Program & Abstract

(Parallel Session)

Introduction to numerical methods for ordinary and partial differential equations

This tutorial introduces the basic concepts of differential equations and how to solve them, or simulate their behaviors in time, using a computer. Key concepts like eigenvalues and stability are explained while solving simple differential equation using MATLAB programming language. Some examples of Hodgkin-Huxley type neuron models and cable equations are also introduced.

Introduction to reinforcement learning and Bayesian inference

The aim of this tutorial is to present the theoretical cores for modeling animal/human action and perception. In the first half of the tutorial, we will focus on "reinforcement learning", which is a theoretical framework for an adaptive agent to learn behaviors from exploratory actions and resulting reward or punishment. Reinforcement learning has played an essential role of understanding the neural circuit and neurochemical systems behind adaptive action learning, most notably the basal ganglia and the dopamine system. In the second half, we will familiarize ourselves with the framework of Bayesian inference, which is critical in understanding the process of perception from noisy, incomplete observations.

• Doya K: Reinforcement learning: Computational theory and biological mechanisms. HFSP Journal, 1(1), 30-40 (2007)
• Free on-line access: http://dx.doi.org/10.2976/1.2732246
• Doya K, Ishii S: A probability primer. In Doya K, Ishii S, Pouget A, Rao RPN eds. Bayesian Brain: Probabilistic Approaches to Neural Coding, pp. 3-13. MIT Press (2007).
• Free on-line access: http://mitpress.mit.edu/catalog/item/default.asp?ttype=2&tid=11106

Gordon Arbuthnott

An exploration of real neuronal systems

A hero of my graduate years Ben DeL. Burns wrote a book called ‘The Uncertain Nervous System’. He claimed he loved writing books because you avoided peer review and could say what you thought.  Most of my contemporaries thought he had lost his mind – nothing was ‘uncertain’ about the nervous system – just unknown! As we consider what you should expect from the real nervous systems you model his insight is worth keeping in mind!

He was Tim Bliss’ Ph.D. supervisor but when Tim discovered LTP with Lomo; Ben said – “we have been looking for the wrong thing all this time!  You don’t learn by co-incidences you learn by consequences!” He (Burns) and Alison Webb began a long series of studies on the measurement of behavior in cats and, for a time, abandoned electrophysiology as he had done it before.  What does all this have to do with computational neuroscience?  Mainly, it has to do with the importance of the right question, but it also has to do with the best strategy in looking for answers.

After some trouble I convinced an editor to accept a purely methodological chapter in a neuroscience text because “the only facts are the methods and their consequences.” The rest is just speculation – important insightful interpretation (speculation?) is exactly what I think computation neuroscience can give us, but it is even more useful if it can be tested by some means against real nervous systems in live animals.

Modeling biochemical reactions, diffusion and reaction-diffusion systems

In my first talk I will use calcium dynamics modeling as a way to introduce deterministic solution methods for reaction-diffusion systems. The talk covers exponentially decaying calcium pools, diffusion, calcium buffers and buffered diffusion, and calcium pumps and exchangers. I will describe properties of buffered diffusion systems and ways to characterize them experimentally. Finally I will compare the different modeling approaches.

In the second talk I will turn towards stochastic reaction-diffusion modeling. Two methods will be described: Gillespie's Stochastic Simulation algorithm extended to simulate diffusion, and particle-based methods. I will briefly describe the STEPS software. I will then describe two applications: stochastic reaction modeling of LTD induction in Purkinje cells and stochastic diffusion modeling of anomalous diffusion in spiny dendrites.

Introduction to modeling neurons and networks

In the first talk I will discuss methods to model morphologically detailed neurons. I will briefly introduce cable-theory, the mathematical description of current flow in dendrites. By discretizing the cable equation we come to compartmental modeling, the standard method to simulate morphologically detailed models of neurons. I will discuss the challenges in fitting compartmental models to experimental data with an emphasis on active properties. The talk will finish with a brief overview of dendritic properties predicted by cable theory and experimental data confirming these predictions.

The second talk will briefly introduce network modeling. I will introduce simpler neuron models like integrate-and-fire neurons and then move on to modeling synaptic currents. I will wrap up with an overview of network connectivity.

• U.S. Bhalla and S. Wils: Reaction-diffusion modeling. In Computational Modeling Methods for Neuroscientists, E. De Schutter ed., MIT Press, Boston. 61–92 (2009)
• E. De Schutter: Modeling intracellular calcium dynamics. In Computational Modeling Methods for Neuroscientists, E. De Schutter ed., MIT Press, Boston. 61–92 (2009)
• G. Antunes and E. De Schutter: A stochastic signaling network mediates the probabilistic induction of cerebellar long-term depression. Journal of Neuroscience 32: 9288–9300. (2012).
• F. Santamaria, S. Wils, E. De Schutter and G.J. Augustine: Anomalous diffusion in Purkinje cell dendrites caused by dendritic spines. Neuron 52: 635–648 (2006).
• Several chapters in Computational Modeling Methods for Neuroscientists, E. De Schutter ed., MIT Press, Boston (2009).
• V. Steuber et al.: Cerebellar LTD and pattern recognition by Purkinje cells. Neuron 54: 121–136 (2007).

Bernd Kuhn 

Optical Imaging Methods

Functional optical imaging has becomes one of the key techniques in neuroscience. I will introduce the most important methods and explain what we can learn from them but also their limitations.

Ion channel physiology and the Hodgkin-Huxley model of neuronal activity

I will give an introduction on ion channels and specifically focus on voltage-gated channels and their dynamics in response to membrane voltage. A combination of different voltage-gated channels is used by neurons to generate fast (about 1 ms) voltage changes. I will talk about the first model by Hodgkin and Huxley describing this electrical activity. I will also talk about more recent additions or fine-tuning of this time-honored model.

• Helmchen, Konnerth: Imaging in Neuroscience, 2011
• Yuste, Lanni, Konnerth: Imaging Neurons, 2000
• Johnston and Wu: Foundation of cellular neurophysiology, MIT press

Upinder Bhalla

Computing with Chemistry (and also electricity as a minor side effect)

The computations of the brain are almost exclusively thought of in electrical terms. This viewpoint misses out on some of the most elaborate and rapid computations in the brain, as well as the most long-lasting ones. Subcellular computations range from the possibly dull but important job of  keeping neurons alive, to building the cells and networks, and most of the key steps in synaptic transmission and plasticity. I will introduce chemical signaling and the kinds of computation that it supports. I'll discuss the kinds of models that represent chemical signaling in different contexts, and introduce how these models are built. I'll then bring in the interface with electrical signaling, and stress how neuronal computation is really a continuum in which all these forms of computation are essential.

Gregory Stephens

An introduction to dynamical systems: from neural activity to organism-scale behavior

My lecture will consist of two parts: an introduction to dynamical systems focused in particular on the power of quantitative analysis and a novel, quantitative approach towards understanding the wiggling, motile behavior of C. elegans.  We apply a low-dimensional yet complete representation of body shape (eigenworms) to construct a principled parameterization of the 2D movements.  Despite it’s simplicity, we show that a linear dynamical model of the eigenworm projections captures long-range temporal correlations and reveals two periodic dynamics, the primary body wave and and an oscillation between the head and body curvature which underlies arcs in the centroid trajectory.  We parameterize the movement phenospace by constructing dynamical systems locally in time and show that variation within this space is remarkably restrained; with increasing window size, a single behavioral mode dominates the variance and represents the coupled control of speed and turning.  The distribution of this primary mode is bimodal, suggesting a correspondence to roaming and dwelling states.  Finally, we apply our behavioral parameterization to show that the worm’s response to a strong impulsive heat shock includes a Hopf-like bifurcation corresponding to an early-time growth of the amplitude of the crawling wave.

Yael Niv

Advanced topics in neural reinforcement learning

Following up on the lecture from Prof. Doya, we will discuss trial by trial model fitting of reinforcement learning models, model comparison, and some recent neural findings using these methods. We will continue to discuss multiple decision making systems in the brain (model free and model based reinforcement learning), and in the third hour, discuss Bayesian inference and partially observable Markov Decision Processes (POMDPs). 

• Niv & Schoenbaum (2008) - Dialogues on prediction errors, Trends Cog Sci 12(7):265-272
• Niv (2009) - Reinforcement learning in the brain - J Math Psychol 53(3):139-154 
• Daw (2011) - Trial by trial data analysis using computational models, in: Delgado, Phelps & Robbins (eds.) Decision Making, Affect, and Learning: Attention and Performance XXIII, Oxford University Press (PDF can be provided)

Ivan Soltesz

Data-driven large-scale modeling of hippocampal networks

The close integration of experimental findings with large-scale, data-driven computational simulations of neuronal networks offers a powerful tool towards the identification of key circuit parameters that control behavior under normal conditions and in various neurological and psychiatric disorders. To this end, we have been developing realistic microcircuit-based network models of the control and injured hippocampus in order to investigate questions related to normal hippocampal microcircuit function and the mechanistic bases of epilepsy. We will discuss the conceptual framework and biological basis of full-scale model development and show specific applications, including computational and experimental results concerning model validation, cell type specific hippocampal chronocircuit properties and the role of hub neurons. The talk will highlight the predictive and analytic power of freely shared, highly realistic, large-scale computational models in understanding normal and abnormal circuit functions.

• Morgan, R.J. & Soltesz, I. Non-random connectivity of the epileptic dentate gyrus predicts a major role for neuronal hubs in seizures. Proceedings of the National Academy of Sciences, USA (2008) 105(16): 6179-6184. 
• Schneider, C., Bezaire, M., & Soltesz, I. Toward a full-scale computational model of the rat dentate gyrus. Frontiers in Neural Circuits. (2012)  6:83.
• Bezaire, M.J., Soltesz, I. Quantitative assessment of CA1 local circuits: Knowledge base for interneuron-pyramidal cell connectivity. Hippocampus (2013) 23: 751-785.

Jason Kerr

Imaging neuronal activity in vivo: how do we make sure we get the right answer and avoid delusion?

This lecture will be divided into 4 sections:

1) Introduction of imaging techniques available and what they can provide                                                    

2) Physics of an image: how do we know what we see is real?

3) Application of imaging tools to real neuroscience problems

4) What’s coming up in the future?

Claudia Clopath

Single neuron modeling

I will cover the different models of Integrate-and-Fire neurons and their properties. 

Tony Prescott

Embodied Computational Neuroscience for Sensorimotor and Social Cognition

Our brains have evolved to guide our interaction with the world about us, both physical and social. This lecture addresses the complementary topics of "embodied computation in the brain" and "robots as a tool for computational neuroscience.” Embodiment has multiple aspects, most notably the notion that sensory and cognitive systems evolved to process data in a form the body needs to execute various tasks, and that some computations are simplified by exploiting the body’s own dynamics. Robotics can be used to test the sufficiency of a brain model to reproduce a target behavior in a way that complements computer simulation.  The lecture will be illustrated by example research on biomimetic and brain-based robots including that from my own laboratory on mammal- and human-like robots.  The aim will be to show how experimental, computational and robotic approaches can operate together to advance our understanding of sensorimotor and social cognition in behaving systems.

• Prescott, T. J., Pearson, M., Mitchinson, B., Sullivan, J. C. W., Pipe, A. G. (2009). Whisking with robots: From rat vibrissae to biomimetic technology for active touch. IEEE Robotics and Automation Magazine, 16(3), 42-50. 
• Gurney, K., Prescott, T. J., Wickens, J., and Redgrave, P. (2004). Computational models of the basal ganglia: from membranes to robots Trends in Neurosciences, 27, 453–459. 

Javier Medina

Can we build an artificial cerebellum for motor control and adaptation?

If there is one circuit in the brain that has captured the imagination of neuroscientists, mathematicians and engineers alike, that is the cerebellum. In the last 50 years, we have made a tremendous amount of progress in elucidating the connectivity, synaptic features and intrinsic properties of neurons in the cerebellum. But are we any closer to understanding how the cerebellar circuit actually works? In my lecture, I will first review computational theories of cerebellar function which are based on the idea that the cerebellum is a neural machine for adapting our movements and keeping them finely tuned. Then, I will describe a number of studies of eyeblink conditioning, a prototypical example of cerebellar-dependent motor learning. These studies provide an excellent opportunity to evaluate some of the key predictions made by cerebellar learning theories, particularly with regards to the neural mechanisms and computations mediating motor timing and error processing. The format of the lecture will be interactive. My goal is not to present a bunch of facts one-by-one, or even to tell a complete story. Instead, I hope to raise a number of questions and open the floor for discussion. At the end, I’d like to ask: do you think we understand the cerebellum well enough to build one?

• Marr, D. A theory of cerebellar cortex. J Physiol 202, 437-70 (1969).
• Medina, J. F. & Mauk, M. D. Computer simulation of cerebellar information processing. Nat Neurosci. 3 Suppl, 1205-1211 (2000).
• Christian, K. M. & Thompson, R. F. Neural substrates of eyeblink conditioning: acquisition and retention. Learn Mem 10, 427-55 (2003).

Hiroyuki Nakahara

Neural computation and social decision-making

A fundamental challenge in social cognition is elucidating how one learns and predicts the mind of others, and what the underlying neural computations and mechanisms are. We approach this challenge by using computational frameworks to map behavioral-level complexity and related neural systems, i.e., extending reinforcement learning theory into the realm of social cognition and combining human fMRI with modeling. In my talk, the first part will cover the basics for using model-based analysis in human fMRI experiments and the second part will describe our research on social decision-making. I will discuss that these efforts will open up new avenues of inquiry, such as identifying critical differences in neural computations for understanding the intentions and behavior of others between people with and without mental disorders.

• Nakahara H. (2014) Multiplexing signals in reinforcement learning with internal models and dopamine. Current Opinion in Neurobiology. 25: 123-129
• Suzuki S, Harasawa H, Ueno K, Gardner JL, Ichinohe N, Haruno M, Cheng K, Nakahara H. (2012) Learning to simulate others’ decisions. Neuron. 74: 1125-1137.
• Nakahara H, Hikosaka O. (2012) Learning to represent reward structure: A key to adapting to complex environments. Neuroscience Research. 74(3-4): 177-183.

Greg Stuart

Single neuron computation

In my presentation I will describe how neurons integrate synaptic inputs. As synaptic inputs are made primarily onto the dendritic tree of neurons the first part of my presentation with focus on the properties of neuronal dendrites and how these properties influence synaptic integration. The second part of my presentation will be on action potential generation, with a focus on the properties of the axon initial segment. If there is time I will finish off by describing some new results we have obtained looking at synaptic integration in vivo during  processing of binocular visual information.

• Gulledge, A.T., Kampa, B.M. and Stuart, G.J. (2005). Synaptic integration in dendritic trees. J. Neurobiol. 64, 75-90.
• Kole, M.H. and Stuart, G.J. (2012) Signal processing in the axon initial segment. Neuron, 73, 235-247.

Taro Toyoizumi

Brain State Modulation by Closed Loop Sensory Feedback

Animals use passive and active sensing to navigate in their environment. Unlike passive sensing, during active sensing the brain processes sensory information resulting from self-generated action.

Neuronal variability and sensitivity reduce during active sensing but it is unknown how brain-environment interactions are involved. Here, we present a theory describing dynamic modulation of neuronal variability and sensitivity by closed-loop sensory feedback from the environment. Our theory on a novel form of neuronal gain control was validated in several species and sensory modalities. These results demonstrate that closed-loop brain-environment interactions are crucial for neuronal dynamics and pave the way for new neurofeedback technologies for therapeutic and behavioral enhancement.