Cortical population activity within a preserved neural manifold underlies multiple motor behaviors

Gallego, Juan Álvaro; Perich, Matthew G.; Naufel, Stephanie; Éthier, Christian; Solla, Sara A.; Miller, Lee E.

doi:10.1038/s41467-018-06560-z

Cited by 253 publications

(300 citation statements)

References 72 publications

Supporting

Mentioning

254

Contrasting

Order By: Relevance

“…Our findings are in line with previous work showing that activity in a large population of neurons is confined to a lower dimensional manifold [Mante et al, 2013, Chaisangmongkon et al, 2017, Remington et al, 2018, Gallego et al, 2018. We found striatal and prefrontal responses encoded the sequence task on a low-dimensional manifold.…”

Section: Discussionsupporting

confidence: 93%

Learning to select actions shapes recurrent dynamics in the corticostriatal system

Márton

Schultz

Averbeck

2019

Preprint

View full text Add to dashboard Cite

Learning to select appropriate actions based on their values is fundamental to adaptive behavior. This form of learning is supported by fronto-striatal systems. The dorsal-lateral prefrontal cortex (dlPFC) and the dorsal striatum (dSTR), which are strongly interconnected, are key nodes in this circuitry. Substantial experimental evidence, including neurophysiological recordings, have shown that neurons in these structures represent key aspects of learning. The computational mechanisms that shape the neurophysiological responses, however, are not clear. To examine this, we developed a recurrent neural network (RNN) model of the dlPFC-dSTR circuit and trained it on an oculomotor sequence learning task. We compared the activity generated by the model to activity recorded from monkey dlPFC and dSTR in the same task. This network consisted of a striatal component which encoded action values, and a prefrontal component which selected appropriate actions. After training, this system was able to autonomously represent and update action values and select actions, thus being able to closely approximate the representational structure in corticostriatal recordings. We found that learning to select the correct actions drove action-sequence representations further apart in activity space, both in the model and in the neural data. The model revealed that learning proceeds by increasing the distance between sequence-specific representations. This makes it more likely that the model will select the appropriate action sequence as learning develops. Our model thus supports the hypothesis that learning in networks drives the neural representations of actions further apart, increasing the probability that the network generates correct actions as learning proceeds. Altogether, this study advances our understanding of how neural circuit dynamics are involved in neural computation, showing how dynamics in the corticostriatal system support task learning.

show abstract

Section: Discussionsupporting

confidence: 93%

Learning to select actions shapes recurrent dynamics in the corticostriatal system

Márton

Schultz

Averbeck

2019

Preprint

View full text Add to dashboard Cite

show abstract

“…We determined the muscle synergies with PCA as we were interested in analyzing the computational implications of motor modularity without regard to the actual physiological implementation (e.g., in the spinal circuitry [15] or in cortex [21]), which is unknown. We observed a decrease in performance when performing the experiments with NNMF, using the solver provided in Kim et al [29].…”

Section: Discussionmentioning

confidence: 99%

The Effects of Motor Modularity on Performance, Learning, and Generalizability in Upper-Extremity Reaching: a Computational Analysis

Borno

Hicks

Delp

2019

Preprint

View full text Add to dashboard Cite

It has been hypothesized that the central nervous system simplifies the production of movement by limiting motor commands to a small set of modules known as muscle synergies. Recently, investigators have questioned whether a low-dimensional controller can produce the rich and flexible behaviors seen in everyday movements. We implemented * Corresponding author: malborno@stanford.edu controllers could also accomplish new tasks-such as reaching to targets on a higher or lower plane, starting from an alternate initial pose, moving at a faster velocity, and holding a light weight-with average errors similar to a full-dimensional controller. We also found that for new tasks, the synergybased controllers converged to good solutions (as defined by a cost-function threshold) with lower computational cost than a full-dimensional controller.Our results show that a modular architecture based on muscle synergies can generate a wide variety of reaching behaviors for a high-dimensional musculoskeletal system without a significant degradation in performance.

show abstract

“…Next, we examined the relationship between the texture and speed representations by assessing the extent to which changes in one interfered with the other. To this end, we computed the proportion of speed-related variance that was captured by each dimension of the texture representation ( Figure 2C), a quantity we refer to as the alignment index (Elsayed et al, 2016;Gallego et al, 2018)(see Methods). We found that speed-driven changes were primarily captured by the first principal component of the texture representation and this relationship was far stronger in the periphery than in cortex (average alignment index, peripheral: 0.80, cortical: 0.46).…”

Section: Population Representations Of Texture and Speedmentioning

confidence: 99%

“…To determine the amount of overlap between the texture space and the speed dimension, we calculated an alignment index (Elsayed et al, 2016;Gallego et al, 2018), which we defined as the amount of speeddriven variance captured by the texture space, normalized by the total amount of speed-related variance in the population response. Specifically, we define the alignment index as:…”

Section: Neural Population Analysesmentioning

confidence: 99%

Emergence of an invariant representation of texture in primate somatosensory cortex

Lieber

Bensmaı̈a

2019

Preprint

View full text Add to dashboard Cite

A major function of sensory processing is to achieve neural representations of objects that are stable across changes in context and perspective. Small changes in exploratory behavior can lead to large changes in signals at the sensory periphery, thus resulting in ambiguous neural representations of objects. Overcoming this initial ambiguity is a hallmark of human object recognition across sensory modalities. Here, we investigate the perception of tactile texture, which is stable across a wide range of exploratory movements of the hand, including changes in scanning speed, despite the strong dependence of the peripheral response on hand movements. To probe the neural basis of speedtolerant texture representations, we scanned a wide range of everyday textures across the fingertips of Rhesus macaques at multiple speeds and recorded the responses evoked in tactile nerve fibers and somatosensory cortical neurons (in Brodmann's areas 3b, 1, and 2). We found that, unlike afferents, individual cortical neurons exhibit a wide range of speed-sensitivities: some neurons are more strongly driven by changes in speed than peripheral afferents, while others exhibit responses that are nearly speed-independent. As a result, compared to the periphery, the cortical population exhibits 1) an increased ability to simultaneously encode independent representations of speed and texture, and 2) an increased ability to account for the speed-tolerant perception of texture in human subjects. Finally, we demonstrate that this separation of speed and texture information is a natural consequence of previously described cortical computations.

show abstract

Cortical population activity within a preserved neural manifold underlies multiple motor behaviors

Cited by 253 publications

References 72 publications

Learning to select actions shapes recurrent dynamics in the corticostriatal system

Learning to select actions shapes recurrent dynamics in the corticostriatal system

The Effects of Motor Modularity on Performance, Learning, and Generalizability in Upper-Extremity Reaching: a Computational Analysis

Emergence of an invariant representation of texture in primate somatosensory cortex

Contact Info

Product

Resources

About