Extended Data Fig. 7: Winner-take-all dynamics in the DRN persist over a wide range of overlap fractions between competing inputs, a phenomenon which depends on recurrent short-term plasticity. Related to network simulations shown in Fig. 7.

a, Schematic of network simulation, with LHb and Area A providing input to 5-HT neurons (top). Bottom contains schematic demonstrating how the overlap between LHb and A inputs onto 5-HT neurons is varied from 0-60%. b, Simulation with small input overlap of o = 0.2. Shown are mean firing rates of A and LHb inputs (top); mean firing rates of A-recipient 5-HT neurons (middle); and raster plot depicting firing rates of 2500 5-HT neurons receiving either LHb input (purple) or A input (orange) (depicted is the 50% of 5-HT neurons receiving direct input from either LHb or A). c, Simulation with larger input overlap of o = 0.6. Layout of panels is identical to in b. d-f quantify the strength of winner-take-all dynamics as a function of input overlap fraction, while short-term plasticity at recurrent connections is intact. d, LHb-evoked change in A-recipient 5-HT neuron responses during the transient phase (0–200 ms), as a function of the rate of A input, for 5 Hz (left) and 20 Hz (right) LHb input. Input overlap is varied (shown in legend). e, LHb-evoked change in A-recipient 5-HT neuron responses during the stationary phase (200–500 ms), as a function of the rate of A input, for 5 Hz (left) and 20 Hz (right) LHb input. Input overlap is varied (shown in legend). f, Strength of winner-take-all effect (quantified as LHb-evoked change in A-recipient 5-HT neuron responses) as A input rate and LHb input rate are varied. Shown are input overlap values of o = 0 (left) and o = 0.4 (right), illustrating winner-take-all dynamics (ie negative 5-HT modulation) for these input fractions. g-i quantify the strength of winner-take-all dynamics as a function of input overlap fraction, while short-term plasticity at recurrent connections has been removed (that is linearized responses). g, LHb-evoked change in A-recipient 5-HT neuron responses during the transient phase (0–200 ms), as a function of the rate of A input, for 5 Hz (left) and 20 Hz (right) LHb input. Input overlap is varied (shown in legend). h, LHb-evoked change in A-recipient 5-HT neuron responses during the stationary phase (200–500 ms), as a function of the rate of A input, for 5 Hz (left) and 20 Hz (right) LHb input. Input overlap is varied (shown in legend). i, Strength of winner-take-all effect (quantified as LHb-evoked change in A-recipient 5-HT neuron responses) as A input rate and LHb input rate are varied. Shown are input overlap values of o = 0 (left) and o = 0.4 (right), illustrating that winner-take-all dynamics are no longer present with moderate input overlap of o = 0.4 when short-term plasticity is removed (compare with f).