Braincurry: A Domain-specific Language for Integrative Neuroscience

This paper describes Braincurry, a domain-specific, declarative language for describing and analysing experiments in neuroscience. Braincurry has three goals: to allow experiments and data analysis to be described in a way that is sufficiently abstract to serve as a definition; to facilitate carrying out experiments by executing such descriptions; and to be directly usable by end users: neuroscientists. We adopted an experimental and incremental approach to the design and implementation of Braincurry, focusing on the neurophysiological response to visual stimuli in locusts as a test case. Braincurry is currently implemented as an embedding in Haskell, which is a highly effective tool for this kind of exploratory language design. The declarative nature of Haskell and its flexible syntax fitted with our goals. We discuss the requirements for a realistic language meeting the above goals, describe the current Braincurry design and how it may be generalised, and explain how some particularly challenging hard real-time requirements were met.