a, Photostimulation paradigm. Extracellular recordings were obtained from Esr1+ VMHvl neurons expressing AAV2 Cre-dependent ChR2 in solitary, awake behaving animals using a modification of a 16-wire electrode bundle micro-drive31 containing an integrated optic fibre. Following a 30-s baseline measurement, photostimulation trials were performed (473 nm, 20 Hz, blue bars) for 30 s using three different pulse-widths (2 ms, 10 ms, and 20 ms). Five trials, each 2 min in length, were recorded for each pulse-width (see c). b, Mean firing rate changes averaged across 12 multi-units (5 trials per unit) in VMHvl during 30-s photostimulation periods. 2 ms, 17.98 ± 2.35 spikes per s, 10 ms, 29.26 ± 3.67 spikes per s, and 20 ms, 28.07 ± 4.65 spikes per s. *P < 0.05, Wilcoxon rank sum test. c, Spiking responses of 12 multi-unit recording channels in VMHvl. Each raster plot represents the average of five trials per channel per pulse-width (2, 10 or 20 ms), arranged in order of response magnitude. The arrangement is the same for the three pulse widths (2 ms, 10 ms and 20 ms). d–f, Peri-stimulus time histograms (PSTHs) illustrating mean firing rate changes averaged over the 12 multi-units shown in c, for photostimulation trials using 20 ms (d), 10 ms (e), or 2 ms (f) light pulse-widths. Data are mean ± s.e.m. See also main Fig. 2d, which presents whole-cell patch-clamp recordings from Esr1+ neurons in VMHvl acute slice preparations, indicating that spike fidelity is close to 100% and statistically indistinguishable between 2 ms and 20 ms light pulse-widths.