Claims 32
From the paper 10.1038/s41586-025-10096-w · claims file · checked 2026-10-21: ready
- roi-counts
Across 388 imaging sessions, the number of ROIs per session ranged from 21 to 286 (95 plus or minus 56, mean and s.d.), and was stable over time.
Across a total of 388 sessions, the number of ROIs per session ranged from 21 to 286 (95 ± 56 (mean ± s.d.)) and was stable over time
Results, The tuning of HD cells is stable over time, paragraph 1; Extended Data Fig. 1dHow it will be judged
ROIs are the CNMF_E segments of each imaging session (Methods, Miniscope recordings), counted per session over all imaging sessions of mice 1 to 10. The session count holds when 388 sessions are found; the paper does not say whether sessions of mice 6 to 10 are included, so the readings are (a) all imaging sessions and (b) mice 1 to 5 only. "Stable over time" has no test in the paper; it is judged per mouse within the Fig. 1 protocol as no monotonic trend, read as a Spearman correlation of ROI count with session number that is not significant at p < 0.05 in a majority of mice, and the correlations are shown.
- edf1e-hd-proportion
In the four-week circular protocol, the median proportion of ROIs detected as HD cells per animal was 0.2, 0.7, 0.4, 0.7 and 0.2, with 72, 42, 197, 48 and 102 ROIs per session on average, over 19, 20, 19, 19 and 20 recordings.
Median proportion ± s.d. per animal = 0.2 ± 0.03, 0.7 ± 0.04, 0.4 ± 0.05, 0.7 ± 0.05, 0.2 ± 0.04. n = 19, 20, 19, 19, 20 recordings per animal, with a respective mean ± s.d. of ROIs detected 72 ± 16, 42 ± 5, 197 ± 18, 48 ± 12, 102 ± 25
Extended Data Fig. 1e legendHow it will be judged
The ROI means use the s.d. given per animal as tolerance; the recording counts must match exactly. The HD cell is as in the opening comment, per session (split-half correlation at least 0.4 and session HDI at least 0.11 among ROIs whose across-session HDI exceeds 0.1). The paper does not say whether the CellReg and six-session conditions apply to this proportion; readings (a) with them and (b) without. Reproduced when, under one reading and one smoothing reading, at least four of the five proportions are within tolerance.
- edf1h-ring
An Isomap projection of the HD cell population activity forms a ring, reflecting the circular representation of head direction.
Low-dimensional projection of HD cell population activity using Isomap (Methods), a technique for... high-dimensional data by preserving its geometric structure, revealed a ring topology
Results, The tuning of HD cells is stable over time, paragraph 2; Extended Data Fig. 1hHow it will be judged
Method from the Methods (Dimensionality reduction). HD cells with split-half correlation above 0.8 and HDI above 0.1 in the session; dF/F smoothed with a Gaussian of sigma 133 ms, normalized (OPEN, readings z-score and division by the maximum), averaged in 0.25 s bins; bins where the mean normalized activity is below 0.1 dropped; Isomap with 50 neighbours to two dimensions. The paper judges the ring by eye with head direction as colour. The criterion is read from Extended Data Fig. 7d, which plots the Isomap angle against head direction, as a circular-circular correlation between the angle of each point around the embedding's centroid and the head direction, significant at p < 0.05 in an example session of each of the five mice. Reproduced when this holds for all five; partially reproduced for three or four. This test and its counts are the reader's, not the paper's, and are recorded as a deviation.
- fig1f-example-stable
In an example mouse, the population change in preferred direction relative to the last session was concentrated around 0 degrees (V-test, v = 17.06, P < 0.001, 18 sessions, median absolute deviation 5.0 degrees).
Population average of across-session angular difference in preferred direction, relative to the last session for one example mouse (V-test, not different from 0°, v = 17.06, two-tailed P < 0.001, n = 18 sessions, median absolute deviation from 0° = 5.0°)
Fig. 1f legend; Results, The tuning of HD cells is stable over time, paragraph 3How it will be judged
Per session, the change in preferred direction relative to the last session of the circular protocol (opening comment). The example mouse is not named in Fig. 1. Readings are (a) animal 4, which Extended Data Fig. 1g names as the animal of Fig. 1, and (b) any mouse with 18 compared sessions. Reproduced when, under one reading, V and the median absolute deviation are within tolerance and the V-test p is below 0.001.
- fig1g-all-mice-stable
Across all five mice, the population change in preferred direction relative to the last session was concentrated around 0 degrees (V-test, v = 75.21, P < 0.001, 84 sessions), with per-mouse means of 15.4, 2.1, -5.5, -7.8 and -6.9 degrees.
Distribution of population-wise angular difference relative to the last session for all five mice (V-test, not different from 0°, v = 75.21, two-tailed P < 0.001, n = 84 sessions). Per mouse... 15.4 ± 24.7°, 2.1 ± 18.7°, −5.5 ± 15.2°, −7.8 ± 23.6°, −6.9 ± 41.5° (mean ± s.d.)
Fig. 1g legendHow it will be judged
As fig1f-example-stable, pooled over the five mice. The per-mouse means use the per-mouse s.d. as tolerance. Reproduced when the pooled V is within tolerance with p below 0.001 and at least four per-mouse means are within tolerance.
- edf4ab-vtest-per-animal
The V statistic of the change in preferred direction relative to the last session was 14.1, 17.1, 17.3, 14.6 and 12.2 per animal, against shuffle medians of 3.0, 3.2, 5.7, 4.8 and 2.0, and 75.2 pooled against a shuffle median of 18.6; each animal's distribution was concentrated around 0 degrees (P < 0.001).
V statistic per animal = 14.1, 17.1, 17.3, 14.6, 12.2. Median of V statistic shuffles per animal = 3.0, 3.2, 5.7, 4.8, 2.0. Left... V-test statistic for pooled animals (75.2) versus the distribution of V statistic for within-day shuffled cell identity (median = 18.6)
Extended Data Fig. 4a,b legendsHow it will be judged
Shuffle of HD cell identity within a day, 1,000 times (Methods, Orientation remapping). The per-animal V of Extended Data Fig. 4b and the Fig. 1f value (17.06) do not agree for animal 4 (14.6); this is recorded, not resolved. Reproduced when at least four of the five per-animal V are within tolerance and each exceeds its shuffle median.
- edf4c-previous-session
The change in preferred direction relative to the previous session was also concentrated around 0 degrees across the five animals (V-test, P < 0.001, 75 sessions; pooled V 69.9 against a shuffle median of 21.9).
Distribution of population- wise angular difference relative to the previous session for all five animals (V-test, not different from 0°, two-tailed p < 0.001, n = 75 sessions)
Extended Data Fig. 4c,d legendsHow it will be judged
As edf4ab-vtest-per-animal, with each session compared to the previous session. Reproduced when the pooled V is within tolerance with p below 0.001 and at least four of the five per-animal V are within tolerance.
- edf2ab-jaccard
The Jaccard index of HD cell detection between sessions of the circular protocol remained high regardless of the delay between sessions.
The Jaccard index remained high regardless of... the delay between sessions, suggesting that most ROIs were... registered and continued to exhibit activity across time
Results, The tuning of HD cells is stable over time, paragraph 2; Extended Data Fig. 2a,bHow it will be judged
Method from the Methods (Jaccard score). Per mouse and pair of sessions, the Jaccard index of the binary vectors of presence (1) or absence (0) of HD tuning in each ROI, over the ROIs registered across the session pairs; the median over all session pairs with the same interval. OPEN, which ROIs count as registered across "all relevant session pairs", readings (a) ROIs registered in both sessions of the pair and (b) ROIs registered in every session of the protocol. The paper gives no test or level for "high". Reproduced when, in at least four of five mice, the median index at the longest interval is at least 80% of that at an interval of one day; this threshold is the reader's and is recorded as a deviation. The Spearman correlation of the median index with interval is shown.
- edf2-no-registration-confound
The shift in preferred direction between the first and last recording was not significantly related to the displacement of the ROI centroid in the field of view (p = 0.290) or to the CellReg score (p = 0.344), while centroid displacement was related to CellReg score (p = 0.004).
the change in preferred direction was not significantly correlated with the quality of cell registration nor the drift of the ROI in the Miniscope field of view
Results, The tuning of HD cells is stable over time, paragraph 3; Extended Data Fig. 2c-eHow it will be judged
A claim of absence, judged on the paper's test, the two-sided Wald test of the slope of a linear regression (as scipy.stats.linregress gives it) at 0.05 Bonferroni-corrected over the tests of the figure. Per HD cell registered in the first and last recording of the circular protocol, the shift in preferred direction (cross-correlation, opening comment), the centroid displacement and the CellReg score. OPEN, readings for which cells give n = 72, (a) pooled over the five mice and (b) the Fig. 1 example mouse; and for the shift, (i) signed and (ii) absolute. Reproduced when the two absence tests are not significant and the displacement against score test is significant; the slopes are shown.
- fig1h-drift-rate
The absolute change in preferred direction grew slowly with the days between recordings, at a drift rate of 7.1 plus or minus 4.4 degrees per week across mice.
we computed the absolute change in preferred direction as a function of days passed between recordings and observed... a slow drift rate (7.1 ± 4.4° per week)
Results, The tuning of HD cells is stable over time, paragraph 3; Fig. 1hHow it will be judged
Per pair of sessions of the circular protocol, the absolute change in preferred direction; OPEN, readings for the value (a) the absolute value of the session-level circular mean over cells and (b) the median over cells of each cell's absolute change. Drift rate per mouse by the readings of the opening comment, averaged over the five mice. Reproduced when the mean is within tolerance and positive under one combination of readings.
- edf4ef-not-diffusive
The drift in preferred direction was not diffusive (not a random walk) but fluctuated around a fixed direction, as shown by the cumulative variance of orientation across days.
Analysis of the cumulative variance of orientation across days suggested that this drift, although modest, was not diffusive (that is, not a random walk) but instead... around a fixed direction
Results, The tuning of HD cells is stable over time, paragraph 3; Extended Data Fig. 4e,fHow it will be judged
Extended Data Fig. 4e plots the mean variance of the absolute change in preferred direction against the number of recordings included, from the first to the last day. The paper gives no test. A random walk predicts a variance growing in proportion to the number of recordings. OPEN, the variance is read (a) as that of the session-level changes relative to the last day over the first k recordings and (b) relative to the first day. Reproduced when, under the reading run, the variance at the full number of recordings is less than half the value a line through the origin fitted to the first five points predicts, in at least four of five mice. This threshold is the reader's and is recorded as a deviation.
- edf4f-change-below-shuffle
The absolute change in overall preferred direction with respect to the last day was smaller than for shuffled cell identities in every animal (median 15.9, 5.1, 7.9, 26.3 and 24.0 degrees; one-tailed Wilcoxon, p from 3.8e-6 to 1.5e-4).
Absolute change in overall preferred direction, per session, with respect to the last day of the experiment in Fig. 1. Median absolute change per animal in degrees... 15.9, 5.1, 7.9, 26.3, 24.0
Extended Data Fig. 4f legendHow it will be judged
Per session the absolute change against the same session's shuffled-identity value. OPEN, the shuffle value per session is read as (a) the median and (b) the mean over 1,000 shuffles. Reproduced when every animal's one-tailed Wilcoxon is below 0.05 and at least three medians are within tolerance.
- edf4g-pairwise-offset-stable
The change in pairwise angular offset across all registered HD cell pairs was strongly biased towards 0, with per-animal medians of 12, 9, 12, 24 and 12 degrees against shuffle medians of 51 to 54 degrees (p < 0.001 per animal, 25,901 pairs).
The change in pairwise angular offset across all registered cell pairs (... of interleaving days) was strongly biased towards 0 (Extended Data Fig. 4g), suggesting highly coherent population dynamics
Results, The tuning of HD cells is stable over time, paragraph 4; Extended Data Fig. 4gHow it will be judged
Method from the Methods (Change in the angular offset) over all session pairs of the circular protocol. The per-animal tolerances are the s.d. that Extended Data Fig. 4g gives with each median. Reproduced when every animal's median is below all 1,000 shuffled medians and at least four medians are within tolerance.
- fig1i-offset-drift
The change in pairwise angular offset drifted more slowly (4.1 plus or minus 4.8 degrees per week) than single cell tuning, though not significantly so.
the change in pairwise angular offset drifted more slowly (4.1 ± 4.8° per week), yet this was not significantly different from single cell tuning drift rate
Results, The tuning of HD cells is stable over time, paragraph 4; Fig. 1iHow it will be judged
Per pair of sessions, the median over cell pairs of the absolute change in offset (Fig. 1i axis, "Median delta angular offset"); drift rate per mouse by the readings of the opening comment. The test against the fig1h-drift-rate values is not named. A claim of absence; OPEN, readings (a) two-sided Wilcoxon signed-rank and (b) paired t-test across the five mice, at 0.05 (with five mice, reading (a) cannot reach p below 0.0625). Reproduced when the mean is within tolerance and neither test is significant; not reproduced when either is. Effect size shown.
- fig1k-decoding-across-days
Decoders of head direction trained on one session and tested on another decoded significantly better than chance for nearly the whole four weeks of the experiment.
we trained and tested non-linear decoders (extreme gradient-boosted trees (Methods)) on every recording session pair and observed statistically significant decoding for nearly the entire duration of the experiment
Results, The tuning of HD cells is stable over time, paragraph 5; Fig. 1j,kHow it will be judged
Decoder as in the opening comment, on every ordered pair of sessions of the circular protocol, error per pair as the median over time bins, chance from time-reversed head direction in the test session (both median and mean error are run, as the Methods and legend differ). The paper does not name its test. OPEN, readings (a) per day difference, one-sided Wilcoxon signed-rank of data against chance over all session pairs of all mice, and (b) per day difference, one-sided Wilcoxon across the five mice of per-mouse means. Reproduced when, under one reading, the error is below chance at p < 0.05 for every day difference up to 20 days except at most one, read as "nearly the entire duration"; partially reproduced when that holds up to 10 days.
- edf4ij-pv-decay
The population vector coherence decayed slowly over the course of the recording.
This is in line with a slow decay in the population vector coherence over the course of the recording
Results, The tuning of HD cells is stable over time, paragraph 5; Extended Data Fig. 4h-jHow it will be judged
Population vector correlation between two sessions with activity in 12 head-direction bins (Extended Data Fig. 4h). OPEN, readings (a) the mean over bins of the correlation across cells of the two sessions' population vectors at the same head-direction bin and (b) the correlation of the two flattened cell by bin matrices. Reproduced when the correlation decreases with the days between sessions (negative Spearman correlation, p < 0.05, in at least four of five mice) and its mean at the longest day difference stays above 0.
- fig2e-offset-across-environments
Across the four environments, HD cells preserved their pairwise angular offsets, with per-animal median changes of 24, 12, 18, 29 and 12 degrees against shuffle medians of 54 to 57 degrees, and a drift rate of 1.9 plus or minus 1.02 degrees per week.
We observed the same tendency of preserved pairwise angular difference in preferred directions across... all sessions and mice (Fig. 2e and Extended Data Fig. 6)
Results, Internal organization across time and environments, paragraph 1; Fig. 2e; Extended Data Fig. 6bHow it will be judged
Over all session pairs of the 40-session multi-environment protocol. The per-animal tolerances are the s.d. that Extended Data Fig. 6b gives with each median. The Methods exclude mouse 2 from this analysis while Extended Data Fig. 6b lists five values; readings (a) mouse 2 excluded, (b) mouse 2 included, (c) mouse 2 with only pairs within one Miniscope epoch. Reproduced when, under one reading, every included animal's median is below all 1,000 shuffled medians and the mean drift rate (opening comment readings) is within tolerance.
- edf5c-stable-within-environment
The tuning of HD cells was stable within each environment, with changes in preferred direction relative to the last day concentrated around 0 degrees in all four, and no difference between environments (repeated-measures ANOVA, F = 0.44, P = 0.73).
The tuning of HD cells was stable within each environment for all mice
Results, Internal organization across time and environments, paragraph 1; Extended Data Fig. 5cHow it will be judged
Per environment, the change in preferred direction of each session relative to the last session of that environment (for mouse 2 in the square, the next-to-last day, as the legend says). The ANOVA is a claim of absence judged on a repeated-measures ANOVA over environments with mice as subjects (n = 5, giving 3 and 12 degrees of freedom) at 0.05; the per mouse and environment value is OPEN, readings (a) mean absolute change and (b) median absolute change. Reproduced when all four V-tests are significant, at least three V are within tolerance and the ANOVA is not significant.
- edf7-ephys-offsets
Electrophysiological recordings confirmed that pairwise angular offsets of HD cells are preserved across environments, with per-animal median changes of 18, 12, 27 and 15 degrees against shuffle medians of 54 degrees (p < 0.001).
To confirm that this long-tailed distribution is not specific to the imaging... we confirmed these results with electrophysiological recordings
Results, Internal organization across time and environments, paragraph 2; Extended Data Fig. 7eHow it will be judged
HD cells with firing rate above 1 Hz and HDI above 0.3; spike-count tuning curves (bins and smoothing OPEN, the imaging values as the reading); offsets compared across the three environment pairs of the day. Tolerances are the s.d. given with each median. Reproduced when every animal's median is below all 1,000 shuffled medians.
- edf7g-waveforms
Electrophysiologically recorded HD cells have broader waveforms than non-HD cells (median trough-to-peak 0.54 against 0.295 ms, half-peak width 0.21 against 0.165 ms).
Note HD cells tend to have broader waveforms, consistent with pyramidal cells. Median HD trough-to-peak = 0.54 ms, median HD half-peak width = 0.21 ms, versus 0.295 and 0.165 respectively for non-HD cells; n = 254 cells
Extended Data Fig. 7g legendHow it will be judged
Mean waveform per unit on its largest channel; trough-to-peak from trough to the following maximum, half-peak width the width at half the trough amplitude. These definitions are not in the paper and are OPEN; they are the standard ones. Reproduced when at least three of the four medians are within tolerance and HD medians exceed non-HD medians for both.
- edf8-same-vs-mixed
For recordings 7 days apart, decoders trained and tested in the same environment tended to be better than across environments, and the change in pairwise offset tended to be smaller, neither significantly (p = 0.063 for both).
Decoders of head direction of mice trained and tested in the same environment one week apart showed a tendency for better (yet non-significant) accuracy than decoders trained across environments with the same delay between sessions
Results, Environmental modulation of the HD cell population, paragraph 1; Extended Data Fig. 8b,cHow it will be judged
Pairs of sessions 7 days apart in the multi-environment protocol, "single" when both are in one environment and "mixed" otherwise; per mouse the mean over pairs. The decoder measure is the variance of the error between predicted and actual direction (OPEN, readings the circular variance and the linear variance of the signed circular error in radians). The tendency claim is reproduced when every mouse is lower for single than mixed (W = 0) under one reading and the means are within tolerance.
- fig2g-environment-classifier
Classifiers of environment identity trained on pairwise correlations of HD cell tuning curves were more accurate than shuffles (proportion correct 0.7 against 0.3; one-sided Wilcoxon, P = 0.031, W = 15).
These classifiers showed higher accuracy than controls (Fig. 2f,g)... the idea that the HD signal in the PoSub is environment specific
Results, Environmental modulation of the HD cell population, paragraph 1; Fig. 2f,gHow it will be judged
Method as in the opening comment. The random half split is OPEN; the criterion runs 100 random splits per mouse and uses the mean proportion correct, and also a single split with a fixed seed, as readings. Reproduced when, under one reading, the one-sided Wilcoxon of data against shuffle across the five mice has p below 0.05 and the means are within tolerance.
- fig2i-random-remapping
The remapping of the HD system between the three environments of one room seemed random across mice (Rayleigh test, P = 0.36 circular to square, 0.61 square to radial maze, 0.032 radial maze to circular).
Notably, this remapping seemed random across mice, although more data would be required to support a true random remapping
Results, Environmental modulation of the HD cell population, paragraph 2; Fig. 2h,iHow it will be judged
Average remapping per mouse and environment pair as in the opening comment, over all session pairs of the two environments, in room coordinates; mouse 2 included (Methods, Inclusion and exclusion criteria). Rayleigh's test over the five mice. Reproduced when the first two tests are not significant at 0.05 and the third is, matching the paper, with z within tolerance; partially reproduced when the first two are not significant but the third pattern differs.
- fig2h-coherent-remapping
While the internal structure of the HD system was largely kept across environments, its absolute orientation was specific to each environment, the HD cells of a mouse shifting together (coherent remapping) compared with shuffled cell identities.
Whereas the internal structure of the HD system was largely preserved across environments, with only minor... its absolute orientation was specific to each environment
Results, Environmental modulation of the HD cell population, paragraph 2; Fig. 2c,d,h,iHow it will be judged
Per mouse and pair of environments, the mean vector length over HD cells of the shift in preferred direction, for each pair of recordings of the two environments (Methods, Orientation remapping), compared with 1,000 shuffles of cell identity within a day. The paper gives no threshold; the 95th percentile of the shuffle distribution, the bound Fig. 3e draws for its controls, is used. Per mouse and environment pair, coherent when the median over recording pairs of the mean vector length exceeds the median over recording pairs of the shuffle 95th percentile. The environment pairs are the three of Fig. 2i (circular to square, square to radial maze, radial maze to circular), the environments of one room whose alignment the paper measures in room coordinates; the open field, in another room, is shown as a variant. Reproduced when this holds for at least 80% of the mouse and environment pairs (mouse 2 included, as for remapping in the Methods), and the average remapping of at least one environment pair differs from 0 by more than 30 degrees in at least three mice, read as an orientation specific to each environment. The 80% and 30 degree thresholds are the reader's, not the paper's, and are recorded as a deviation.
- fig3de-probe-memory
After a delay without visits, the orientation of the HD system in each environment was retained, drifting slowly with the delay; it was concentrated around 0 degrees for delays under 1, 2 and 4 weeks (V-test, P = 0.002, 0.02, 0.03) but not at 6 weeks (P = 0.191), with mean absolute changes of 3.6, 23.9, 39.6 and 70.0 degrees.
the average alignment of the HD system to external environments across mice drifted gradually, albeit slowly, as the interval between sessions increased, becoming significantly... from 0° only at the 6-week delay
Results, HD system forms long-term memories of orientation, paragraph 1; Fig. 3dHow it will be judged
Per mouse (mice 1 to 4; mouse 5 excluded, Methods) and environment, the remapping between the last visit of the two-week protocol and the probe trial (Methods, Orientation remapping), HD cells with CellReg score above 0.6 detected in at least two sessions. The delays are those of the environment order, open field under 1 week, radial maze 2 weeks, square 4 weeks, circular 6 weeks (OPEN, the probe order is not given; the delays are taken from the session dates). Reproduced when the V-tests are significant at 0.05 for the first three delays and not at 6 weeks, and the mean absolute change increases with delay; partially reproduced when the 6-week test is not significant but one earlier delay also fails.
- fig3e-mvl-delay
The mean vector length of the remapping angles across mice decreased with delay (0.998, 0.79, 0.79 and 0.536 for under 1, 2, 4 and 6 weeks), compared with shuffled HD cell identities.
Mean vector length of animal remapping angles (that is, probe versus last visit of each environment, for n = 4 mice) as a function of delay between visits compared to controls (shuffled HD cell identity)
Fig. 3d (the mvl values printed in the panel) and Fig. 3eHow it will be judged
Mean vector length of the four mice's remapping angles at each delay, from fig3de-probe-memory. Reproduced when at least three of the four values are within tolerance and the 6-week value is the lowest.
- fig3fg-probe-drift
The absolute change in individual preferred directions drifted with delay at 10.8 plus or minus 0.1 degrees per week, while the pairwise angular offsets stayed coherent, drifting at 2.5 plus or minus 0.1 degrees per week.
We further tested this drift by assessing the absolute changes in individual neuronal preferred directions and observed a similar drift from the original direction (Fig. 3f). Of note, the population structure maintained its coherence across these recordings
Results, HD system forms long-term memories of orientation, paragraph 1; Fig. 3f,gHow it will be judged
The legend calls its spread (0.1, for both rates) the s.d. over mice; a spread this small for four mice is doubtful, so 10% of the value is used as tolerance, and whether the rates fall within 0.1 of the values is also shown. Values against delay in weeks, drift rate per mouse by the readings of the opening comment, mean over mice 1 to 4. The Methods exclude mouse 2 from the 6-week delay "of pairwise angular offset (Fig. 3f)"; readings (a) mouse 2 excluded at 6 weeks from the pairwise offset (Fig. 3g) and (b) from the preferred direction (Fig. 3f). Reproduced when both rates are within tolerance under one combination of readings; partially reproduced when the preferred-direction rate exceeds the pairwise-offset rate but a value is outside tolerance.
- edf9-interference
After four further weeks of visits to the four environments in random order, the system showed reduced retention of the original preferred direction, though changes relative to week 1 stayed centred on 0 degrees (V = 11.3, 10.1, 10.8), and pairwise offsets drifted at a mean of 1.6 degrees per week.
this orientation memory may be affected by cognitive load as after four additional weeks of extensive experience with the different... the systems showed reduced retention of the original preferred direction
Results, HD system forms long-term memories of orientation, paragraph 1; Extended Data Fig. 9b,dHow it will be judged
V-tests against 0 degrees over 20, 20 and 18 sessions (weeks 2, 3 and 4). The legend gives no p and says only that the V-tests are significant; p below 0.05 is the reading. "Reduced retention" has no test in the paper. OPEN, it is read as the mean over mice of the absolute change in preferred direction relative to week 1 at week 4 (Extended Data Fig. 9e) being larger than the mean absolute change at the 4-week probe delay of fig3de-probe-memory. Reproduced when the V statistics and drift rate are within tolerance and that comparison holds.
- fig4d-single-visit-memory
After a single visit, the orientation of the HD cell population was retained four weeks later as well as with weekly visits; the change in preferred direction from first to last session was smaller than shuffles in both groups (one-sided Wilcoxon, P = 0.031) and did not differ between groups (P = 0.438; weekly 12.4, single 6.6 degrees).
We observed a highly maintained absolute orientation of the HD cell... for both environments
Results, Memory from a single visit is stable over weeks, paragraph 1; Fig. 4dHow it will be judged
Mice 6 to 10; per mouse and environment, the absolute change in preferred direction between the first and the last session in that environment, HD cells detected in at least two sessions with CellReg score above 0.8 (Methods; the 0.6 score is allowed for Fig. 3 only); shuffle of cell identity within a day, 1,000 times (Methods, Orientation remapping). OPEN, the shuffle value per mouse, readings (a) the median and (b) the mean over the 1,000 shuffles. Reproduced when both one-sided tests have p below 0.05, the two-sided group test is not significant at 0.05 (a claim of absence on the paper's test), and the means are within tolerance.
- fig4e-decoder-single-visit
A decoder of head direction trained on the first and tested on the last session was accurate for both environments, with no difference between the weekly and single groups (weekly 35.9, single 43.9 degrees median error; P = 0.438).
This was further supported by the high accuracy of a decoder of head direction for both environments (not significantly different from other animals)
Results, Memory from a single visit is stable over weeks, paragraph 1; Fig. 4eHow it will be judged
Decoder as in the opening comment, trained on the first and tested on the last session of each environment for mice 6 to 10. The group test is a claim of absence on the paper's test, two-sided Wilcoxon at 0.05. "High accuracy" is read as every mouse's error below its time-reversed chance. The comparison with "other animals" has no named test or data and is not tested. Reproduced when the means are within tolerance, the group test is not significant and every error is below chance.
- edf10c-environment-identity
The change in preferred direction from first to last visit did not differ between the two environments (circular mean 1.6 against 5.1 degrees; two-sided Wilcoxon, p = 1.0, n = 10).
Change in preferred direction first versus last visit, classified by environment identity. Circular mean ± s.d. in degrees... 1.6 ± 14.1, 5.1 ± 8.0. Two-sided Wilcoxon signed-rank p = 1.0, W statistic = 7.0, n = 10
Extended Data Fig. 10c legendHow it will be judged
A claim of absence on the paper's test, a two-sided Wilcoxon signed-rank at 0.05, paired by mouse across the two environments of mice 6 to 10 (n = 10 read as 5 mice by 2 environments). The legend does not say which environment is which; the readings are both assignments of square and circular. Reproduced when the test is not significant and the means are within tolerance under one assignment.
- edf10b-per-animal-change
Per animal of the single-visit protocol, the circular mean change in preferred direction relative to the first visit was 4.7, -0.4, -15.5, 5.4 and -6.2 degrees.
Circular mean ± s.d. per animal with pooled environments in degrees... 4.7 ± 3.4, −0.4 ± 5.8, −15.5 ± 34.7, 5.4 ± 10.8, −6.2 ± 10.9
Extended Data Fig. 10b legendHow it will be judged
Per session of mice 6 to 10, the change in preferred direction relative to the first visit of its environment, pooled over both environments. The mapping of the legend's order to mice 6 to 10 is taken as ascending. Reproduced when at least four of five are within tolerance.
Pipeline
- Paper
- done (1 of 1)
- Claims
- done
- Claims check
- done (ready)
- Attempt
- todo
- Questions
- none
- Verified
- none