Inverse Problem

The inverse problem in brain emulation

In its most general form, brain emulation is an inverse problem. One seeks to find a model that makes good predictions that match observations about the system. When the goal is purely behavioral then these are external observations, but if the goal of brain emulation includes modeling experiential characteristics then one must take care to give adequate attention to internal structure and system states.

Unfortunately, the formal problem of brain emulation has not yet received much attention in the literature. Additionally, approaches and corresponding formulations are constrained by techniques that are available to obtain brainwide data at high resolution. The most promising techniques available at this time, which are applicable from invertebrates through to mammals and humans, rely primarily on acquiring morphological data with some potential additional labeling during sample preparation, e.g. high-throughput connectomics. For this reason, we will focus mostly on system identification as a process of functionalizing connectome-constrained models in the following sections.

Anticipated pre-print release: May 31, 2025