Functional network reorganization during learning in a brain-computer interface paradigm

Abstract

Efforts to study the neural correlates of learning are hampered by the size of the network in which learning occurs. To understand the importance of learning-related changes in a network of neurons, it is necessary to understand how the network acts as a whole to generate behavior. Here we introduce a paradigm in which the output of a cortical network can be perturbed directly and the neural basis of the compensatory changes studied in detail. Using a brain-computer interface, dozens of simultaneously recorded neurons in the motor cortex of awake, behaving monkeys are used to control the movement of a cursor in a three-dimensional virtual-reality environment. This device creates a precise, well-defined mapping between the firing of the recorded neurons and an expressed behavior (cursor movement). In a series of experiments, we force the animal to relearn the association between neural firing and cursor movement in a subset of neurons and assess how the network changes to compensate. We find that changes in neural activity reflect not only an alteration of behavioral strategy but also the relative contributions of individual neurons to the population error signal.brain-machine interface ͉ neural prosthetics ͉ perturbation ͉ population vector algorithm A wealth of evidence associates learning with changes in neural activity (for reviews, see refs. 1-4). For example, the tuning functions of neurons in the motor cortex can change when monkeys adapt to perturbations that interfere with the execution (5-7) or visual feedback (8 -10) of their movements. Although the observed changes in neural activity in these studies are closely associated with the behavioral manifestation of learning, it is difficult to interpret the behavioral significance of the neural changes because the precise relationship between neural activity and behavioral output is unknown. However, recent developments in brain-computer interface technology now make it possible to control the activity of a cursor in a three-dimensional (3D) virtual environment using the spiking activity of ensembles of simultaneously recorded motor cortical units (11)(12)(13)(14)(15)(16). This ''braincontrol'' paradigm is unique in that the behavior, cursor movement, is solely the result of neural activity in the population under study. Thus, any mismatches between desired cursor motion and decoded cursor motion can only be corrected by altering the activity of these recorded neurons.The brain-computer interface also allows for a unique kind of perturbation that targets selective subsets of neurons. By altering the way that the firing activity of a subset of neurons is decoded and mapped into cursor movement, it is possible to test whether the tuning functions of the perturbed subset selectively change to compensate for the global error signal, or whether the entire population changes together.To study the response to selective perturbations of this neural network, monkeys were trained to perform center-out movements to eight equally spaced targets in a 3D virtual environm...

Neuronal Correlates of Memory Formation in Motor Cortex after Adaptation to Force Field

“…With VFB, tuning amplitude increased in CF cells, whereas without VFB, it increased in WF cells. The different responses between these two cell groups parallel the suggestion of a reweighting strategy (Jarosiewicz et al, 2008), which, in the case here, applies to "preferred" cell groups depending on the feedback context in which learning took place. Increasing the tuning amplitude increases the contribution of the cell to the population vector.…”

“…Acquisition of motor skills involves a process of initial encoding of new motor outputs in response to novel sensorimotor associations (Wise et al, 1998;Gandolfo et al, 2000;Li et al, 2001;Paz et al, 2003;Lebedev et al, 2005;Jarosiewicz et al, 2008;Zach et al, 2008;Arce et al, 2010). Unless disrupted, this may be followed by a process of consolidation in which labile motor memories may evolve into a more stable form, becoming less susceptible to interference and allowing for later retrieval (Stickgold and Walker, 2005;Nader and Hardt, 2009;Robertson, 2009).…”

Combined Adaptiveness of Specific Motor Cortical Ensembles Underlies Learning

“…Predicting the hand's displacement by the force field, the neuronal population specifies a new intended target direction such that with the force field, the hand reaches the LD. Thus, the rate modulation could reflect reaiming, as suggested for visuomotor rotation (Jarosiewicz et al, 2008). Indeed, the population directional signal in field reaches pointed away from the LD (Fig.…”

Mechanisms of Neurofeedback: A Computation-theoretic Approach

“…They focused on a study by Jarosiewicz et al (2008) in which monkeys were trained to control a cursor in a 3D virtual reality environment by modifying the activity in the motor cortex. Legenstein et al's computational model did not include a division between cortical and striatal areas, but consisted of two functional pools of neurons, where neurons in the first pool send input to neurons in the second pool.…”