Search

Disturbance forces facilitate motor learning, but theoretical explanations for this counterintuitive phenomenon are lacking. Smooth arm movements require predictions (inference) about the force-field associated with a workspace. The Free Energy Principle (FEP) suggests that such 'active inference' is driven by 'surprise'. We used these insights to create a formal model that explains why disturbance might help learning. In two experiments, participants undertook a continuous tracking task where they learned how to move their arm in different directions through a novel 3D force field. We compared baseline performance before and after exposure to the novel field to quantify learning. In Experiment 1, the exposure phases (but not the baseline measures) were delivered under three different conditions: (i) robot haptic assistance; (ii) no guidance; (iii) robot haptic disturbance. The disturbance group showed the best learning as our model predicted. Experiment 2 further tested our FEP inspired model. Assistive and/or disturbance forces were applied as a function of performance (low surprise), and compared to a random error manipulation (high surprise). The random group showed the most improvement as predicted by the model. Thus, motor learning can be conceptualised as a process of entropy reduction. Short term motor strategies (e.g. global impedance) can mitigate unexpected perturbations, but continuous movements require active inference about external force-fields in order to create accurate internal models of the external world (motor learning). Our findings reconcile research on the relationship between noise, variability, and motor learning, and show that information is the currency of motor learning., Competing Interests: The authors have declared that no competing interests exist.

Previously we developed a method that supports active movement generation to allow practice with improvement of good compliance control in tracing and drawing. We showed that the method allowed children with motor impairments to improve at a 3D tracing task to become as proficient as typically developing children and that the training improved 2D figure copying. In this study, we expanded the training protocol to include a wider variety of ages (5-10-year-olds) and we made the figures traced in training the same as in figure copying, but varied the scale of training and copying figures to assess the generality of learning. Forty-eight children were assigned to groups trained using large or small figures. All were tested before training with a tracing task and a copying task. Then, the children trained over five sessions in the tracing task with either small or large figures. Finally, the tracing and copying tasks were tested again following training. A mean speed measure was used to control for path length variations in the timed task. Performance on both tasks at both baseline and posttest varied as a function of the size of the figure and age. In addition, tracing performance also varied with the level of support. In particular, speeds were higher with more support, larger figures and older children. After training, performance improved. Speeds increased. In tracing, performance improved more for large figures traced by children who trained on large figures. In copying, however, performance only improved significantly for children who had trained on small figures and it improved equally for large and small figures. In conclusion, training by tracing smaller figures yielded better learning that was not, however, specific to the scale of drawn figures. Small figures exhibit greater mean curvature. We infer that it yielded better general improvement.

The role of the monocular-flow-based optical variable τ in the perception of the time to contact of approaching objects has been well-studied. There are additional contributions from binocular sources of information, such as changes in disparity over time (CDOT), but these are less understood. We conducted an experiment to determine whether an object's velocity affects which source is most effective for perceiving time to contact. We presented participants with stimuli that simulated two approaching squares. During approach the squares disappeared, and participants indicated which square would have contacted them first. Approach was specified by (a) only disparity-based information, (b) only monocular flow, or (c) all sources of information in normal viewing conditions. As expected, participants were more accurate at judging fast objects when only monocular flow was available than when only CDOT was. In contrast, participants were more accurate judging slow objects with only CDOT than with only monocular flow. For both ranges of velocity, the condition with both information sources yielded performance equivalent to the better of the single-source conditions. These results show that different sources of motion information are used to perceive time to contact and play different roles in allowing for stable perception across a variety of conditions.

Zannoli, Cass, Alais, and Mamassian (2012) found greater audiovisual lag between a tone and disparity-defined stimuli moving laterally (90-170 ms) than for disparity-defined stimuli moving in depth or luminance-defined stimuli moving laterally or in depth (50-60 ms). We tested if this increased lag presents an impediment to visually guided coordination with laterally moving objects. Participants used a joystick to move a virtual object in several constant relative phases with a laterally oscillating stimulus. Both the participant-controlled object and the target object were presented using a disparity-defined display that yielded information through changes in disparity over time (CDOT) or using a luminance-defined display that additionally provided information through monocular motion and interocular velocity differences (IOVD). Performance was comparable for both disparity-defined and luminance-defined displays in all relative phases. This suggests that, despite lag, perception of lateral motion through CDOT is generally sufficient to guide coordinated motor behavior.

The kinematics of upper limb movements and the coordination of eye and hand movements are affected by ageing. These age differences are exacerbated when task difficulty is increased, but the exact nature of these differences remains to be established. We examined the performance of 12 older adults (mean age = 74) and 11 younger adults (mean age = 20) on a multi-phase prehension task. Participants had to reach for a target ball with their preferred hand, pick it up and place it in a tray, then reach for a second target ball and place that in the same tray. On half the trials (stabilising condition), participants were required to hold the tray just above the surface of the table with their non-preferred hand and keep it as still as possible. Hand and eye movements were recorded. Older adults took longer to complete their movements and reached lower peak velocities than the younger adults. Group differences were most apparent in the stabilising condition, suggesting that the added complexity had a greater effect on the performance of the older adults than the young. During pickup, older adults preferred to make an eye movement to the next target as soon as possible, but spent longer fixating the current target during placement, when accuracy requirements were higher. These latter observations suggest that older adults employed a task-dependent eye movement strategy, looking quickly to the next target to allow more time for planning and execution when possible, but fixating on their hand and successful placement of the ball when necessary.

Many children have difficulty producing movements well enough to improve in perceptuo-motor learning. We have developed a training method that supports active movement generation to allow improvement in a 3D tracing task requiring good compliance control. We previously tested 7-8 year old children who exhibited poor performance and performance differences before training. After training, performance was significantly improved and performance differences were eliminated. According to the Dynamic Systems Theory of development, appropriate support can enable younger children to acquire the ability to perform like older children. In the present study, we compared 7-8 and 10-12 year old school children and predicted that younger children would show reduced performance that was nonetheless amenable to training. Indeed, the pre-training performance of the 7-8 year olds was worse than that of the 10-12 year olds, but post-training performance was equally good for both groups. This was similar to previous results found using this training method for children with DCD and age-matched typically developing children. We also found in a previous study of 7-8 year old school children that training in the 3D tracing task transferred to a 2D drawing task. We now found similar transfer for the 10-12 year olds., (Copyright © 2015 Elsevier B.V. All rights reserved.)

Locomoting-to-reach to a target is a common visuomotor approach behavior that consists of two nested component actions: locomotion and reaching. The information and control strategies that guide locomotion and reaching in isolation are well studied, but their interaction during locomoting-to-reach behavior has received little attention. We investigated the role of proportional rate control in unifying these components into one action. Individuals use this control strategy with hand-centric disparity-based τ information to guide seated reaching (Anderson and Bingham in Exp Brain Res 205:291-306. doi: 10.1007/s00221-010-2361-9 , 2010) and use it with sequential information to perform targeted locomotion to bring an outstretched arm and hand to a target; first with eye-centric τ information and then hand-centric τ information near the target (Anderson and Bingham in Exp Brain Res 214:631-644. doi: 10.1007/s00221-011-2865-y , 2011). In the current study, participants performed two tasks: locomoting to bring a rigidly outstretched arm and hand to a target (handout), and locomoting to initiate and guide a reach to a target (locomoting-to-reach). Movement trajectories were analyzed. Results show that participants used proportional rate control throughout both tasks, in the sequential manner that was found by Anderson and Bingham (Exp Brain Res 214:631-644. doi: 10.1007/s00221-011-2865-y , 2011). Individual differences were found in the moment at which this information switch occurred in the locomoting-to-reach task. Some participants appeared to switch to proportional rate control with hand-τ once the hand came into view and others switched once the reaching component was complete and the arm was fully outstretched. In the locomoting-to-reach task, participants consistently initiated reaches when eye-τ specified a time-to-contact of 1.0 s. Proportional rate control provides a solution to the degrees-of-freedom problem in the classic manner, by making multiple things one.

Mon-Williams and Bingham (2011) developed an affordance model of the spatial structure of reaches-to-grasp. With a single free parameter (P), the model predicted the safety margins (SMs) exhibited in maximum grasp apertures (MGAs), during the approach of a hand to a target object, as a function of an affordance measure of object size and a functional measure of hand size. An affordance analysis revealed that object size is determined by a diagonal through the object, called the maximum object extent. Mon-Williams and Bingham provided no theoretical account for the empirically determined values of P. We now address this question. Snapp-Childs and Bingham (2009) augmented Warren's (1984) geometric affordance scaling model with a dynamical component determined by the stability of the motor performance. Because P was found to vary with the speeds of reaches, we incorporated a measure of the variability of performance into the model to yield predictions of P. We also found that P varied with gender. In respect to the size of safety margins, women were more conservative in taking risks then men. Finally, following Warren (1984), the classic paradigm for testing affordance models is to test the scaling relations with both small and large participants. We tested small- and large-handed men and small- and large-handed women and found that the new parameter free model successfully accounted for the spatial structure of reaches-to-grasp.

We recently found that older adults show reduced learning rates when learning a new pattern of coordinated rhythmic movement. The purpose of this study was to extend that finding by examining the performance of all ages across the lifespan from the 20 s through to the 80 s to determine how learning rates change with age. We tested whether adults could learn to produce a novel coordinated rhythmic movement (90° relative phase) in a visually guided unimanual task. We determined learning rates to quantify changes in learning with age and to determine at what ages the changes occur. We found, as before, that learning rates of participants in their 70 s and 80 s were half those of participants in their 20 s. We also found a gradual slow decline in learning rate with age until approximately age 50, when there was a sudden drop to a reduced learning rate for the 60 though 80 year olds. We discuss possible causes for the "50 s cliff" in perceptuo-motor learning rates and suggest that age related deficits in perception of complex motions may be the key to understanding this result.

The role of static eyeheight-scaled information in perceiving the passability of and guiding locomotion through apertures is well established. However, eyeheight-scaled information is not the only source of visual information about size and passability. In this study we tested the sufficiency of two other sources of information, both of which are available only to moving observers (ie are dynamic) and specify aperture size in intrinsic body-scaled units. The experiment was conducted in an immersive virtual environment that was monocularly viewed through a head-mounted display. Subjects walked through narrow openings between obstacles, rotating their shoulders as necessary, while head and shoulder position were tracked. The task was performed in three virtual environments that differed in terms of the availability of eyeheight-scaled information and the two dynamic sources of information. Analyses focused on the timing and amplitude of shoulder rotation as subjects walked through apertures, as well as walking speed and the number of collisions. Subjects successfully timed and appropriately scaled the amplitude of shoulder rotation to fit through apertures in all three conditions. These findings suggest that visual information other than eyeheight-scaled information can be used to guide locomotion through apertures.