Reberlab Cognitive Neuroscience of Learning and Memory

Explicit knowledge influences consolidation but not immediate performance in implicit skill learning

Sanchez, D.J., & Reber, P.J. (2011) Society for Neuroscience.

Perceptual-motor sequence learning has often been used as a task for dissociating the neural mechanisms and operating characteristics of the explicit and implicit memory systems. Although neuropsychological and behavioral evidence indicate separable memory systems in the brain, the interaction of knowledge representations across the systems is likely to be vital in everyday activities. For instance, motor skill acquisition typically relies on explicit, declarative knowledge of what to do, which precedes the development of the implicit, procedural knowledge necessary for improving performance. Using the Serial Interception Sequence Learning (SISL) task, the effect of explicit knowledge on skill learning was examined in order to identify how the interaction of knowledge representations may benefit (or inhibit) performance. The SISL task utilizes perceptually-guided responding to intercept moving cues, and mimics real world skill expertise in requiring accurate timing and order between motor responses. A repeating sequence of cues can be covertly embedded in the task, which is then learned implicitly by participants. Across two experiments, adding explicit pre-instruction about the 12-item sequence prior to training produced no benefit in acquisition or performance in this task, even when explicit knowledge was exceptionally robust. In a second set of experiments examining skill learning consolidation, participants trained on two sequences (SeqA and SeqB) consecutively on one day and 48 hours later received training on a third sequence (SeqC), followed by a test of all three sequences. Retroactive interference between the first two trained sequences (from SeqB to SeqA) was affected by whether participants had explicit knowledge of the initial sequence (SeqA). Participants with relatively higher concomitant explicit knowledge of SeqA exhibited better retention at test for all three sequences than those with lower explicit knowledge (measured by sequence recognition). This relationship between recognition and performance was not found for either SeqB or SeqC. Curiously, while the ability to recognize this particular sequence predicted higher levels of sequence-specific performance improvements, it did not predict higher levels of corresponding explicit knowledge for the other sequences. Although explicit sequence knowledge did not directly translate to a benefit in initial acquisition or performance, a positive effect of explicit knowledge on motor skill consolidation may be related to the ability of the medial temporal lobe memory system in sorting and storing information representations without catastrophic interference.

Bojinov, H., Sanchez, D., Reber, P., Boneh, D., & Lincoln, P.

Cryptographic systems often rely on the secrecy of cryptographic keys given to users. However, many schemes cannot resist coercion attacks where the user is forcibly asked by an attacker to reveal the key. These attacks, known as rubber hose cryptanalysis, are often the easiest way to defeat cryptography. We present a new approach to protecting against coercion attacks using the concept of implicit learning from cognitive psychology. Implicit learning refers to learning of patterns without any conscious knowledge of the learned pattern. We use a carefully crafted computer game to plant a secret password in the participant’s brain without the participant having any conscious knowledge of the trained password. While the planted secret can be used for authentication, the participant cannot be coerced into revealing it since he or she has no conscious knowledge of it. We performed a number of user studies using Amazon’s Mechanical Turk to verify that participants can successfully re-authenticate over time and that they are unable to reconstruct or even recognize short fragments of the planted secret.

Sanchez, D.J., Wesley, A.H., & Reber, P.J.

Although implicit skill learning occurs incidentally and without conscious awareness of what is learned, the rate and effectiveness of learning may still be affected by variation in the cognitive state of the learner during practice. Ego depletion theory states that humans possess a limited store of cognitive resources that, when depleted, results in deficits in self-regulation and cognitive control. While the specific mechanism of ego depletion is not known, its effects on processes associated with dopaminergic function suggest the possibility that ego depletion might affect the dopamine-gated plasticity that is hypothesized to support implicit sequential skill learning. In a first experiment this idea was tested using an assessment of pre-experimental depletion given prior to training on the Serial Interception Skill Learning (SISL) task and relative depletion state was found to predict implicit learning rate. In a second experiment, ego depletion was manipulated by having participants complete a depleting task prior to a standard implicit learning protocol. Depleted participants exhibited less learning than did non-depleted controls. In a third experiment, depletion was administered after training and was not found to interfere with skilled performance, suggesting the effect of ego depletion is specific to learning. These results indicate that in both training and educational contexts, ego depletion should be avoided prior to practice to maximize training gains even from rote or repetitive practice.

Submitted to CABN.

Sanchez, D.J. & Reber, P.J.

Skills are typically learned by initial explicit instruction followed by repetitive practice to hone and improve performance. Memory systems theory provides a framework for characterizing the basis of this process whereby declarative memory plays a large initial role in instruction and implicit learning subsequently becomes important during practice. However, prior studies dissociating these two types of memory have not indicated how these systems interact during the skill learning process. The types of memory could operate largely independently, with explicit memory merely providing an initial scaffolding to guide future practice. Alternately, there could be cooperative interactions between the systems by which they form a shared representation that drives performance. Using an implicit perceptual-motor sequence learning task, skill learning was compared when explicit instruction was available or under typical incidental learning conditions. In Experiment 1, sequence pre-instruction did not lead to improved skill learning, but explicit memory for the sequence was poor after practice. In Experiment 2, improved instruction led to better explicit knowledge, but as in Experiment 1, sequence learning was robust and equivalent for both conditions. The lack of an instruction benefit suggests that during skill learning, implicit and explicit memory operate independently. Initial instruction appears to mainly serve to guide initial action sequence performance so that it can be made faster, more accurate and more fluid by implicit learning during repetitive practice.

Submitted to Memory & Cognition

Title: Coordinated action and timing responses separated across hands are integrated in sequence learning.

Area: Cognitive Neuroscience of Learning and Memory.

Problem or Major Purpose: Skill learning relies on a flexible explicit set of actions to perform which provides scaffolding for the relatively inflexible implicit representation supporting performance improvements through practice. A key question regarding implicit knowledge is to what extent the timing and order of sequential movements is specific and inflexible to what was practiced. Sanchez, Gobel, & Reber (2010) previously used the Serial Interception Sequence Learning (SISL) task to demonstrate implicit learning in healthy patients with little trace of associated explicit knowledge. In the SISL task, participants’ attempt to make a precisely-timed motor response to cues scrolling down a monitor toward one of four target zones. The cues follow a covertly-embedded repeating sequence of cue order and inter-cue timing. Other work examining the integration of sequential response order and timing with the SISL task has shown that these two sources of information are integrated during learning (Gobel, Sanchez, & Reber, 2011). This produces a surprisingly inflexible knowledge representation which resists transfer to very similar motor sequences.

Procedure: To test the hypothesis that sequence inflexibility arose from the need to combine order and timing information into a single keypress response, SISL learning was examined with a guitar-shaped manipulandum that separated action selection and response timing across hands and required a bimanually-coordinated response on each trial. Twenty-eight Northwestern University undergraduates (17 F, Mean Age = 21.3 years) participated for course credit. Participants completed 2880 trials of training on the SISL task, followed by a test where the order of cue responses and inter-cue timing were separately manipulated from the trained sequence in order to assess transfer to sequences with novel timing or novel order.

Results: Participants exhibited sequence-specific performance improvements for only the trained sequence and performance was equivalent to an unpracticed sequence if either timing or order was disrupted. Separate examination of each hand individually also failed to show any evidence of partial transfer from the trained sequence.

Conclusions and Implications: When response timing and order are both necessary for coordinated sequence performance, they become integrated in the motor plan that is necessary for expression even when expressed largely through different hands. These results have broader implications to educational training whereby rote practice of cognitive or motor skills may result in a hyper-specific, inflexible knowledge representation.

Mini-Abstract: Skill learning was examined with a guitar-shaped manipulandum that separated action selection and response timing across hands and required a bimanually-coordinated response on each trial. Participants exhibited sequence-specific performance improvements for only the trained sequence and performance was equivalent to an unpracticed sequence if either timing or order was disrupted.

Working memory training gains and transfer to other cognitive functions

Gigler, K.L. & Reber, P.J.

Recent research demonstrating improvements in working memory (WM) capacity has challenged the idea that WM capacity is an immutable cognitive trait. Indeed, relatively modest training protocols have been shown to lead to significant improvement. Because WM is a core cognitive process, increasing capacity has the potential to enhance performance on a wide range of cognitive functions. A critical question is the degree to which gains exhibited on the WM training task transfer to other cognitive processes. To test for transfer, the CogState cognitive assessment battery was completed by participants both before and after 10 hours of WM training with a novel training protocol. The CogState assessment battery includes a collection of tests that measure long-term memory function, executive function, attention and processing speed. During training, participants completed 2000 training trials of a visuospatial working memory task. Each trial consists of two phases: the presentation phase, during which participants see a sequence of moving visual cues and must hold that sequence in WM, and the response phase, during which participants attempt to replicate the sequence. The training is adaptive, adjusting the length of presented sequences based on performance in order to keep training near each individual’s WM span. Participants showed reliable improvement in WM span on both the trained task and a separate, non-trained assessment of visuospatial WM. Significant improvements were also observed on CogState measures of non-verbal long-term memory, attention, and processing speed, indicating that WM training can produce considerable gains in cognitive functioning beyond merely domain-specific WM improvement.

Sequence-specific and non-specific gains in working memory following cognitive training

Gigler, K.L. & Reber, P.J.

Working memory (WM) refers to the ability to hold a limited amount of information in mind for a short period of time and is a core cognitive component important for many higher-level cognitive functions, including problem solving and language comprehension. An increasing volume of research indicates that individual WM capacity can be enhanced through training, potentially improving cognitive performance in a variety of domains; however, a major challenge to realizing the value of this approach lies in the tendency of WM gains to be domain-specific, limited to only trained material. The current research utilized a novel visuo-spatial WM training task based on a game-like sequence learning task. It is comprised of two phases wherein a sequence of moving visual cues is first observed and then replicated following a 2-second delay, during which the sequence must be held in WM. Training was adaptive in that the length of the sequences increased as participants improved, increasing demands on WM and maintaining a challenging and engaging level of difficulty. Participants completed 450 trials of training over 2 hour-long sessions across 2 days, with a repeating sequence of spatial locations covertly embedded in 20% of trials. Increased WM capacity was found after training for both repeating and novel sequences, although observed gains were larger for the repeating sequence. The sequence-specific improvements reflect the potential for hyper-specific gains in WM capacity following training, analogous to domain-specific improvements, while the task-general gains indicate the potential for expanding general WM function through cognitive training using this task.

Training on a game-like working memory task can improve visuo-spatial working memory capacity

Gigler, K.L. & Reber, P.J.

The question of transfer has emerged in studies of cognitive training that have found improvements in the trained cognitive processes, but inconsistent transfer of these improvements to other cognitive functions. Similar findings exist in the working memory (WM) literature; while domain-specific increases in WM capacity, or span, have been observed, such increases do not generalize to other domains. Because WM is known to be associated with individual differences in performance on tasks from problem solving to language comprehension, it is possible that succesfully training domain-general working memory could transfer to many other cognitive processes as well. Using a novel visuo-spatial WM training task, the current research demonstrates that working memory training can produce both general and domain-specific improvements following two hours of practice. The SISL-WM task is based on a game-like sequence learning task used to study skill learning. The task is made up of two phases, the first of which consists of the participant observing a sequence of moving visual cues across a computer screen that indicate a sequence of motor responses. After a 2-second delay interval, during which time the sequence must be held in working memory, the participant attempts to correctly replicate the cued motor response sequence. The task is adaptive in that the presented sequences increase in length as the participant improves with training and decrease in length if performance is poor, maintaining an acceptably challenging level of task difficulty. The number of cues in a given sequence is considered the participant’s current working memory span for the task. Sequences consist of either randomly ordered cues (80% of trials) or of fragments drawn from a covertly repeating 12-cue sequence (20% of trials). Participants completed 2 separate, hour-long sessions of adaptive training on the SISL-WM task. Working memory span on the task reliably increased from the beginning to end of training, improving for both random and repeated sequences. Furthermore, these individuals demonstrated significant improvement in working memory span on a different, untrained visuospatial WM task, showing some generalization of training gains. This demonstrates that the improvement in working memory span seen in training was not task-specific and benefitted viuospatial working memory more generally. These results indicate the viability of this visuospatial working memory task as a training task, and further suggest that training of working memory can indeed lead to generalizable cognitive improvement.

Coordinated action selection and timing responses separated across hands are integrated in sequence learning

Daniel J. Sanchez, Eric N. Yarnik, Paul J. Reber

Previous work examining the integration of sequential response order and timing with the Serial Interception Sequence Learning (SISL) task has shown that these two sources of information are integrated during learning (Gobel, Sanchez, & Reber, 2011). This produces a surprisingly inflexible knowledge representation which resists transfer to very similar motor sequences. To test the hypothesis that this inflexibility arose from the need to combine order and timing information into a single keypress response, SISL learning was examined with a manipulandum (simulated guitar) that separated action selection and response timing across hands and required a bimanually coordinated response on each trial. Participants completed 2880 trials of training on the SISL task in which precisely-timed motor responses were made to cues moving along one of four columns that followed a covertly embedded repeating sequence of cue order and inter-cue timing. At test, the order of cue responses and inter-cue timing were separately manipulated from the trained sequence in order to assess transfer to sequences with novel timing or novel order. Participants exhibited sequence-specific performance improvements for only the trained sequence and performance was equivalent to an unpracticed sequence if either timing or order was disrupted. Separate examination of each hand individually also failed to show any evidence of partial transfer from the trained sequence. When response timing and order are both necessary for accurate coordinated sequence performance, they become integrated in the motor plan that is necessary for expression even when expressed largely through different hands.

A comparison of implicit perceptual-motor skill learning in individuals with mild cognitive impairment and Parkinson’s disease

Gobel, E. W., Blomeke, K. M., Weintraub, S., & Reber, P. J.

Skilled performance of complex motor skills requires learning a specific order of movements with precise timing. The Serial Interception Sequence Learning (SISL) task has been used to study the neural basis of this type of perceptual-motor learning. During the SISL task, participants make computer keypress responses that coincide with the timed passage of spatial cues that move through specified target regions on the computer screen. Participants are not told that the sequence of cues repeats itself, but implicit learning of the sequence is reflected by improvement in performance of the practiced sequence with repetition. Learning the sequence without awareness suggests that this type of memory does not depend on the medial temporal lobe memory system affected in Mild Cognitive Impairment (MCI) and Alzheimer’s disease (AD). Instead, this type of learning may depend on the basal ganglia and corticostriatal circuits. These subcortical systems, rather than medial temporal regions, are implicated in Parkinson’s disease (PD). To assess the importance of these different memory systems for learning in the SISL task, cognitively healthy older participants (N = 15, mean age 70.6 years), individuals with a clinical diagnosis of MCI (N = 10, mean age 77.8), and those with a diagnosis of PD (N = 15, mean age 63.7 years) performed a 3-button unimanual version of the SISL task, with the initial speed of the moving cues set to an appropriate level of difficulty for each participant. During training, cues generally followed a 12-item repeating sequence over 1,440 training trials (11-39 min depending on individual speed). On a subsequent test phase, the cognitively healthy and MCI participant groups both exhibited reliable implicit sequence knowledge. The PD patient group, while not showing a task performance deficit, failed to show reliable sequence learning. However, the PD group was heterogenous: 2 exhibited strong sequence learning, 3 showed moderate (but reliable) learning, and 10 did not appear to learn. None of the groups could reliably identify the trained sequence, consistent with the unavailability of this form of learning to conscious recollection. These results are consistent with our hypothesis that serial interception sequence learning is not dependent on the declarative memory system impaired in MCI, but likely depends on the basal ganglia.

Training working memory using a novel visuospatial task

Gigler, K. L. & Reber, P. J.

Though cognitive psychologists have long thought working memory span immutable, compelling recent evidence suggests that it is possible to improve other memory systems through various cognitive training programs. Furthermore, work done with older adult populations indicates that these individuals benefit from such memory training as much, if not more so, than younger adults. Because working memory is innately tied to executive function and other higher order cognitive processes which decline sharply in older adults, the training and improvement of this system could lead to transfer to various important cognitive processes in this population. Though truly successful working memory training has not yet been described, several laboratories have identified important and trainable aspects of working memory. The current research utilizes a novel visuospatial working memory task created with such previous research in mind. The task is designed in the style of a “replicate” task, with a first phase in which participants learn a sequence of items to be held in working memory, and a second phase in which participants attempt to replicate this learned task. A pilot study was conducted with an undergraduate population in which 21 participants completed 2 sessions of training on this task. A comparison of working memory span as measured through task performance found a significant increase in span across training, with the average span of individuals increasing from an initial span of 6.34 (SE = .23) to a span of 7.03 (SE = .37), p < .004. Furthermore, these individuals demonstrated significant improvement in working span on a visuospatial task used to test transfer of training gains, from an average pre-training span of 5.86 (SE = .51) to an average post-training span of 6.86 (SE = .46), p < .018. These results suggest that the novel task is a strong candidate for use in an older adult population, and future research in the laboratory will test the efficacy of the task in a longer-term training study with older adults. The incorporation of further measures of transfer into pre- and post-training assessment will aid in determining the extent of transfer of training gains to more general cognitive function.

Cognitive depletion has a negative impact on the rate of implicit perceptual-motor sequence learning

Abigail H. Wesley, Daniel J. Sanchez, Paul J. Reber

Ego-depletion theory states that humans possess a limited store of cognitive resources that, when depleted, produce deficits in self-regulation or cognitive control. Depletion effects on implicit learning, which is not thought to require cognitive control, have not previously been reported. However, if depletion reflects transiently lower levels of dopamine, ego-depletion might be associated with slower learning for tasks dependent on dopamine-gated plasticity in cortico-striatal circuits. The relationship between ego-depletion and implicit learning was examined by comparing participants’ levels of cognitive depletion with sequence learning performance. Participants first completed the Stroop Task to assess depletion, measured as the reaction time difference between control and incongruent trials. Participants then performed the Serial Interception Sequence Learning (SISL) task. The SISL task is a perceptual-motor sequence learning task whereby circular cues scroll across a computer screen toward targets, and participants attempt to press the correct key when a cue fits within the target zone. Participants received 2880 trials of training on a covertly embedded 12-item second-order conditional sequence, followed by tests of both implicit and explicit sequence knowledge. Implicit sequence knowledge was assessed as the percent correct difference between performance on the trained sequence and novel sequences. A negative correlation was found between the interference effect and the amount of implicit learning exhibited, with a slightly stronger relationship observed for participants who did not demonstrate explicit knowledge of the sequence. These results show that ego depletion may lead to slower implicit learning, implying this process is not as automatic as previously hypothesized.

Gobel, E.W., Parrish, T.B., & Reber, P.J. (2011). NeuroImage.

Learning of complex motor skills requires learning of component movements as well as the sequential structure of their order and timing. Using a Serial Interception Sequence Learning (SISL) task, participants learned a sequence of precisely timed interception responses through training with a repeating sequence. Functional MRI data were collected during performance of the known sequence and compared with activity evoked during novel sequences of actions, novel timing patterns, or both. Reduced activity was observed during the practiced sequence in a distributed bilateral network including extrastriate occipital, parietal, and premotor cortical regions. These reductions in evoked activity likely reflect improved efficiency in visuospatial processing, spatio-motor integration, motor planning, and motor execution for the trained sequence, which is likely supported by nondeclarative skill learning. In addition, the practiced sequence evoked increased activity in the left ventral striatum and medial prefrontal cortex, while the posterior cingulate was more active during periods of better performance. Many prior studies of perceptual-motor skill learning have found increased activity in motor areas of frontal cortex (e.g., motor and premotor cortex, SMA) and striatal areas (e.g., the putamen). The change in activity observed here (i.e., decreased activity across a cortical network) may reflect skill learning that is predominantly expressed through more accurate performance rather than decreased reaction time.

Combining Physical Exercise and Repetition-Lag Training to Improve Everyday Memory Function in Older Adults

Gigler, K. L., Jennings, J. J., Dagenbach, D., Katula, J. A., Dove, R. W., & Stark, S.

Cognitive and physical activity interventions have been shown to improve older adults’ performance on lab-based measures of memory and executive function. However, less work has examined the real-world benefits of these two types of interventions, particularly when administered in conjunction with one another. The present study, which was part of the Seniors Health and Activity Research Program pilot trial, explored the independent and combined effects of repetition-lag memory training and aerobic exer- cise training on the frequency of self-reported memory errors in older adults. Following 4 months of training, a significant decrease in errors was observed for both the repetition-lag-only group and the combined exercise and repetition-lag group, with no benefit for the exercise-only participants. These results suggest that repetition-lag training may be more efficacious than physical exercise for reducing everyday memory errors and that cognitive and physical activity training may not produce
additive effects when administered together.

Enhancing motor memory for a melodic sequence by re-playing the melody during sleep

James W. Antony, Eric W. Gobel, Justin K. O’Hare, Paul J. Reber & Ken A. Paller

Northwestern University

A steadily increasing body of evidence supports a role for sleep in memory reactivation and consolidation. Memory traces are thought to be spontaneously reactivated during sleep, enhancing storage and improving subsequent memory performance. This natural process can apparently be triggered by auditory or olfactory stimuli during sleep, if those stimuli had previously been linked with learning (Rasch et al., 2007; Rudoy et al., 2009; Smith & Weeden, 1990). We now report that motor memories for a 12-item musical sequence, played using four fingers on four keys in time with moving visual cues (as in the video game, “Guitar Hero”), can be improved by presenting the corresponding tone sequences during sleep. Sixteen participants learned two repeating melodies composed with either four high tones or four low tones. Training involved 40 repetitions of each sequence in interleaved blocks. Performance was tested before and after a 90-minute afternoon nap. Responses were scored as correct only if made with the correct key and within the target time window. Performance on the learned sequences was superior to performance on novel sequences. Crucially, one musical sequence was played softly through a speaker 20 times during a 4-minute segment of slow-wave sleep, without the participants’ knowledge. After the nap, but not before, performance was significantly better on the cued sequence than on the non-cued sequence. These results demonstrate that motor memories can be selectively enhanced during sleep, most likely because reinstating the tone sequence reactivated corresponding representations of the visual cues and/or the learned motor sequence.

Consolidation in implicit sequence learning: Retroactive interference effects modulated by concomitant explicit knowledge

Paul J. Reber¹, Daniel J. Sanchez¹, David Fraser²;¹Northwestern University, ²Chatham University

Memory consolidation is a vital process by which knowledge representations become stable and long-lasting. While well-studied in declarative memory, implicit skill learning also appears to undergo a consolidation process after the training period. Immediately after learning a new motor sequence, knowledge may be vulnerable to retroactive interference from learning of another novel sequence. To assess whether retroactive interference is observed during implicit perceptual-motor sequence learning, participants learned three different 12-item second-order conditional sequences over two sessions using the Serial Interception Sequence Learning (SISL) task. In the SISL task, cues scroll vertically towards targets on a computer screen. Participants attempt to press a corresponding key when a cue reaches its target zone. During the first session participants trained on a first sequence (A), which was immediately followed by training on a second sequence (B). Either 24 or 48 hours later, participants returned for a second session and trained on a third sequence (C), followed by tests for implicit and explicit sequence knowledge for all three sequences. Retroactive interference for sequence A was observed in participants who had low explicit knowledge of sequence A, but not in participants with high explicit knowledge. No interference was observed for sequence B, for which there was a substantial delay period prior to learning sequence C. These results suggest that implicit sequence learning requires a consolidation period to avoid retroactive interference, but this interference effect can be ameliorated by concomitant explicit sequence knowledge.

Cognitive depletion has a negative impact on the rate of implicit perceptual-motor sequence learning

Abigail H. Wesley, Daniel J. Sanchez, Paul J. Reber

Ego-depletion theory states that humans possess a limited store of cognitive resources that, when depleted, produce deficits in self-regulation or cognitive control. Depletion effects on implicit learning, which is not thought to require cognitive control, have not previously been reported. However, if depletion reflects transiently lower levels of dopamine, ego-depletion might be associated with slower learning for tasks dependent on dopamine-gated plasticity in cortico-striatal circuits. The relationship between ego-depletion and implicit learning was examined by comparing participants’ levels of cognitive depletion with sequence learning performance. Participants first completed the Stroop Task to assess depletion, measured as the reaction time difference between control and incongruent trials. Participants then performed the Serial Interception Sequence Learning (SISL) task. The SISL task is a perceptual-motor sequence learning task whereby circular cues scroll across a computer screen toward targets, and participants attempt to press the correct key when a cue fits within the target zone. Participants received 2880 trials of training on a covertly embedded 12-item second-order conditional sequence, followed by tests of both implicit and explicit sequence knowledge. Implicit sequence knowledge was assessed as the percent correct difference between performance on the trained sequence and novel sequences. A negative correlation was found between the interference effect and the amount of implicit learning exhibited, with a slightly stronger relationship observed for participants who did not demonstrate explicit knowledge of the sequence. These results show that ego depletion may lead to slower implicit learning, implying this process is not as automatic as previously hypothesized.

Eric W. Gobel^1,2, Kelsey M. Blomeke², Sandra Weintraub^3,4, and Paul J. Reber^2
1Interdepartmental Neuroscience Program, ²Department of Psychology, ³Cognitive Neurology and Alzheimer’s Disease Center, and ⁴Departments of Psychiatry and Neurology
Northwestern University, Evanston, IL, and Northwestern Feinberg School Of Medicine, Chicago, IL

Cognitive Neuroscience Society Annual Meeting 2011

Skilled performance of complex motor skills requires learning a specific order of movements with precise timing. The Serial Interception Sequence Learning (SISL) task has been used to study this type of perceptual-motor sequence learning. During SISL, participants make keypress responses to coincide with the passage of a moving spatial cue though a target zone. Participants are not told that the cues follow a repeating sequence, but implicit sequence knowledge is observed through sequence-specific performance enhancement. Implicit learning with little concomitant explicit knowledge of the sequence has been demonstrated in undergraduate populations. This implicit-explicit memory dissociation is likely enhanced by the continuous performance demands and video-game-like interface. However, it is not known if SISL will be an effective tool for examining sequence learning in older adults given the overall task difficulty. Cognitively healthy elderly participants (mean age 70.0 years, range 63 – 76) performed a modified SISL task designed to provide an appropriate level of difficulty. The number of possible responses was reduced to three (making the task unimanual) and an initial performance pre-assessment identified an appropriate cue velocity for each participant. As in prior research, cues followed a 12-item repeating sequence during 80% of training trials. On a subsequent test phase, participants exhibited implicit sequence knowledge but no ability to recognize or recall the trained sequence. This cognitively healthy elderly sample demonstrated reliable implicit learning and would serve as a suitable comparison group for patient populations with impaired memory function, such as those with MCI, Alzheimer’s disease, and Parkinson’s disease.

Who reads the instructions, anyway? Explicit knowledge benefits perceptual-motor sequence learning independent of instruction.

Sanchez, D.J., Wesley, A.H., & Reber, P.J.

Motor skill training, outside of the laboratory, is typically accompanied by explicit instruction for the sequence of motor movements to be learned. However, expert performance is often unaccompanied by awareness of the sequence being executed, suggesting a role for implicit learning. Perceptual-motor sequence learning was assessed with, and without, explicit pre-instruction in order to examine the effect of explicit instruction on skill learning. Participants in the explicit instruction condition observed the 12-item training sequence five times before training, whereas in the implicit condition no mention was made of the repeating sequence. Both groups received 2880 trials of training with the Serial Interception Sequence Learning (SISL) task, followed by tests of implicit and explicit sequence knowledge. In the SISL task, cues scroll vertically towards targets on a computer screen and participants attempt to press a corresponding key when a cue reaches its target zone. Explicit instruction did not reliably affect sequence learning overall, but did lead to high levels of explicit knowledge in many, but not all, of these participants. Some participants in the implicit training condition acquired significant explicit knowledge as well. Participants were divided post-hoc into high and low explicit knowledge groups by the median of the sequence recognition test scores. Participants with greater explicit knowledge exhibited better learning of the embedded repeating sequence than those with only implicit knowledge, after an initial period of similar performance. This suggests that even with a continuous performance task requiring precisely timed responses, explicit knowledge provides a benefit to procedural learning. (249)

Gobel, E.W., Sanchez, D.J., & Reber, P.J. (2011). Journal of Experimental Psychology: Learning, Memory, and Cognition.

The expression of expert motor skills typically involves learning to perform a precisely timed sequence of movements (e.g., language production, music performance, athletic skills).  Research examining incidental sequence learning has previously relied on a perceptually-cued task that gives participants exposure to repeating motor sequences but does not require timing of responses for accuracy.  Using a novel perceptual-motor sequence learning task, learning a precisely timed cued sequence of motor actions is shown to occur without explicit instruction.  Participants learned a repeating sequence through practice and showed sequence-specific knowledge via a performance decrement when switched to an unfamiliar sequence.  In a second experiment, the integration of representation of action order and timing sequence knowledge was examined.  When either action order or timing sequence information was selectively disrupted, performance was reduced to levels similar to completely novel sequences. Unlike prior sequence-learning research that has found timing information to be secondary to learning action sequences, when the task demands require accurate action and timing information an integrated representation of these types of information is acquired.  These results provide the first evidence for incidental learning of fully integrated action and timing sequence information in the absence of an independent representation of action order, and suggest that this integrative mechanism may play a material role in the acquisition of complex motor skills.

Eric W. Gobel, Paul J. Reber; Northwestern University
Society for Neuroscience Annual Meeting 2009, Chicago, IL. 21 Oct 2009.

Most well practiced perceptual-motor skills that we execute are composed of a learned sequence of movements that must be performed in a particular order with specific timing. Experiments described here support the idea that order and timing are interdependent and integrated in perceptual-motor sequence learning. Furthermore, there are distinct regions of the brain that process the separate order and timing components and their integration.

To study the neural underpinnings of order and timing in implicit perceptual-motor sequence learning, a new paradigm was developed emphasizing precise timing in the response to a moving visual cue and allowing for timing information to be embedded in the response sequence. In this paradigm, circular cues move upwards on the screen toward four target regions. Participants are instructed to press a corresponding key when a cue is centered within the target region. In a behavioral experiment, participants learned a repeating 12-item sequence of responses with embedded timing information: responses were separated by either 350ms or 700ms in a repeating pattern. Overall performance on the task improved over time, as measured by decreasing error rates. When a novel pseudo-random series of responses was required instead of the repeating sequence, the error rate increased significantly, showing that participants had acquired implicit knowledge of the specific practiced sequence.

After learning, participants performed four transfer conditions, which maintained either the ordinal sequence, the timing sequence, both sequences, or neither sequence. Participants exhibited high levels of performance only when the test sequence was identical to the training conditions, maintaining both the order and inter-item timing. When either the order of actions or the timing intervals between actions was changed, participants’ performance dropped to a level equivalent to that of a completely novel sequence. These results demonstrate that timing information is tightly integrated with order information in a learned perceptual-motor sequence.
Functional imaging of the transfer conditions was performed to identify the brain circuits mediating performance of the order and timing integration components of the practiced sequence. Results suggest that cerebellar and anterior precuneus activity is correlated with executing an integrated order and timing sequence, while posterior precuneus and anterior cingulate activity is correlated with executing a practiced order of movements, independent of timing.

Sanchez, D.J., Gobel, E.W., Reber, P.J. CNADC 16th Annual Alzheimer Day. May 6th, 2010.

The majority of actions we take to accomplish everyday activities are based around routine behaviors. The procedural learning involved in developing a routine behavior, or skill, relies on an implicit memory system preserved in patients with Alzheimer’s disease. Rooted in the basal ganglia and connected cortico-striatal circuits, this preserved learning may support cognitive rehabilitation that utilizes abilities that AD patients retain. It is therefore critical to understand the capacity and operating characteristics of this type of memory. Procedural learning can be isolated from other types of memory in a perceptual-motor sequence learning task, such as the serial interception sequence learning (SISL) task. Based on the well-studied serial reaction time (SRT) task, the SISL task requires participants to make continuous sequences of keypress responses to cues scrolling vertically up a computer screen. Participants are instructed to time their response to coincide with a cue reaching a stationary target on the opposite end of the screen. Participants are not told that the cues appear in a repeating sequence (typically 12 items long), but they do show gradual improvements in performance that reflect learning of the repeating sequence.

Three experiments are reported that use the SISL task to examine procedural learning dependent on cortico-striatal circuits. In the first experiment, participants learned to perform the SISL task more accurately after practice with a repeating sequence but exhibited little or no explicit, conscious memory of the repeating sequence. This result shows that learning in this task is independent of the explicit memory system that is impaired in Alzheimer’s disease. In a second experiment, participants learned a repeating sequence with an embedded timing structure. When presented with the same sequence of motor responses but novel inter-item timing, performance was as poor as with a novel sequence. The lack of transfer indicates that this type of procedural learning integrates both action selection and timing. In a third experiment, participants exhibited significant learning of much longer sequences than have previously been studied; up to 80 items long. The ability to learn such long sequences indicates that the cortico-striatal learning system has a greater learning capacity than has previously been appreciated. By furthering our understanding of the function and capacities of this learning system, future work may be able to develop a cognitive rehabilitation approach that builds on this preserved memory function and improves general cognitive function in memory-disordered patients.

Sanchez, D.J. & Reber, P.J. Cognitive Neuroscience Society Meeting 2010.

Perceptual-motor sequence learning has an extensive research history, but most studies have relied on using short sequences of five to twelve items. Utilizing a new Serial Interception Sequence Learning (SISL) task, the implicit learning of longer motor sequences was examined to explore the learning rate and limit of information complexity that can be learned in short training sessions. In the SISL task, cues scroll vertically towards targets on a computer screen. Participants attempt to press a corresponding key when a cue reaches its target zone. Participants were not told that the cues followed a repeating sequence. Twenty percent of the training trials were in a random order to obscure the repetitions. To maintain task difficulty over practice, the velocity of the cues increased as task performance improved. Implicit knowledge was assessed by comparing performance on the SISL task for the trained sequence against novel sequences. Training began with 30-item sequences and subsequent groups received successively longer sequence lengths. After the implicit test, participants were informed about the repeating sequence and performed recognition and verbal recall tests. For recognition, participants rated five sequences as to how likely it was that each sequence was the one that had been practiced. Even with extended-length sequences, participants exhibited sequence-specific learning within an hour of training. The steep decline in recognition performance for sequences exceeding 30 items suggests that implicit memory systems are primarily driving the perceptual-motor sequence learning. These results provide broader insight into the capabilities of the brain regions responsible for implicit learning.

Sanchez, D.J, Gobel, E.W., & Reber, P.J. (2010). Psychonomic Bulletin & Review.

Memory-impaired patients express intact implicit perceptual-motor sequence learning, but it has been difficult to obtain a similarly clear dissociation in healthy participants. When explicit memory is intact, participants acquire some explicit knowledge and performance improvements from implicit learning may be subtle. Therefore, it is difficult to determine whether performance exceeds what could be expected based on the concomitant explicit knowledge. Using a challenging new sequence learning task, robust implicit learning is found in healthy participants with virtually no associated explicit knowledge. Participants trained on a repeating sequence randomly selected from a set of five.  On a performance test of all five sequences, performance was best on the trained sequence and two-thirds of the participants exhibited individually reliable improvement (by χ²). Participants could not reliably indicate which sequence had been trained by either recognition or recall. Only by expressing their knowledge via performance were participants able to indicate which sequence they had learned.

Sanchez, D.J. & Reber, P.J. Society for Neuroscience 2009.

The sequence of motor operations involved in performing a procedural task is often difficult to describe verbally even when it can be easily performed after training. Previous research using the serial reaction time task has shown group level dissociations between implicit and explicit perceptual-motor sequence knowledge in amnesic patients and young, healthy participants. Using a new perceptual-motor sequence learning task, we examined the dissociation between performance and explicit sequence knowledge in individual participants.

The task consists of circular cues scrolling vertically towards four ring-shaped targets placed at equidistant horizontal locations. Participants were instructed to press a corresponding key when the cue reached its matching target and were not told that the cues followed a repeating sequence. Each participant was assigned to one of five 12-item sequences. The intervals between cues were gradually shortened across training to maintain a challenging level of difficulty and 20% of training cues were random to minimize explicit knowledge. Test consisted of the five sequences and learning was assessed by comparing error rates from the learned sequence against the novel sequences. Participants were then asked to verbally report any sequence information they may have noticed and completed a recognition test. For the recognition test, the five sequences were shown and participants rated how likely it was that each sequence had been the one practiced.

An analysis of error rates across test indicated that the learned sequence was the sequence with the fewest total errors in 21 of 30 participants. For 20 of these participants, the error rate on the practiced sequence was significantly lower than the four novel sequences at an individual participant level. On the recognition task, only four of 30 participants (13%) gave their practiced sequence the highest rating of the five sequences. In 70% of participants, observation of their motor performance indicated which of the five sequences had been trained while verbal report and recognition of the sequences was at chance across the group.

Practice with a repeating perceptual-motor sequence in this task is shown to produce an implicit representation that identifies which particular sequence was learned in an individual participant even when that participant can neither report nor explicitly recognize the same information. This finding reinforces the encapsulation between implicit and explicit sequence knowledge and further suggests that perceptual-motor sequence information can be encoded and retrieved from the motor system without the awareness of the participant.