The Abstraction Atlas charts the principal simplifications by which computational neuroscience has learned to describe neural tissue, from ion channels to networks of areas.
Its skeleton is settled and largely perennial knowledge; like elementary algebra, refined over centuries, its essentials have withstood decades of scrutiny and are unlikely to change, and therefore reward being owned thoroughly.
Each model on the map has a deliberate omission. Rather than oversimplifying, the omissions are simplifications that add clarity and explanatory power. Each edge records what a simplification retains, what it forfeits, and the conditions under which its conclusions remain trustworthy.
Enter at whichever node you happen to arrive, then move inward towards detail or outward towards parsimony, following the dashed links wherever plasticity, modulation or noise alter the picture.
Fluency across these levels is useful for anyone who thinks seriously about brains, and the surest safeguard against mistaking a model for the tissue it describes.
simplificationA step along the spine that replaces a model with a cheaper one, keeping some properties and deliberately giving up others.compositionA step that builds a larger system out of many copies of the level before, taking nothing away from the model itself.side branchA step off the spine towards a functional or structural description, such as artificial networks, attractors or connectivity graphs.modulator, shown on selectionA process, such as plasticity, neuromodulation or noise, that acts on several levels at once and changes how they behave rather than replacing the model.● established ◐ likely ○ unverified † disputed