In 18 mice (DANDI 000785), earlier and more odor-selective piriform responses of OB-projecting than mPFC-projecting neurons also occurred in unlabelled neurons of the same mice

Question: a preprint (Daste et al. 2025) reports that piriform neurons that project to the olfactory bulb (OB-p) respond and code odors earlier than neurons that project to mPFC (mPFC-p). Each mouse has only one retrograde label, so projection class is mixed with mouse, injection site and imaging field; in the 8-odor set, 4 of 5 OB-p mice were also imaged one year before all mPFC-p mice. Approach: we used the public data (DANDI 000785 v0.260520.1603; 18 mice, 61 sessions, 7,327 cells in passive sessions). The mouse was the unit of analysis, and the unlabelled (untagged) neurons in the same field of view were the reference in each mouse. We used exact permutation tests over mice, a planted-shift positive control and the preprint-style pipeline on both cell sets. Results: preprint-style tests reproduced the claim (pooled pairs p = 0.0005; 100-resample pseudo-population tests p down to 1e-29). The same resample test gave p = 4e-9 before odor onset, at chance accuracy, and p down to 3e-11 with untagged neurons only. At mouse level (10 OB-p vs 8 mPFC-p mice), untagged neurons of OB-p mice responded earlier (half-rise latency -0.42 s, 95% CI -0.65 to -0.18, p = 0.007) and were more odor-selective in 0-2 s (omega squared +0.015, p = 0.045), by about the same amount as labelled neurons (-0.31 s; +0.014). The labelled-minus-untagged index did not differ between groups (latency +0.11 s, CI -0.18 to +0.40; selectivity -0.001, p = 0.92). The untagged difference also occurred inside the 2021 and 2023 cohorts (p = 0.036). In Go/NoGo sessions (4 vs 4 mice), odor-phase valence decoding was higher in OB-p mice for labelled and untagged neurons alike; only the late action-phase index favoured mPFC-p neurons (p = 0.029, uncorrected, smallest possible p). Conclusion: most of the reported projection-class difference is a property of the labelled mice, not of the projection class. Limitations: few mice; the index is inconclusive in the first second (OB-p favoured, p = 0.35); the planted control shows that a mouse-level test detects projection-specific shifts near 1 s but not near 0.25 s. Same-mouse dual-colour labelling would remove the confound.

piriform cortex · olfaction · projection neurons · retrograde labelling · feedback · two-photon imaging · odor coding · response latency · pseudoreplication · mouse-level statistics · reanalysis · structure-function · DANDI 000785