Ruben A. Tikidji-Hamburyan, Ph.D.

Read about my research in

Physics News Highlights

by American Institute of Physics

RESEARCH EXPERIENCE and BACKGROUND

Trained as a physicist, in 2000 I made a switch from experimental Laser Physics to Theoretical and Computational Neuroscience, by carrying out a Doctorate in Neurocybernetics at the A.B.Kogan Research Institute for Neurocybernetics (KRINC), Southern Federal University (SFedU) in Russia. Since then, my research has been focused on different aspects of information processing, information coding, signal detection and network formation in the nervous system, through modeling the nervous tissue at different levels of its organization.

I received my Ph.D. in Computer Science (Technical Science) under the mentorship of Prof. Boris M. Valdimirski (Mathematics and Computer Science) and Dr. Lubov N. Podladchikova (Neuroscience). My thesis consisted of developing a simplified mathematical model for a single neuron, developing a biologically plausible learning rule and reproducing spike activity of thalamic relay cells and inhibitory neurons of reticular formation in both waking and non-REM sleep.

After graduation, I developed my career in collaboration with Dr. Podladchikova at the Laboratory of Neuroinformatics of Sensory and Motor Systems, KRINC SFedU. The goal of my work was to understand information processing at the level of small and large neural assemblies. I worked on several projects, including developing detailed biophysical models and simulations of single neurons, small local networks, large-scale networks as well as phenomenological numerical models and theoretical / analytical models in the field of Neuroscience. The majority of these results were published in Russian peer-reviewed journals.

In 2007, I was awarded a Principal Investigator position at KRINC, SFedU (equal to a tenured position) and became the head of a laboratory for Detailed Analysis and Modeling of Neurons and Neural Networks at the same institution. In my lab, we used networks of single- and multi-compartmental neuron models to study the binaural signal processing in brainstem auditory nuclei, the nonlinearity of activity-dependent synaptic plasticity based on kinetics of calmodulin-calcium complexes, and the stability of gamma oscillations in recurrent networks, among other projects.

To increase my international experience and to reach a higher level of scientific research, in 2011 I accepted a postdoctoral fellow position under the mentorship of Prof. Carmen Canvier at the Health Sciences Center of Louisiana State University (LSUHSC), New Orleans. In 2015 I accepted a senior postdoctoral position under the mentorship of Prof. Tarek El-Ghazawi at the Institute of Massively Parallel Applications and Computing Technologies, School of Engineering and Applied Science, George Washington University (IMPACT, SEAS, GWU), Washington, D.C. In 2017, I returned to Prof. Canavier’s laboratory as a senior postdoctoral researcher. Thus, I have completed more than 8 years as a post-doctoral researcher in the USA.

SELECTED PROJECTS

Brain network formation in a critical developmental period

Connectivity patterns and neuron types are the critical components of information processing in brain networks. They define both the network activity and the repertoire of available network dynamics. On the other hand, these connectivity patterns depend upon both network activity and experience. This implies that the network activity and connectivity have to be in synergy. However, the mechanisms by which this synergy is achieved are not clear yet.

Although there is an extensive body of research of activity-dependent refinement in the feed-forward networks, in my research in collaboration with Dr. Jason W. Triplett from Center for Neuroscience Research, Children’s National Health System, we developed the first in the field framework for activity-dependent feedback formation. Previously, it was shown that retinal inputs instruct the alignment of V1 inputs in a manner dependent on spontaneous correlated activity, co-called retinal waves (Triplett et al., 20091). However, the mechanisms underlying activity-dependent instruction of map alignment remain unclear. In this project, I developed two novel models of visual map alignment in the superior colliculus (SC) - a midbrain nucleus that receives input from the retina - and the primary visual cortex (V1) to regulate goal-directed eye-movements.

Each of my models utilizes a distinct activity-dependent strategy. In both models, alignment is achieved via a Hebbian “fire together, wire together” mechanism. In the first model, V1 axons make a synapse with SC neurons sharing common retinal-driven firing patterns. In the second model, V1 axons can contribute to the firing of SC neurons, and thus influence the correlative activity of post-synaptic partners. The models accurately reflect the organization of the visual map alignment observed in wild type, transgenic and combination mutant mouse models. However, each model exhibits distinct behaviors in a test similar to a bifurcation analysis of dynamical systems. These findings suggest that the models not only can be differentiated empirically, but also allow us to make further predictions about the nature of visual map alignment depending on which activity-dependent mechanism is used. This work was published in the PLoS Computational Biology in 2016.

Signal Detection and Information Coding in Brainstem Auditory Nuclei.

The overall aim of this research is to understand signal detection and information decoding at the peripheral level of sensory information processing and to map this understanding to different psychoacoustical and psychophysical phenomena.

One of the accomplishments of my laboratory in Russia was in the modeling of Interaural Time Difference (ITD) processing, where we discovered a novel mechanism of highly accurate ITD detection by a population of sluggish, noisy neurons which receive excitatory and inhibitory (EI) binaural projections. The theoretical bases and main results of this study were published in Physical Review Letters2. The key finding suggests that a relatively large population of EI cells can detect time differences in input signals that are significantly shorter than any time constant in the system. Moreover, noise and heterogeneity play an important role in this mechanism, improving accuracy of detection, a phenomenon which is similar to stochastic resonance, but in the population domain.

We continued to develop the work by hypothesizing that ITD can be detected, coded and processed differently for different sounds. To check this hypothesis, my former PhD student Viacheslav Vasilkov and I have been working on a large-scale biophysical model of brainstem binaural auditory nuclei. This model incorporates an auditory nerve representation (Zilany & Carney, 2010, 2014) and biophysical bushy-cell model of ventral cochlear nuclei (VCN, Rothman & Manis, 2003; Xie & Manis, 2013). We extended the single EI neuron model and calibrated it to replicate original lateral superior olive (LSO), principal cell recordings from Wu & Kelly (1991). Currently, the model consists of more than 15,000 neurons with several million connections. The project webpage http://absm.r-a-r.org is currently under a reconstruction and will be available soon. The latest version of source code developed for this project is hosted at GitHub: https://github.com/rat-h/auditory-brainstem-model.

Computations in Oscillatory Networks

One of the most intriguing implications of our results from modeling of ITD detection is the possibility that inhibitory/excitatory networks (widely present in all parts of the nervous system) are universal processors for information encoded by temporal disparity in arriving spikes. From this perspective, oscillations, and more specifically synchronization, in recurrent networks provide an ideal framework for the multiplexing and enhancement of such calculations. I have been actively researching this field with Prof. Canavier at LSUHSC, using the Coupled Oscillators Theory as a basis. In contrast to the research conducted at my laboratory in Russia, my research at LSUHSC takes a more theoretical approach. I focus on the study of processes in networks of oscillators, for which the Phase Resetting Curve (PRC) approach is widely used. The section below discusses three projects in more detail

First, I developed the theory of cross-frequency synchronization in a feed-forward network. This result is critical for understanding to what extent a population of high-frequency oscillators may be synchronized or phase-locked by an external low-frequency input such as theta-gamma nesting. Some predictions of this theory were verified in a specially designed experiment carried out in Dr. Sonia Gasparini’s laboratory at LSUHSC. These theoretical results and experimental observations were published in the special issue on “Computations in Oscillating Neuronal Networks” in the journal of Network: Computation in Neural Systems in 2014

Second, I developed a theoretical framework to analyze the effect of inhibitory feed-back on oscillation variability. This framework can provide insight into optimal feedback schemes for minimizing variability to increase reliability, or for maximizing variability to increase the flexibility of network oscillators. The framework consisted of several models whose predictions were highly accurate. For example, the stationary distributions predict an experimentally observed mean and standard deviation of network period with an error 1.69% and 16.63% correspondingly. These theoretical results and experimental observations were published in the Journal of Neurophysiology in 2015.

Finally, another of my projects has uncovered a new mechanism of precise and robust synchronization in the gamma band. This finding shows a critical role of type 2 neuron excitability in gamma rhythm robustness and implies significant extension for Inter-Neuron Gamma (ING) mechanisms. The feasibility of the proposed mechanism was verified experimentally, in direct model-driven experiments conducted by Prof. John A. White at the University of Utah. The theoretical part of this work suggests a way to distinguish between this mechanism and other proposed mechanisms. This work was published in the Journal of Neuroscience in 2015. An extension of this model shows that the frequency of resonant gamma oscillations depends on delays and synaptic strength. These results were published in the Physical Review E in 2017 and in the Journal of Neurophysiology in 2019.

FUTURE PLANS

I will focus on questions concerning the coding, transmission, storage, retrieval, and processing of information in single neurons and neural networks, as well as role of information in network formation.

Network Formation in Development. A critical assumption of my models of cortical input alignment in SC is that the cortical neurons just repeat retinal waves, which is not completely correct. The puzzling part of the circuit is the thalamocortical loop which has a functional recurrent coupling during this stage of development. It is known from experimental observations3 that the thalamocortical loop can exhibit oscillations in response to retinal input. Moreover, the retinal input may or may not trigger these oscillations, which results in slow-activity transients with associated rapid oscillations in the LFP or short bursts with non-associated LFP oscillation, correspondingly4. On the other hand it is not clear how these oscillations interact with maturation of Hebbian plasticity5. As a first step, I plan to develop a model of gamma rhythm formation during thalamocortical circuit development. There are at least two critical questions which this model will be able to address. The first is to identify which type of oscillations the thalamocortical loop can maintain at different stages of its development and the conditions for such oscillations. The second crucial question is to understand the information function of these oscillations, and how they contribute to corticocortical and corticosubcortical network development. In a big picture, my results suggested that there may be spatial bistability in overall corticocortical, corticosubcortical connectivity, which is separated from other possible connectivity patterns by unstable manifolds. Revealing biophysical mechanisms of such bi-(multi-)stability is the ultimate goal of my efforts in this project. This work will be developed through collaboration with Dr. Colonnese from the Neuroscience Institute, GWU and Dr. Triplett from Center for Neuroscience Research, Children’s National Health System.

Computations in oscillatory networks. The focus of this research is the study of transient synchronization events (for example gamma-bursting during theta-gamma nesting) examining the functions of transient gamma synchronization, rather than the mechanisms of synchronization per se. I plan to expand my current model of PIR-PING networks, developed with Prof. Canavier6, to study possible information processing during transient gamma synchronization. The line of experimental evidence in rodents7 and humans8 suggest that strong phase-amplitude coupling (PAC) during theta-nested gamma oscillations is required for successful encoding and retrieval of episodic memory. However, it is not clear what are the mechanisms of such memory. I will examine several suggested mechanisms of information processing in PING network under conditions of PAC transient synchronization. Two major questions will be in focus of this work: the first, what kind of neural code can benefit from transient synchronization in the gamma range and the second, what possible mechanisms of information encoding and retrieval can be based upon transient synchronization. The further extension of these questions may cover also mechanisms of information processing (unrelated with memory formation/readings), such as detection of spike patterns, pattern binding or differentiation, as well as inference, which can be performed based upon transient synchronization in an excitatory/inhibitory network.

Signal Detection and Information Coding in Brainstem Auditory Nuclei. Our preliminary results indicate that auditory subsystems may use an optimal mixture of different neural codes to represent the spatial position of a sound source. I plan to submit another theoretical article to open a discussion on the role of noise and heterogeneity in the mechanisms of sound source localization and the robustness of different mechanisms. The role of heterogeneity has been discussed intensively in the literature9, but the question of an optimal level of noise and heterogeneity has not been studied well yet. At the same time, I plan to finish developing a large-scale biophysical model of auditory nuclei. The model will be expanded in several ways. First, the inhibitory inputs to spherical bushy cells in the anteroventral division of the cochlear nucleus (AVCN) will be introduced to accommodate recent experimental observations10. Second, LSO neurons will be recalibrated against recently obtained data on click trains which revealed the critical role of h-channel in spatial coding11. Third, the network of medial superior olive (MSO) with bilateral excitatory and inhibitory inputs and the networks of medial/lateral nucleus of the trapezoid body (MNTB/LNTB) will be implemented in addition to the LSO and AVCN networks. Finally, we will seek the original inter-ear recordings from studies of cat’s head-related transfer function (HRTF), to feed our model with both real recorded stimuli and artificially stimuli constructed from HRTF. This model will be used to test existing and generate new hypotheses of possible mechanisms of sound source localization in mammals. The overall goal of this work is a better understanding of signal processing and information decoding in the auditory system.


  1. Triplett, J.W., Owens, M.T., Yamada, J., Lemke, G., Cang, J., Stryker, M.P., and Feldheim, D.A. (2009). Retinal input instructs alignment of visual topographic maps. Cell 139, 175-185. ↩︎

  2. This paper was highlighted in Synopsis: ‘Telling Left From Right’ by the American Institute of Physics as one of the most interesting publications of the week. ↩︎

  3. Murata, Y., & Colonnese, M. T. (2016). An excitatory cortical feedback loop gates retinal wave transmission in rodent thalamus. Elife5, e18816. ↩︎

  4. Colonnese, M. T. (2014). Rapid developmental emergence of stable depolarization during wakefulness by inhibitory balancing of cortical network excitability. Journal of Neuroscience34(16), 5477-5485. ↩︎

  5. Martens, M. B., Celikel, T., & Tiesinga, P. H. (2015). A developmental switch for hebbian plasticity. PLoS computational biology, 11(7), e1004386. ↩︎

  6. Code is available at ModelDB: https://senselab.med.yale.edu/ModelDB/showmodel.cshtml?model=239177#tabs-1 ↩︎

  7. Lopez-Pigozzi, D., Laurent, F., Brotons-Mas, J. R., Valderrama, M., Valero, M., Fernandez-Lamo, I., ... & de la Prida, L. M. (2016). Altered oscillatory dynamics of CA1 parvalbumin basket cells during theta–gamma rhythmopathies of temporal lobe epilepsy. eNeuro, 3(6). ↩︎

  8. Lega, B., Burke, J., Jacobs, J., & Kahana, M. J. (2014). Slow-theta-to-gamma phase–amplitude coupling in human hippocampus supports the formation of new episodic memories. Cerebral Cortex, 26(1), 268-278. ↩︎

  9. Yuan, C. W., Khouri, L., Grothe, B., & Leibold, C. (2014). Neuronal adaptation translates stimulus gaps into a population code. PloS one, 9(4), e95705; Brette, R. (2012). Computing with neural synchrony. PLoS computational biology, 8(6), e1002561. ↩︎

  10. Keine, C., Rübsamen, R., & Englitz, B. (2016). Inhibition in the auditory brainstem enhances signal representation and regulates gain in complex acoustic environments. Elife, 5, e19295. ↩︎

  11. Beiderbeck, B., Myoga, M. H., Müller, N. I., Callan, A. R., Friauf, E., Grothe, B., & Pecka, M. (2018). Precisely timed inhibition facilitates action potential firing for spatial coding in the auditory brainstem. Nature communications9(1), 1771. ↩︎