Computational Neuroscience :: Syllabus

Term:	Fall 2015
Instructor:	Daniel Meliza (cdm8j)
Class times:	Fr 9:00-11:30A, Gilmer 225
Collab site:	PSYC 7559 Comp Neurosci
Office Hours:	by appointment
Last revised:	7/2015

What is this course about?

In order to survive and reproduce in a complex and highly variable physical world, animals have to extract information from sensory inputs, make decisions about how to allocate their resources, and perform tightly choreographed sequences of motor behaviors. Brains have evolved to perform these computations reliably, flexibly, and efficiently. Our goal is to understand how these computations are implemented by circuits of neurons.

Course objectives

Our work in this course will focus on methods for quantifying, visualizing, and predicting neural activity. At the end of this course, you will be comfortable processing different kinds of neural and behavioral data, and analyzing these data using a broad range of phenomenological and biophysical models. Specifically, you will be able to:

• Identify the kinds of data obtained from different experimental techniques in systems and circuit neuroscience
• Describe and visualize the statistics of neural activity
• Use linear models to estimate sensory receptive fields and predict responses to novel stimuli
• Use decoding models to quantify information carried in single neurons and populations of neurons
• Use dynamical systems models to represent the biophysics of neurons and circuits
• Assess the advantages and disadvantages of different frameworks and levels of analysis for specific kinds of data and questions

Meeting the objectives

This course uses readings, short lectures, in-class discussion, and computer exercises to introduce major topics in computational neuroscience, survey and evaluate recent techniques, and provide practical experience in processing and analyzing neural data. You’ll receive feedback on your progress towards the objectives through class participation, homework assignments, and a capstone project in which you develop a novel analysis of a publicly-available neural dataset. These activities will contribute to your final score as follows:

In-class discussion and paper presentations (35%)

Each week we will read several research papers, reviews, or textbook chapters related to the topic under discussion. You should carefully read and analyze all the assigned material before class meets each week. Bring your reading materials and notes to class, and be prepared to ask questions of the instructor and each other. The class meeting will include a short lecture by the instructor followed or preceded by a student-led discussion of a paper related to the previous week’s topic.

Take-home analysis assignments (35%)

Each week you will be assigned an exercise in model building, data analysis, or model fitting. Many of these assignments will require computer programming to implement. You may use any language of your choice. Suggested languages with strong support for scientific computing include MATLAB, Python, and R. Assignments are posted each Friday and are due the following Thursday evening. Unless otherwise noted, you must complete the assignments without using third-party toolkits or packages for neural analysis. You may reference the source code, but you must implement the analysis yourself. You may work in teams, but each member of the team must submit a copy of his or her own work.

Capstone project (30%)

Data from published papers is in the public domain, and many experimental neurobiology labs make their recordings available online or will share them on request. Your final assignment is to apply a computational method from the course to a dataset from the literature that has not previously been analyzed using your chosen methodology. You’ll design and implement the analysis, then prepare oral and written reports. If your analysis yields an interesting result, you may be able to submit a manuscript for publication.

There are several steps in this assignment:

• Midway through the semester you’ll propose an analysis plan to the class
• Once the analysis plan is approved, obtain the data
• Prepare an initial report on the organization of the data (and metadata, if any). If necessary, contact the source of the data to obtain additional information needed for your analysis.
• Prepare an oral progress report
• Submit your analysis code and a final written report

What materials will you need?

Introductory readings will be from the following texts:

• Dayan, P and Abbott, LF (2001) Theoretical Neuroscience. MIT Press
• Sterratt D, Graham B, Gillies A, and Willshaw D (2011). Principles of Computational Modelling in Neuroscience. Cambridge University Press.

These books will be placed on course reserve, or are available electronically through the course Collab page or VIRGO.

Additional readings will be posted on the course Collab page.

In addition, the following texts are recommended as primers on relevant math and programming topics

• Spence, Insel, and Friedberg, Elementary Linear Algebra. Prentice Hall.
• Eugene Izhikevich, Dynamical Systems in Neuroscience. MIT Press.
• Hirsch, Smale, and Devaney, Differential Equations, Dynamical Systems, and an Introduction to Chaos. Springer.

What is the class schedule?

This schedule will be updated with readings and presentations as the semester progresses. You may propose alternate topics or ask to spend more time on existing topics.

See the Assignments Page for links to weekly assignments.

Date	Topic	Background Reading	Paper
8/28	Neural data, descriptive analysis	DA1
9/4	Linear systems analysis	DA2	Kilgard et al. J Neurophys 2001
9/11	Receptive fields	Schwartz et al. 2009	Haag and Borst, J Neurosci 1997
9/18	Receptive fields and non-white stimuli	Theunissen et al. 2001	Touryan et al, J Neurosci 2002
9/25	Dynamical systems and neuron biophysics	S2	Atencio et al, Neuron 2008
10/2	Data assimilation (Henry Abarbanel)	Toth et al. 2011	Meliza et al, Biol Cybern 2014
		Knowlton et al. 2014
10/9	Ion channels	S - 5.1-5.4, Appendix B
	Present analysis plans
10/16	No class (Society for Neuroscience Conference)
10/23	Phenomenological neuron models	S8	Drion et al, PNAS 2015
		Izhikevich, Ch 3 (optional)
	Initial data report due
10/30	Decoding models and information theory	DA: 3.1-3.2, 4	Lynch and Houghton, Front Neuroinformat 2015
11/6	Parametric decoding models	Bates, Ch 1	Osborne et al, J Neurosci 2004
			Rose and Hindmarsh, Proc B 1989
11/13	Population codes	DA: 3.3-3.5, 7.1-7.3	Jeanne et al, Neuron 2013
11/20	Circuit models	DA: rest of Ch 7 OR S: 9	Britten et al, Vis Neurosci 1996
11/27	No class (Thanksgiving break)
12/4	Spike sorting
12/15	Final analysis report due