In the first post on this blog, I want to discuss my motivations for writing about Theory in Neuroscience. My honest motivation is that I can learn more about the role of theory in neuroscience (the best way to learn is by doing?). But why should I or we care?
We should care because, collectively, we all want to understand the brain. By understanding, what I am referring to is be able to describe mechanisms for the function or phenomenon of interest. This would require that the multiple datapoints pertaining to the system can eventually build a cohesive and consistent picture. For me, this naturally invokes the need to have a “theoretical” model in mind, which would enable explicitly stating the assumptions built into our understanding, and to identify the relevant variables. This model built from available data can guide us into the unknown, helping us define the gaps and how they can be addressed most effectively.
In addition, there are a few external factors that suggest that this exploration would be a learning exercise worth carrying out. One of them is the series of interviews of theorists and experimentalists conducted at COSYNE 16 which confirms that the theory-experiment interplay is yet to be properly understood and developed. The interview participants agree that this interplay is fundamentally necessary to building a useful scientific understanding of the brain. Another is several articles in recent years suggesting the need to think about theoretical motivations of our experiments (whether at the circuit level), or at the level of behaviour), so that we can make sense of all the data we are collecting in neuroscience. I am fairly new in the field to fully grasp how big the big data explosion is, still, these articles feel cautionary. From what I hear, they reflect a sentiment that is certainly not new, but perhaps far more relevant now with our access to a multitude of new and fancy tools to map and dissect and reconstruct nervous systems.
Another factor that has been pointed out in the Theory Matter series mentioned above, and in an excellent perspective piece on the role of theory in neuroscience by Larry Abbott is that gathering every possible datapoint may not be as helpful in building understanding. Both of these highlight the idea that the map is not the territory, and thus establishing generalizable rules and models require a theoretical, model-driven approach. To determine what matters, and what doesn’t matter is almost more important than characterizing systems fully. This sentiment seems to echo from students of neuroscience even at very early stages of their careers. I was at an undergraduate student breakfast recently, and one of the students asked the visiting speaker – “how do we know what do we need to know”? The number of unknowns seem so high, and it’s not clear what manipulations are necessary and sufficient. And I find this line of thinking exciting, and something we don’t spend too much time pondering on in our daily research lives.
To end for now, I will list here a few questions and thoughts that I gathered from the COSYNE series that highlight the broad questions on this topic. Beyond that, though, my core interest in the topic is to understand the process adopted by scientists who work at this interface (or students who want to). How do theorist-experimentalist collaborations begin and foster? And how do young scientists (students) learn to think and work at this interface from the get-go?
Notes from Theory Matters at COSYNE 16 (videos available here)
-
What is the role of theory in neuroscience?
-
What kinds of theories would benefit from experimentation?
-
Did theory have an impact on neuroscience?
-
Why do we need to promote theory in neuroscience?
-
What computational tools made a significant impact in neuroscience?
-
What do we need to (do) bring theorists and experimentalists together?
-
How is it like to be in an experimental lab and theory lab at the same time? (Be willing to ask dumb questions)
-
Did theories help your experiments?
-
There is a strong need for models and theories to capture the heterogeneity of cell-types and neural responses (Anne Churchland)
-
In terms of an experimentalist who moved into theory it’s trying to understand the basic phenomenon, the underlying principle of the experimental question of interest (Bradley Voytek)
-
Experiments can establish “necessary” conditions for phenomenon, but experimental neuroscience cannot, but theory can, establish “sufficient” conditions. Together, these can work out the necessary and sufficient conditions, which is exactly what we want to understand to understand how nature works. (Eugene Izhikevich)
-
We rely on theory mostly to explain our data and characterize it better, but we really want theory to do is to make predictions to drive experiments (Benjamin Scholl)
One of the interviewees said that for experimentalists and theorists working together, they should “be willing to ask dumb questions” – I hope to do exactly that to people who work at this interface.