Computational Neuroscience FAQ

We use computational neuroscience, which is an interdisciplinary field that powerfully combines the natural biology of the brain with machine learning and artificial intelligence. I operate in this space to creatively design and generate artificial models of the brain and circuit functions. Theorists like myself use mathematics and ground it in biological data to develop these simple models that replicate natural processes and ask if the same principles are used by the biological brain. In this way, computational neuroscience enables critical insights about brain-wide communication within a species and fundamental operating principles across species— all while staying true to the basic biology of the brain. Read on to learn more about what computational neuroscience can do, the background and skills that are helpful, and career trajectories.

  • Many of the brain's most powerful and intriguing features are difficult, if not impossible, to replicate and study in a laboratory setting. Computational neuroscience is the interdisciplinary study of the nervous system's development, structure, physiology, information processing, and cognitive abilities. It is at the intersection of computer science, neuroscience, cognitive psychology, physics, engineering, mathematics, and statistics. We at the lab run experiments that test our artificial models of the brain to gain insight into neurobiological mechanisms.

    Most computational neuroscientists focus on theory rather than conducting experiments in a lab. Computational neuroscientists use mathematical multi-scale models, theoretical analysis, and simulations of neural function from the perspective of molecules, cells, and networks. Importantly, our models are informed by actual biological data and principles, allowing for significant improvements in the study, diagnosis, and treatment of brain diseases. Computational neuroscience unpacks a toolbox filled with processing architectures and theoretical techniques that are flexible and suitable to pursue my questions about how the mind works. To learn more about my endeavors in computational neuroscience and how our research is changing the field, you can explore the press page covering our work.

    Learn more:

    A Brief Introduction to Computational Neuroscience Part 1, Towards Data Science

    Comics about computational neuroscience

  • The tools we use as computational neuroscientists and machine learning researchers overlap, but there is sometimes little overlap in goals. The ML field is interested in artificial systems that imitate intelligent human behavior, or artificial intelligence. Distinctively, the AI field primarily has an engineering goal — for example, to design an artificial system that outperforms the previous artificial system for a specific purpose or application. When you reverse engineer such models, they reveal how an artificial system solves a problem or optimizes its performance subsequently. In my lab, we engineer recurrent neural networks (RNNs) and take advantage of the natural features of this type of model as the substrate to solve the problem of how we learn.

    My intellectual conviction in the confection of mathematical hypotheses of brain function is an example of the role of computational neuroscientists. Biologists and neuroscientists have a different goal from computer engineers: to understand how the biological system solves various problems. Here the goal is to discover mechanisms and to use models to make predictions that can improve data collection and test different hypotheses about brain function. So overall, the goal is to understand how a biological brain functions in health and disease. I can artificially design a simplified model of a behaving organism or a pair of neurons and run experiments using collected biological data to extrapolate theories of the underlying neurobiology.

    Nature may have evolved to solve a problem optimally, but this is different from an artificial agent that can solve a problem perfectly. As neuroscientists, we're interested in the biologically plausible way to solve the problem using only the messy and imperfect machinery biology has access to, such as neurons and synapses. Our research takes physical measurements like neuronal activation, synaptic architecture, and behavioral decision-making to create models that can be observed and replicated by experimentalists.

    Therefore, computational neuroscience and machine learning make good partners but remain in slightly different spaces because our goals are often different. Collaborations between these fields have the potential to be incredibly powerful. We are already seeing evidence of this, and my prediction is that this trend will only grow in the next few years. You can check out my projects to learn more about our collaborations with experimentalists combining computational neuroscience and machine learning.

  • Neuroscience and computer science are closely connected in many ways and can work together to provide insight into the multitude of factors that influence human behavior. For once, both fields are exponentially growing these days. We know very little about how deep the rabbit hole goes both in the search for a unifying theory of how the mind works and in the acceleration of artificial processing systems. The intersection of these fields is computational neuroscience, and many computational neuroscience training programs are interdisciplinary, allowing students to study neurobiology and computer science. Please read about our current funding to learn how the Rajan Lab uses neuroscience and computer science and the conceptual framework inspiring our latest studies.

  • My research is always inclusive of the material sciences because my ideas include theories from physics, chemistry, and biology data, among others. Computational neuroscience intersects all the sciences, a feature that accelerates and facilitates discovery for my team. Therefore, computational neuroscientists can come from a diverse set of academic backgrounds, from data analysts and biologists to pure mathematicians and computer scientists. I myself attained a physics background before starting graduate school in a neuroscience program.

    No matter what, computational neuroscientists should have a strong background in - or a strong desire to learn more of - biology, academic writing, coding, and mathematics.

    An undergraduate degree is required to pursue further study as a computational neuroscientist. Majors such as bioengineering, computer science, neuroscience, and physics are common choices to prepare for graduate studies in computational neuroscience. If you are interested in going to graduate school, undergraduate research experience is an excellent way to gain exposure to many elements of computational neuroscience.

    Coding is also a fundamental part of computational neuroscience, so I encourage learning one or more languages that computational neuroscientists use. For example, the Rajan Lab BRAINY crew has varied career trajectories before and after graduate school.

  • A general understanding of computational neuroscience goes a long way. People from different backgrounds, including non-scientific groups, have the capacity and enthusiasm to acquire a basic knowledge of how to apply coding tools to answer scientific questions with biological data. Please read below to learn about the most popular computational languages in neuroscience, and check out resources for learning code I created intending to make computational neuroscience accessible to all. You can follow along with my COSYNE 2021 (Computational and Systems Neuroscience) lecture series on the foundational elements of RNNs and their applications to neuroscience. Finally, find fun coding exercises, problem sets, and other educational resources in my GitHub.

    A brief rundown of computational languages:

    Python → Python is a high-level, general-purpose, object-oriented programming language. Its design philosophy emphasizes code readability. Researchers can use Python to write clear, logical code for small- and large-scale projects. Python is open source.

    MATLAB → MATLAB is a proprietary, multi-paradigm programming language and numeric computing environment developed by MathWorks. MATLAB allows matrix manipulations, plotting functions and data, implementing algorithms, creating user interfaces, and interfacing with programs written in other languages. One major downside of MATLAB is the high cost of software licenses.

    • Learn MATLAB on EdX

    • Free MATLAB courses on Udemy


    Julia     → Julia is a high-level, high-performance, dynamic programming language. While it is a general-purpose language and can be used to write any application, many features are well suited for numerical analysis and computational science.

    R → R is a programming language for statistical computing and graphics. R is trendy among biologists (and neuroscientists) for data analysis. It is open source, meaning that it is free to use.

  • I work precisely at the interface between artificial intelligence, machine learning, and experimental neuroscience. The natural features of an artificial network inform my work, whose processing I can exploit to study learning, memory, and decision-making. Other topics exist under the umbrella of computational neuroscience, such as the brain-computer interface, robotics, and artificial genomic networks. My work, along with the research of others, covers a broad spectrum of topics that range from math to psychology to genetics.

    Topics within the field of computational neuroscience include:

    • Single-neuron modeling

    • Development and direction of neural circuits

    • The behavior of the neural network

    • Learning and memory

    • Computational cognitive neuroscience

    • Connectomics

    Check out our science comics to learn more about the diversity of computational neuroscience topics, gain insights into the brain, and have a good laugh.

  • No, you don't need a medical degree / MD to go into computational neuroscience. There are a variety of paths that can lead to a career in computational neuroscience. Most computational neuroscientists opt for a Ph.D. because a medical degree is not research-focused, so a Ph.D. is better preparation.

    An MD is a better option for those wanting to pursue neuroscience more clinically, in which case an MD and further study in neurology are likely required. Computational psychiatry is a growing field that may be a good fit.

  • Computational neuroscientists form a strong and ever-expanding community across the globe. My career trajectory was –and still is– enhanced by my commitment to this community. Societies and journals host and promote teaching, mentoring, and networking opportunities for researchers at any level and are the primary vehicle to engage in scientific discussions with peers. Some peer-reviewed, field-specific journals in which computational neuroscientists often publish:

    Societies/organizations that computational neuroscientists are often part of include:

    Visit our publications page to read our peer-reviewed research, and follow us on social media to catch our next science stories and be a part of the conversation!

  • The field of computational neuroscience is a great launching pad for many career opportunities. Every day a new app comes out, algorithms alter behavior, coding becomes art, hospitals use augmented reality, and advertisers use big data, among other recent adoptions of technology into the mainstream. Moreover, a large industrial sector is moving towards using artificially intelligent processing systems, and career options for computational neuroscientists are popping up in every corner. In addition to academic research, there are lots of career options with a degree in computational neuroscience, including:

    • Research positions

    • Teaching at the post-secondary level

    • Pharmaceutical or medical device development and sales

    • Development of AI technologies

    • Machine learning and software development