Yarden Cohen

Yarden Cohen

Researcher of living and artificial neural systems,
behavior, memory, and computation.

Publications

Large-scale cellular-resolution imaging of neural activity in freely behaving mice

Published in bioRxiv, 2021

Miniaturized microscopes for head-mounted fluorescence imaging are powerful tools for visualizing neural activity during naturalistic behaviors, but the restricted field of view of first-generation ‘miniscopes’ limits the size of neural populations accessible for imaging. Here we describe a novel miniaturized mesoscope offering cellular-resolution imaging over areas spanning several millimeters in freely moving mice. This system enables comprehensive visualization of activity across entire brain regions or interactions across areas. Read more

Recommended citation: Leman DP, Chen IA, Bolding KA, Tai J, Wilmerding LK, Gritton HJ, Cohen Y, Yen WW, Perkins LN, Liberti WA, Kilic K, Han X, Cruz-Martín A, Gardner TJ, Otchy TM, Davison IG (2021) "Large-scale cellular-resolution imaging of neural activity in freely behaving mice". bioRxiv, https://doi.org/10.1101/2021.01.15.426462 https://www.biorxiv.org/content/10.1101/2021.01.15.426462v1

A novel approach to the empirical characterization of learning in biological systems

Published in bioRxiv, 2021

Movement control ties brain activity to measurable external actions in real time, providing a useful tool for both neuroscientists interested in the emergence of stable behavior and biomedical engineers interested in the design of neural prosthesis and brain-machine interfaces. We approach the question of motor skill learning by introducing artificial errors through a novel perturbative scheme amenable to analytic examination in the linearized regime close to the desired behavior. Numerical simulations then demonstrate how to probe the learning dynamics in both linear and nonlinear systems. These findings stress the usefulness of analyzing responses to deliberately induced errors and the importance of properly designing such perturbation experiments. Our approach provides a novel generic tool for monitoring the acquisition of motor skills. Read more

Recommended citation: Cohen Y, Cvitanovic P, Solla SA (2021) "A novel approach to the empirical characterization of learning in biological systems". bioRxiv https://www.biorxiv.org/content/10.1101/2021.01.10.426118v1

The geometry of neuronal representations during rule learning reveals complementary roles of cingulate cortex and putamen

Published in Neuron, 2020

To study neural dynamics during learning of new and diverse concepts, I trained two monkeys to perform the same classification task my human subjects carried and recorded neurons in the dACC and the Striatum as the animals learned eight novel classification rules. To examine dynamics, I developed a description of rules and neural representations of the visual stimuli that allowed tracking dynamics in geometrical terms - unifying sessions of learning different rules. This framework allowed teasing apart potential roles of the different brain areas and to predict future behavior from the neural state. Read more

Recommended citation: Cohen Y, Schneidman E, Paz R (2020) "The geometry of neuronal representations during rule learning reveals complementary roles of cingulate cortex and putamen". Neuron. https://doi.org/10.1016/j.neuron.2020.12.027 https://www.cell.com/neuron/pdf/S0896-6273(20)31031-X.pdf

Hidden neural states underlie canary song syntax

Published in Nature, 2020

Motor skills with long-range sequence dependencies are common in complex behaviors, with speech the richest example. In general, the neural mechanisms underlying long-range motor sequence dependencies are unknown. Using miniaturized head-mounterd fluorescence microscopes and genetic tools I revealed coding of such memory-dependent syntactic properties in singing canaries. Read more

Recommended citation: Cohen Y, Shen J, Semu D, Leman DP, Liberti WA III, Perkins N, and Gardner TJ (2020). "Hidden neural states underlie canary song syntax." Nature 582, 539–544. https://www.nature.com/articles/s41586-020-2397-3

TweetyNet: A neural network that enables high-throughput, automated annotation of birdsong

Published in bioRxiv (under review, eLife), 2020

Many studies involving variable birdsong, like that of Bengalese finches and canaries, require that experimenters annotate syllbles - the basic components of vocal sequences. These studies are currently hindered by the lack of automation means to scale up analyses. We developed TweetyNet, an artificial neural network for automated annotation. This algorithm learns features from data, and does not require segmented syllables to predict annotations. TweetyNet allowed us to annotate many more songs of individual complex singers than previously demonstrated, with high accuracy across individuals and across species. This accuracy allowed fully-automated analyses, saved most of the labor, and revealed novel details of canary syntax in a new strain. Read more

Recommended citation: Cohen Y, Nicholson DA, Gardner TJ (2020) "TweetyNet: A neural network that enables high-throughput, automated annotation of birdsong". bioRxiv (under review, eLife) https://www.biorxiv.org/content/10.1101/2020.08.28.272088v1.full.pdf

Amorphous Silicon Carbide Platform for Next Generation Penetrating Neural Interface Designs

Published in Micromachines, 2018

Following the development work we tested more geometries of electrode arrays and included additional case studies in rodents. Read more

Recommended citation: Deku F, Frewin C, Stiller A, Cohen Y, Aqeel S, Joshi-Imre A, Black B, Gardner TJ, Pancrazio JJ, and Cogan SF (2018) "Amorphous Silicon Carbide Platform for Next Generation Penetrating Neural Interface Designs". Micromachines, 9(10), 480. https://www.mdpi.com/2072-666X/9/10/480

Calcium imaging in canary (serinus canaria) HVC reveals latent states supporting behavioral sequencing with long range history dependence

Published in CCNeuro, 2018

We observe that the prior state or past syllable information is revealed in calcium activity during a fixed sequence of four canary phrases – showing that Ca2+ signals do not just reflect the current state or current transition. These properties, as well as the existence of neurons with calcium activity, locked to the same phrase type in a subset of phrase sequence histories, may be a signature of new forms of phrase-level hidden states in HVC that, with further investigation, will allow us to refine models of syntax control for species that sing complex songs. Read more

Recommended citation: Cohen Y, Shen J, Semu D, Otchy TM & Gardner TJ (2018) "Calcium imaging in canary (serinus canaria) HVC reveals latent states supporting behavioral sequencing with long range history dependence". in 2018 Conference on Cognitive Computational Neuroscience. doi:10.32470/CCN.2018.1133-0. https://ccneuro.org/2018/proceedings/1133.pdf

Amorphous silicon carbide ultramicroelectrode arrays for neural stimulation and recording

Published in Journal of neural engineering, 2018

Foreign body response to indwelling cortical microelectrodes limits the reliability of neural stimulation and recording, particularly for extended chronic applications in behaving animals. In collaboration with Stuart Cogan's lab at UT Dallas, we developed microelectrode arrays based on amorphous silicon carbide, providing chronic stability and employing semiconductor manufacturing processes to create arrays with small shank dimensions. My role in the project was to design electrode geometries, to test their electrochemical properties ex-vivo, and to test them by acute and chronic in-vivo recording in zebra finches. Read more

Recommended citation: Deku F, Cohen Y, Joshi-Imre A, Kanneganti A, Gardner TJ, and Cogan SF (2018) "Amorphous Silicon Carbide Ultramicroelectrode Arrays for Neural Stimulation and Recording". J. Neural Eng. 15, 016007. https://iopscience.iop.org/article/10.1088/1741-2552/aa8f8b

High-order feature-based mixture models of classification learning predict individual learning curves and enable personalized teaching

Published in PNAS, 2013

Concept-based classification learning tasks are commonly used to explore learning strategies in humans. The universal and individual traits of learning such tasks reflect our cognitive abilities and have been of interest both psychophysically and clinically. I designed and performed psychophysical experiments in which subjects learned to classify binary sequences according to deterministic rules of different complexity. To capture the wide variety of behavior, I developed reinforcement learning models using a mixture of stimulus features and a gradient based learning rule. Fitting models to individuals revealed the importance of their priors, their use of high order features, and suggested that the dynamics may take a very simple form. To validate the models, I demonstrated their ability to predict future behavior and support personally optimized guided learning. Read more

Recommended citation: Cohen Y, Schneidman E (2013) "High-order feature-based mixture models of classi cation learning predict individual learning curves and enable personalized teaching". Proc Natl Acad Sci USA 110:684689. https://www.pnas.org/content/110/2/684