Shigehiro Miyachi

Parallel neural networks for learning sequential procedures

Recent studies have shown that multiple brain areas contribute to different stages and aspects of procedural learning. On the basis of a series of studies using a sequence-learning task with trial-and-error, we propose a hypothetical scheme in which a sequential procedure is acquired independently by two cortical systems, one using spatial coordinates and the other using motor coordinates. They are active preferentially in the early and late stages of learning, respectively. Both of the two systems are supported by loop circuits formed with the basal ganglia and the cerebellum, the former for reward-based evaluation and the latter for processing of timing. The proposed neural architecture would operate in a flexible manner to acquire and execute multiple sequential procedures.

Differential roles of monkey striatum in learning of sequential hand movement.

Abstractu2002To study the role of the basal ganglia in learning of sequential movements, we trained two monkeys to perform a sequential button-press task (2×5 task). This task enabled us to examine the process of learning new sequences as well as the execution of well-learned sequences repeatedly. We injected muscimol (a GABA agonist) into different parts of the striatum to inactivate the local neural activity reversibly. The learning of new sequences became deficient after injections in the anterior caudate and putamen, but not the middle-posterior putamen. The execution of well-learned sequences was disrupted after injections in the middle-posterior putamen and, less severely, after injections in the anterior caudate/putamen. These results suggest that the anterior and posterior portions of the striatum participate in different aspects of learning of sequential movements.

Differential activation of monkey striatal neurons in the early and late stages of procedural learning

The basal ganglia is a key structure for procedural learning. To examine in what aspects of procedural learning the basal ganglia participate, we recorded from striatal neurons (phasically active neurons) in monkeys while the animals were performing a sequential button press task (the 2×5 task) and compared the neuronal activity between two conditions: (1) while learning new sequences and (2) while executing overlearned sequences. Among 147 neurons recorded in two monkeys, 45 neurons were activated preferentially for new sequences (new-preferring neurons), 34 for overlearned sequences (learned-preferring neurons), and 68 were activated non-selectively (non-selective neurons). New-preferring neurons were more abundant in the “association” region [association striatum (AS); caudate nucleus and rostral putamen anterior to the anterior commissure], while the learned-preferring neurons were more abundant in the “sensorimotor” region [sensorimotor striatum (SM); putamen posterior to the anterior commissure]. In addition to the learning dependency, the AS and SM neurons were activated in different task periods: many AS neurons were activated during the delay period, while the SM neurons were more activated with reaching and button presses. These data, together with the data from our previous blockade study, suggest that the “association” and “sensorimotor” regions of the basal ganglia contribute preferentially to the early and late stages of procedural learning, respectively.

Characteristics of a long-term procedural skill in the monkey

Abstractu2002The purpose of this study was to characterize the nature and structure of procedural memory. We have previously studied the process of learning sequential behavioral procedures using monkeys. The monkey’s task was to press five consecutive pairs of buttons (indicated by illumination) in the correct order for every pair, which he had to find by trial-and-error in a block of trials. The whole sequence was called a “hyperset”; each pair was called a “set”. We first examined whether monkeys learned to perform a hyperset as a single sequence or learned the order of button-presses individually for each set. To answer this question, we generated hypersets that were the same as the hypersets that had been extensively learned except that the order of the sets was reversed. The performance of these “reversed hypersets” was much worse than the performance of the original learned hypersets and was similar to the performance of new hypersets, as regards both the number of errors and the performance time. The result suggests that monkeys learned a hyperset as a sequence. To examine whether the learned performance was specific to the hand used for practice, we had monkeys use the same hand throughout the long-term practice of each hyperset, and then tested the opposite hand. The performance using the opposite hand was worse than the performance using the trained hand, but was better than the performance for new hypersets. This indicates that the memory for the sequential procedure is only partially accessible to the hand that was not used for the practice.

Long-term retention of motor skill in macaque monkeys and humans.

Abstract. Remarkable human performance, such as playing the violin, is often based on motor skills that, once acquired, are retained for a long time. To examine how motor skills are retained, we trained monkeys and humans extensively to perform many visuomotor sequences and examined their performance after a long retention period of up to 18 months. For both monkeys and humans, we found strong evidence for long-term retention of motor skills. Each of the monkey subjects initially learned 6–18 sequences of button presses extensively by trial-and-error for up to 18 months. After a long retention period, they were asked to perform the previously learned (OLD) sequences together with completely new (NEW) sequences. The performance for OLD sequences was much better than for NEW sequences in terms of accuracy (assessed by the number of errors to criterion) and speed (assessed by the performance time). However, the retention was interfered with in two conditions, but in selective manners: (1) Learning of other sequences during the retention period interfered with accuracy, but not speed, of performance; (2) Inter-manual transfer was absent for speed, but not accuracy, of performance. The human subjects performed basically the same task as the monkeys. Each subject initially learned one sequence of 20 button presses by trial-and-error during an 8–10xa0day learning session. After 16 months, they were asked to perform the previously learned sequence (OLD sequence) and additional sequences including RECENT sequences (learned one day before) and NEW sequences. Their performance was considerably better on OLD and RECENT sequences than NEW sequences. Whereas the number of errors (reflecting accuracy) was lower for RECENT than for OLD sequences, the performance time (reflecting speed) was shorter for OLD than for RECENT sequences. Interestingly, the subjects were unaware that they had experienced OLD sequences. The results suggest that a motor skill is acquired and retained in two different forms, accuracy and speed. This occurs separately but concurrently. This conclusion is consistent with the hypothesis that at least two neural mechanisms operate independently to represent a motor skill.

Characteristics of sequential movements during early learning period in monkeys.

Abstract. We previously demonstrated that the organization of a learned sequential movement, after long-term practice, is based on the entire sequence and that the information pertaining to the sequence is largely specific to the hand used for practice. However, it remained unknown whether these characteristics are present from the beginning of learning. To answer the question, we examined the performance of four monkeys for the same sequential procedure in the early stage of learning. The monkeys task was to press five consecutive pairs of buttons (which were illuminated), in a correct order for every pair, which they had to find by trial-and-error during a block of trials. We first examined whether the memory of a sequential procedure that was learned once was specific to the hand used for practice. The second time that the monkeys attempted to learn a novel sequence, they were required to use either the same hand they used the first time or the opposite hand. The number of errors decreased to a similar degree in the same-hand condition and in the opposite-hand condition. The performance time decreased in the same-hand condition, but not in the opposite-hand condition. The results suggest that, in the early stage of learning, memory of the correct performance of a sequential procedure is not specific to the hand originally used to perform the sequence (unlike the well-learned stage, where the transfer was incomplete), whereas memory of the fast performance of a sequential procedure is relatively specific to the hand used for practice (like the well-learned stage). We then examined whether memory of a sequential procedure depends on the entire sequence, not individual stimulus sets. For the second learning block, we had the monkey learn the sequence in the same or reversed order. In the reversed order, the order within each set was identical, but the order of sets was reversed. The number of errors decreased in both the same-order and reversed-order conditions to a similar degree for two out of four monkeys; the decrease was larger in the same-order condition for the other two monkeys. For all monkeys, the performance time decreased in the same-order condition, but not in the reversed-order condition. The results suggest that the memory structure for correct performance varies among monkeys in the early stage of learning (unlike the well-learned stage, where the memory of individual sets was consistently absent). On the other hand, memory of the fast performance of a sequential procedure is relatively specific to the learned order used for practice (like the well-learned stage).

Corticostriatal projections from distal and proximal forelimb representations of the monkey primary motor cortex

Corticostriatal projections from one distal and two proximal subregions in the forelimb representation of the primary motor cortex (MI) were examined in the macaque monkey. The distal and proximal subregions in the anterior bank of the central sulcus (distal and proximal-bank subregions) and the proximal subregion in the surface of the precentral gyrus (proximal-surface subregion) of the MI were identified using intracortical microstimulation. Different anterograde tracers were then injected into two of these three forelimb subregions of the MI. In the ipsilateral putamen, the distribution areas of corticostriatal fibers from the distal, proximal-bank and proximal-surface subregions were arranged from ventrolateral to dorsomedial in this order. These corticostriatal input zones were largely segregated from one another.

Procedural Learning in the Monkey

We devised a behavioral paradigm (sequential button press task) for monkeys in order to test the hypothesis that the basal ganglia are crucial for procedural learning. Upon pressing of a home key, two of 16 (4×4) LED buttons (called ‘set’) were illuminated and the monkey had to press them in a predetermined order which he had to find by trial and error. A total of 5 sets (called ‘hyperset’) were presented in a fixed order for completion of a trial.