BODB Introduction

The Brain Operation Database is explicitly designed to provide a framework for the linkage of neurobiological data and computational modeling via the general operating principles of the brain which provide a unifying perspective on a host of diverse studies. The 3 main conceptual entities to be stored in BODB are models, brain operating principles (BOPs), and summary data (SD). The system currently provides 3 main features: (i) a repository for article information in bibliography format; (ii) a repository for BOPs and Models, enhanced with discussion-based text annotation; and (iii) a repository for Talairach-based brain-imaging experimental data along with a visualization tool for brain-imaging data analysis.

A brain operating principle (BOP) presents some general operational principle that will provide a unifying principle embodied in, usually, a wide range of neural systems and the corresponding models. Examples include winner-take-all, somatotopic layered computation, and Hebbian learning, to name just a few. In general, a given BOP will be supported by many experimental and simulation data – though, of course such data may be used to refine and bound the validity of specific BOPs. Brain Operating Principles capture the gist of the brain mechanisms that abstract away from their heterogeneities, as a way to organize experiments, models and technological spin-off of brain mechanisms.

We define a brain model as either (a) an explicit computational model at, e.g., the level of biologically realistic brain regions and neural networks of how a particular brain system operates, or (b) a more abstract conceptual model. To be more specific, the former tries to emulate the function of a biologically established brain system, matching simulation results to data from experimental protocols. The simulation results will be used to contrast and compare with a certain collection of experiments as well as predict results for further ones. By contrast, a conceptual model may provide diagrams and comments suitable for later development into a computational model; but which is not yet specific enough to reach an implementation level.

A Summary of Experimental Data (SED) encapsulates a fragment of knowledge in a particular domain, encapsulating empirical findings from one or more studies in a form that is amenable for use in the support or testing of a certain model or BOP. Note that a summary datum could encapsulate any set of related experimental results along with the associated protocol. Thus, brain structure connectivity, neurophysiological or neurochemical data, or a table in Talairach coordinates from an fMRI brain imaging study could be presented in this way. [An issue with this paradigm would be when scientist A works on his model and inserts SED A to support his model while scientist B works inserts SED B to support her model. If SED A is in fact the same as (or a minor variant of) SED B, they (or an editor) will have to discover this fact and replace each of these SEDs by the same integrated version.

A Summary of Simulation Results (SSR) summarizes a basic set of simulation results from a model. A Model entry should include explicit analysis of appropriate SEDs either with design features of the model or with SSRs for the model.