Fig. 1: Neurons with unstable tuning properties form a stable representational map.

a, Schematic of the longitudinal Ca²⁺ imaging experiment in the mouse auditory cortex. On day 6, sham ablation was performed in the control cohort as a reference for other experimental cohorts described later (yellow arrow; see Methods); rAAV, recombinant AAV. b, Example imaging plane showing broad GCaMP6m and H2B::mCherry expression in layer 2/3 neurons; scale bar, 100 μm. c, Exemplary responses to auditory stimuli (19 PTs, 2–45 kHz; 15 CSs) from the neurons indicated in b. F, fluorescence. d, Normalized response profiles across imaging days for neurons categorized as sound responsive and sorted by stimulus, with the highest response amplitude on day 5. The gray arrow/bar marks the sham procedure in this control cohort. Response profiles of the same 1,156 neurons are shown in all panels (every fourth cell for illustrative purpose); NS, not significant. e, Normalized correlation of single-neuron stimulus tuning between day 5 and the other days (data are shown as mean ± s.e.m. across nine mice). Data were analyzed by one-way ANOVA (***P < 0.001) and two-sided paired t-test (day 3 versus day 7, day 1 versus day 9; P > 0.12). f, Same as d, but neurons were categorized as sound responsive and sorted by the highest response amplitude on each day (n indicates the number of significantly sound-responsive neurons on a given day). g, Same as e but baseline-normalized single-neuron correlation per day (baseline: average of days 1–5). Data were analyzed by one-way ANOVA; P = 0.77. h, Similarity matrix of population response vectors across stimuli, averaged across all FOVs per day (n = 59 FOVs). i, Normalized correlations from h averaged across diagonal (dark green) and all nondiagonal (light green) elements relative to baseline (data are shown as mean ± s.e.m. across nine mice). Data were analyzed by two-sided t-test with a false discovery rate (FDR) correction (*P < 0.05). See Supplementary Table 1 for details.

Source data