Neural Circuits, Vision, and Computation

## Recording neural activity **Electrophysiology** measures the electrical activity of neurons. A **multielectrode array** places many electrodes under a piece of retina so that dozens or hundreds of ganglion cells can be recorded at once. Each cell's output is a **spike train** — the list of times it fired. A **raster plot** shows spike trains as rows of tick marks: each row is one repeat (trial) of the same stimulus, and each tick is a spike. If ticks line up vertically across rows, the cell responds reliably at that moment. Averaging spikes across trials in small time bins gives the **firing rate**. **Two-photon imaging** uses fluorescent indicators to watch activity in many identified cells, including interneurons that are hard to reach with electrodes, at the cost of slower time resolution. ## Choosing stimuli **Controlled stimuli** (flashes, gratings, white noise) make it easy to estimate exactly which inputs drive a cell. **Natural stimuli** reveal how a circuit behaves in the conditions it evolved for, but are harder to analyze. Good studies often use both. ## Relationship versus mechanism Recording tells you about a **relationship**: when the stimulus does X, cell Y responds. That is valuable but correlational. To test whether a specific cell or synapse is *necessary* for the computation, scientists use **perturbation**: silence, activate, or block a component and measure what changes. recording → relationship perturbation → stronger mechanistic test If suppressing one inhibitory interneuron abolishes a ganglion cell's object-motion selectivity, that is much stronger evidence for its role than finding that the two cells are active at the same time. ## Replicates Repeating a stimulus many times on the same piece of tissue gives **technical replicates**: they measure reliability. Repeating the experiment in retinas from different animals gives **biological replicates**: they show the finding generalizes. Many trials from one retina are not the same as many animals. ## Further reading - [Baccus Lab research overview](https://baccuslab.github.io/research/) - [Baccus Lab publications](https://baccuslab.github.io/publications/)