Daniel Ferris @Ferris_vball@qoto.org

Author summary In this study, we aimed to better understand motor learning using a neuro-computational approach. While previous work emphasized different learning regimes of the various components, the main novelty of our study is the interplay of its components. Notably, we show that the model accounts well for motor adaptation data of experiments that involve a cognitive strategy.

Using EEG and local field potentials (LFPs) as an index of large-scale neural activities, research has been able to associate neural oscillations in different frequency bands with markers of cognitive functions, goal-directed behavior, and various neurological disorders. While this gives us a glimpse into how neurons communicate throughout the brain, the causality of these synchronized network activities remains poorly understood. Moreover, the effect of the major neuromodulatory systems (e.g., noradrenergic, cholinergic, and dopaminergic) on brain oscillations has drawn much attention. More recent studies have suggested that cross-frequency coupling (CFC) is heavily responsible for mediating network-wide communication across subcortical and cortical brain structures, implicating the importance of neurotransmitters in shaping coordinated actions. By bringing to light the role each neuromodulatory system plays in regulating brain-wide neural oscillations, we hope to paint a clearer picture of the pivotal role neural oscillations play in a variety of cognitive functions and neurological disorders, and how neuromodulation techniques can be optimized as a means of controlling neural network dynamics. The aim of this review is to showcase the important role that neuromodulatory systems play in large-scale neural network dynamics, informing future studies to pay close attention to their involvement in specific features of neural oscillations and associated behaviors.

Microsaccades (MSs) are commonly associated with spatially directed attention, but how they affect visual processing is still not clear. We studied MSs in a task in which the animal was randomly cued to attend to a target stimulus and ignore distractors, and it was rewarded for detecting a color change in the target. We found that the enhancement of firing rates normally found with attention to a cued stimulus was delayed until the first MS directed towards that stimulus. Once that MS occurred, attention to the target was engaged and there were persistent effects of attention on firing rates for the remainder of the trial. These effects were found in the superficial and deep layers of V4 as well as the lateral pulvinar and IT cortex. Although the tuning curves of V4 cells do not change depending on the locus of spatial attention, we found pronounced effects of MS direction on stimulus representations that persisted for the length of the trial in V4. In intervals following a MS towards the target in the RF, stimulus decoding from population activity was substantially better than in intervals following a MS away from the target. Likewise, turning curves of cells were substantially sharper following a MS towards the target in the RF. This sharpening appeared to result from both a “refreshing” of the initial transient sensory response to stimulus onset, and a magnification of the effects of attention in this condition. MSs to the target also enhanced the neuronal response to the behaviorally relevant target color change and led to faster reaction times. These results thus reveal a major link between spatial attention, object processing and its coordination with eye movements. ### Competing Interest Statement The authors have declared no competing interest.

Artificial neural networks (ANNs) are a powerful class of computational models for unravelling neural mechanisms of brain function. However, for neural control of movement, they currently must be integrated with software simulating biomechanical effectors, leading to limiting impracticalities: (1) researchers must rely on two different platforms and (2) biomechanical effectors are not generally differentiable, constraining researchers to reinforcement learning algorithms despite the existence and potential biological relevance of faster training methods. To address these limitations, we developed MotorNet, an open-source Python toolbox for creating arbitrarily complex, differentiable, and biomechanically realistic effectors that can be trained on user-defined motor tasks using ANNs. MotorNet is designed to meet several goals: ease of installation, ease of use, a high-level user-friendly API, and a modular architecture to allow for flexibility in model building. MotorNet requires no dependencies outside Python, making it easy to get started with. For instance, it allows training ANNs on typically used motor control models such as a two joint, six muscle, planar arm within minutes on a typical desktop computer. MotorNet is built on TensorFlow and therefore can implement any network architecture that is possible using the TensorFlow framework. Consequently, it will immediately benefit from advances in artificial intelligence through TensorFlow updates. Finally, it is open source, enabling users to create and share their own improvements, such as new effector and network architectures or custom task designs. MotorNet's focus on higher order model and task design will alleviate overhead cost to initiate computational projects for new researchers by providing a standalone, ready-to-go framework, and speed up efforts of established computational teams by enabling a focus on concepts and ideas over implementation. ### Competing Interest Statement The authors have declared no competing interest.

Summary: Three-dimensional digital methods useful for the study of tetrapod locomotor biomechanics have matured in recent decades. Here, we summarise their progress, relevance and potential.