start on motor lit review

Author: rcoreilly

citedrefs.json

@@ -456,7 +456,7 @@ Furthermore, the [[Rubicon]] goal-driven control framework provides a critical o
 ## Vestibulo-ocular reflex
 
 {id="figure_vor-anatomy" style="height:30em"}
-![Anatomy of the VOR reflex, which is driven by vestibular signals from the semicircular canals, which project into the vestibular nuclear complex (VNC), and from there into optical motor areas. All primary inputs are _ipsilateral_ (e.g., the right ear projects to the right hemisphere), and the VOR involves the subset of VNC neurons that project _contralaterally_ to drive the _nucleus abducens_ optical motor interneurons that cause eye movents in the compensatory, opposite direction. There are also ipsilateral pathways, which are inhibited by the contralateral ones, as well as cross-hemisphere inhibitory competition. This is a classic opponent process organization. The VNC neurons in this circuit are the functional equivalent of the cerebellar nucleus neurons in other areas, and thus direct plasticity in therse neurons is the ultimate consolidated result of learning. The cerebellar-cortex modulation of the gain on this reflex arises from changes in the tonic neural activity onto the VNC neurons from Purkinje neurons (e.g., disinhibiting increases gain), and in synaptic inputs to these VNC neurons.](media/fig_cerebellum_vor_anatomy.png)
+![Anatomy of the VOR reflex, which is driven by vestibular signals from the semicircular canals, which project into the vestibular nuclear complex (VNC), and from there into optical motor areas. All primary inputs are _ipsilateral_ (e.g., the right ear projects to the right hemisphere), and the VOR involves the subset of VNC neurons that project _contralaterally_ to drive the _nucleus abducens_ optical motor interneurons that cause eye movents in the compensatory, opposite direction. There are also ipsilateral pathways, which are inhibited by the contralateral ones, as well as cross-hemisphere inhibitory competition. This is a classic opponent process organization. The VNC neurons in this circuit are the functional equivalent of the cerebellar nucleus neurons in other areas, and thus direct plasticity in therse neurons is the ultimate consolidated result of learning. The cerebellar-cortex modulation of the gain on this reflex arises from changes in the tonic neural activity onto the VNC neurons from Purkinje neurons (e.g., disinhibiting increases gain), and in synaptic inputs to these VNC neurons. Adapted from [wikimedia.org](https://commons.wikimedia.org/wiki/File:Vestibulo-ocular_reflex_EN.svg).] (media/fig_cerebellum_vor_anatomy.png)
 
 {id="figure_vor" style="height:30em"}
 ![Central elements of VOR gain modulation in the cerebellum. The violet colored boxes at the top reflect brainstem control areas where descending cortical and basal ganglia motor inputs drive coordinated motor behavior. The `Head Ctrl` area drives neck rotation, but also by default provides inputs into the `VOReflex` motor reflex area that modulates the VOR compensatory eye movement circuitry (as shown in prior figure). The `Eye Ctrl` area supports direct saccadic control over eye position, which automatically suppresses the VOR, by inhibiting the VOReflex. The CNiIO (predictive inhibition) and visual motion (specific sensory input) drive error spiking in the associated IO neurons, which then drive learning in the FTNc (flocculous target neurons, contralateral) in the VNC (vestibular nuclear complex), which are the functional equivalent of the cerebellar nucleus (CNeDn) neurons. Floccular Purkinje neurons (PCDn) also learn (LTD, downbound) from the IO and thus disinhibit these FTNc neurons.](media/fig_cerebellum_vor.png)

content/cerebellum.md

@@ -35,7 +35,11 @@ The dot product operation (also known as the _inner product_ or _scalar product_
 
 [[#figure_face-dim-prjn]] illustrates this projection operation in the context of the [[faces simulation]], as discussed in [[categorization]]. This projection operation organizes and systematizes the inputs along dimensions of behavioral importance, for example projecting a face input along dimensions of emotion and gender in the case shown in the figure, which you can explore in the [[faces simulation]].
 
-Another linear algebra framing of this operation is in terms of the synaptic weight matrix being composed of **basis vectors**, where each row of the matrix defines a different _basis_ or _axis_ of a new, _rotated_ version of the space defined by the the input activity vector. Thus, the hidden layer neurons are creating a _transformation_ of the input space into this new rotated space, which amplifies certain things while filtering out others. 
+## Basis space
+
+Another linear algebra framing of this operation is in terms of the synaptic weight matrix being composed of **basis vectors**, where each row of the matrix defines a different _basis_ or _axis_ of a new, _rotated_ version of the **space** defined by the the input activity vector. Thus, the hidden layer neurons are creating a _transformation_ of the input space into this new rotated space, which amplifies certain things while filtering out others.
+
+Furthermore, this basis space can help with the [[curse of dimensionality]] by using fewer basis vectors to capture most of the important dimensions of variance, relative to the high dimensionality of the original space. For example, original dimensions that are largely redundant (e.g., the shapes of the eyes and the mouth used in expressing different emotions) can be combined together into one basis vector.
 
 ## Eigenvectors and singular value decomposition
 

content/linear-algebra.md

@@ -3,9 +3,36 @@ Categories = ["Neuroscience", "Cognition"]
 bibfile = "ccnlab.json"
 +++
 
-The **motor** system in the brain involves many different areas, and in many ways, the entire brain exists in service of producing motor output. For example, Daniel Wolpert has made the point that the sea squirt eats its own brain once it no longer needs to move around. More specific discussion of various aspects of motor control can be found in [[reinforcement learning]], [[basal ganglia]], [[cerebellum]], and the [[Rubicon]] framework. This page provides a high-level overview and fills in a few missing pieces.
+The **motor** system in the brain involves many different areas, and in many ways, the entire brain exists in service of producing motor output. For example, Daniel Wolpert has made the point that the sea squirt eats its own brain once it no longer needs to move around. More specific discussion of various aspects of motor control can be found in [[reinforcement learning]], [[basal ganglia]], [[cerebellum]], and the [[Rubicon]] framework. This page provides a high-level overview of the challenges and solutions for motor control, and some details about the underlying physiology of muscles and how the spinal cord and brainstem provide a systematic basis for motor control.
 
-For a more detailed treatment from experts in this area, see [[@^ArberCosta22]].
+<!--- For a more detailed treatment from experts in this area, see [[@^ArberCosta22]]. -->
+
+## Dimensionality reduction, coordination, and muscle synergies
+
+As in all aspects of neural function, the central problem in motor control is managing the [[curse of dimensionality]]: there are exponentially many possible combinations of muscle activations over time. How does the brain reduce this huge space down to the relatively small subset of muscle activations necessary for an animal's survival?
+
+This is another instance of the fundamental [[search]] problem, and in the case of the motor system, [[evolution]] has done a lot of the work by building in complex **muscle synergies** into the spinal cord and brainstem, where individual **interneurons** (excitatory and inhibitory) activate **spatiotemporal patterns** of muscle activation. From this lower-dimensional _repertoire_ of elements (i.e., a set of [[linear algebra#basis space]]), complex patterns of motor behavior are constructed ([[@Bernstein67]]; [[@Bernstein96]]; [[@TreschSaltielBizzi99]]; [[@dAvellaSaltielBizzi03]]; [[@TingMcKay07]]; [[@LatashLevinScholzEtAl10]]; [[@BrutonODwyer18]]; [[@Latash18]]).
+
+Another critical problem that is solved by these muscle synergies is the **coordination** across many different muscles that is required to accomplish any given motor action. The contraction of any given muscle creates a variety of physical consequences for other muscles, including altering the basic center of gravity of the entire organism. Thus, any given action must take all of these consequences into account, and ensure that all of the individual muscle contractions are indeed synergistic, and not working at cross purposes.
+
+A variety of different terms have been used in the literature to refer to these synergies, including **reflexes** ([[@Sherrington10]]), **central pattern generators** ([[@GrillnerElManira20]]; [[@GrillnerZangger79]]), and _force fields_ ([[@GiszterMussa-IvaldiBizzi93]]).
+
+{id="figure_synergies" style="height:30em"}
+![Three muscle synergies extracted from frog kicking behavior. a) Activation of different muscles (vertical axis) over time (horizontal axis), accomplishing given overall motor action (moment arms, aligned by corresponding column order of the synergies): _HE_ hip extension; _HF_ hip flexion; _KE_ knee extension; _KF_ knee flexion; _AE_ ankle extension; _AF_ ankle flexion. b) Activation strengths over time for each synergy component ($C_{1-3}$) needed to reconstruct 6 different overall motor actions. These patterns differ primarily in magnitude, not temporal pattern. From d'Avella et al., 2003](media/fig_muscle_synergies_davella_etal_03.png)
+
+As a concrete example, a detailed analysis of spinal-level muscle synergies in the frog leg system revealed a set of 4 synergies that could be combined with different activation strengths to explain a wide range of overall motor response patterns ([[@TreschSaltielBizzi99]]). A subsequent analysis allowing for contributions from the entire brain in intact frogs showed how the timing and activation modulation of 3 different spatiotemporal muscle synergies ([[#figure_synergies]]) can explain this same space of motor responses ([[@dAvellaSaltielBizzi03]]). 
+
+Thus, the final motor behavior pattern depends on the integrated contributions of multiple levels of control, with the spinal muscle synergies implemented by interneurons providing the lowest level, and higher levels progressively adding their own direct synergies as well as broader coordination across the basic spinal elements ([[@GrillnerElManira20]]). For example, an analysis of correlated motor units in hand movement showed _last order_ (i.e., just upstream of the muscle activation) inputs from the pontomedullary reticular formation, the magnocellular red nucleus, and primary motor cortex (M1), in addition to the basic spinal cord interneurons ([[@XuMawaseSchieber24]]).
+
+## Feedback control across levels
+
+The other critical element of dimensionality reduction in the motor system arises from the interactions with the environment, which is also highly variable and thus represents a major additional source of variance and combinatorial explosion. There is considerable evidence that the muscle synergies implemented by spinal cord interneurons incorporate direct feedback control mechanisms, that automatically compensate for environmental perturbations ([[@WimalasenaPandarinathAuYong25]]; [[@ConwayHultbornKiehn87]]; [[@AngelGuertinJimenezEtAl96]]; [[@AlvarezFyffe07]]). 
+
+Unfortunately, the ability to understand these feedback control mechanisms has been impaired by the difficulty in categorizing and mapping the connectivity of the spinal interneurons ([[@SenguptaBagnall23]]). Thus, extant circuit-based models tend to exclude such mechanisms ([[@RybakDoughertyShevtsova15]]; [[@McCreaRybak08]]). Nevertheless, more abstract state-space analyses of large populations of spinal interneurons have the potential to reveal the presence and dynamical implications of these feedback control mechanisms ([[@WimalasenaPandarinathAuYong25]]).
+
+The rather underdeveloped understanding of feedback control in motor systems contrasts with the theoretical clarity of the mathematically-based hierarchical control known as **perceptual control theory (PCT)** ([[@Powers73]]; [[@Powers73a]]; [[@Cools85]]; [[@Yin14a]]; [[@BarterYin21]]). This framework is consistent with the principle of **equilibrium-point** motor control, which postulates that motor control signals specify a target length for each muscle, rather than dynamical variables such as force ([[@FeldmanLevin09]]; [[@GribbleOstrySanguinetiEtAl98]]).
+
+While the simplest form of the equilibrium-point hypothesis is likely incorrect, the available evidence strongly suggests that muscle synergies incorporate their own internal feedback control dynamics that make them intrinsically robust, and adaptive, in the face of environmental perturbations.
 
 ## Frontal cortex learning via subcortical efferent copy signals
 
@@ -22,8 +49,6 @@ This hypothesis is consistent with the "big loop" of descending and ascending mo
 
 ## Hierarchical cascades of predictive controllers
 
-This goes in motor, but brief summary here.
-
 The residual signals provide control knobs for higher levels of control! Key example of VOR vs. saccades etc!  Saccade is an error signal from perspective of VOR. Subsumption / override and neural mechanisms supporting that.
 
 output of lower level is input to next higher level -- higher level learns to predict the residuals in lower level, using broader / higher-level context that explains the perturbations.
@@ -34,7 +59,15 @@ Key Powers et al point: you can do all of motor control in sensory space!
 
 What does cerebellum need to handle this?
 
-## Detailed muscle dynamics
+## Descending muscle control signals
+
+_Size principle_ [[@Henneman85]]; [[@LucaErim94]]: orderly activation of small to large according to total force as signalled by increasingly "strong" descending activity -- otherwise too high-dimensional. Smaller units have lower threshold, and fire at higher rates than larger units. This also optimizes fatigue effects. The size principle entails the _common-drive_ hypothesis, which is that a single unidimensional  descending control signal projects to all motor units for a given overall muscle, with the specific units activated according to the strength of the drive signal.
+
+This characterizes general story, but in primates, there is more flexible control [[@MarshallGlaserTrautmannEtAl22]], associated also with a larger number of secondary motor areas in primates that have more focal, specialized projections to specific limbs etc [[@StrickDumRathelot21]].
+
+Population coding of cortical neurons: [[@AflaloGraziano06a]] show that final hand posture accounts for most of the variance, but other factors such as speed, curvature of space, distance and force were also coded. This postural aspect, otherwise known as a _spatial synergy_ ([[@OverduindAvellaRohEtAl15]]), provides a simple model of muscle control where, regardless of the starting muscle configuration, the control signal specifies a _final configuration_ (i.e., pattern of total contraction) across all of the relevant muscles.
+
+Where more complex sequences of motor actions are required, a temporal sequence of postural configurations is specified -- a sequence of "poses" -- otherwise known as a _spatiotemporal synergy ([[@OverduindAvellaRohEtAl15]])
 
 [[@^MeyerSmithWright82]] synthesize psychophysical literature on speed and accuracy of motor movements, to develop a symmetric impulse control model that specifies the force and duration parameters as curves with an initial acceleration phase for the first half, followed by a symmetric deceleration phase in the second half. Both the force and time parameters of these curves can be controlled by people. There is evidence that ballistic movements are made below around 260 ms, with multiple iterations of visually-corrected movement updates happening after that, time permitting. See also [[@MeyerSmithKornblumEtAl90]].
 

content/motor.md

@@ -5,7 +5,7 @@ bibfile = "ccnlab.json"
 
 The web of interconnected [[neuron]]s in the brain is known as a **network**, and there are a number of important [[computational cognitive neuroscience#emergent phenomena]] that arise in these networks, associated with different patterns of connectivity present in the [[neocortex]]. We can only really understand brain function in terms of these networks and populations of neurons.
 
-The _feedforward_ flow of neural activity from lower layers of neurons that are closer to the sensory inputs up to higher layers can be understood in terms of [[categorization]], where detailed sensory patterns are systematically transformed into more abstract categories that provide a more efficient [[linear algebra|basis]] for behavior.
+The _feedforward_ flow of neural activity from lower layers of neurons that are closer to the sensory inputs up to higher layers can be understood in terms of [[categorization]], where detailed sensory patterns are systematically transformed into more abstract categories that provide a more efficient [[linear algebra#basis space]] for behavior.
 
 Thus, the network level of analysis is focused on the nature of [[representation]]s, i.e., the properties of the patterns of neuron firing in a given area of the network. [[Distributed representations]] are critical for efficient encoding of high-dimensional information, as elaborated in the discussion of [[combinatorial-vs-conjunctive]] encodings. These different types of representations have important implications for [[generalization]] performance: how well the system can process novel inputs and situations.