Bulletin of the American Physical Society
APS March Meeting 2016
Volume 61, Number 2
Monday–Friday, March 14–18, 2016; Baltimore, Maryland
Session P35: Physics of Sensorimotor Neural Circuits IIFocus
|
Hide Abstracts |
Sponsoring Units: DBIO Chair: Alex Koulakov, Cold Springs Harbor Laboratory Room: 338 |
Wednesday, March 16, 2016 2:30PM - 2:42PM |
P35.00001: Old and new news about single-photon sensitivity in human vision Philip Nelson It is sometimes said that ``our eyes can see single photons,'' when in fact the faintest flash of light that can reliably be reported by human subjects is closer to 100 photons. Nevertheless, there is a sense in which the familiar claim is true. Experiments conducted long after the seminal work of Hecht, Shlaer, and Pirenne in two distinct realms, those of human psychophysics and single-cell physiology, now admit a more precisem conclusion to be drawn about our visual apparatus. Finding a single framework that accommodates both kinds of result is a nontrivial challenge, and one that sets severe quantitative constraints on any model of dim-light visual processing. I will present one such model and compare it to a recent experiment. [Preview Abstract] |
Wednesday, March 16, 2016 2:42PM - 2:54PM |
P35.00002: Matching tutors and students: effective strategies for information transfer between circuits Tiberiu Tesileanu, Vijay Balasubramanian, Bence Olveczky Many neural circuits transfer learned information to downstream circuits: hippocampal-dependent memories are consolidated into long-term memories elsewhere; motor cortex is essential for skill learning but dispensable for execution; anterior forebrain pathway (AFP) in songbirds drives short-term improvements in song that are later consolidated in pre-motor area RA. We show how to match instructive signals from tutor circuits to synaptic plasticity rules in student circuits to achieve effective two-stage learning. We focus on learning sequential patterns where a timebase is transformed into motor commands by connectivity with a `student' area. If the sign of the synaptic change is given by the magnitude of tutor input, a good teaching strategy uses a strong (weak) tutor signal if student output is below (above) its target. If instead timing of tutor input relative to the timebase determines the sign of synaptic modifications, a good instructive signal accumulates the errors in student output as the motor program progresses. We demonstrate song learning in a biologically-plausible model of the songbird circuit given diverse plasticity rules interpolating between those described above. The model also reproduces qualitative firing statistics of RA neurons in juveniles and adults. [Preview Abstract] |
Wednesday, March 16, 2016 2:54PM - 3:06PM |
P35.00003: Sensory stimuli reduce the dimensionality of cortical activity Luca Mazzucato, Alfredo Fontanini, Giancarlo La Camera Neural ensembles in alert animals generate complex patterns of activity. Although cortical activity unfolds in a space whose dimension is equal to the number of neurons, it is often restricted to a lower dimensional subspace. Dimensionality is the minimal number of dimensions that accurately capture neural dynamics, and may be related to the computational tasks supported by the neural circuit. Here, we investigate the dimensionality of neural ensembles from the insular cortex of alert rats during periods of `ongoing' (spontaneous) and stimulus-evoked activity. We find that the dimensionality grows with ensemble size, and does so significantly faster during ongoing compared to evoked activity. We explain both results using a recurrent spiking network with clustered architecture, and obtain analytical results on the dependence of dimensionality on ensemble size, number of clusters, and pair-wise noise correlations. The theory predicts a characteristic scaling with ensemble size and the existence of an upper bound on dimensionality, which grows with the number of clusters and decreases with the amount of noise correlations. To our knowledge, this is the first mechanistic model of neural dimensionality in cortex during both spontaneous and evoked activity. [Preview Abstract] |
Wednesday, March 16, 2016 3:06PM - 3:42PM |
P35.00004: Patterns in nature shape human visual perception. Invited Speaker: Ann Hermundstad The statistical regularities of natural signals are known to shape the first stages of sensory processing. In the visual system, an efficient representation of light intensity begins in retina, where statistical redundancies are removed via spatiotemporal decorrelation. Much less is known, however, about the efficient representation of complex features in higher visual areas. I will discuss how the central visual system, operating with different goals and under different constraints, makes efficient use of resources to extract meaningful features from complex visual stimuli. [Preview Abstract] |
Wednesday, March 16, 2016 3:42PM - 3:54PM |
P35.00005: Nonlinear Bayesian cue integration explains the dynamics of vocal learning Baohua Zhou, Samuel Sober, Ilya Nemenman The acoustics of vocal production in songbirds is tightly regulated during both development and adulthood as birds progressively refine their song using sensory feedback to match an acoustic target. Here, we perturb this sensory feedback using headphones to shift the pitch (fundamental frequency) of song. When the pitch is shifted upwards (downwards), birds eventually learn to compensate and sing lower (higher), bringing the experienced pitch closer to the target. Paradoxically, the speed and amplitude of this motor learning decrease with increases in the introduced error size, so that birds respond rapidly to a small sensory perturbation, while seemingly never correcting a much bigger one. Similar results are observed broadly across the animal kingdom, and they do not derive from a limited plasticity of the adult brain since birds can compensate for a large error as long as the error is imposed gradually. We develop a mathematical model based on nonlinear Bayesian integration of two sensory modalities (one perturbed and the other not) that quantitatively explains all of these observations. The model makes predictions about the structure of the probability distribution of the pitches sung by birds during the pitch shift experiments, which we confirm using experimental data. [Preview Abstract] |
Wednesday, March 16, 2016 3:54PM - 4:06PM |
P35.00006: Neurocontrol in sensory cortex Jason Ritt, Anirban Nandi, Joseph Schroeder, ShiNung Ching Technology to control neural ensembles is rapidly advancing, but many important challenges remain in applications, such as design of controls (e.g. stimulation patterns) with specificity comparable to natural sensory encoding. We use the rodent whisker tactile system as a model for active touch, in which sensory information is acquired in a closed loop between feedforward encoding of sensory information and feedback guidance of sensing motions. Motivated by this system, we present optimal control strategies that are tailored for underactuation (a large ratio of neurons or degrees of freedom to stimulation channels) and limited observability (absence of direct measurement of the system state), common in available stimulation technologies for freely behaving animals. Using a control framework, we have begun to elucidate the feedback effect of sensory cortex activity on sensing in behaving animals. For example, by optogenetically perturbing primary sensory cortex (SI) activity at varied timing relative to individual whisker motions, we find that SI modulates future sensing behavior within 15 msec, on a whisk by whisk basis, changing the flow of incoming sensory information based on past experience. [Preview Abstract] |
Wednesday, March 16, 2016 4:06PM - 4:18PM |
P35.00007: Dynamical encoding of looming, receding, and focussing.~ Andre Longtin, Stephen Elisha Clarke, Leonard Maler This talk will discuss a non-conventional neural coding task that may apply more broadly to many senses in higher vertebrates. We ask whether and how a non-visual sensory system can focus on an object. We present recent experimental and modeling work that shows how the early sensory circuitry of electric sense can perform such neuronal focusing that is manifested behaviorally. This sense is the main one used by weakly electric fish to navigate, locate prey and communicate in the murky waters of their natural habitat. We show that there is a distance at which the Fisher information of a neuron's response to a looming and receding object is maximized, and that this distance corresponds to a behaviorally relevant one chosen by these animals. Strikingly, this maximum occurs at a bifurcation between tonic firing and bursting. We further discuss how the invariance of this distance to signal attributes can arise, a process that first involves power-law spike frequency adaptation. The talk will also highlight the importance of expanding the classic dual neural encoding of contrast using ON and OFF cells in the context of looming and receding stimuli. ~~ [Preview Abstract] |
Wednesday, March 16, 2016 4:18PM - 4:30PM |
P35.00008: Songbird Respiration is Controlled by Multispike Patterns at Millisecond Temporal Resolution Caroline Holmes, Kyle Srivastava, Michiel Vellema, Coen Elemans, Ilya Nemenman, Samuel Sober Although the importance of precise timing of neural action potentials (spikes) is well known in sensory systems, approaches to motor control have focused almost exclusively on firing rates. Here we examined whether precise timing of spikes in multispike patterns has an effect on the motor output in the respiratory system of the Bengalese finch, a songbird. By recording from single motor neurons and the muscle fibers they innervate in freely behaving birds, we find that the spike trains are significantly non-Poisson, suggesting that the precise timing of spikes is tightly controlled. We further find that even a one millisecond shift of an individual spike in a multispike pattern predicts a significantly different air sac pressure. Finally, we provide evidence for the causal relation between precise spike timing and the motor output in this organism by stimulating the motor system with precisely timed patterns of electrical impulses. We observe that shifting a single pulse by as little as two milliseconds elicits differences in resulting air sac pressure. These results demonstrate that the precise timing of spikes does play a role in motor control. [Preview Abstract] |
Wednesday, March 16, 2016 4:30PM - 4:42PM |
P35.00009: The primary visual cortex in the neural circuit for visual orienting Li Zhaoping The primary visual cortex (V1) is traditionally viewed as remote from influencing brain's motor outputs. However, V1 provides the most abundant cortical inputs directly to the sensory layers of superior colliculus (SC), a midbrain structure to command visual orienting such as shifting gaze and turning heads. I will show physiological, anatomical, and behavioral data suggesting that V1 transforms visual input into a saliency map to guide a class of visual orienting that is reflexive or involuntary. In particular, V1 receives a retinotopic map of visual features, such as orientation, color, and motion direction of local visual inputs; local interactions between V1 neurons perform a local-to-global computation to arrive at a saliency map that highlights conspicuous visual locations by higher V1 responses. The conspicuous location are usually, but not always, where visual input statistics changes. The population V1 outputs to SC, which is also retinotopic, enables SC to locate, by lateral inhibition between SC neurons, the most salient location as the saccadic target. Experimental tests of this hypothesis will be shown. Variations of the neural circuit for visual orienting across animal species, with more or less V1 involvement, will be discussed. [Preview Abstract] |
Wednesday, March 16, 2016 4:42PM - 4:54PM |
P35.00010: Distributed multisensory integration in a recurrent network model through supervised learning He Wang, K. Y. Michael Wong Sensory integration between different modalities has been extensively studied. It is suggested that the brain integrates signals from different modalities in a Bayesian optimal way. However, how the Bayesian rule is implemented in a neural network remains under debate. In this work we propose a biologically plausible recurrent network model, which can perform Bayesian multisensory integration after trained by supervised learning. Our model is composed of two modules, each for one modality. We assume that each module is a recurrent network, whose activity represents the posterior distribution of each stimulus. The feedforward input on each module is the likelihood of each modality. Two modules are integrated through cross-links, which are feedforward connections from the other modality, and reciprocal connections, which are recurrent connections between different modules. By stochastic gradient descent, we successfully trained the feedforward and recurrent coupling matrices simultaneously, both of which resembles the Mexican-hat. We also find that there are more than one set of coupling matrices that can approximate the Bayesian theorem well. Specifically, reciprocal connections and cross-links will compensate each other if one of them is removed. Even though trained with two inputs, the network's performance with only one input is in good accordance with what is predicted by the Bayesian theorem. [Preview Abstract] |
Wednesday, March 16, 2016 4:54PM - 5:06PM |
P35.00011: Learning tinnitus J. Leo van Hemmen Tinnitus, implying the perception of sound without the presence of any acoustical stimulus, is a chronic and serious problem for about 2\% of the human population. In many cases, tinnitus is a pitch-like sensation associated with a hearing loss that confines the tinnitus frequency to an interval of the tonotopic axis. Even in patients with a normal audiogram the presence of tinnitus may be associated with damage of hair-cell function in this interval. It has been suggested that homeostatic regulation and, hence, increase of activity leads to the emergence of tinnitus. For patients with hearing loss, we present spike-timing-dependent Hebbian plasticity (STDP) in conjunction with homeo\-stasis as a mechanism for ``learning'' tinnitus in a realistic neuronal network with tonotopically arranged synaptic excitation and inhibition. In so doing we use both dynamical scaling of the synaptic strengths and altering the resting potential of the cells. The corresponding simulations are robust to parameter changes. Understanding the mechanisms of tinnitus induction, such as here, may help improving therapy. [Preview Abstract] |
Wednesday, March 16, 2016 5:06PM - 5:18PM |
P35.00012: Simulating Visual Learning and Optical Illusions via a Network-Based Genetic Algorithm Theodore Siu, Miguel Vivar, Troy Shinbrot We present a neural network model that uses a genetic algorithm to identify spatial patterns. We show that the model both learns and reproduces common visual patterns and optical illusions. Surprisingly, we find that the illusions generated are a direct consequence of the network architecture used. We discuss the implications of our results and the insights that we gain on how humans fall for optical illusions [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700