Table 2. Statistics Table.

We used a Generalized Linear Mixed-Effects (GLME) model and Wilcoxon signed-rank tests to compute the statistics for the data.

For Figure 1, Figure 2B,C,E,G,I,J,L,N; the data (‘table‘) had four columns: cell, sound level, laser power, output. The formula used was (Matlab): glme=fitglme(table,‘output ~ sound + laser + sound*laser + (1|cell)’);

For Figure 3D,K, Figure 4 and Figure 5, the data (‘table‘) had three columns: cell, laser power, output. The formula used was (Matlab): glme=fitglme(table,‘output ~ laser + (1|cell)’);

For Figure 3E,H,K,N, the data (‘table‘) had four columns: cell, sound level difference, laser power, output. The formula used was (Matlab): glme=fitglme(table,‘output ~ sounddiff + laser + sounddiff*laser + (1|cell)’);

For Figure 2F,M, we compared each sound amplitude across different light conditions using Wilcoxon tests.

Comparison	Figure	N	Test	Test Statistic	p-value	Effect size
FIGURE 1
SST neuron with SST activation	Fig 1D	10 repeats	GLME	tlaser=7.89 tsound=0.34 tlas er:sound=−0.55 DF = 206	***plaser=1.8e-13 psound=0.74 plaser:sound=0.58	ηlaser2=0.58 ηsound2=3.8e-3 ηlaser:sound2=1.5e-2
Sound-increasing neuron with SST activation	Fig 1E	10 repeats	GLME	tlaser=0.33 tsound=12.37 tlaser:sound=−8.34 DF = 206	plaser=0.74 ***psound=1.2e-26 ***plaser:sound=1.0e-14	ηlaser2=2.3e-3 ηsound2=0.84 ηlaser:sound2=0.78
VIP neuron with VIP activation	Fig 1F	10 repeats	GLME	tlaser=5.40 tsound=0.93 tlaser:sound=−2.56 DF = 206	***plaser=1.8e-7 psound=0.35 *plaser:sound=1.1e-2	ηlaser2=0.39 ηsound2=2.8e-2 ηlaser:sound2=0.25
Sound-increasing neuron with VIP activation	Fig 1G	10 repeats	GLME	tlaser=2.45 tsound=3.06 tlaser:sound=1.11 DF = 206	*plaser=1.5e-2 **psound=2.5e-3 plaser:sound=0.27	ηlaser2=0.12 ηsound2=0.24 ηlaser:sound2=5.8e-2
Control: VIP neurons with laser activation	Fig 1H	54 cells	GLME	tlaser=1.27 tsound=2.99 tlaser:sound=−0.11 DF = 11336	plaser=0.20 **psound=2.8e-3 plaser:sound=0.91	ηlaser2=6.5e-4 ηsound2=5.5e-3 ηlaser:sound2=1.1e-5
Control: All neurons (VIP excluded) with laser activation	Fig 1I	492 cells	GLME	tlaser=0.46 tsound=12.23 tlaser:sound=3.27 DF = 103316	plaser=0.64 ***psound=2.1e-34 **plaser:sound=1.1e-3	ηlaser2=8.5e-6 ηsound2=9.0e-3 ηlaser:sound2=9.8e-4
FIGURE 2
SST neurons with SST activation	Fig 2B	132 cells	GLME	tlaser=36.91 tsound=1.32 tlaser:sound=0.16 DF = 27716	***plaser=3.1e-291 psound=0.19 plaser:sound=0.88	ηlaser2=0.14 ηsound2=3.2e-4 ηlaser:sound2=6.7e-6
All non-SST neurons with SST activation	Fig 2C	2152 cells	GLME	tlaser=1.27 tsound=10.75 tlaser:sound=−6.35 DF = 451916	plaser=0.20 ***psound=5.9e-27 ***plaser:sound=2.2e-10	ηlaser2=1.5e-5 ηsound2=1.6e-3 ηlaser:sound2=8.5e-4
Sparseness with SST activation	Fig 2D	None: 2033 Med: 2029 High: 1994	GLME	tlaser=5.21 DF = 5702	***plaser=2.0e-7	ηlaser2=4.5e-3
Activity sparseness with SST activation	Fig 2E	13 populations	GLME	tlaser=3.20 tsound=0.87 tlaser:sound=−0.79 DF = 230	**plaser=1.6e-3 psound=0.38 plaser:sound=0.43	ηlaser2=0.22 ηsound2=1.3e-2 ηlaser:sound2=2.5e-2
SST: Decoding accuracy of a linear SVM decoder with laser activation	Fig 2F	13 populations	GLME	tlaser=−3.92 tsound=−2.84 tlaser:sound=1.92 DF = 269	***plaser=1.1e-4 **psound=4.9e-3 plaser:sound=5.6e-2	ηlaser2=0.19 ηsound2=0.16 ηlaser:sound2=0.11
Activity sparseness from baseline with SST activation	Fig 2G	13 populations	GLME	tlaser=−0.89 tsound=3.28 tlaser:sound=2.01 DF = 269	plaser=0.37 **psound=1.2e-3 *plaser:sound=4.5e-2	ηlaser2=6.1e-3 ηsound2=0.11 ηlaser:sound2=6.7e-2
VIP neurons with VIP activation	Fig 2I	226 cells	GLME	tlaser=41.50 tsound=2.71 tlaser:sound=−1.96 DF = 47456	***plaser=0 **psound=6.7e-3 *plaser:sound=4.95e-2	ηlaser2=0.12 ηsound2=9.3e-4 ηlaser:sound2=7.3e-4
All non-VIP neurons with VIP activation	Fig 2J	3095 cells	GLME	tlaser=34.18 tsound=11.32 tlaser:sound=8.41 DF = 649946	***plaser=7.8e-256 ***psound=1.1e-29 ***plaser:sound=4.1e-17	ηlaser2=6.0e-3 ηsound2=1.0e-3 ηlaser:sound2=8.4e-4
Sparseness with VIP activation	Fig 2K	None: 2979 Med: 2921 High: 3020	GLME	tlaser=2.78 DF = 8350	**plaser=5.5e-3	ηlaser2=7.0e-4
Activity sparseness with VIP activation	Fig 2L	16 populations	GLME	tlaser=1.46 tsound=−1.18 tlaser:sound=0.17 DF = 284	plaser=0.15 psound=0.24 plaser:sound=0.86	ηlaser2=2.8e-2 ηsound2=1.2e-2 ηlaser:sound2=6.4e-4
VIP: Decoding accuracy of a linear SVM decoder witηlaser activation	Fig 2M	13 populations	GLME	tlaser=0.69 tsound=−3.58 tlaser:sound=−0.11 DF = 332	plaser=0.49 ***psound=4.0e-4 plaser:sound=0.91	ηlaser2=4.3e-3 ηsound2=0.15 ηlaser:sound2=2.7e-4
Activity sparseness from baseline power with VIP activation	Fig 2N	16 populations	GLME	tlaser=8.02 tsound=4.01 tlaser:sound=0.76 DF = 332	***plaser=1.8e-14 ***psound=7.5e-5 plaser:sound=0.45	ηlaser2=0.24 ηsound2=0.11 ηlaser:sound2=6.4e-3
FIGURE 3
Separation angle from 0dB at each laser power – SST activation	Fig 3E	15 angles, 13 recordings	GLME	tlaser=8.80 tΔsound=12.37 tlaser:Δsound=−7.44 DF = 581	***plaser=1.6e-17 ***pΔsound=2.3e-31 ***plaser:Δsound=3.6e-13	ηlaser2=0.30 ηΔsound2=0.58 ηlaser:Δsound2=0.43
Separation angle from 0dB at each laser power – VIP activation	Fig 3H	15 angles, 16 recordings	GLME	tlaser=−2.75 tΔsound=7.32 tlaser:Δsound=0.73 DF = 716	**plaser=6.1e-3 ***pΔsound=6.9e-13 plaser:Δsound=0.47	ηlaser2=3.0e-2 ηΔsound2=0.26 ηlaser:Δsound2=5.2e-3
Vector length at each laser power – SST activation	Fig 3K	21 lengths, 13 recordings	GLME	tlaser=−4.71 tΔsound=5.71 tlaser:Δsound=−3.57 DF = 815	***plaser=2.9e-6 ***pΔsound=1.6e-8 ***plaser:Δsound=3.8e-4	ηlaser2=6.1e-2 ηΔsound2=0.16 ηlaser:Δsound2=9.1e-2
Vector length at each laser power – VIP activation	Fig 3N	21 lengths, 16 recordings	GLME	tlaser=2.50 tΔsound=4.03 tlaser:Δsound=4.84 DF = 1004	*plaser=1.2e-2 ***pΔsound=6.1e-5 ***plaser:Δsound=1.5e-6	ηlaser2=9.7e-3 ηΔsound2=4.8e-2 ηlaser:Δsound2=9.0e-2
FIGURE 4
Sigmoid fit – offset – with SST activation	Fig 4D	None: 109 Med: 103 High: 64	GLME	tlaser=0.83 DF = 274	plaser=0.41	ηlaser2=2.4e-3
Sigmoid fit – offset – with VIP activation	Fig 4D	None: 267 Med: 239 High: 269	GLME	tlaser=1.74 DF = 773	plaser=8.1e-2	ηlaser2=3.6e-3
Sigmoid fit – range – with SST activation	Fig 4E	None: 109 Med: 103 High: 64	GLME	tlaser=1.24 DF = 274	plaser=0.22	ηlaser2=3.1e-3
Sigmoid fit – range – with VIP activation	Fig 4E	None: 267 Med: 239 High: 269	GLME	tlaser=3.11 DF = 773	**plaser=1.9e-3	ηlaser2=9.4e-3
Sigmoid fit – midpoint – with SST activation	Fig 4F	None: 109 Med: 103 High: 64	GLME	tlaser=2.65 DF = 274	**plaser=8.6e-3	ηlaser2=2.1e-2
Sigmoid fit – midpoint – with VIP activation	Fig 4F	None: 267 Med: 239 High: 269	GLME	tlaser=0.88 DF = 773	plaser=0.38	ηlaser2=5.9e-4
Sigmoid fit – width – with SST activation	Fig 4G	None: 109 Med: 103 High: 64	GLME	tlaser=−0.019 DF = 274	plaser=0.99	ηlaser2=8.8e-7
Sigmoid fit – width – with VIP activation	Fig 4G	None: 267 Med: 239 High: 269	GLME	tlaser=0.56 DF = 773	plaser=0.57	ηlaser2=2.2e-4
FIGURE 5
Gaussian fit – offset – with SST activation	Fig 5D	None: 224 Med: 175 High: 130	GLME	tlaser=−3.16 DF = 527	**plaser=1.7e-3	ηlaser2=1.8e-2
Gaussian fit – offset – with VIP activation	Fig 5D	None: 243 Med: 278 High: 310	GLME	tlaser=0.71 DF = 829	plaser=0.48	ηlaser2=5.3e-4
Gaussian fit – range – with SST activation	Fig 5E	None: 224 Med: 175 High: 130	GLME	tlaser=5.41 DF = 527	***plaser=9.4e-8	ηlaser2=3.8e-2
Gaussian fit – range – with VIP activation	Fig 5E	None: 243 Med: 278 High: 310	GLME	tlaser=5.60 DF = 829	***plaser=2.9e-8	ηlaser2=1.7e-2
Gaussian fit– mean – with SST activation	Fig 5F	None: 224 Med: 175 High: 130	GLME	tlaser=0.35 DF = 527	plaser=0.73	ηlaser2=1.9e-4
Gaussian fit – mean – with VIP activation	Fig 5F	None: 243 Med: 278 High: 310	GLME	tlaser=3.34 DF = 829	***plaser=8.6e-4	ηlaser2=7.4e-3
Gaussian fit – standard deviation – with SST activation	Fig 5G	None: 224 Med: 175 High: 130	GLME	tlaser=0.63 DF = 527	plaser=0.53	ηlaser2=7.3e-4
Gaussian fit – standard deviation – with VIP activation	Fig 5G	None: 243 Med: 278 High: 310	GLME	tlaser=3.96 DF = 829	***plaser=8.1e-5	ηlaser2=1.7e-2