Dendritic Integration of Sensory Evidence in Perceptual Decision-Making

Abstract

Perceptual decisions require the accumulation of sensory information to a response criterion. Most accounts of how the brain performs this process of temporal integration have focused on evolving patterns of spiking activity. We report that subthreshold changes in membrane voltage can represent accumulating evidence before a choice. αβ core Kenyon cells (αβ_c KCs) in the mushroom bodies of fruit flies integrate odor-evoked synaptic inputs to action potential threshold at timescales matching the speed of olfactory discrimination. The forkhead box P transcription factor (FoxP) sets neuronal integration and behavioral decision times by controlling the abundance of the voltage-gated potassium channel Shal (K_V4) in αβ_c KC dendrites. αβ_c KCs thus tailor, through a particular constellation of biophysical properties, the generic process of synaptic integration to the demands of sequential sampling.

Keywords: Drosophila melanogaster; decision-making; drift-diffusion model; forkhead box P transcription factors; membrane biophysics; neural integrator; olfaction; potassium channel; synaptic integration.

Graphical abstract

Figure 1

FoxP Controls the Responsiveness of αβ_c KCs to Antennal Nerve Stimulation (A) The P element insertion (red) in the mutant FoxP^5-SZ-3955 allele maps to the alternatively spliced exon of the FoxP-RC isoform. (B) Levels of polysome-bound FoxP mRNA in αβ_c or α’β’ KCs of wild-type (WT) flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing FoxP^RNAi (blue), relative to the geometric mean of three marker gene transcripts (circles, biological replicates; bars, means ± SEM; see Table S1 for sample sizes). One-way ANOVA detected a significant genotype effect on FoxP-RC levels in αβ_c KCs (p < 0.0001); asterisks denote significant differences from wild-type in post hoc comparisons. (C) Antennal nerve stimulation and two-photon imaging in vivo. ORN, olfactory receptor neuron; PN, projection neuron. (D) NP7175-GAL4-driven GCaMP6m expression in αβ_c KCs. (E) Ca²⁺ transients in αβ_c KC dendrites evoked by 0.5 s epochs of antennal nerve stimulation at the indicated frequencies of individual wild-type (black, n = 10) or homozygous FoxP^5-SZ-3955 mutant flies (red, n = 10). Solid traces represent group averages. (F) Peak ΔF/F as a function of stimulation frequency in αβ_c KC dendrites (calyx, top) or axons (center) or in α’β’ KC axons (bottom) of wild-type (black) or homozygous FoxP^5-SZ-3955 mutant flies (red). Dashed lines in the schematics mark the approximate positions of the imaging planes. Data are means ± SEM. Asterisks indicate significant differences between the stimulus-response curves of wild-type and FoxP^5-SZ-3955 mutant flies (p < 0.0001, F test). See also Figures S1 and S2 and Tables S1 and S5.

Figure S1

Expression Patterns of GAL4 Lines, Related to Figure 1 (A–D) Maximum intensity projections of confocal image stacks of the central brains of flies expressing UAS-CD8::GFP (cyan) under the control of NP7175-GAL4 (A) or NP6024-GAL4 (B) in αβ_c KCs; under the control of VT030604-GAL4 in α’β’ KCs (C); or under the control of OK107-GAL4 in all KCs (D). Synaptic structures were counterstained with an antibody against discs large (magenta).

Figure S2

GCaMP6m Signals in KCs, Related to Figure 1 (A–C) Raw two-photon images of GCaMP6m fluorescence in dendrites (A) or axons (B) of αβ_c KCs or axons of α’β’ KCs (circled regions of interest in C) at the indicated stimulation frequencies, in wild-type flies (top) and homozygous FoxP^5-SZ-3955 mutants (bottom). At the end of each imaging experiment, neurons were bulk-depolarized by elevating the extracellular KCl concentration to 100 mM (right). t tests failed to detect significant differences between the peak changes in GCaMP6m fluorescence evoked by 100 mM KCl in wild-type flies (black) and homozygous FoxP^5-SZ-3955 mutants (red) (p ≥ 0.5245).

Figure 2

FoxP Determines the Biophysical Properties of αβ_c KCs (A) Targeted whole-cell recording from αβ_c KCs expressing NP7175-GAL4-driven CD8::GFP (magenta). ORN, olfactory receptor neuron; PN, projection neuron. Synaptic structures were counterstained with an antibody against discs large (gray). (B) Resting membrane potentials of αβ_c KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue) (circles, individual KCs; bars, means ± SEM). One-way ANOVA failed to detect a significant difference between genotypes (p = 0.0511). (C–E) Voltage responses (C) of αβ_c KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue) to ramps of depolarizing current. Cells were held at –75 ± 2 mV at the start of the ramp. FoxP-deficient αβ_c KCs initiate action potentials at the same membrane voltage as wild-type cells (D, p = 0.7143) but reach threshold later during the ramp (E, p < 0.0001). Circles, individual KCs; bars, means ± SEM; asterisks, significant differences from wild-type in post hoc comparisons following one-way ANOVA or Kruskal-Wallis test. (F–H) Voltage responses (F) of αβ_c KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue) to steps of depolarizing current. FoxP-deficient αβ_c KCs have lower input resistances (R_m) (G, p < 0.0001) and shorter membrane time constants (τ_m) than wild-type cells (H, p < 0.0001). Circles, individual KCs; bars, means ± SEM; asterisks, significant differences from wild-type in post hoc comparisons following Kruskal-Wallis tests. (I) Spike frequencies evoked by depolarizing current injections into αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). The cells were held at an initial membrane potential of –70 ± 5 mV, at which spiking is suppressed. Data are means ± SEM. F test detected a significant difference between the current-spike frequency functions of wild-type and FoxP-deficient αβ_c KCs (p < 0.0001). (J) Targeted whole-cell recording from α’β’ KCs expressing VT030604-GAL4-driven CD8::GFP (magenta). ORN, olfactory receptor neuron; PN, projection neuron. Synaptic structures were counterstained with an antibody against discs large (gray). (K) Resting membrane potentials of α’β’ KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing VT030604-GAL4-driven FoxP^RNAi (blue) (circles, individual KCs; bars, means ± SEM). One-way ANOVA failed to detect a significant difference between genotypes (p = 0.6016). (L–N) Voltage responses (L) of α’β’ KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing VT030604-GAL4-driven FoxP^RNAi (blue) to ramps of depolarizing current. FoxP-deficient α’β’ KCs initiate action potentials at the same membrane voltage (M, p = 0.8062) and with the same latency as wild-type cells (N, p = 0.7373). Circles, individual KCs; bars, means ± SEM; asterisks, significant differences from wild-type in post hoc comparisons following one-way ANOVA or Kruskal-Wallis test. (O–Q) Voltage responses (O) of α’β’ KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing VT030604-GAL4-driven FoxP^RNAi (blue) to steps of depolarizing current. FoxP-deficient and wild-type α’β’ KCs have identical input resistances (R_m) (P, p = 0.9497) and membrane time constants (τ_m) (Q, p = 0.9771). Circles, individual KCs; bars, means ± SEM. (R) Spike frequencies evoked by depolarizing current injections into α’β’ KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing VT030604-GAL4-driven FoxP^RNAi (blue). Data are means ± SEM. F test failed to detect a significant difference between the current-spike frequency functions of wild-type and FoxP-deficient α’β’ KCs (p = 0.0660). See also Figure S3.

Figure S3

FoxP Does Not Regulate Synaptic Transmission to αβ_c KCs, Related to Figure 2 (A) Example transmembrane currents (left, holding potential –90 mV) and mean frequency of spontaneous EPSCs (right) of αβ_c KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). Kruskal-Wallis test failed to detect a significant difference among genotypes (p = 0.8432). Circles, individual KCs; bars, means ± SEM. (B–D) EPSC waveforms at a holding potential of –90 mV (B; shaded and solid lines, cell and population averages) in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). Cell numbers are indicated in (C). Kruskal-Wallis tests failed to detect significant differences of peak EPSC amplitudes (C, p = 0.2669) or decay time constants (D, p = 0.0507) among genotypes. Circles, individual KCs; bars, means ± SEM. (E) Average action potential waveforms (left) and afterhyperpolarization amplitudes (right) of αβ_c KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). Circles, individual KCs; bars, means ± SEM. Kruskal-Wallis test detected a significant genotype effect (p = 0.0058); asterisks denote significant differences from wild-type in post hoc comparisons. (F) Example membrane potential traces of αβ_c KCs in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). (G–I) EPSP waveforms (E; solid and shaded lines, means ± SEM) in wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). Cell numbers are indicated in (H). FoxP-deficient αβ_c KCs have lower mEPSP amplitudes (H, p < 0.0036) and shorter decay time constants (τ_decay) than wild-type cells (I, p = 0.0299). Circles, individual KCs; bars, means ± SEM; asterisks, significant differences from wild-type in post hoc comparisons following one-way ANOVA or Kruskal-Wallis test. In 5 αβ_c KCs of wild-type flies, 1 αβ_c KC of a FoxP^5-SZ-3955 mutant, and 1 αβ_c KC of a FoxP^RNAi fly, a satisfactory single-exponential fit to the decaying phase of the EPSP could not be found; these cells were excluded from the analyis of τ_decay in (I).

Figure 3

FoxP Regulates a Voltage-Dependent, Ba²⁺-Sensitive Potassium Current of αβ_c KCs (A) Potassium current densities, evoked by voltage steps from a holding potential of –100 mV to the indicated probe potentials, in αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). (B) A-type potassium current densities in FoxP-deficient αβ_c KCs are increased relative to wild-type cells (B, p < 0.0001). Circles, means ± SEM; asterisks, significant differences from wild-type in post hoc comparisons following two-way repeated-measures ANOVA. (C) A-type potassium currents evoked by stepping αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue) from variable holding potentials (–120 mV to –30 mV) to a probe potential of +50 mV. (D) Steady-state activation and inactivation curves of A-type potassium currents in αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). Circles, means ± SEM; solid lines, Boltzmann fits. Kruskal-Wallis tests failed to detect significant genotype effects on half-activation voltage (p = 0.0806), activation slope factor (p = 0.1996), half-inactivation voltage (p = 0.4735), and inactivation slope factor (p = 0.8752). (E) Average potassium current densities in response to a single voltage pulse before (colored lines) and after (gray lines) the bath application of 150 μM Ba²⁺ in αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). (F) Average peak current densities before (left) and after (right) Ba²⁺ application. Data are means ± SEM. Two-way repeated-measures ANOVA detected a significant effect of Ba²⁺ treatment (p < 0.0001) and a significant genotype × treatment interaction (p = 0.0012); the asterisk denotes significant differences in peak current densities before and after the addition of Ba²⁺. Kruskal-Wallis tests detected a significant difference in peak current densities among wild-type and FoxP-deficient αβ_c KCs before (p = 0.0034), but not after (p = 0.2968), the addition of Ba²⁺. See also Figure S4.

Figure S4

A-Type Potassium Currents of αβ_c KCs Show Inactivation Kinetics and Toxin Sensitivities Characteristic of Shal, Related to Figure 3 (A) Exponential fits (solid lines) to average A-type potassium currents (shaded lines) evoked by stepping αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue) from a holding potential of –100 mV to the indicated probe potentials. (B) Inactivation time constants τ of I_A in αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue) as functions of voltage. Two-way repeated-measures ANOVA failed to detect a significant interaction between genotype and the voltage-dependence of τ (p = 0.8943). (C) Pressure application of phrixotoxin-2 to the soma of an αβ_c KC. Average potassium current densities in response to a single voltage pulse before (colored traces) and after (gray traces) the application of phrixotoxin-2 to αβ_c KCs of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing NP7175-GAL4-driven FoxP^RNAi (blue). (D) Average peak current densities before (left) and after (right) the application of phrixotoxin-2. Data are means ± SEM. Two-way repeated-measures ANOVA detected significant effects of genotype and toxin (p < 0.0001 for both effects) and a significant genotype × toxin interaction (p = 0.0050); the asterisk denotes significant differences in peak current densities before and after the application of phrixotoxin-2.

Figure 4

FoxP Represses the Dendritic K_V Channel Shal in αβ_c KCs (A and B) Levels of polysome-bound ion channel transcripts in αβ_c (A) or α’β’ KCs (B) of homozygous FoxP^5-SZ-3955 mutants (red) and flies expressing NP7175-GAL4- or VT030604-GAL4-driven FoxP^RNAi (blue), relative to corresponding transcript levels in αβ_c or α’β’ KCs of wild-type flies (circles, biological replicates; bars, means ± SEM; see Table S1 for sample sizes). Asterisks denote significant differences from wild-type in post hoc comparisons following one-way ANOVA (Bonferroni-corrected p < 0.0016). (C) NP6024-GAL4-driven expression of membrane-bound mCherry (magenta) and GFP-tagged Shal (cyan) in αβ_c KCs. See also Figure S5 and Tables S1 and S5.

Figure S5

FoxP Represses Shal in αβ_c KCs, Related to Figure 4 (A and B) Levels of polysome-bound ion channel transcripts in αβ_c KCs (A) or α’β’ KCs (B) of wild-type flies (black), homozygous FoxP^5-SZ-3955 mutants (red), or flies expressing FoxP^RNAi (blue), relative to the geometric mean of three marker gene transcripts (Gpdh, Tbp, and Ef1α100E; circles, biological replicates; bars, means ± SEM; see Table S1 for sample sizes). Note the logarithmic scale. One-way ANOVA detected a significant genotype effect on Shal levels in αβ_c KCs (p < 0.0001); asterisks denote significant differences from wild-type in post hoc comparisons.

Figure 5

FoxP Tunes the Integrative Properties of αβ_c KCs (A) Antennal nerve stimulation and targeted whole-cell recording from KCs expressing NP7175-GAL4- or VT030604-GAL4-driven CD8::GFP. ORN, olfactory receptor neuron; PN, projection neuron. (B) Sequentially recorded voltage (top) and transmembrane current responses (bottom, holding potential –70 mV) of the same αβ_c (left) or α’β’ KC (right) in a wild-type fly during antennal nerve stimulation at 10 Hz (orange marks); the averages of 119 (left) and 17 (right) miniature EPSCs recorded in the same cells are shown for comparison (mEPSC). (C) Examples of spike rasters (top) and voltage responses (bottom) of 5 αβ_c KCs in wild-type flies (left) and homozygous FoxP^5-SZ-3955 mutants (center) and of 4 α’β’ KCs in wild-type flies (right) during antennal nerve stimulation at the indicated frequencies, in control conditions (black or red) and after the sequential addition of 150 μM Ba²⁺ to block Shal (gray) and 1 μM tetrodotoxin (TTX) to block action potentials (blue). Each row in the rasters depicts a different KC. Stimulus artifacts were removed for clarity. (D) Stimulus-response curves of αβ_c KCs in wild-type flies (left) and homozygous FoxP^5-SZ-3955 mutants (center) and of α’β’ KCs in wild-type flies (right) during antennal nerve stimulation in control conditions (black or red) and after the sequential addition of 150 μM Ba²⁺ to block Shal (gray) and 1 μM TTX to block action potentials (blue). Data are means ± SEM. F tests detected significant effects of 150 μM Ba²⁺ on the stimulus-response curves of αβ_c KCs in wild-type flies (p = 0.0288) and homozygous FoxP^5-SZ-3955 mutants (p < 0.0001), but not of α’β’ KCs in wild-type flies (p = 0.4186). See also Figure S6.

Figure S6

EPSCs Evoked in αβ_c KCs by Antennal Nerve Stimulation: Comparison with Miniature EPSCs, Related to Figure 5 (A) Average transmembrane currents (holding potential –90 mV) of αβ_c KCs in wild-type flies (black) or homozygous FoxP^5-SZ-3955 mutants (red). EPSCs evoked by 50 μs voltage pulses (left) were blocked by 1 μM tetrodotoxin (TTX; center); to minimize the fraction of transmission failures, stimulation voltages exceeded those at which EPSCs were first detected by ∼25%. The peak currents of spontaneously occurring miniature EPSCs in 1 μM TTX (mEPSCs, right) averaged 80–89% of those of eEPSCs. (B) Peak eEPSC currents in the absence and presence of 1 μM TTX (left and center) versus peak mEPSC currents in 1 μM TTX (right), in αβ_c KCs of wild-type flies (black) and homozygous FoxP^5-SZ-3955 mutants (red). Circles, individual EPSCs; bars, means ± SEM. Two-way ANOVA detected a significant effect of release mode (evoked versus spontaneous, p < 0.0001) but not of genotype (p = 0.6094).

Figure 6

Shal Currents in αβ_c KCs Determine Reaction Times (A) Measurement of reaction times and decision accuracies. Two odor streams converge in a 7-mm-wide decision zone (orange) at the center of a narrow chamber. Flies are trained to avoid 20 ppm of 4-methylcyclohexanol (MCH, gray) and must discriminate the reinforced from a lower MCH concentration (2–18 ppm); the concentration ratio during testing determines the difficulty of the task. The time spent in the decision zone is quantified as the reaction time. (B–D) Example traces (top) of position on the long chamber axis versus time of flies of the indicated genotypes in the decision zone: homozygous FoxP^5-SZ-3955 mutants expressing dominant-negative Shal in αβ_c KCs (B), wild-type flies overexpressing functional Shal in αβ_c KCs (C), or wild-type flies overexpressing functional Shal in α'β' KCs (D), along with their respective parental controls; the MCH concentration ratio is 0.9. Accuracy (bottom left) and reaction time (bottom right) of flies of the indicated genotypes as functions of the MCH concentration ratio. Data are means ± SEM (see Table S2 for sample sizes); asterisks denote significant differences of experimental flies from both parental controls (accuracy, Dunn’s test following one-way Kruskal-Wallis test; reaction time, Kolmogorov-Smirnov test with Bonferroni-corrected p < 0.00625). See also Table S2.

Figure 7

The First Odor-Evoked αβ_c KC Spike Predicts Behavior (A) Odor stimulation and targeted whole-cell recording from KCs expressing NP7175-GAL4- or VT030604-GAL4-driven CD8::GFP. ORN, olfactory receptor neuron; PN, projection neuron; MCH, 4-methylcyclohexanol. (B and C) MCH concentrations (top) and membrane voltages (bottom) of up (orange), down (pale orange), and unresponsive (gray) αβ_c (B) or α’β’ KCs (C) during repeated steps between base (2 ppm) and peak MCH concentrations (20 ppm); pie charts indicate the proportions of functional KC classes encountered in wild-type flies and homozygous FoxP^5-SZ-3955 mutants. (D) Examples of spike rasters and voltage responses of up αβ_c KCs in wild-type flies (black, top) and homozygous FoxP^5-SZ-3955 mutants (red) and of α’β’ KCs in wild-type flies (black, bottom) during ten odor intensity cycles between a variable base (2–18 ppm) and a constant peak (20 ppm) MCH concentration at 1 Hz. Measured MCH concentration time courses at the different base-to-peak ratios are displayed on top. The spike rasters are sorted, in ascending order from the bottom, by the latency of the first spike. Filled and open squares in the margins mark trials whose first spikes occur after odor concentration changes in the preferred or null (“correct” or “incorrect”) directions (Figure S7B). (E) Cumulative frequency distributions of spike latencies after stimulus onset of up αβ_c KCs in wild-type flies (black) and homozygous FoxP^5-SZ-3955 mutants (red) at the MCH concentration ratios indicated on top. (F) Examples of membrane voltages preceding the first αβ_c KC spike after stimulus onset in wild-type flies (black) and homozygous FoxP^5-SZ-3955 mutants (red), at the MCH concentration ratios indicated on top. The traces are aligned to the upstroke of the action potential (peak of the first derivative) and depict a period of ≤ 1 s. (G) Voltage gradients from stimulus onset to the upstroke of the first αβ_c KC spike in wild-type (black) and FoxP^5-SZ-3955 mutant flies (red) at the indicated MCH concentration ratios. Circles, individual trials; bars, means ± SEM. Two-way repeated-measures ANOVA detected significant effects of odor contrast (p < 0.0001) and genotype (p < 0.0001) and a significant genotype × contrast interaction (p < 0.0001); asterisks denote significant differences between genotypes. (H) Drift-diffusion model of evidence accumulation. Homozygous FoxP^5-SZ-3955 mutants (red) exhibit lower drift rate (v) and noise (σ²) than do wild-type flies (black). (I) Neurometric predictions of accuracy (left) and decision time (right; decision time = reaction time – residual time) as functions of MCH concentration ratio. The predictions are based on the timing and fidelity of the first MCH-evoked up αβ_c KC spikes in wild-type flies (black) and homozygous FoxP^5-SZ-3955 mutants (red). Data are means ± SEM (see Table S3 for sample sizes); asterisks, significant differences from wild-type (accuracy, Mann-Whitney U test; reaction time, Kolmogorov-Smirnov test; both with Bonferroni-corrected p < 0.0125). Shaded bands represent 95% confidence intervals of accuracy and decision time measurements; solid lines depict the fit of a drift-diffusion model to these behavioral data. (J) Average membrane potential variances of αβ_c KCs in wild-type flies (black) and homozygous FoxP^5-SZ-3955 mutants (red) as functions of base MCH concentration. Data are means ± SEM. Two-way ANOVA detected a significant genotype effect (p < 0.0164). See also Figure S7 and Tables S3 and S4.

Figure S7

Average Spike Latencies of αβ_c KCs Are Independent of Odor Stimulus Waveform, Related to Figure 7 (A) MCH concentrations (top) and membrane voltages (bottom) of up (orange), down (pale orange), and unresponsive αβ_c KCs (gray) in wild-type flies during a single 10 s step between base (2 ppm) and peak MCH concentrations (20 ppm) and back. (B) Voltage responses of an up αβ_c KC during 10 odor intensity cycles between a variable base (2–18 ppm) and a constant peak (20 ppm) MCH concentration. Only the first five of the 10 odor stimulation cycles are shown. MCH base-to-peak ratios are indicated on the left; filled and open squares mark trials whose first spikes (arrowheads) occur after odor concentration changes in the preferred or null (“correct” or “incorrect”) directions. (C) Neurometric predictions of decision times (decision time = reaction time – residual time) as a function of MCH concentration ratio. The predictions are based on the timing of the first MCH-evoked spike in response to a single 10 s concentration step (filled circles) or 10 odor intensity cycles at 1 Hz (open circles). Kolmogorov-Smirnov tests with Bonferroni-corrected p < 0.0125 failed to detect a significant difference between stimulation protocols (p = 0.7559). Data are means ± SEM (see Table S4 for sample sizes). (D) MCH stimulus waveforms, examples of spike rasters, and voltage responses during a single 10 s odor intensity step (top) or 10 odor intensity cycles at 1 Hz (bottom) between a variable base (2–18 ppm) and a constant peak (20 ppm) MCH concentration and back. Measured MCH concentration time courses at the different base-to-peak ratios are displayed on top. The spike rasters are sorted, in ascending order from the bottom, by the latency of the first spike. (E) Cumulative frequency distributions of spike latencies for a single 10 s concentration step (solid lines) and 10 odor intensity cycles at 1 Hz (dashed lines), at the MCH concentration ratios indicated on top.