Controlling, Mapping, and Modeling Neural Circuits

## Scaling up Behavior is not produced by a single neuron. The levels build: cell → local circuit → brain region → distributed network → behavior Modern methods make the higher levels measurable. **Multiregion recording** with many electrodes or wide-field imaging captures **large-scale activity** from several brain areas at once. ## Population dynamics With hundreds of neurons recorded, researchers describe **population activity**: the combined pattern across cells at each moment. That pattern is a **neural state** — a point in a space with one axis per neuron. As time passes, the state moves, tracing a **neural trajectory**. Different behaviors, choices, or internal states can correspond to different trajectories. Signals are often **distributed**: information about a choice or a state may appear across many regions, not in one "center." ## Models **Computational models** — for example, networks of simulated units — can be fitted to reproduce recorded dynamics. Good models compress data, make predictions, and suggest mechanisms. But a mathematical model that fits neural dynamics does **not** automatically establish true biological connectivity. Many different networks can produce similar activity. A fitted model generates *hypotheses* about connections and mechanisms, which then need testing by anatomy (mapping) and perturbation. ## The full picture The course arc comes together: **molecule → cell type → circuit → network → behavior** Mapping methods (CLARITY, STARmap) locate cells and molecules; recordings measure dynamics; perturbations test causality; models integrate them and propose new experiments. Each covers what the others miss. ## The course framework GENES → CELL TYPES → CIRCUITS → NETWORK DYNAMICS → BEHAVIOR, studied by four complementary approaches: **map**, **record**, **perturb**, and **model**. Understanding a neural circuit requires several of these ways of observing and testing the same biological system. ## Further reading - [Deisseroth Lab (Stanford)](https://web.stanford.edu/group/dlab/) - Developmental xenocortication using human-derived organoids in mice. *Nature* (2026) — collaborative work by multiple Stanford groups and investigators.