Friday, May 05, 2017

Systems Neuroscience Highlights: April 2017

Lots of great systems neuroscience this month. It was hard to narrow it down, but three papers really stood out.

Cognitive Neuroscience


Eichenbaum -- The role of the hippocampus in navigation is memory. J. Neurophys.. [Pubmed] Most of us have wondered about the relationship between the two main views of the hippocampus: on one hand, the hippocampus is key for long-term memory formation, and on the other hand we have the view from the place field, where the hippocampus contains a map that is used for navigation. In this wide-ranging review article, Eichenbaum forcefully argues that the hippocampus is not specialized for spatial navigation per se, but for the construction of memories of relatively highly organized complex information in space and time (i.e., episodic memories). He argues that context-dependent spatial features are just one of many complex relational features to which the hippocampus is sensitive as it serves its role in memory function.
    This review article is notable partly because it is a rich source of references that outsiders probably don't know about. For instance, if you are really familiar with an environment, you can still navigate it even with hippocampal lesions (https://www.ncbi.nlm.nih.gov/pubmed/15723062). Also, an imaging study in humans suggests there may be a grid-like parcellation of abstract conceptual spaces, not just geometric space (https://www.ncbi.nlm.nih.gov/pubmed/27313047). Note I can't endorse all these studies, as I have yet to read or evaluate them; but it is useful to have all this intriguing stuff in one place as food for thought.

Motor Control

Giovannucci et al -- Cerebellar granule cells acquire a widespread predictive feedback signal during motor learning. Nat. Neurosci.. [Pubmed] Using calcium imaging, they recorded from populations of granule cells, the input cells of the cerebellum, during eyeblink conditioning (recall in eyeblink conditioning you associate a cue, such as a light, with an air puff to the eye, and eventually that cue will evoke a blink). As animals acquired the behavioral response to the new cue, this was reflected in the emergence of signals within the granule cells that predicted oncoming eyeblinks.
    What is really amazing in this study is that they recorded from populations of some of the smallest cells in the brain for multiple days in a row, in awake animals. I'm not surprised that the cerebellum acquired eyeblink-related control signals during training; what is most impressive to me is the raw experimental expertise involved here, and the potential this model system has for helping us dissect forward model theories of motor control.

Shadmehr -- Distinct neural circuits for control of movement vs. holding still. J. Neurophys.. [Pubmed] A fun review article by Shadmehr that focuses on the eye movement system. There are different mechanisms at play for movement versus holding still, even though from the perspective of the muscles in your eyes, holding still is "just as much an active process as movement" (Shadmehr, quoting Robinson, 1970). Could this be a general principle? After reviewing the evidence from the eye-movement system in some detail, Shadmehr discusses whether the same principles might hold for neuronal control of head movement, arm movement, and navigation.
    This intriguing possibility could help shed light on apparent discrepancies between pre-movement preparatory activity observed in M1 (when animals are still) and activity observed during movement. This topic has received a lot of attention lately from Mark Churchland's lab (see [1], [2]).

Tuesday, April 04, 2017

Systems Neuroscience Highlights: March 2017

First post of monthly highlights from the systems neuroscience literature. My goal is to point out cool stuff that people might not ordinarily see, so I will try not to just include Nature and Science papers. I will typically highlight three to five papers a month,  but this includes some February spillover so is a little longer. I will post by the fifth of each month.

Sensory Coding

Shi et al -- Retinal origin of direction selectivity in the superior colliculus. Nature Neuroscience [Pubmed] The authors used optogenetic stimulation to show that the motion-selectivity of superficial superior colliculus neurons is inherited entirely from the direction selectivity of retinal ganglion cells that project there.

Cognitive Neuroscience

Yackle et al -- Breathing Control Center Neurons That Promote Arousal in Mice. Science. [Pubmed] The CPG that controls breathing contains a small subpopulation of neurons that projects to the locus coeruleus, which releases noradrenaline (i.e., sympathetic activation for fight/flight). Removing this subset of neurons apparently did not influence the ability of mice to breath, but did make them especially chill. Take-home lesson: if you want to calm down, stop breathing.

Motor Control

Shadmehr -- Learning to Predict and Control the Physics of Our Movements. J Neurosci. [Pubmed]  Interestingly, this month there were quite a few papers related to the forward model framework in motor control (for a review, see Shadmehr and Krakaur's Error correction, sensory prediction, and adaptation in motor control (2010)). This paper from Shadmehr is an excellent summary of his many seminal contributions to this framework over the years. It focuses on his research on our ability to learn to manipulate objects with our hands, which involves quickly learning their unique dynamical signatures.


Maeda et al -- Foot placement relies on state estimation during visually guided walking. J. Neurophys. [Pubmed] The second notable paper from the forward-model theoretic framework. How do we walk when we wear prismatic lenses that render visual feedback unreliable? This paper suggests that subjects learn to weight internally generated predictions more than the resulting noisy and unreliable visual feedback. Similar results have been seen before in reaching tasks (e.g., K├Ârding and, Wolpert, 2004). However, this is a cool use of distorting lenses to demonstrate such effects during walking, which is typically thought to rely on mindless CPGs.


Confais et al -- Nerve-Specific Input Modulation to Spinal Neurons during a Motor Task in the Monkey. J. Neurosci. [Pubmed]  When we move, we activate our own sensory transducers. What keeps our sensory systems from getting overwhelmed by such self-generated sensory signals?  Following up on Seki et al (2004), this paper shows that there are sensory-nerve specific patterns of modulation (both excitation and inhibition) of somatosensory responses in the spinal cord during voluntary wrist movements. The sign of modulation sometimes depended on the particular direction of movement of the wrist. This is a beautiful model system for the study of the effects of corollary discharge.

Chaisanguanthum et al -- Neural Representation and Causal Models in Motor Cortex. J. Neurosci. [Pubmed] An excellent paper straddling classical motor control theories of Georgopoulos and friends, and some modern ideas from a horde that has been attacking such ideas recently. They construct a simple mathematical model of the sensorimotor transformation required to perform a center-out reaching task, and show that movement variability will be minimized when the output neurons that directly drive behavior are tuned to velocity. Indeed, they discover just such a population in their data (using a somewhat rough-hewn spike-width criterion to individuate subclasses of cortical neurons). While the model in this paper is simple, it is a welcome counterweight to the recent overreactions against Georgopoulos. Hopefully it is the first of many studies that will ultimately absorb previous work in a principled way.

Why am I being so pro-Georgopoulos? I'm not: I'm just surprised that people have recently been so dismissive of Georgopoulos, to the point where it seems they are just attacking a straw man. Students of motor control were never so locked into the velocity-tuning framework that they thought it would apply to all neurons (for an excellent review, see Kalaska, 2009). Further, is anyone that surprised at nonstationarities in the system? That is, was anyone really surprised that neurons don't show the same tuning properties seconds before an animal starts moving, when recording in brain regions whose primary function is to directly control movement? The sensory systems literature is absorbing nonstationarities and dynamics without all this fanfare. What's up, motor control?