Optimal reaching subject to computational and physical constraints reveals structure of the sensorimotor control system.

Optimal feedback control provides an abstract framework describing the architecture of the sensorimotor system without prescribing implementation details such as what coordinate system to use, how feedback is incorporated, or how to accommodate changing task complexity. We investigate how such details are determined by computational and physical constraints by creating a model of the upper limb sensorimotor system in which all connection weights between neurons, feedback, and muscles are unknown. By optimizing these parameters with respect to an objective function, we find that the model exhibits a preference for an intrinsic (joint angle) coordinate representation of inputs and feedback and learns to calculate a weighted feedforward and feedback error. We further show that complex reaches around obstacles can be achieved by augmenting our model with a path-planner based on via points. The path-planner revealed "avoidance" neurons that encode directions to reach around obstacles and "placement" neurons that make fine-tuned adjustments to via point placement. Our results demonstrate the surprising capability of computationally constrained systems and highlight interesting characteristics of the sensorimotor system.

A comparison of probe geometries for neuronal localization.

Extracellular electrical recordings capture the spiking activity of multiple neurons in the vicinity of a probe. Typically, the features of interest in these recordings are action potentials and their timing. However, for planar probes that span tens or hundreds of neurons, it is possible to identify relative spatial locations of neurons. Such spatial information may be useful for reconstructing local network structure or for improving the quality of spike sorting. We propose a Bayesian modification of a dipole-based method for estimating neural positions from waveforms recorded on multi contact probes and investigate how sensitive it is to prior knowledge about the equivalent dipole sizes of neurons and the geometry of the recording probe. In addition, we determine the probe spacing and number of contacts which produce optimal localization accuracy within the class of planar, circularly symmetric contact configurations.

A computational perspective on coordinate systems for motor control.

It is currently unknown what coordinate system or systems the primate motor cortex uses to represent movement, although experimental evidence has suggested several candidates. In order to understand how the physical geometry of the arm combines with computational constraints to influence the optimal choice of coordinate system, we construct a two-dimensional, physics-based arm model and couple it to a linear model of the motor cortex. The cortical model is provided with target positions and real time feedback of the current hand position in two different coordinate systems: cartesian and joint angle. We then optimize the parameters of the model subject to penalties on neural connectivity and muscle and neural energy use. We find that the optimized model strongly prefers to work in the joint angle coordinate system, suggesting that for neurons whose activity is closely tied to muscle activation, this is computationally the most efficient coordinate system in which to represent movement.

Alpha Power in the Cingulate Cortex reflects Accumulated Winnings During Gambling in Humans.

When making bets one's level of attention determines how much they may win. The cingulate cortex is a brain region associated with attention and may influence behaviors during gambling. With data gathered from the cingulate cortex in humans implanted with depth electrodes for clinical purposes while performing a gambling task of high card, we determine a relationship between neural correlates of attention and accumulated winnings. Specifically, we analyze how changes in alpha power (8-12 Hz) in the CC relate to accumulated winnings. We compared three subjects with different betting strategies: Reflexive (betting low on cards 2, 4, and 6), Logical (varying how they bet on card 6), and Illogical (betting randomly on all cards). We found that alpha power encodes attention in the cingulate cortex and relates to their accumulated winnings, especially in the illogical subject who had the least winning.

Neural Encoding of Reaches in a Linear Cortical Model.

To effectively control the arm, motor cortical neurons must produce complex patterns of activation that vary with the position and orientation of the arm and reach direction. In order to better understand how such a finely tuned dynamical system could arise and what its basic organizing principles are, we develop a model of the motor cortex as a linear dynamical system with feedback coupled to a two-joint model of the macaque arm. By optimizing the connections between neural populations with respect to an objective function that penalizes error between hand and target, as well as neural and muscular energy use, we show that certain properties of the motor cortex, such as muscle synergies, can naturally be obtained. We also demonstrate that the optimization process produces a stable neural system in which targets in the physical space are mapped to attracting fixed points in the neural state space. Finally, we show that this optimization process produces neural units with complex spatial and temporal activation patterns.

High Frequency Activity in the Orbital Frontal Cortex Modulates with Mismatched Expectations During Gambling in Humans.

During gambling, humans often begin by making decisions based on expected rewards and expected risks. However, expectations may not match actual outcomes. As gamblers keep track of their performance, they may feel more or less lucky, which then influences future betting decisions. Studies have identified the orbitofrontal cortex (OFC) as a brain region that plays a significant role during risky decision making in humans. However, most human studies infer neural activation from functional magnetic resonance imaging (fMRI), which has a poor temporal resolution. In particular, fMRI cannot detect activity from neuronal populations in the OFC, which may encode specific information about how a subject reacts to mismatched outcomes. In this preliminary study, four human subjects participated in a gambling task while local field potentials (LFPs), captured at a millisecond resolution, were recorded from the OFC. We analyzed high-frequency activity (HFA: >70 Hz) in the LFPs, as HFA has been shown to correlate to activation of neuronal populations. In 3 out of 4 subjects, HFA in OFC modulated between matched and mismatched trials as soon as the outcome of each bet was revealed, with modulations occurring at different times and directions depending on the anatomical location within the OFC.

Correlates of Attention in the Cingulate Cortex During Gambling in Humans.

People make decisions multiple times on a daily basis. However, some decisions are easier to make than others and perhaps require more attention to ensure a positive outcome. During gambling, one should attempt to compute the expected rewards and risks associated with decisions. Failing to allocate attention and neural resources to estimate these values can be costly, and in some cases can lead to bankruptcy. Alpha-band (8-12 Hz) oscillatory power in the brain is thought to reflect attention, but how this influences financial decision making is not well understood. Using local field potential recordings in nine human subjects performing a gambling task, we compared alpha-band power from the cingulate cortex (CC) during trials of low and high attention. We found that alpha-band power tended to be higher during a 2 second window after a fixation cue was shown in low attention trials.

Classification of Stereo-EEG Contacts in White Matter vs. Gray Matter Using Recorded Activity.

For epileptic patients requiring resective surgery, a modality called stereo-electroencephalography (SEEG) may be used to monitor the patient's brain signals to help identify epileptogenic regions that generate and propagate seizures. SEEG involves the insertion of multiple depth electrodes into the patient's brain, each with 10 or more recording contacts along its length. However, a significant fraction (≈ 30% or more) of the contacts typically reside in white matter or other areas of the brain which can not be epileptogenic themselves. Thus, an important step in the analysis of SEEG recordings is distinguishing between electrode contacts which reside in gray matter vs. those that do not. MRI images overlaid with CT scans are currently used for this task, but they take significant amounts of time to manually annotate, and even then it may be difficult to determine the status of some contacts. In this paper we present a fast, automated method for classifying contacts in gray vs. white matter based only on the recorded signal and relative contact depth. We observe that bipolar referenced contacts in white matter have less power in all frequencies below 150 Hz than contacts in gray matter, which we use in a Bayesian classifier to attain an average area under the receiver operating characteristic curve of 0.85 ± 0.079 (SD) across 29 patients. Because our method gives a probability for each contact rather than a hard labeling, and uses a feature of the recorded signal that has direct clinical relevance, it can be useful to supplement decision-making on difficult to classify contacts or as a rapid, first-pass filter when choosing subsets of contacts from which to save recordings.

Re-curation and rational enrichment of knowledge graphs in Biological Expression Language.

The rapid accumulation of new biomedical literature not only causes curated knowledge graphs (KGs) to become outdated and incomplete, but also makes manual curation an impractical and unsustainable solution. Automated or semi-automated workflows are necessary to assist in prioritizing and curating the literature to update and enrich KGs. We have developed two workflows: one for re-curating a given KG to assure its syntactic and semantic quality and another for rationally enriching it by manually revising automatically extracted relations for nodes with low information density. We applied these workflows to the KGs encoded in Biological Expression Language from the NeuroMMSig database using content that was pre-extracted from MEDLINE abstracts and PubMed Central full-text articles using text mining output integrated by INDRA. We have made this workflow freely available at https://github.com/bel-enrichment/bel-enrichment.

Hippocampal anatomy supports the use of context in object recognition: a computational model.

The human hippocampus receives distinct signals via the lateral entorhinal cortex, typically associated with object features, and the medial entorhinal cortex, associated with spatial or contextual information. The existence of these distinct types of information calls for some means by which they can be managed in an appropriate way, by integrating them or keeping them separate as required to improve recognition. We hypothesize that several anatomical features of the hippocampus, including differentiation in connectivity between the superior/inferior blades of DG and the distal/proximal regions of CA3 and CA1, work together to play this information managing role. We construct a set of neural network models with these features and compare their recognition performance when given noisy or partial versions of contexts and their associated objects. We found that the anterior and posterior regions of the hippocampus naturally require different ratios of object and context input for optimal performance, due to the greater number of objects versus contexts. Additionally, we found that having separate processing regions in DG significantly aided recognition in situations where object inputs were degraded. However, split processing in both DG and CA3 resulted in performance tradeoffs, though the actual hippocampus may have ways of mitigating such losses.