Noise-invariant representations of sound emerge along the canonical cortical hierarchy. Tomas Suarez Omedas and Ross S. Williamson, 2026

Documentation for the data and code used for this publication

Neural Data

All data used in this publication is stored in the repository [insert link / doi]. Each subpopulation is stored in a separate .mat file, each containing a single variable with a consistent structure. The fields are described below.

allNeural

Field	Type	Description
animal_id	string	Identifier of the mouse from which this field of view was recorded
zscores	double	Trial-by-trial mean z-scored spike rate during the stimulus-evoked window (10 frames after stimulus onset)
zscores_spont	double	Trial-by-trial mean z-scored spike rate during the spontaneous activity window (15 frames before stimulus onset)
dff	double	Trial-by-trial mean ΔF/F during the stimulus-evoked window (10 frames after stimulus onset)
dff_spont	double	Trial-by-trial mean ΔF/F during the spontaneous activity window (15 frames before stimulus onset)
freq_sequence	double	Ordered list of pure tone frequencies presented during the session
freq_index	int	Trial-by-trial index into freq_sequence indicating which frequency was presented on each trial
level_sequence	double	Ordered list of pure tone intensities presented during the session
level_index	int	Trial-by-trial index into level_sequence indicating which intensity was presented on each trial
session_sequence	double	Ordered list of session types defined by background noise condition; by convention, the first entry is always "No-BN"
session_index	int	Trial-by-trial index into session_sequence indicating the background noise condition on each trial
xy_pix	double	Spatial coordinates (in pixels) of each neuron within the field of view
mean_Img	double	Mean fluorescence image of the field of view
imagingParams	struct	Imaging acquisition parameters; particularly relevant for analyses involving physical distances between neurons (see below)

imagingParams

Field	Type	Description
pixelX	int	Number of pixels along the horizontal axis of the field of view
pixelY	int	Number of pixels along the vertical axis of the field of view
objective	string	Microscope objective used for this recording
zoom	int	Galvo-resonant zoom level used for this recording
pixelSizeUM	double	Pixel size in µm; used to convert pixel coordinates to physical distances

Code

Data analysis is organized across five analysis scripts, each saving its output to a temporary folder. A set of dedicated plotting scripts then read from these folders to generate the figure panels. The tables below link each script to its corresponding figure.

Analysis Scripts

Script	Folder	Associated figure	Description
matchedComparison	Single-Cell-Summaries	Fig. 2 Fig. S1 Fig. S2	Computes average tuning curves and fits RMA regression models
calculateMI_AllPairs	Single-Cell-Summaries	Fig. 3	Computes pairwise mutual information across all neuron pairs
findAllNCs	Single-Cell-Summaries	Fig. 4	Computes signal and noise correlations
decodeAllSessions	Masked-Noise-Decoding	Fig. 5 Fig. S4	Runs detection and discrimination decoding analyses
findRepSpace	Manifold-Analysis	Fig. 6A–C	Characterizes the geometry of the neural representational manifold
findManGeo	Manifold-Analysis	Fig. 6D–F	Quantifies manifold size and related geometric metrics

Plotting Scripts

Script	Folder	Figure panels
plotAllMeanTuning	Single-Cell-Summaries	Fig. 2B
plotBetasAcrossTypes	Single-Cell-Summaries	Figs. 2F, G
plotMIAcrossTypes	Single-Cell-Summaries	Figs. 3C–E
plotAllPairCorr	Single-Cell-Summaries	Figs. 4C–F
plotAllAccuracies	Masked-Noise-Decoding	Figs. 5C, D, G, H
plotGeometryOutput	Manifold-Analysis	Figs. 6B, C
plotManifoldMetrics	Manifold-Analysis	Figs. 6D–F