not involve any conscious hypothesis testing based on explicit information or instructions, and (3) learned information is more complex than just a simple stimulus-response association.12,15 The hallmark of implicit motor learning is the capacity to acquire skill through physical practice without conscious recollection of what elements of performance improved. A classic example illustrating this process is learning to ride a bicycle. Improved performance is manifested by fewer falls, yet the ability to explicitly express “what” procedures are being used to avoid falling is almost impossible.
Nevertheless, in many learning situations, explicit awareness of task characteristics may shape performance. Green and Flowers16 present the situation of a baseball hitter watching the pitcher for physical cues regarding the nature of an upcoming pitch. The hitter can use this declarative knowledge to select an appropriate procedural motor plan to improve chances for success. In this situation, information from distinct memory systems can be integrated to facilitate successful task completion. The focus of the remainder of this article is the interaction between implicit and explicit systems during motor skill acquisition. To highlight where focal lesions may disrupt these processes and how the interaction between implicit and explicit might be supported neuroanatomically, the next section is devoted to a short discussion the neuroanatomical substrates that support the declarative and procedural systems.
NEURAL SUBSTRATES SUBSERVING THE DECLARATIVE AND PROCEDURAL SYSTEMS
An exhaustive review of the neural regions that support memory systems is beyond the scope of this article. However, because physical therapists rehabilitate individuals with damage to various brain regions, an understanding of the neuroanatomical underpinnings of declarative and procedural systems may aid in the diagnosis of memory deficits, identification of individuals who may demonstrate disrupted patterns of learning, and consideration of when explicit information might help or hurt implicit motor skill acquisition.
A critical feature of the declarative memory system is the capability to transfer new, short-term memories into long-term ones. Damage to the neural regions that support declarative learning and memory may disrupt the ability to form new long-term memories. This condition, known as amnesia, can be a side effect of surgery (as in the case of H.M.) or may also be a consequence of encephalitis or Alzheimer's disease.
The Medial Temporal Lobe
It has been robustly demonstrated that damage to the medial temporal lobe (hippocampal formation and associated cortex) severely disrupts the declarative memory system (for examples, see references 2,12,13). These types of lesions essentially dissociate the declarative and procedural memory systems. For example, amnesic patients can invoke the procedural memory system through implicit learning of a probabilistic classification task without demonstrating any explicit memory for the training session.2,13
The Prefrontal Cortex
Activation in a region in the prefrontal cortex, the dorsolateral prefrontal cortex (DLPFC) likely guides conscious, explicit hypothesis testing during learning17,18 and also appears to be critical when explicit knowledge is either gained19 or provided.20 Likely, activation of DLPFC enables factual information or declarative knowledge to be held in working memory and subsequently integrated with other visuospatial information to guide motor performance.
Generally, the formation of declarative memory during motor sequence learning is a multifaceted process critically supported by the medial temporal lobe. However, patients with hippocampal and parahippocampal damage do not lose their long-term memories1,21; thus, declarative memory storage must occur outside the medial temporal lobe. Via their role in hypothesis testing and working memory, regions in the prefrontal cortex appear to be critical for the acquisition, but not the neural encoding, of explicit knowledge. Most important for the formation of declarative memories is the medial temporal lobe structures (hippocampus and adjacent cortex) and damage to these areas eliminates both the ability to encode and retrieve declarative knowledge.
No single or focal region of the brain appears to subserve all implicit motor-sequence learning. Rather, it is evident that three broadly defined brain regions are critical for implicit motor-sequence learning: the cerebellum, basal ganglia, and sensorimotor cortical areas.
During motor learning, the cerebellum monitors and updates movements online using sensory feedback.22–24 The cerebellum is ideally situated to coordinate the sensory information essential for guiding movement during implicit motor learning.25,26 Neuroanatomical connectivity confirms the receipt of both sensory information by the cerebellum from the dorsal visual system (area v5 and parietal area 7) and motor output from the motor cortex (layer V pyramidal cells via the pons)27; each are considered important for guiding and executing movements.28 Efferent outflow from the cerebellum affects the cortex bilaterally.29 In this view, the cerebellum is not necessary to generate movement; however, it is critical for optimizing movement, which facilitates implicit motor learning.
After a cerebellar stroke, participants can demonstrate improved overall implicit motor performance (ie, decreased tracking errors) indicating preserved implicit motor learning.4 Despite retaining the ability to learn some features of an implicit motor task after a cerebellar stroke, residual deficits in movement timing do not appear to improve. After a cerebellar stroke, implicit learning can be attributed to better spatial tracking accuracy rather than to temporal accuracy; normal controls show gains in both components. Consider this analogous to improvements in the spatial but not temporal components of a baseball bat swing; the ball is always missed as the swing, though spatially perfect, occurs late. Taken together with past work, these data illustrate a cerebellar role in learning to time implicit motor responses rather than plan the spatial extent of movement.
The Basal Ganglia
Numerous studies have demonstrated decreased implicit learning in individuals with diseases affecting, or damage to, the basal ganglia.4,5,30–33 Data from these studies indicates that during implicit motor skill learning, the basal ganglia likely operate to coordinate neural processes across brain regions.34–36 Numerous reciprocal connections exist between cortical regions and the basal ganglia, including the prefrontal cortex.37,38
At least five neuroanatomically separate, reciprocal basal ganglia–thalamocortical circuits allow the basal ganglia to affect multiple cortical functions.39,40 During motor learning, the “motor” circuit (putamen, thalamus, supplementary motor area [SMA], and premotor cortex [PMC]), most directly affects movement.39–42 Other “complex” circuits have been delineated, including one composed of interconnections between the caudate, thalamus, and DLPFC.37 This complex circuit may facilitate high-level integrative functions and modify action plans during motor learning.43,44
Critical for switching among different motor responses, the basal ganglia might represent one interface between explicit information in declarative memory and the implicit motor plan. Once a movement is started, individuals with Parkinson's disease have great difficulty stopping or transitioning to another response,45 suggesting that damage to the basal ganglia disrupts response selection. Several lines of evidence demonstrate that the basal ganglia are integrally involved in the advanced preparation of plans for movement.32,46,47 Further, we have shown that unilateral basal ganglia stroke does not eliminate the capacity for implicit motor learning but rather (1) slows the rate of change in motor performance and (2) disrupts the ability to benefit from explicit instructions during both the performance and learning of motor skills.3,5 The importance of the basal ganglia during motor learning was recently illustrated by the finding that stroke in this region disrupts the relationship between explicit information and implicit motor learning regardless of task type (ie, discrete versus or continuous).5
The Cortical Motor Areas
Three cortical motor areas mediate implicit learning: the primary motor cortex (M1), SMA, and PMC. Putative roles for M1 during implicit motor-sequence learning include movement initiation and fine motor coordination.48,49 Robust M1 activation during implicitly learned motor sequences is evident as compared to activity during random ones.50,51 These differences in cortical activity reflect the generation of task-specific motor processing within M1 that may include the direction of the next movement52 and/or the required force output.53
PMC mediates the transition between different implicitly learned movements in an ordered sequence, particularly when movements depend on spatial working memory.19,54 However, when explicit awareness of an implicit motor sequence is prevented, PMC is not active and appears to play a limited role in implicit motor learning.20,55,56 PMC, therefore, seems to makes a larger contribution to sequence learning when movements are being directed by external cues or explicit information.
In contrast, SMA appears to increase its activity when explicit knowledge is unavailable and learning is purely implicit or internally driven. Single-cell recordings from monkey SMA show that some cells are only active during internally guided sequences; further, activity in other subsets of neurons is associated with the production of previously learned sequences of action.54 Thus, the SMA participates in the selection of responses when choices are based on internal information or implicit learning.
The functional interplay between the SMA and PMC demonstrates their unique contributions to implicit motor learning. PMC is highly active early in practice when explicit strategies are being used to learn the sequence. Then, as sequence production became more automatic, PMC activity decreases and SMA activity increases.17 In sum, activation in these regions depends on available information; PMC uses external explicit information and the SMA coordinates internally guided implicit responses.
Previously, we found that providing explicit instruction to individuals with stroke that affected the motor cortical areas disrupted implicit motor sequence learning.57 Interestingly, motor cortical regions, in particular the PMC, have strong connections with the prefrontal regions associated with explicit memories (ie, DLPFC) and are also richly and reciprocally interconnected with the caudate nucleus of the basal ganglia.42,58 It is quite likely that damage to, and in regions associated with, the PMC results in disrupted integration of explicit information into planned sequences of movement. Taken together, it appears that disruption to either the motor cortical areas or the basal ganglia leads to diminished ability to take advantage of explicit instructions during implicit motor-sequence practice. In the next section, we consider why this may be the case.
INTERACTIONS BETWEEN EXPLICIT AND IMPLICIT SYSTEMS
A theme is emerging in the motor skill learning literature suggesting that implicit and explicit learning can (1) be separated and operate in isolation or (2) affect and interact with one another.59 Previous data demonstrate that it cannot be assumed that the mixing of explicit and implicit is always helpful,3,57,59–61 although for healthy controls who are performing relatively simple tasks, it can be beneficial.3,5,60–62 However, we and others have shown that providing explicit instructions before to task practice can be detrimental to motor skill acquisition as compared to allowing the implicit system to learn task regularities without the imposition of external guidance.3,5,57,60,61,63 This finding is certainly not new; in 1892, Bliss63 and subsequently in 1935, Boder64 described interference effects of instructions on motor learning. Disrupted implicit learning after the adoption of conscious, explicit experimenter suggested strategies has been demonstrated for both motor skills in healthy populations61,65 and individuals with stroke,3,5,57 as well as in cognitive learning.60,66 Attempting to apply an overt control strategy based on explicit instructions can result in the production of a less efficient movement pattern and impede the development of a correct and accurate strategy for success.3,5,13,57,61 Wulf and Weigelt67 suggested that sometimes the cognitive demand of instructions may disrupt the formation of the implicit motor plan. This may occur because the rules necessary for successful task completion are not ones that can be expressed explicitly. Learners are essentially being told to search for and use rules that they were not likely to find60 and/or not likely to find useful.61
Neurobiologically, one explanation for the disruptive effect of explicit instruction may be neural competition between the implicit and explicit systems during learning.66 This may be due to incompatible demands during motor learning; the need for access to flexible knowledge maintained by the medial temporal lobe stands at odds with the necessity of fast, automatic responses supported by the basal ganglia, motor cortical regions, and cerebellum. During motor skill learning rapid, reciprocal changes in activation in the medial temporal lobe are seen in coordination with similar patterns of activity in the striatum; the net result is that only one of these brain regions is ever on at any given time. Our previous findings of diminished ability to use explicit information during implicit learning after a basal ganglia stroke raise the possibility that damage to even a portion of the striatum disrupts the coordinated alternation of activation between brain regions and alters implicit motor learning.3
However, it is also possible to internally develop explicit knowledge during physical task practice. Acquired explicit knowledge for task regularities is thought to autogenously develop via afferent feedback and/or knowledge of results. To date, it is not known how explicit information that is provided by an experimenter or therapist differently affects implicit motor learning as compared to explicit information that is acquired through physical practice. That is, how is explicit knowledge of task characteristics used by learning and memory systems when it is not provided by an external source but rather is developed from information processed through internal feedback systems? Unlike experimenter- or therapist-delivered explicit information, there is relatively low cognitive load on the learner when he or she uses autogenously developed explicit knowledge. In this case, it may be that explicit knowledge of “declarable” task regularities may not be detrimental to learning. In the following section of this article, we present data that suggests that indeed this may be the case; we found that acquired explicit knowledge of a repeating motor sequence did not disrupt learning but instead altered the variability of performance during acquisition practice.
The Influence of Gaining Autogenous Explicit Knowledge on Implicit Learning
In this section, we present new data that extend the story of how the explicit and implicit systems interact during motor skill learning. As a part of ongoing work, we are characterizing the impact of autogenous explicit knowledge during implicit motor skill practice of a continuous tracking task. This is in contrast to previous work that has considered the influence of explicit instruction delivered by an external source (eg, an experimenter, teacher, coach, physical therapist) in advance of practice. In the present study, we allowed participants to gain explicit knowledge of task characteristics only through the use of internally generated feedback; no specific explicit instruction regarding the presence or composition of a repeating sequence was delivered. We expected that our sample of young, healthy individuals would over the course of practice sometimes develop explicit knowledge of the presence of the repeating motor sequence in our task and that consistent with our previous work, this explicit knowledge might not necessarily benefit implicit learning. We found that indeed some participants (but not all) did autogenously acquire explicit knowledge, but were surprised to note that the impact of declarative awareness was not necessarily detrimental for implicit learning and was more apparent during practice than at retention.
To examine how varied explicit knowledge affected motor learning, we recruited nine healthy participants (mean age, 27.3 years; range, 24–33). All participants had vision equivalent to 20/30 or better. Hand dominance was determined using the Edinburgh Inventory. Institutionally approved informed consent was obtained from all participants.
All participants engaged in 10 blocks of 10 trials of a 30-second continuous tracking task68 over 2 days of acquisition practice. To assess motor learning of the repeated sequence, participants returned on a third day for retention testing. At retention, all participants completed two tracking trials that each contained a random and repeated sequence. Seated before a computer monitor, participants manipulated a nearly frictionless lever with a push and pull motion using motion of shoulder/elbow flexion and extension. Lever position controlled the vertical excursion of a cursor in the horizontal center of a computer screen placed in front of participants. Participants were instructed to track the target as it moved from up and down on the screen via movements of the lever.
The pattern of the targets movement was constructed using a method modified from Wulf and Schmidt.69 For each trial, a unique target pattern was assembled from two component sine-cosine wave segments, or epochs: a repeated epoch (15 seconds), similar to the one used by Wulf and Schmidt,69 and a novel epoch (15 seconds) constructed from random coefficients (Fig. 2). The order in which novel and repeated target epochs were presented within each trial was randomized. Therefore, during training, participants were exposed to the repeated epoch 100 times and also to 100 novel, randomly generated epochs. We provided no explicit instruction regarding the presence of a repeating pattern. However, to maintain motivation, participants were provided summary knowledge of results (KR) at the end of each trial regarding their tracking accuracy as a percentage of time spent within a bandwidth of ±5 degrees of the target.
To reduce the contribution of vision to tracking accuracy, visual feedback on the computer screen was linearly faded over the first 20 trials on training day 1 and maintained at a low level on days 2 and 3. This manipulation was designed to encourage participants to focus on their movements rather than on visual feedback regarding cursor position. Additionally, direct visualization of the arm was occluded by an opaque cloth extending over participant's upper body. Table 1 outlines the experimental design.
To assess acquired explicit knowledge, all participants next watched 10 recognition trials in which they were shown, but did not track with the arm, a 15-second pattern and asked to respond with “yes” or “no” as to whether they recognized the waveform. Seven of these patterns were novel, foil patterns. Participants who reported recognizing two of three truly repeating patterns and not recognizing four of seven foil patterns (demonstrating better than chance discrimination) were considered to have gained explicit knowledge of the sequence.
Raw position data were smoothed using a 100-millisecond moving average. To test motor learning, we calculated the root mean square error (RMSE) (RMSE = SQRT(Σ (xi Xi)2/n), where xi = probe position and Xi = target position) separately for repeating sequence and random epochs across of each block practice data. To determine the impact of the acquired explicit knowledge for the repeating sequence on movement variability, coefficient of variation (COV) (COV = SD (Xi)/mean(xi), where xi are the epoch-respective 10 trials of each block) was calculated for each training block. COV was not calculated at retention because of the small number of trials. Participants were divided into implicit or explicit awareness groups post hoc, according to the results of explicit knowledge tests.
The preliminary data presented here are part of a larger, ongoing study of the role of proprioception in motor learning. Participants in each of our two groups (distributed equally) received vibration to their upper arm. Because there was no effect of this manipulation on either practice or retention data for this group of participants, this factor was not considered during data analysis. Box plot analysis identified one participant as having atypical error during acquisition. This individual was removed from further analysis.
To ensure that baseline differences did not bias our between group results, we compared initial performance of the two groups with a one-factor analysis of variance (ANOVA) (knowledge [implicit, explicit]) using random epoch tracking error from block 1 as the dependent measure. To determine whether the groups responded differently to practice, a three-factor ANOVA (knowledge × epoch [random, sequence] × practice block [3–10]) with repeated-measures correction for epoch and block was performed separately for our two dependent measures, RMSE and COV. Block 3 represents the point where visual feedback had been faded to such a point where all participants had limited vision of the on screen cursor that represented their movements and was therefore used as the first data point for analyses. To evaluate skill learning at retention, we performed a two-factor ANOVA (knowledge × block [3, retention]) with repeated-measures correction for block on repeated epoch RMSE data. SPSS v.15 (SPSS Inc., Chicago, IL) was used for statistical analyses; significance was set at α = 0.05 for all tests.
In all, five of eight participants acquired explicit knowledge of the repeated sequence as indicated by their ability to discern between and correctly identify both truly repeated and novel sequences. Generally, all participants reduced tracking error with practice. However, the pattern of between group differences in COV demonstrated that the implicit group was initially more variable in their responses, whereas acquired explicit knowledge resulted in more variability late in practice. Interestingly, higher variability during practice did not affect learning; both implicit and explicit groups demonstrated sequence-specific learning at retention.
At baseline, all participants displayed similar tracking error, evidenced by a lack of an effect of knowledge at block 1 for tracking error (F < 1) (Fig. 3). Similar to previous studies using a continuous tracking paradigm,68–70 participants reduced error on the repeated epoch with practice as demonstrated by an interaction of epoch × block (F7,42 = 4.384, p = 0.001).
The benchmark of motor learning is a relatively permanent improvement in performing the skill.71 To assess knowledge-based differences from initial performance to retention testing, we performed a two-factor ANOVA (block [3, retention], knowledge [explicit, implicit]) with repeated-measures correction for block on sequence epoch RMSE. Only a significant main effect of block was detected (F1,6 = 15.923, p = 0.007), indicating that individuals learned the sequence regardless of acquisition of explicit sequence awareness (Fig. 4; see retention data).
During acquisition, participants displayed differing patterns of performance variability as evidenced by an interaction of knowledge and block (F7,42 = 2.823, p = 0.017; Fig. 5). Visual inspection suggests this was primarily a result of increased early variability in the implicit group on day 1, in combination with higher variability for the explicit group at the conclusion of day 2 of practice.
DISCUSSION OF EXPERIMENTAL DATA
Much of the previous literature regarding the impact of explicit knowledge on implicit learning either focused on feedback or instruction that was delivered by an experimenter (for reviews, see references 72,73). Few studies consider autogenous development of explicit knowledge, perhaps because participants in most implicit paradigms either do not gain explicit awareness for repeated sequences or are eliminated from data analyses if they do. Experimentally it is often the goal to maintain separation between the implicit and explicit systems; there are numerous reports of absent explicit knowledge for repeated sequences in continuous tracking paradigms despite extended practice68,70,74 in combination with improved performance and implicit learning. In the clinic, however, it is likely that explicit and implicit knowledge develop in parallel,75 and this was the situation that we sought to consider.
In the preliminary results presented here, participants acquired explicit knowledge of the repeating sequence more often than has been previously reported. Although all began training naive to the presence of a repeated sequence, 60% of participants were ultimately able to discriminate both truly repeated and novel target patterns at a better-than-chance level. Other studies have reported lower rates of development of autogenic explicit knowledge following continuous tracking practice.68,69 One factor that explains why a higher percentage of participants in the present study gained explicit knowledge may be a reduction in the amount time spent practicing a random epoch. Commonly, continuous tracking studies embed the repeated sequence between two novel distracter sequences.68,69 However, in both the present study and Shea et al,70 the ratio of exposure to the repeated versus novel sequence was 1:1 or greater; both reported higher levels of acquired explicit awareness than has been otherwise noted. Another possible reason for the higher levels of acquired explicit knowledge in the present study may be an altered strategy on the part of the participants. With visualization of the body occluded and visual feedback of the cursor representing participant movement severely reduced, individuals may have devoted more attentional resources to observing target movements. Other work has shown that directed attention to external, environmental goals enhances learning76,77 and that attentional capability may relate to motor learning.78 Hence, it is possible that individuals who have limited sensory information as a result of central neurologic or peripheral receptor damage may rely on other types of information such as the development of explicit knowledge to facilitate implicit motor learning.
It has been established that explicit and implicit learning systems can and often do operate independently.2 Poldrack et al66 have suggested that even when developing in parallel, these distinct memory systems compete for neural resources. The medial temporal lobe, which is essential for the formation of explicit memories, is responsible for learning spatial characteristics, whereas the basal ganglia–dominated implicit memory system governs the development of stimulus-response relationships. In the present experiment, we minimized stimulus-response feedback (typically managed by the basal ganglia as a part of the implicit system). It is perhaps not surprising that some participants gained explicit knowledge of the sequence, as 90% of stimulus-response feedback (ie, lever position) was withheld during practice. Restricting stimulus-response feedback (both extrinsic via limited feedback and intrinsic via restricted vision) may have forced participants to rely more heavily on learning spatial relationships mediated by the explicit system.
In our data, participants improved their continuous tracking performance with practice regardless of whether they became aware of target repetition. What was particularly interesting was the varied time course of the pattern of each group's COV data. Finding higher variability early in practice, as we did for the implicit group, was not surprising and likely relates to initial exposure to the task.79,80 Discovering that the explicit knowledge group's performance was more variable late in practice suggests explicit knowledge may stimulate different tracking strategies. Indeed, flexible memory representation is one hallmark of the explicit system,67 and it may be important in discovering whether there is more than one correct movement solution for any given task. The learning benefit of variability in practice is one of the predictions of schema theory81 and has been demonstrated repeatedly in healthy populations.74,82 In these cases, task variability during skill acquisition was structured by the researcher. Our data suggest that performance variability was either a consequence or a component of both group's learning strategy at different times during practice. Recently, it has been proposed that variability is not only helpful but essential for motor learning and performance. Stergiou et al79 have suggested that both in learning and performance, the motor system strives for an optimal level of variability. During learning, variability results in the development of a flexible motor plan for movement. From our data, it appears that regardless of whether higher variability was noted early or late in practice, motor learning occurred for the repeated sequence.
In sum, these preliminary data suggest that acquiring explicit knowledge through practice does not interfere with implicit learning in the same fashion as has been described by other work where explicit information was delivered by an experimenter.60,61 It appears that there may be a critical difference in the utility of explicit knowledge that is discovered as compared to delivered. One important caveat may be that the helpfulness of explicit information likely depends on the population, the task, and also how explicit knowledge is gained. Thus, there is no single “rule” concerning how useful or deleterious explicit information is for implicit motor skill learning. Although the data we present here are from healthy young individuals, they do mimic the common clinical situation of reduced sensory information as a result of neurologic injury. In our experiment, when faced with minimal intrinsic feedback to update and correct movement, such as might occur with neurologic pathology, some participants were able to explicitly identify and exploit regularities that allowed them to learn the repeated sequence. The distinct time course of variability in motor output shown for the explicit and implicit groups demonstrates the differences between these learning systems. Clinically, it suggests that caution should be exercised when modifying the course of a therapeutic intervention based on variability of motor performance; it appears that no linear relationship exists between variations in motor skill performance and learning.
GENERAL CONCLUSIONS AND CLINICAL IMPLICATIONS
How explicit instructions interact with and affect implicit motor learning is an important issue for physical therapists. Often in the rehabilitation setting, physical therapists attempt to substitute the delivery of explicit knowledge via instructions regarding “how to” accomplish a task for implicit motor skill practice in an effort to speed learning and enhance recovery of function. Because the impact of explicit instructions depends on multiple issues including (to name a few) the presence or absence of a lesion,61,63 specific lesion location,3,4,57 and implicit task characteristics,5,60,61,83,84 it is difficult if not impossible to form one conclusion regarding its utility. It is tempting to determine that the net impact of explicit instructions is negative; however, this conclusion is overly simplistic and not verified by the literature. Taken together, the data suggest that in the context of rehabilitation, the delivery (or not) of explicit instructions is simply another variable that physical therapists may manipulate in an effort to facilitate motor learning and recovery. The preliminary data presented in this article support this contention. It appears that during motor skill practice, even when sensory information is greatly restricted, individual learners can discover the correct solution to a movement problem using either their implicit, explicit, or a combination of these two memory systems; each strategy may lead to motor skill learning.
1. Scoville WB, Milner B. Loss of recent memory after bilateral hippocampal lesions. 1957. J Neuropsychiatry Clin Neurosci. 2000;12:103–113.
2. Knowlton BJ, Mangels JA, Squire LR. A neostriatal habit learning system in humans. Science. 1996;273:1399–1402.
3. Boyd LA, Winstein CJ. Providing explicit information disrupts implicit motor learning after basal ganglia stroke. Learn Mem. 2004;11:388–396.
4. Boyd LA, Winstein CJ. Cerebellar stroke impairs temporal but not spatial accuracy during implicit motor learning. Neurorehabil Neural Repair. 2004;18:134–143.
5. Boyd L, Winstein C. Explicit information interferes with implicit motor learning of both continuous and discrete movement tasks after stroke. J Neurol Phys Ther. 2006;30:46–59.
6. Squire LR. Memory and Brain. New York: Oxford University Press, 1987.
7. Willingham DB, Stadler MA, Frensch PA. Implicit learning and motor skill in older subjects: an extension of the processing speed theory. The Handbook of Implicit Learning. Thousand Oaks: Sage, 1998:573–594.
8. Frensch PA.. One concept, multiple meanings: on how to define the concept of implicit learning. In: Stadler MA, Frensch PA, eds. The Handbook of Implicit Learning. Thousand Oaks: Sage, 1998:47–106.
9. Buchner A, Funke J, Berry DC. Negative correlations between control performance and verbalizable knowledge-indicators for implicit learning in-process control tasks. Q J Exp Psychol A 1995;48:166–187.
10. Squire LR, Zola-Morgan S. The medial temporal lobe memory system. Science. 1991;253:1380–1385.
11. Corkin S. Acquisition of motor skill after bilateral medial temporal-lobe excision. Neuropsychologia. 1968;6:255–265.
12. Nissen MJ, Bullemer P. Attentional requirements of learning: evidence from performance measures. Cogn Psychol. 1987;19:1–32.
13. Reber PJ, Squire LR. Encapsulation of implicit and explicit memory in sequence learning. J Cogn Neurosci. 1998;10:248–263.
14. Doyon J. Skill learning. In: Schmahmann JD, ed. The Cerebellum and Cognition. San Diego: Academic Press, 1997:273–294.
15. Segar CA. Implicit learning. Psychol Bull. 1994;115:163–196.
16. Green TD, Flowers JH. Comparison of implicit and explicit learning processes in a probabilistic task. Percept Mot Skills. 2003;97:299–314.
17. Toni I, Krams M, Turner R, et al. The time course of changes during motor sequence learning: a whole- brain fMRI study. Neuroimage. 1998;8:50–61.
18. Sakai K, Hikosaka O, Miyauchi S, et al. Transition of brain activation from frontal to parietal areas in visuomotor sequence learning. J Neurosci. 1998;18:1827–1840.
19. Grafton ST. Mapping memory systems in the human brain. Neurosciences. 1995;7:157–163.
20. Honda M, Deiber MP, Ibanez V, et al. Dynamic cortical involvement in implicit and explicit motor sequence learning—a PET study. Brain. 1998;121:2159–2173.
21. Glisky EL, Schacter D. Acquisition of domain-specific knowledge in organic amnesia: training for computer-related work. Neuropsychologia. 1987;25:893–906.
22. Gao JH, Parsons LM, Bower JM, et al. Cerebellum implicated in sensory acquisition and discrimination rather than motor control. Science. 1996;272:545–547.
23. Bower JM. Is the cerebellum sensory for motor's sake, or motor for sensory's sake: the view from the whiskers of a rat? Prog Brain Res. 1997;114:463–496.
24. Jueptner M, Ottinger S, Fellows SJ, et al. The relevance of sensory input for the cerebellar control of movements. Neuroimage. 1997;5:41–48.
25. Blakemore SJ, Frith CD, Wolpert DM. The cerebellum is involved in predicting the sensory consequences of action. Neuroreport. 2001;12:1879–1884.
26. Brown IE, Bower JM. The influence of somatosensory cortex on climbing fiber responses in the lateral hemispheres of the rat cerebellum after peripheral tactile stimulation. J Neurosci. 2002;22:6819–6829.
27. Schmanhmann JD, Pandya DN. The cerebrocerebellar system. In: Schmahmann JD (ed). The Cerebellum and Cognition. Boston: Academic Press; 1993:31–60.
28. Stein JF, Glickstein M. Role of the cerebellum in visual guidance of movement. Physiol Rev. 1992;72:976–1015.
29. Molinari M, Ramachandran VS. Cerebellum. In: Encyclopedia of the Human Brain. Volume 1. San Diego: Academic Press, 2003:611–627.
30. Pascual-Leone A, Grafman J, Clark K, et al. Procedural learning in Parkinson's disease and cerebellar degeneration. Ann Neurol. 1993;34:594–602.
31. Doyon J, Gaudreau D, Laforce R Jr, et al. Role of the striatum, cerebellum, and frontal lobes in the learning of a visuomotor sequence. Brain Cogn. 1997;34:218–245.
32. Jennings PJ. Evidence of incomplete motor programming in Parkinson's disease. J Mot Behav. 1995;27:310–324.
33. Jackson GM, Jackson SR, Harrison J, et al. Serial reaction time learning and Parkinson's disease: evidence for a procedural learning deficit. Neuropsychologia. 1995;33:577–593.
34. Poldrack RA, Prabhakaran V, Seger CA, et al. Striatal activation during acquisition of a cognitive skill. Neuropsychology. 1999;13:564–574.
35. Willingham DB, Koreshetz WJ, Treadwell JR, et al. Comparison of Huntington's and Parkinson's disease patient's use of advance information. Neuropsychology. 1995;9:39–46.
36. Willingham DB, Salidis J, Gabrieli JDE. Direct comparison of neural systems mediating conscious and unconscious skill learning. J Neurophysiol. 2002;88:1451–1460.
37. Middleton FA, Strick PL,. New concepts about the organization of basal ganglia output. Adv Neurol. 1997;74;57–68.
38. Cavada C, Goldman-Rakic PS. Topographic segregation of corticostriatal projections from posterior parietal subdivisions in the macaque monkey. Neuroscience. 1991;42:683–696.
39. Alexander GE, Crutcher MD. Functional architecture of basal ganglia circuits. Trends Neurosci. 1990;13:266–271.
40. Alexander GE, DeLong MR, Strick PL, et al. Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annu Rev Neurosci. 1986;9:357–381.
41. Alexander GE, DeLong MR, Crutcher MD. Do cortical and basal ganglionic motor areas use “motor programs” to control movement? Behav Brain Sci. 1992;15:656–665.
42. Selemon LD, Goldman-Rakic PS. Longitudinal topography and interdigitation of corticostriatal projections in the rhesus monkey. J Neurosci. 1985;5:776–794.
43. Braver TS, Cohen JD, Nystrom LE, et al. A parametric study of prefrontal cortex involvement in human working memory. Neuroimage. 1997;5:49–62.
44. Rypma B, Prabhakaran V, Desmond JE, et al. Load-dependent roles of frontal brain regions in the maintenance of working memory. Neuroimage. 1999;9:216–226.
45. Harrington DL, Haaland KY. Sequencing in Parkinson's disease: abnormalities in programming and controlling movement. Brain. 1991;114:99–115.
46. Harrington DL, Haaland KY, Rosenbaum DA, et al. Sequencing and timing operations of the basal ganglia. In: Rosenbaum DA, Collyer CE (eds). Timing of Behavior: Neural, Psychological and Computational Perspectives. Cambridge: MIT Press, 1998:35–61.
47. Stelmach GE, Worringham CJ, Strand EA. The programming and execution of movement sequences in Parkinson's disease. Int J Neurosci. 1987;36:55–65.
48. Berridge KC, Whishaw IQ. Cortex, striatum and cerebellum: control of serial order in a grooming sequence. Exp Brain Res. 1992;90:275–290.
49. Barone P, Joseph JP. Prefrontal cortex and spatial sequencing in macaque monkey. Exp Brain Res. 1989;78:447–464.
50. Kawashima R, Roland PE, O'Sullivan BT. Fields in human motor areas involved in preparation for reaching, actual reaching, and visuomotor learning: a positron emission tomography study. J Neurosci. 1994;14:3462–3474.
51. Shibasaki H, Sadato N, Lyshkow H, et al. Both primary motor cortex and supplementary motor area play an important role in complex finger movement. Brain. 1993;116:1387–1398.
52. Kakei S, Hoffman DS, Strick PL. Muscle and movement representations in the primary motor cortex. Science. 1999;285:2136–2139.
53. Mullebacher W, Ziemann U, Boroojerdi B, et al. Role of the human motor cortex in rapid motor learning. Exp Brain Res. 2001;136:431–438.
54. Mushiake H, Inase M, Tanji J. Neuronal activity in the primate premotor, supplementary, and precentral motor cortex during visually guided and internally determined sequential movements. J Neurophysiol. 1991;66:705–718.
55. Hazeltine E, Grafton ST, Ivry R. Attention and stimulus characteristics determine the locus of motor-sequence encoding a PET study. Brain. 1997;120:123–140.
56. Grafton ST, Hazeltine E, Ivry R. Functional mapping of sequence learning in normal humans. J Cogn Neurosci. 1995;7:497–510.
57. Boyd LA, Winstein CJ. Impact of explicit information on implicit motor-sequence learning following middle cerebral artery stroke. Phys Ther. 2003;83:976–989.
58. Cavada C, Goldman-Rakic PS. Posterior parietal cortex in rhesus monkey: II. Evidence for segregated corticocortical networks linking sensory and limbic areas with the frontal lobe. J Comp Neurol. 1989;287:422–445.
59. Willingham DB. Becoming aware of motor skill. Trends Cogn Sci. 2001;5:181–182.
60. Reber AS. Implicit learning of synthetic languages: the role of instructional set. J Exp Psychol Hum Learn Mem. 1976;2:88–94.
61. Green TD, Flowers JH. Implicit versus explicit learning processes in a probabilistic, continuous fine-motor catching task. J Mot Behav. 1991;23:293–300.
62. Curran T, Keele SW. Attentional and nonattentional forms of sequence learning. J Exp Psychol Learn Mem Cogn. 1993;19:189–202.
63. Bliss CB. Investigations in reaction time and attention. Studies from the Yale Psychology Laboratory. 1892;1:1–55.
64. Boder DP. The influence of concomitant activity and fatigue upon certain forms of reciprocal hand movements and its fundamental components. Comp Psychol Monogr. 1935;11.
65. Verdolini-Marston K, Balota DA. Role of elaborative and perceptual integrative processes in perceptual-motor performance. J Exp Psychol Learn Mem Cogn. 1994;20:739–749.
66. Poldrack RA, Clark J, Pare-Blagoev EJ, et al. Interactive memory systems in the human brain. Nature. 2001;414:546–550.
67. Wulf G, Weigelt C. Instructions about physical principles in learning a complex motor skill: to tell or not to tell. Res Q Exerc Sport. 1997;68:362–367.
68. Pew RW. Levels of analysis in motor control. Brain Res. 1974;71:393–400.
69. Wulf G, Schmidt RA. Variability of practice and implicit motor learning. J Exp Psychol Learn Mem Cogn. 1997;23:987–1006.
70. Shea CH, Park JH, Braden HW. Age-related effects in sequential motor learning. Phys Ther. 2006;86:478–488.
71. Salmoni AW, Schmidt RA, Walter CB. Knowledge of results and motor learning: a review and critical reappraisal. Psychol Bull. 1984;95:355–386.
72. Schmidt RA, Lee TD. Augmented feedback. In: Motor Control and Learning: A Behavioral Emphasis. Champaign, IL: Human Kinetics, 1999:323–355.
73. Hodges NJ, Franks IM. Modelling coaching practice: the role of instruction and demonstration. J Sports Sci. 2002;20:793–811.
74. Wulf G, Schmidt RA. Variability of practice and implicit motor learning. J Exp Psychol Learn Mem Cogn. 1997;23:987–1006.
75. Goedert KM, Willingham DB. Patterns of interference in sequence learning and prism adaptation inconsistent with the consolidation hypothesis. Learn Mem. 2002;9:279–292.
76. Wulf G, Shea C, Park JH. Attention and motor performance: preferences for and advantages of an external focus. Res Q Exerc Sport. 2001;72:335–344.
77. Wulf G, McNevin N, Shea CH. The automaticity of complex motor skill learning as a function of attentional focus. Q J Exp Psychol A. 2001;54:1143–1154.
78. McDowd JM, Craik FI. Effects of aging and task difficulty on divided attention performance. J Exp Psychol Hum Percept Perform. 1988;14:267–280.
79. Stergiou N, Harbourne RT, Cavanaugh JT. Optimal movement variability—a new theoretical perspective for neurologic physical therapy. J Neurol Phys Ther. 2006;30:120–129.
80. Schmidt RA, Lee TD. Motor Control and Learning: A Behavioral Emphasis. Champaign, IL: Human Kinetics, 1999.
81. Schmidt RA. A schema theory of discrete motor skill learning. Psychol Rev. 1975;82:225–260.
82. Catalano JF, Kleiner BM. Distant transfer in coincident timing as a function of variability of practice. Percept Mot Skills. 1984;58:851–856.
83. Boyd LA, Quaney BM, Pohl PS, et al. Learning implicitly: effects of task and severity after stroke. Neurorehabil Neural Repair. 2007;in press.
84. Reber PJ, Stark CE, Squire LR. Contrasting cortical activity associated with category memory and recognition memory. Learn Mem. 1998;5:420–428.