A5029269972- Neural dynamics and brain function
- Neural Networks and Applications
- Advanced Memory and Neural Computing
- Neuroscience and Neural Engineering
- Retinal Development and Disorders
- Neuroendocrine regulation and behavior
- Neurobiology and Insect Physiology Research
- Memory and Neural Mechanisms
- Neuroscience and Neuropharmacology Research
- stochastic dynamics and bifurcation
- Photoreceptor and optogenetics research
- Animal Behavior and Reproduction
- Olfactory and Sensory Function Studies
- Visual perception and processing mechanisms
- Advanced Chemical Sensor Technologies
- Evolutionary Game Theory and Cooperation
- Genetics, Aging, and Longevity in Model Organisms
- Neural and Behavioral Psychology Studies
- Neuroscience, Education and Cognitive Function
- Fish Ecology and Management Studies
- Evolutionary Algorithms and Applications
- Gaussian Processes and Bayesian Inference
- Receptor Mechanisms and Signaling
- Neural Networks and Reservoir Computing
- Retinal Imaging and Analysis
Weizmann Institute of Science
2015-2024
New York University
2017
Princeton University
2002-2007
Hebrew University of Jerusalem
1997-2002
A key issue in understanding the neural code for an ensemble of neurons is nature and strength correlations between how these are related to stimulus. The complicated by fact that there not a single notion independence or lack correlation. We distinguish three kinds: (1) activity independence; (2) conditional (3) information independence. Each measure: cells, cells given stimulus, synergy about respectively. show measures form interrelated framework evaluating contributions signal noise...
The firing reliability and precision of an isopotential membrane patch consisting a realistically large number ion channels is investigated using stochastic Hodgkin-Huxley (HH) model. In sharp contrast to the deterministic HH model, biophysically inspired model reproduces qualitatively different characteristics spike in response DC fluctuating current input neocortical neurons, as reported by Mainen & Sejnowski (1995). For inputs, timing highly unreliable; are significantly increased for...
Maximum entropy models are the least structured probability distributions that exactly reproduce a chosen set of statistics measured in an interacting network. Here we use this principle to construct probabilistic which describe correlated spiking activity populations up 120 neurons salamander retina as it responds natural movies. Already groups small 10 neurons, interactions between spikes can no longer be regarded perturbations otherwise independent system; for 40 or more pairwise need...
Information is carried in the brain by joint activity patterns of large groups neurons. Understanding structure and function population neural codes challenging because exponential number possible dependencies among We report here that for ~100 retinal neurons responding to natural stimuli, pairwise-based models, which were highly accurate small networks, are no longer sufficient. show sparse nature code, higher-order interactions can be easily learned using a novel model very low-order...
Entropy and information provide natural measures of correlation among elements in a network. We construct here the theoretic analog connected functions: irreducible $N$--point is measured by decrease entropy for joint distribution $N$ variables relative to maximum allowed all observed $N-1$ variable distributions. calculate ``connected information'' terms several examples, show that it also enables decomposition carried population about an outside source.
Social behavior in mammals is often studied pairs under artificial conditions, yet groups may rely on more complicated social structures. Here, we use a novel system for tracking multiple animals rich environment to characterize the nature of group and interactions, show strongly correlated mice. We have found that minimal models only individual traits pairwise correlations between are not enough capture behavior, but include third-order interactions give very accurate description group....
In retina and in cortical slice the collective response of spiking neural populations is well described by "maximum-entropy" models which only pairs neurons interact. We asked, how should such interactions be organized to maximize amount information represented population responses? To this end, we extended linear-nonlinear-Poisson model single include pairwise interactions, yielding a stimulus-dependent, maximum-entropy model. found that as varied noise level distribution network inputs,...
In vision, two mixtures, each containing an independent set of many different wavelengths, may produce a common color percept termed “white.” audition, frequencies, perceptual hum “white noise.” Visual and auditory whites emerge upon conditions: when the mixture components span stimulus space, they are equal intensity. We hypothesized that if we apply these same conditions to odorant “whiteness” in olfaction as well. selected 86 molecules olfactory space individually diluted them point about...
Neural populations encode information about their stimulus in a collective fashion, by joint activity patterns of spiking and silence. A full account this mapping from to neural is given the conditional probability distribution over codewords sensory input. For large populations, direct sampling these distributions impossible, so we must rely on constructing appropriate models. We show here that population 100 retinal ganglion cells salamander retina responding temporal white-noise stimuli,...
Sensory information is represented in the brain by joint activity of large groups neurons. Recent studies have shown that, although number possible patterns and underlying interactions exponentially large, pairwise-based models give a surprisingly accurate description neural population patterns. We explored architecture maximum entropy functional interaction networks response populations retinal ganglion cells, adult tiger salamander retina, responding to natural artificial stimuli. found...
Odor identity is coded in spatiotemporal patterns of neural activity the olfactory bulb. Here we asked whether meaningful information could also be read from global population response. We applied standard statistical methods dimensionality-reduction to 12 previously published studies using seven different species. Four reported receptor activity, glomerulus and one projection-neurons. found two linear axes that accounted for more than half variance response across The first axis was...
The manner in which groups of neurons represent events the external world is a central question neuroscience. Estimation information encoded by small has shown that many neural systems, cells carry mildly redundant information. These measures average over all activity patterns population. Here, we analyze population code salamander and guinea pig retinas quantifying conveyed specific multicell patterns. Synchronous spikes, even though they are relatively rare highly informative, convey less...
Information is carried in the brain by joint spiking patterns of large groups noisy, unreliable neurons. This noise limits capacity neural code and determines how information can be transmitted read-out. To accurately decode, must overcome this identify which are semantically similar. We use models network encoding to learn a thesaurus for populations neurons vertebrate retina responding artificial natural videos, measuring similarity between population responses visual stimuli based on they...
The effect of the detailed connectivity a neural circuit on its function and resulting behavior organism is key question in many systems. Here, we study for nociception C. elegans, which composed same neurons two sexes that are wired differently. We show nociceptive sensory respond similarly sexes, yet animals display sexually dimorphic behaviors to aversive stimuli. To uncover role downstream network topology shaping behavior, learn simulate models replicate observed use them predict simple...
Ising models with pairwise interactions are the least structured, or maximum-entropy, probability distributions that exactly reproduce measured correlations between spins. Here we use this equivalence to construct describe correlated spiking activity of populations 40 neurons in retina, and show account for observed higher-order correlations. By first finding a representative ensemble networks can create synthetic 120 neurons, find increasing size operate closer critical point start...
Ising models with pairwise interactions are the least structured, or maximum-entropy, probability distributions that exactly reproduce measured correlations between spins. Here we use this equivalence to construct describe correlated spiking activity of populations 40 neurons in salamander retina responding natural movies. We show account for observed higher-order correlations, and groups 10 more can no longer be regarded as small perturbations an independent system. then network ensembles...
Collective navigation and swarming have been studied in animal groups, such as fish schools, bird flocks, bacteria, slime molds. Computer modeling has shown that collective behavior of simple agents can result from interactions between the agents, which include short range repulsion, intermediate alignment, long attraction. Here we study bacteria-inspired smart complex terrains, with adaptive depend on performance. More specifically, each agent adjusts its other according to local...
Whisking mediated touch is an active sense whereby whisker movements are modulated by sensory input and behavioral context. Here we studied the effects of touching object on whisking in head-fixed rats. Simultaneous whiskers C1, C2, D1 were tracked bilaterally their compared. During free-air whisking, protractions typically characterized a single acceleration-deceleration event, amplitude velocity correlated, whisk duration correlated with neither nor velocity. Upon contact object, second...
Individual computations and social interactions underlying collective behavior in groups of animals are great ethological, behavioral, theoretical interest. While complex individual behaviors have successfully been parsed into small dictionaries stereotyped behavioral modes, studies largely ignored these findings; instead, their focus was on inferring single, mode-independent interaction rules that reproduced macroscopic often qualitative features group behavior. Here we bring two approaches...