Retention and interference of learned dexterous manipulation: interaction between multiple sensorimotor processes

Abstract

Fu Q, Santello M. Retention and interference of learned dexterous manipulation: interaction between multiple sensorimotor processes. J Neurophysiol 113: 144 -155, 2015. First published October 1, 2014; doi:10.1152/jn.00348.2014.-An object can be used in multiple contexts, each requiring different hand actions. How the central nervous system builds and maintains memory of such dexterous manipulations remains unclear. We conducted experiments in which human subjects had to learn and recall manipulations performed in two contexts, A and B. Both contexts involved lifting the same L-shaped object whose geometry cued its asymmetrical mass distribution. Correct performance required producing a torque on the vertical handle at object lift onset to prevent it from tilting. The torque direction depended on the context, i.e., object orientation, which was changed by 180°object rotation about a vertical axis. With an A 1 B 1 A 2 context switching paradigm, subjects learned A 1 in the first block of eight trials as indicated by a torque approaching the required one. However, subjects made large errors in anticipating the required torque when switching to B 1 immediately after A 1 (negative transfer), as well as when they had to recall A 1 when switching to A 2 after learning B through another block of eight lifts (retrieval interference). Classic sensorimotor learning theories attribute such interferences to multi-rate, multistate error-driven updates of internal models. However, by systematically changing the interblock break duration and within-block number of trials, our results suggest an alternative explanation underlying interference and retention of dexterous manipulation. Specifically, we identified and quantified through a novel computational model the nonlinear interaction between two sensorimotor mechanisms: a shortlived, context-independent, use-dependent sensorimotor memory and a context-sensitive, error-based learning process. sensorimotor; manipulation; interference; retention DEXTEROUS MANIPULATION is thought to rely on building an internal representation of the task or object dynamics that can be updated through trial-by-trial learning to achieve a stable performance (Flanagan et al. 2001;Nowak et al. 2007;Salimi et al. 2000). However, the underlying sensorimotor mechanisms are not well understood. The properties of the learned representation of a motor task can be evaluated by generalization or retrieval protocols in which subjects are asked to generalize a previously learned task to a new context or recall the learned task in the same context. Until recently, such protocols have been mostly applied to reaching tasks (for review, see , which has led to the establishment of several models of sensorimotor learning. For instance, error-based learning suggests that the central nervous system (CNS) updates the internal model of the task by using errors made in the previous trial(s) (Smith et al. 2006), whereas model-free learning is based on reinforcement and/or usedependent plasticity (Huang et al. 2011). Desp...

Credit assignment between body and object probed by an object transportation task

“…The partition of learning between the body and the object is consistent with recent advance in object-manipulation studies investigating sensorimotor memory, which was originally referred to as the memory of an object’s physical properties 27 . Researchers have found that memory of previous actions affects object manipulation 13 , 28 , 29 . For instance, after forcefully squeezing an unrelated object people was biased to use a large grip force to lift a familiar object 13 .…”

“…We postulate that object learning lasts longer as it is akin to the concept of object permanence 37 – 39 , while body learning changes more rapidly. Early studies on object grasping have found that people retain the knowledge of weight distribution for more than 24 hours 2 , 40 , while the influence of recent actions only lasts for a brief duration 28 . With accumulating evidence that motor learning consists of components of different time scales 41 – 43 , our findings suggest that during object manipulation body learning and object learning might be associated with a fast and a slow time scale, respectively.…”

On neuromechanical approaches for the study of biological and robotic grasp and manipulation

“…Later studies provided further evidence supporting the concept of multiple sensorimotor mechanisms and how their different time scales may interfere with generalization or retrieval of previously learned manipulation [230]-even when the object being manipulated is the same. A recent study has provided evidence for the co-existence of contextdependent and independent learning processes [204], which would operate similarly to those described for adaptation of reaching movements [231]. The advantage of context-dependent representations of manipulation is that they can be recalled when the object has strong contextual cues (i.e., object geometry and perhaps other perceptual attributes).…”

Modeling Previous Trial Effect in Human Manipulation through Iterative Learning Control

“…The fact that the generalized model had the best performance is consistent with the recent trend in neuroscience research, where motor adaptation can be attributed to a combination of parallel neural mechanisms which differ in time-space and sensorimotor memory usage. [5,37,38] In addition, our model can capture within-trial temporal evolution of human motor control during trial-by-trial adaptations, whereas most previous studies consider the motor output and error signals only as scalars for each trial.…”