Claims 50
From the paper A flexible hippocampal population code for experience relative to reward · claims file · checked 2026-10-21: ready
- fig1h-lick-ratio-improves
After each reward switch, every switch mouse improved its anticipatory licking before the new reward zone and kept that precise licking, as measured by the anticipatory lick ratio over blocks of ten trials.
All mice improved and retained precise licking preceding the new reward zone (Fig. 1h ; n = 11 mice), demonstrating accurate spatial learning and memory
Results, Monitoring neural activity during a reward learning task, paragraph 2; Figure 1hHow it will be judged
The anticipatory lick ratio as the Methods define it (Quantification of licking behavior), (lick in - lick out)/(lick in + lick out), lick in being the mean lick rate in the 50 cm before the reward zone start and lick out the mean rate outside that zone and the reward zone, licks binned in 10 cm bins and divided by occupancy, trials with sensor errors (more than 30% of frames with a cumulative lick count above 2) removed; computed per trial and averaged over blocks of ten trials within each reward zone, as Fig1_Ext1.md:474-690 and behavior.py:876-951 do (np.array_split into ceil(n/10) blocks). The paper reports no test and does not say what "improved" is measured against. Readings (a) for each mouse and each switch day, the mean ratio of the last block after the switch exceeds that of the first block after the switch; (b) for each mouse, the mean over its switch days of (last block after minus first block after) is positive. Reproduced when all 11 mice improve under a reading and, for "retained", each mouse's mean ratio over the first block of the day after each switch day (a stay day, or the next switch day's pre-switch trials) is above 0. Partially reproduced when 9 or 10 mice meet it.
- place-cell-fraction
On switch days, 459 +/- 263 of 954 +/- 453 imaged cells per session were place cells, 48.5 +/- 14.5% (mean +/- s.d. across 11 mice and seven switch sessions).
we identified place cells as cells with significant spatial information (SI) either before or after the reward switch (means ± s.d. across 11 mice, seven switch sessions... 459 ± 263 place cells out of 954 ± 453 cells imaged; 48.5 ± 14.5% identified as place cells)
Results, Moving reward induced remapping spanning the environment, paragraph 1How it will be judged
Place cells as the Methods define them (Place cell identification), from deconvolved activity in 45 bins of 10 cm at speed of 2 cm/s or more, SI as the Methods' formula, 100 shuffles per cell circularly shifting position against activity independently on each trial by ~1 s to the trial length, significant when the cell's SI exceeds 95% of the shuffled SI of all cells of the animal in that session and trial set pooled (code, spatial.py:39-56 and 1260-1289, make_multi_anim_sess.md:79-141), in trial set 1 or 2. Putative interneurons excluded first. Means over the 77 mouse and switch-day sessions. The values match when the three means are within tolerance. The paper's "means +/- s.d. across 11 mice, seven switch sessions" can also be read as the mean over mice of per-mouse means; that reading is shown as a variant.
- remap-category-fractions
After a switch within an environment, 21.4 +/- 7.9% of place cells kept a field at the same track position (track-relative, TR), 11.2 +/- 7.0% lost their field (disappearing), 6.8 +/- 4.0% gained one (appearing), 4.8 +/- 3.3% followed the reward within 50 cm of both zone starts (remap near reward) and 15.6 +/- 3.6% remapped with fields far from reward (remap far from reward).
a subset of place cells maintained a stable field at the same track location (‘track-relative’ (TR) cells, 21.4 ± 7.9% of place cells, mean ± s.d. across six switches within environment, 11 mice ... We observed cells with place fields that disappeared (‘disappearing’; 11.2 ± 7.0%), appeared (‘appearing’; 6.8 ± 4.0%) or precisely followed the reward location ... (‘remap near reward’; 4.8 ± 3.3%) ... (‘remap far from reward’; 15.6 ± 3.6%
Results, Moving reward induced remapping spanning the environment, paragraph 1; Figure 2a; Extended Data Fig. 2a-dHow it will be judged
Categories as the Methods define them (Remapping category definitions) on the six within-environment switch days (3, 5, 7, 10, 12, 14; the code takes day != 8, Fig2_Ext2_identifyRewardRelative.md:365-408), over place cells, with spatial peaks from the trial-averaged unsmoothed binned dF/F (Spatial peak firing identification). TR, significant SI before and after, peaks no more than 50 cm apart; disappearing, significant before only, mean binned dF/F after below the 50th percentile of the per-trial mean dF/F before; appearing, significant after only, mean after above the mean before plus one s.d. of the per-trial means; near reward, significant before and after, both peaks within 50 cm of their zone starts; far, significant before and after, not TR, a peak more than 50 cm from at least one zone start. The Methods remove RR cells (claim rr-cell-fractions) from the appearing and disappearing groups only; the code removes them from every category (dayData.py:2156, exclude_rr_from_others=True). Readings (a) the Methods' removal, (b) the code's. The values match under a reading when all five are within tolerance.
- ed2cd-switch-remapping-intermediate
A reward switch within a constant environment induced more remapping than a stay day with a fixed reward and environment, and less than a reward switch combined with a novel environment.
At the population level, a reward switch within a constant environment induced more remapping than a fixed reward and environment 28 , 29 but less remapping than the introduction of a novel environment
Results, Moving reward induced remapping spanning the environment, paragraph 1; Extended Data Fig. 2b-dHow it will be judged
As the Extended Data Fig. 2b,c legend gives it. For each mouse and day, the unsmoothed 2D histogram of place-cell peaks (cells with significant SI before and after) in 10 cm bins, before against after the switch (first against second half on stay days), as a fraction of those cells; "diagonal" is peaks within 50 cm of the same position before and after, excluding near-reward bins, "near reward" peaks within 50 cm of both zone starts, "all other remapping" the rest. Each mouse's mean over stay days (1, 2, 4, 6, 9, 11, 13), switch-within days (3, 5, 7, 10, 12, 14) and day 8; logit-transformed; pairwise paired two-sided t-tests, Holm-corrected within the category (pingouin pairwise_ttests, Fig2_Ext2_identifyRewardRelative.md:1292-1666). Reproduced when "all other remapping" is ordered stay < switch within < across environment and "diagonal" stay > across environment and switch within > across environment, each at corrected p < 0.05; partially reproduced when the order holds for one category only. The code's diagonal band (np.tri, offsets of -4 to +5 bins) is shown as a variant of the legend's symmetric 50 cm.
- ed2f-teleport-distance
The distance run in the teleport zone did not predict the next lap's spatial firing error for ~93% of RR cells; 6.7% of RR cells had a significant Spearman correlation on trials before the switch and 7.5% after.
The distance run in the teleport zone did not predict spatial firing variability on the following lap in the vast majority (~93%) of RR cells
Results, Reward-relative remapping both near and far from reward, paragraph 1; Extended Data Fig. 2e,fHow it will be judged
As the Methods give it (Relationship between distance run in the teleport zone and trial-to-trial variability), for each RR cell and trial set, the distance run in the teleport period after each trial (speed integrated from the first teleport frame to the last frame before the track restarts) against the next trial's signed spatial peak error (the trial's peak minus the set's trial-averaged peak, circular error converted to cm), Spearman correlation, significant at p < 0.05 (Extended Data Fig. 2f). The code holds no version of this analysis. The values match when both percentages are within tolerance.
- fig2e-rr-remapping-above-chance
On the first switch day the fraction of remapping place cells whose firing kept the same position relative to reward exceeded a random-remapping shuffle.
We found that the fraction of place cells that exhibited RR remapping across animals (Fig. 2d ) exceeded chance on the first switch day (Fig. 2e , top)
Results, Reward-relative remapping both near and far from reward, paragraph 2; Figure 2d,eHow it will be judged
As the Methods give it (Quantification of RR remapping compared to chance). Cells with significant SI before and after the switch, TR cells (peaks within 0.698 rad, 50 cm) removed; peaks re-computed on circular coordinates (0-450 cm to -pi..pi) and expressed relative to each zone start; the circular difference of relative peaks (after minus before), pooled over mice, histogrammed in 45 bins of 2pi/45 as a fraction of cells; the shuffle keeps each pre-switch peak and circularly shifts the post-switch trial-averaged map 1,000 times by 0-44 bins. The above-chance fraction is the summed excess over the per-bin 95th percentile of the shuffle, over bins whose centres lie within 0.698 rad of zero (dayData.py:1541-1741). Reproduced when on day 3 at least one such bin exceeds the shuffle's 95th percentile and the above-chance fraction is positive. The code's 2pi/44 map bins and 0 to nbins-1 offsets are shown as a variant.
- fig2h-rr-remapping-grows
The above-chance fraction of RR remapping cells grew linearly with task experience over the seven switch days.
Above-chance fraction of RR remapping cells grows linearly with task experience
Results, Reward-relative remapping both near and far from reward, paragraph 2; Figure 2h; Extended Data Fig. 2hHow it will be judged
The above-chance fraction as in fig2e-rr-remapping-above-chance, per switch day. Figure 2h, pooled over mice per day, a linear regression on switch number 1-7 (regression.linear_reg, p from scipy.stats.linregress, Fig2_Ext2_identifyRewardRelative.md:665-680). Extended Data Fig. 2h, per mouse and day, the logit of the fraction in a linear mixed-effects model with switch index as fixed effect and a random intercept per mouse (statsmodels mixedlm, REML, Wald test; fractions of exactly 0 or 1 dropped, Fig2_Ext2_identifyRewardRelative.md:744-755). The paper prints neither p. Reproduced when both slopes are positive at p < 0.05; partially reproduced when one is.
- fig2eg-mean-rr-position-after-reward
The mean reward-relative firing position of cells remapping within 50 cm relative to reward was significantly after the reward zone start, 0.441 rad on the first switch day (n = 453 cells) and 0.292 rad on the last (n = 645 cells).
The mean of the distribution of RR firing positions was significantly greater than zero (Fig. 2e , bottom), indicating that more RR place fields are in locations following reward delivery
Results, Reward-relative remapping both near and far from reward, paragraph 2; Figure 2e,g (bottom) legendHow it will be judged
The cells of fig2e-rr-remapping-above-chance whose relative peaks before and after lie within 0.698 rad of each other, pooled over mice; each cell's value is the circular mean of its two relative peaks (dayData.py:1714-1719); pycircstat mtest against 0 at alpha 0.05 gives the 95% interval. Reproduced when on both days the mean is positive, the interval excludes zero, and the means are within tolerance.
- fig2i-rr-remapping-spans-track
After excluding cells with peaks within increasing distances on either side of both reward zone starts, above-chance RR remapping remained, spanning nearly the whole track by the last switch day.
With these exclusions, we observed that above-chance RR remapping spans nearly the entire track length by the last switch day
Results, Reward-relative remapping both near and far from reward, paragraph 3; Figure 2iHow it will be judged
As fig2e-rr-remapping-above-chance, after removing cells with peaks within an exclusion distance of both zone starts, for exclusion distances of 10 to 190 cm in 10 cm steps (Fig2_Ext2_identifyRewardRelative.md:775-795). The paper gives no number for "nearly the entire track". Reproduced when on day 14 the above-chance fraction is positive at every exclusion distance up to 190 cm, the largest the code tests; partially reproduced when it is positive up to at least 50 cm (the exclusion of Extended Data Fig. 2i-l) but not to 190 cm. The largest distance with a positive fraction is reported for days 3 and 14.
- ed2jl-rr-remapping-beyond-50cm
Excluding cells within 50 cm of both reward zone starts, RR remapping still exceeded the shuffle, and the mean reward-relative position of the remaining RR remapping cells was 1.145 rad (n = 285) on the first switch and 1.245 rad (n = 303) on the last.
The fraction of cells remapping relative to reward at distances greater than ±50 cm still exceeds the shuffle
Extended Data Fig. 2i-l legendHow it will be judged
As fig2e-rr-remapping-above-chance and fig2eg-mean-rr-position-after-reward with cells within 50 cm (0.698 rad) of both zone starts removed. Reproduced when on both days the above-chance fraction is positive, the circular-mean interval excludes zero and the expected values are within tolerance.
- ed2m-more-fields-after-reward
More RR cells fired after than before the reward zone start, after excluding cells within distances up to 50 cm on the first switch and up to 80 cm on the last.
more RR firing fields were located at positions following than preceding the reward zone at distances up to ±80 cm
Results, Reward-relative remapping both near and far from reward, paragraph 3; Extended Data Fig. 2mHow it will be judged
RR cells (claim rr-cell-fractions) pooled over mice on days 3 and 14, excluding cells whose post-switch relative peak is within each exclusion distance (10 to 190 cm in 10 cm steps) of the zone start; statsmodels proportions_ztest of the number after the start against all, value 0.5, at the legend's level p < 0.00263 (0.05/19, Fig2_Ext2_identifyRewardRelative.md:1292-1666). Reproduced when the after fraction exceeds 0.5 at that level for every exclusion distance from 10 to 50 cm on day 3 and from 10 to 80 cm on day 14; partially reproduced when it does so on one day.
- ed2n-far-remapping-no-increase
After excluding cells within 50 cm of the reward zone start, the above-chance fraction of RR remapping cells did not increase significantly across days.
However, there was no significant increase in remapping at these distances across days (Extended Data Fig. 2n )
Results, Reward-relative remapping both near and far from reward, paragraph 3; Extended Data Fig. 2nHow it will be judged
A claim of absence. The above-chance fraction of ed2jl-rr-remapping-beyond-50cm per switch day, linear regression on switch number 1-7 as in Figure 2h (the legend names a linear regression; the code holds no version of panel n). Reproduced when the slope's p is 0.05 or more; not reproduced when the slope is positive at p < 0.05. Slope and p reported either way.
- rr-cell-fractions
Under the additional criteria, RR cells were 16.3 +/- 5.3% of place cells over all switch days, 13.7 +/- 3.9% on the first switch and 19.4 +/- 5.2% on the last; non-RR remapping cells were 11.9 +/- 3.0%, and 31.4 +/- 14.3% of place cells fell in no category.
We implemented additional criteria ( Methods and Extended Data Fig. 2o–q ) to identify a robust subpopulation of RR cells for further analysis (mean ± s.d., 16.3 ± 5.3% of place cells averaged over all switch days; 13.7 ± 3.9% on the first switch, 19.4 ± 5.2% on the last switch). We refer to cells that remapped independently of RR position as ‘non-RR remapping’ (11.9 ± 3.0% of all place cells). A remaining 31.4 ± 14.3% of place cells did not show sufficiently stereotyped remapping patterns
Results, Reward-relative remapping both near and far from reward, paragraph 3How it will be judged
RR cells as the Methods define them (Criteria to define RR cells). Place cells (significant SI before or after), interneurons excluded, whose peaks relative to the zone start before and after lie within 0.698 rad (50 cm) of each other, and whose trial-averaged reward-aligned dF/F before and after has a cross-correlation peak above the 97.5th percentile of 500 shuffles (each post-switch trial circularly shifted by 1 to 45 bins) at a lag within 5 bins of zero. Non-RR remapping, the "remap far from reward" cells less the RR cells (code). Unclassified, place cells in none of the categories (RR, TR, disappearing, appearing, near, far). Over the 77 switch sessions for the RR fractions. The text does not say which days the non-RR and unclassified percentages cover; readings (a) all seven switch days, (b) the six within-environment days, as the code does for non-RR. The values match under a reading when all five are within tolerance. The code's shuffle of 1 to 44 bins and 2pi/44 bins is shown as a variant.
- ed2pq-rr-reward-aligned-correlation
RR cells had higher pre- against post-switch correlations of their reward-aligned activity maps than TR and non-RR remapping cells, and TR cells had higher correlations on the original track axis than RR and non-RR cells.
How it will be judged
For each cell, the Pearson correlation of its trial-averaged maps before and after the switch, reward-aligned (circular coordinates rotated to the zone starts) and in track coordinates, pooled over 11 mice and 7 switch days. Reproduced when the four rank-sum tests are in the stated direction at p < 0.001.
- fig3b-decode-rr-position
A decoder trained before the switch decoded reward-relative position above shuffle from RR, TR and non-RR remapping cells when tested before the switch, but only RR cells decoded it above shuffle on trials after the switch.
However, when we tested on trials after the reward switch, only the RR population decoded the animals’ RR position better than shuffle (Fig. 3b , bottom)
Results, Reward-relative remapping both near and far from reward, paragraph 3; Figure 3bHow it will be judged
As the Methods give it (Decoding of RR position). The circular-linear regression decoder of decode.py (CircularRegression, ridge 1e-4, tol 1e-5, 100 iterations, Fig3_Decoder.md:302) on deconvolved activity of the RR, TR or non-RR cells at speed above 2 cm/s, predicting position relative to the zone start; score cos(y - y_hat); tenfold cross-validation; data downsampled to equal occupancy per RR position bin (2pi/45) within trials before the switch or the last 30 trials after; 100 shuffles circularly shifting each cell's timeseries independently. Per session, the mean score against the mean of the shuffles; two-sided Wilcoxon signed-rank over sessions per subpopulation, Bonferroni over the three subpopulations; effect size Hedges' g (pingouin pairwise_ttests, effsize hedges, Fig3_Decoder.md:982). Reproduced when all three exceed shuffle at corrected p < 0.05 tested before, RR exceeds shuffle tested after, and TR and non-RR do not (p of 0.05 or more, or real below shuffle); the effect sizes are compared with tolerance as well. Code variants shown, all post-switch trials in place of the last 30 (Fig3_Decoder.md:173, 423), 2pi/44 bins (Fig3_Decoder.md:177), contiguous test folds (decode.py:124).
- fig3c-decode-range
Decoding of reward-relative position from RR cells, z-scored to the shuffle, exceeded 2 over most of the environment, from -104.5 +/- 20.1 cm to +152.7 +/- 22.9 cm relative to the reward zone start (mean +/- s.e.m. across switch days).
Moreover, this above-shuffle decode of RR position using RR cells was significant across the majority of the environment (Fig. 3c ; range of z -scored decode >2... −104.5 ± 20.1 cm to +152.7 ± 22.9 cm relative to reward zone start; mean ± s.e.m. across switch days)
Results, Reward-relative remapping both near and far from reward, paragraph 3; Figure 3cHow it will be judged
The decoder of fig3b-decode-rr-position, trained before and tested after the switch, its score per RR-position bin z-scored to the session's shuffle (mean and s.d. over shuffles, Fig3_Decoder.md:644-646) and averaged over mice per switch day. Neither the Methods nor the code say how the range limits are found. Readings (a) per day, the most negative and most positive bin centres with mean z > 2; (b) per day, the limits of the contiguous run of bins with z > 2 that contains the zone start. Each then averaged over the seven switch days with its s.e.m. The values match under a reading when both limits are within tolerance.
- fig4c-rr-sequences-preserved
RR cells formed sequences whose firing order was preserved across the reward switch, above a cell-identity shuffle in 75 of 77 sessions.
Sequence preservation (the circular–circular correlation coefficient between peak firing positions pre-switch versus post-switch) was higher than the shuffle of cell identities for nearly all animals and days (75 out of 77 sessions)
Results, Preserved behavioral timescale sequences relative to reward, paragraph 1; Figure 4a-cHow it will be judged
As the Methods give it (Sequence detection and quantification). Unsmoothed binned dF/F in circular track coordinates; cells sorted by their peaks on odd trials before the switch; sequence positions the peaks of the even-before and the after trial averages (all post-switch trials, placeCellPlot.py:1428); circular-circular correlation (Kempter et al. 2012, corr_cc); 1,000 permutations of cell identities after the switch; P = (n of shuffled |rho| >= observed |rho| + 1)/1,001; significant at P < 0.05. Sessions with fewer than 5 cells count as not significant (placeCellPlot.py:1565). The value matches when the count is within tolerance.
- fig4f-tr-sequences-preserved
TR cells likewise formed sequences preserved across the switch, above shuffle in 76 of 77 sessions.
The TR place cells likewise constructed robust behavioral timescale sequences (76 out of 77 sessions) relative to the ends of the track
Results, Preserved behavioral timescale sequences relative to reward, paragraph 1; Figure 4d-fHow it will be judged
As fig4c-rr-sequences-preserved, for TR cells (RR cells excluded, placeCellPlot.py:1317).
- fig4i-nonrr-sequences-not-preserved
Sequential firing was not preserved for most non-RR remapping cells; 61 of 77 sessions did not exceed the shuffle.
sequential firing was not preserved for most of the non-RR remapping cells (61 out of 77 sessions did not exceed shuffle
Results, Preserved behavioral timescale sequences relative to reward, paragraph 1; Figure 4g-iHow it will be judged
As fig4c-rr-sequences-preserved, for non-RR remapping cells; the count is of sessions with P of 0.05 or more, sessions with fewer than 5 cells included as not significant.
- ed5a-tr-fewer-across-environments
The environment change reduced the proportion of TR cells, 8.4 +/- 3.1% of place cells on day 8 against 21.4 +/- 7.9% on switches within an environment.
However, the environment change significantly reduced the proportion of cells identified as TR (Extended Data Fig. 5a ; 8.4 ± 3.1% of place cells across environments on day 8 versus 21.4 ± 7.9% within; mean ± s.d., n = 11 mice)
Results, Preserved behavioral timescale sequences relative to reward, paragraph 1; Extended Data Fig. 5aHow it will be judged
TR fraction of place cells per mouse and switch day (RR cells excluded); pingouin Friedman test over the seven switch days, then pairwise two-sided Wilcoxon signed-rank tests, Holm-corrected (Fig2_Ext2_identifyRewardRelative.md:465-485). Reproduced when the Friedman test gives p < 0.05, day 8 is lower than each other switch day at corrected p < 0.05, and the two values are within tolerance (the 21.4% over the 66 within-environment mouse-days pooled, as Fig2_Ext2_identifyRewardRelative.md:429-440 computes it).
- ed5f-rr-no-decrease-across-environments
The proportion of RR cells did not decrease upon the environment switch on day 8.
there was no decrease in the proportion of RR cells upon the environment switch (Extended Data Fig. 5f )
Results, RR representation increased with learning, paragraph 2; Extended Data Fig. 5fHow it will be judged
A claim of absence. The legend of panel f refers to panel a ("as in ( a )"), and the code runs the same tests on the RR fraction (Fig2_Ext2_identifyRewardRelative.md:493-511), a Friedman test over the seven switch days and pairwise two-sided Wilcoxon signed-rank tests, Holm-corrected. Reproduced when no post hoc comparison shows day 8 lower than another switch day at corrected p < 0.05; not reproduced when one does. The Friedman statistic, the day 8 and other-day medians are reported either way.
- fig4j-rr-overrepresent-reward
RR sequences spanned the whole track but over-represented a region from ~50 cm before to ~70 cm after the reward zone start, with the highest density following the start of the zone.
RR sequences spanned the entire track but over-represented a region from ~50 cm before to ~70 cm after the reward zone start, with the highest density of cells following the start of the reward zone
Results, RR representation increased with learning, paragraph 1; Figure 4jHow it will be judged
As the Methods give it (Sequence detection and quantification) and the code (Fig4_Ext4_sequences.md:200-454). Each RR cell's post-switch peak relative to the zone start, per mouse and day the count per 2pi/45 bin divided by the mouse's place-cell count, averaged over mice; the expected uniform level is the mean fraction of place cells in the sequence divided by 45. The tolerance is one bin (~10 cm), the resolution of the measure, as 10% of the values is finer than a bin. Readings for the days (a) the mean over the seven switch days, (b) each day. Reproduced under a reading when the bins above the uniform level form a region whose limits are within tolerance and the densest bin lies at or after the zone start.
- fig4m-rr-density-near-reward-grows
The density of RR cells within 50 cm of the reward zone start increased with task experience, while the fraction of place cells more than 50 cm from it stayed stable over days.
This density near reward increased with task experience, while the fraction of place cells >50 cm from the reward zone start remained stable over days
Results, RR representation increased with learning, paragraph 1; Figure 4j,mHow it will be judged
The per-mouse fraction of place cells that are RR cells with post-switch peaks within 50 cm on either side of the zone start (near) or beyond (far), per switch day (Fig4_Ext4_sequences.md:702-731). Linear mixed-effects model, fraction ~ 1 + switch, switch 1-7 continuous, random intercept per mouse, REML, two-sided Wald test (plotUtils.py:209-219). The code fits the raw and the logit fraction (Fig4_Ext4_sequences.md:791-794), and the Methods say the raw fraction was usually used and the logit confirmed it; readings (a) raw, (b) logit. Reproduced under a reading when the near slope is positive at p < 0.05 and the far slope has p of 0.05 or more.
- fig4no-licking-speed-increase
Over switch days mice increased their post-switch anticipatory lick rate and their running speed.
In parallel, mice increased their post-switch anticipatory lick rate (Fig. 4k,n ) and running speed over days (Fig. 4l,o )
Results, RR representation increased with learning, paragraph 1; Figure 4k,l,n,oHow it will be judged
Per mouse and switch day, the trial-averaged post-switch lick rate (10 cm bins) and speed (2 cm bins), averaged over bins within 50 cm on either side of the zone start, or beyond (Fig4_Ext4_sequences.md:743-761); the mixed model of fig4m-rr-density-near-reward-grows on each. Lick rate is read in the near window ("anticipatory"). The paper does not say which window the speed result rests on; readings (a) within 50 cm, (b) beyond 50 cm. Reproduced when the near lick-rate slope is positive at p < 0.05 and, under a reading, the speed slope is positive at p < 0.05.
- fig4p-lick-precision-sequence-precision
On a given day, the circular variance of the animal's licking positions correlated with the circular variance of the RR sequence, both before and after the switch.
Circular variance in licking positions significantly correlates with RR sequence variance before (left) and after the switch (right)
Results, RR representation increased with learning, paragraph 1; Figure 4pHow it will be judged
Per session (77), the astropy circular variance of RR cells' relative peaks, against the circular variance of licking positions relative to the zone start, before and after the switch; linear regression pooled over sessions (scipy linregress, Fig4_Ext4_sequences.md:540-550). Readings for lick variance (a) the circular variance of the positions of all binned licks in the trial set (the legend's wording), (b) the code's circular mean over trials of each trial's circular variance (Fig4_Ext4_sequences.md:293-301). Reproduced under a reading when the slope is positive at p < 0.05 both before and after.
- ed4hij-sequence-variance-decreases
The variance of the RR sequence and of post-switch licking positions decreased across switch days, while the mean reward-relative position of the sequence did not change.
The variance of both the RR distribution and the post-switch licking decreased across switch days (Extended Data Fig. 4h,i ) but the mean position of the sequence did not change (Extended Data Fig. 4j )
Results, RR representation increased with learning, paragraph 1; Extended Data Fig. 4h-jHow it will be judged
Per mouse and switch day the circular variance and circular mean of RR relative peaks and the licking variance as in fig4p-lick-precision-sequence-precision; the mixed model of fig4m-rr-density-near-reward-grows on switch index (Fig4_Ext4_sequences.md:589-599). Reproduced when the RR variance and the post-switch lick variance have negative slopes at p < 0.05 and the mean position's slope has p of 0.05 or more (the paper's test for the absence). The code fills mouse-days without RR cells with 0 (Fig4_Ext4_sequences.md:229-270); main reading drops them, the code's zeros shown as a variant. The lick-variance readings of fig4p apply.
- fig4qr-tr-track-ends
TR sequences tended to over-represent the ends of the track, and TR density far from the track ends decreased across days.
TR sequence density decreased at positions far from the track ends across days, consistent with selective stabilization of place fields near the landmarks
Results, RR representation increased with learning, paragraph 1; Figure 4q,rHow it will be judged
TR cells' post-switch peaks in track coordinates, density per 2pi/45 bin per mouse and day as in fig4j-rr-overrepresent-reward. Reproduced when the mean density over mice and days in the bins within 50 cm of either track end exceeds the uniform level, and the mixed model of fig4m-rr-density-near-reward-grows on the fraction of place cells that are TR cells more than 50 cm from the track ends has a negative slope at p < 0.05. The raw and logit readings of fig4m apply.
- ed4l-disappearing-decrease
Disappearing, appearing and non-RR remapping cells encoded a more uniform representation; the fraction of disappearing cells decreased with experience both near and far from reward, while non-RR remapping and appearing fractions did not change significantly.
Both near ( ≤ 50 cm, top) and far from ( > 50 cm, bottom) the start of the reward zone, the fraction of place cells with disappearing fields decreases across experience. Non-RR remapping and appearing fractions did not show significant changes over experience
Results, RR representation increased with learning, paragraph 1; Extended Data Fig. 4k,lHow it will be judged
Per mouse and switch day the fraction of place cells of each category with peaks within or beyond 50 cm of the zone start, the mixed model of fig4m-rr-density-near-reward-grows. Reproduced when the disappearing slope is negative at p < 0.05 in both windows and the non-RR remapping and appearing slopes have p of 0.05 or more in both (the paper's test for the absence). "More uniform" is not tested further.
- ed5-teleport-fractions
In sessions imaged through the teleport, 32.7 +/- 10.8% of TR cells and 12.7 +/- 7.1% of RR cells peaked in the teleport period on days with a constant environment, against 41.9 +/- 5.4% and 27.2 +/- 5.7% on day 8, with no significant effect of day for either.
How it will be judged
As the Methods give it (Analysis of teleport periods). In sessions imaged through the teleport (2P imaging, Methods), place cells re-identified with 10 cm bins from -50 cm (tunnel start) to the largest jitter bin with fewer than ten trials missing (up to 580 cm), and TR and RR re-categorized from them; the fraction peaking in the tunnel or jitter, mean +/- s.d. over the 7 mice with any teleport sampling, and a Friedman test of day over the 4 mice sampled on all days. The code holds no version of this analysis. The legends do not say whether the constant-environment value pools mouse-days or averages per mouse first; readings (a) mouse-day rows, as the code does elsewhere, (b) per-mouse means. Reproduced under a reading when the four values are within tolerance and both Friedman tests have p of 0.05 or more.
- fig5b-rr-remain-rr-increases
Across experience, RR cells were increasingly likely to remain RR on the next switch day.
We found that across experience, RR cells were increasingly likely to remain RR on consecutive switches (Fig. 5a,b )
Results, Dynamic cell recruitment into the RR population, paragraph 1; Figure 5bHow it will be judged
Cells matched across days as the Methods give it (intersection over union of ROI pixels, with a threshold per dataset such that the best match's IoU always exceeds any second-best match), for the six consecutive switch-day pairs (3-5, 5-7, 7-8, 8-10, 10-12, 12-14). The fraction is the number of day-A RR cells that are RR on day B over the day-A RR cells matched to day B and a place cell there, set to missing below 5 such cells (Fig5_Ext6_TrackingCellsAcrossDays.md:123-141, 380-397). Mixed model, fraction ~ 1 + pair index 0-5, random intercept per mouse, REML, Wald test, no logit (Fig5_Ext6_TrackingCellsAcrossDays.md:608-625). Reproduced when the slope is positive at p < 0.05. Where the dataset holds no cross-day matching, the matching is done by the Methods' rule and said so.
- fig5d-nontr-to-rr-increases
An increasing fraction of non-TR place cells (appearing, disappearing and non-RR remapping) converted into RR cells on the next switch day over experience.
Fraction of ‘non-TR’ place cells (combined appearing, disappearing and non-RR remapping) that convert into RR cells on subsequent switch days increases over task experience
Results, Dynamic cell recruitment into the RR population, paragraph 1; Figure 5dHow it will be judged
As fig5b-rr-remain-rr-increases, for day-A non-TR place cells (RR excluded) becoming RR. Reproduced when the slope is positive at p < 0.05.
- fig5f-nonpc-to-rr-increases
The fraction of non-place cells converting to RR cells on the next switch day showed a modest increase over experience.
Fraction of non-place cells converting to RR cells on subsequent switch days shows a modest increase over task experience
Results, Dynamic cell recruitment into the RR population, paragraph 1; Figure 5fHow it will be judged
As fig5b-rr-remain-rr-increases, for day-A cells without significant SI in either trial set becoming RR. Reproduced when the slope is positive at p < 0.05.
- fig5h-tr-to-rr-constant
TR cells were sometimes recruited into the RR population, at a rate that did not change over days.
TR cells were sometimes also recruited, though this occurred at a constant rate over days
Results, Dynamic cell recruitment into the RR population, paragraph 1; Figure 5g,hHow it will be judged
A claim of absence, as fig5b-rr-remain-rr-increases for day-A TR cells becoming RR. Reproduced when the slope has p of 0.05 or more; not reproduced when p < 0.05. Slope and mean fraction reported either way.
- fig5i-rr-to-tr-decreases
The fraction of RR cells converting to TR cells decreased slightly over experience.
Fraction of RR cells converting to TR cells decreases slightly over task experience
Figure 5i legendHow it will be judged
As fig5b-rr-remain-rr-increases for day-A RR cells becoming TR, without the 5-cell minimum (Fig5_Ext6_TrackingCellsAcrossDays.md:362-365). Reproduced when the slope is negative at p < 0.05.
- fig5jkl-tr-no-recruitment
The TR population did not recruit more cells from non-TR place cells or non-place cells over days, and TR cells were not more likely to remain TR across experience.
By contrast, the TR population did not recruit more cells from any other population (Fig. 5i–k ) and did not show an increased likelihood of remaining TR across experience (Fig. 5l )
Results, Dynamic cell recruitment into the RR population, paragraph 1; Figure 5j-lHow it will be judged
A claim of absence, as fig5b-rr-remain-rr-increases for day-A non-TR place cells, non-place cells and TR cells becoming TR (Figure 5j, k, l). Reproduced when none of the three slopes is positive at p < 0.05; not reproduced when one is. Slopes and p reported either way. Figure 5i, which the Results cite here, is claim fig5i-rr-to-tr-decreases.
- ed6a-rr-near-reward-stable
RR cells that first fired closer to reward were more likely to keep stable RR tuning across days, and TR cells closer to reward were more likely to remap relative to the track and become RR.
Furthermore, RR cells that originally fired closer to reward were more likely to have stable RR tuning over days, and TR cells closer to reward were more likely to remap with respect to the track and become RR
Results, Dynamic cell recruitment into the RR population, paragraph 1; Extended Data Fig. 6aHow it will be judged
As the Extended Data Fig. 6a legend and Fig5_Ext6_TrackingCellsAcrossDays.md:672-838 give it. For cells matched across consecutive switch-day pairs, categorized on day A, the absolute cross-day change in the pre-switch peak relative to reward (RR cells) and in track position (TR cells), binned by the day-A distance from the zone start in 10 cm bins (mean +/- s.e.m. over pooled cells, smoothed with a 10 cm s.d. Gaussian for display). The paper reports no test and no boundary between near and far, so how "closer to reward" is judged is open. Readings (a) the change within 50 cm of the zone start against beyond it (the near and far boundary of Figure 4m), two-sided Wilcoxon rank-sum over pooled cells at p < 0.05; (b) a linear regression of the change on the day-A absolute distance from the zone start over pooled cells (the test Extended Data Fig. 6c uses for the neighbouring panel), at p < 0.05. Reproduced when under both readings the RR reward-relative change is smaller near reward (a) or grows with distance (b), and the TR track change is larger near reward (a) or falls with distance (b); not reproduced when both readings fail; blocked as methods_underspecified when they disagree. The binned curves are reported.
- ed6c-rr-shift-toward-reward
RR cells that stayed RR across switch-day pairs shifted their peaks slightly toward the reward zone start over days, their initial position being weakly anticorrelated with the change.
RR cells near reward shifted their peak firing even closer to the reward zone start over days
Results, Dynamic cell recruitment into the RR population, paragraph 1; Extended Data Fig. 6b,cHow it will be judged
RR cells RR on both days of a pair, day-A relative peak within 80 cm of the zone start, the signed cross-day change against the day-A position, linear regression (legend; code at Fig5_Ext6_TrackingCellsAcrossDays.md:841-978). Reproduced when the slope is negative at p < 0.05.
- ed6-across-day-drift
Across-day sequence correlations of RR and TR cells were generally lower than within-day sequence correlations, consistent with representational drift.
Although sequence preservation across switch days was greater than chance in some sessions, correlation coefficients across days were generally lower (Extended Data Fig. 6j–m ) than within-day sequences (Fig. 4c,f )
Results, Dynamic cell recruitment into the RR population, paragraph 2; Extended Data Fig. 6j-m; Figure 4c,fHow it will be judged
Across-day correlation as the Methods give it (Sequence detection and quantification), the sort from day-A trials before the switch, correlating day-A after-switch peaks with day-B before-switch peaks of matched cells, at least 5 cells. The Results cite panels j-m, which hold two cell sets; readings for the cells (a) cells followed agnostic of their day-B category (panels j for RR, l for TR), (b) cells keeping their category on day B (panels k, m). The paper names no test for "generally lower"; the comparison of medians is a reading, not the paper's. Reproduced under a cell reading when the median across-day rho over mouse-pairs is below the median within-day rho of fig4c-rr-sequences-preserved for RR cells and of fig4f-tr-sequences-preserved for TR cells.
- ed7-no-anatomical-bias
RR and TR cells showed no reliable bias in their medio-lateral, antero-posterior or superficial-deep distribution within CA1.
However, we found no reliable bias in the medio-lateral, antero-posterior or superficial-deep distributions of either subpopulation (Extended Data Fig. 7 )
Results, Dynamic cell recruitment into the RR population, paragraph 2; Extended Data Fig. 7How it will be judged
A claim of absence, judged on the paper's tests (Extended Data Fig. 7b,c,h legends). Per mouse (and per plane in m17 and m18), cells pooled over the seven switch days, a two-sided Wilcoxon rank-sum test of RR against TR ROI centres of mass in ML and AP, at the legend's Bonferroni level p < 0.0038, and of estimated depth in single-plane mice at p < 0.0055. The code applies no correction (Ext7_Anatomical_Map.md:605-672). Counts tolerate one test, the smallest unit. Reproduced when in each axis the count of significant mice or planes is no more than the paper's count plus the tolerance (3 of 13 for ML, 2 of 13 for AP, 2 of 9 for depth); not reproduced otherwise. The direction of each significant test is reported.
- fig6e-rr-prefer-reward
RR cells with peaks after the reward zone start fired more after rewards than after omissions before the switch; the median reward versus omission index was above 0 in every mouse and significantly so in 9 of 11 mice.
At the population level, the RR cells fired significantly more following rewards compared to omissions, both before the reward switch (Fig. 6d,e ) and after
Results, Encoding of reward proximity versus movement covariates, paragraph 2; Figure 6eHow it will be judged
As the Methods give it (Analysis of rewarded versus omission trials; Time warp modeling; Reward versus omission index). Trial sets before the switch with the zone at A or B and at least three omission trials; deconvolved activity in 10 cm bins at speed of 2 cm/s or more, each cell divided by its session mean; affinewarp PiecewiseWarping with three knots fit to the trial speed profiles (rewarded and omission together; warp_reg_scale 1e-6, smoothness_reg_scale 20, 50 iterations and 200 warp iterations, Fig6_Ext8_Rewarded_vs_Omission.md:333-337) and applied to the activity; RO index = sum(r - o)/sum(r + o) over bins of the trial-averaged warped activity; per mouse, cells pooled over days, two-sided one-sample Wilcoxon signed-rank test against 0 at p < 0.0045 (the legend's Bonferroni level; the code applies none). RR cells with peaks after the zone start; readings (a) the Methods, track peak between the zone start and the track end, (b) the code, positive circular offset from the zone start (Fig6_Ext8_Rewarded_vs_Omission.md:169). Reproduced under a reading when the count of mice with a positive median at that level is within tolerance; the count tolerates one mouse, the smallest unit.
- ed8k-rr-prefer-reward-after-switch
After the switch too, RR cells fired more following rewards than omissions, with a positive median reward versus omission index significant in 9 of 11 mice.
Reward vs. omission index across all switch days where the reward zone was at “A” or “B”, on the trials after the switch
Results, Encoding of reward proximity versus movement covariates, paragraph 2; Extended Data Fig. 8kHow it will be judged
As fig6e-rr-prefer-reward on trial sets after the switch (zone at A or B after the switch). Reproduced under a reading of fig6e when the count is within tolerance.
- ed8df-time-warp-speed-mse
Time warping reduced the mean squared error across trial-by-trial speed profiles, and the speed error was higher after the switch than before.
Time warp model fitting significantly reduces MSE across trial-by-trial speed profiles compared to the original speed data
Extended Data Fig. 8d,f legendsHow it will be judged
The warp of fig6e-rr-prefer-reward per trial set; MSE between each trial's speed profile and the mean over trials, raw and warped. Reproduced when warped MSE is below raw at p < 0.05 and raw MSE after exceeds before at p < 0.05 (two-sided Wilcoxon signed-rank over sessions).
- glm-well-fit-cells
The GLM fit well (fraction deviance explained above 0.15 on held-out trials) 33% of place cells, 4,106 of 7,027 TR cells, 2,122 of 5,979 RR cells and 1,748 of 4,314 non-RR remapping cells; mean fraction deviance explained was 0.32 for TR cells and 0.29 for RR cells.
Only well-fit neurons (fraction deviance explained > 0.15) 37 were included in further analysis (33% of place cells across 11 mice and seven switch days; TR cells, 4,106 out of 7,027; RR cells, 2,122 out of 5,979; non-RR remapping cells, 1,748 out of 4,314
Results, Encoding of reward proximity versus movement covariates, paragraph 3; Extended Data Fig. 9b,cHow it will be judged
As the Methods give it (GLM). Poisson GLM at ~15.5 Hz, speed of 2 cm/s or more; 45 cosine bases of track position, 45 of RR position, 45 of rewarded x position, 7 B-spline bases (degree 3, 5 knots, on quantile-transformed values) for each of speed, acceleration (5-sample Gaussian) and licks (2-sample Gaussian), z-scored; trials split 85/15 by GroupShuffleSplit separately for rewarded-before, rewarded-after and omission trials, random_state 30 (Fig7_Ext9_GLM.md:284); fivefold cross-validation grouped by trial over the training trials; GLM_Tensorflow_2 GLM_CV with group lasso, lambda 10^linspace(-1, -6, 11), Adam, learning rate 0.01, exponential link, lambda chosen by best cross-validated deviance (Fig7_Ext9_GLM.md:292-302); FDE = 1 - dev_model / dev_null on the held-out trials. The response is the deconvolved activity (the code scales it by 10, which leaves FDE unchanged). The Methods give "0.32 +/- 0.13 for TR cells" and "0.29 +/- 0.11 for RR cells" next to 0.10 for all place cells and do not say over which cells; readings for the FDE means (a) the well-fit cells, (b) all cells of the subpopulation. The values match under a reading when all are within tolerance.
- fig7de-glm-top-predictors
Track position was the top GLM predictor for 81.3% of TR cells; for RR cells, RR position gave the highest relative contribution and was the top predictor for 32.8%, followed by speed (21.3%), track position (17.6%) and reward (14.0%); ~56% of well-fit RR cells had RR position as first or second predictor.
Track position was the top predictor for 81.3% of TR cells (Fig. 7e ) ... For RR cells, RR position provided the highest relative contribution at the population level (Fig. 7d ) and the top predictor for 32.8% of RR cells, followed by speed (21.3%), linear track position (17.6%) and the receipt of reward (14.0%) ... Overall, ~56% of RR neurons well-fit by the GLM had RR position as a first or second predictor
Results, Encoding of reward proximity versus movement covariates, paragraph 4; Figure 7d,e; Extended Data Fig. 9d,eHow it will be judged
The well-fit cells of glm-well-fit-cells, pooled over mice and switch days. Relative contribution by ablation as the Methods give it, (dev_ablated - dev_full)/(dev_null - dev_full) on the cross-validation data, each variable's coefficients zeroed, binned by track position and averaged over bins (Fig7_Ext9_GLM.md:426-512). Top predictor, the variable with the largest contribution; second, the largest of the rest (Fig7_Ext9_GLM.md:854-1052). Figure 7d, the variable with the highest median contribution in each subpopulation must be track position for TR and RR position for RR. Reproduced when both Figure 7d orderings hold and all six values are within tolerance; partially reproduced when the orderings hold and the RR and TR top-predictor rankings match but some values fall outside tolerance.
- fig8e-rr-remap-precedes-behavior
The RR population's remap trial preceded the licking remap trial by about two trials for backward (2.1 +/- 0.6) and forward (1.7 +/- 0.7) reward moves, preceded the speed remap trial by 4.4 +/- 1.1 trials for backward moves, and was synchronous with speed for forward moves (0.5 +/- 0.5).
Neural remapping preceded changes in licking by approximately two trials regardless of whether the reward was moved backward or forward on the track (backward... 2.1 ± 0.6 trials (mean ± s.e.m.), neural before licking; forward... 1.7 ± 0.7 trials), but only preceded changes in speed when reward was moved backward (4.4 ± 1.1 trials, neural before speed) and was synchronous with speed when reward was moved forward (0.5 ± 0.5 trials before speed)
Results, RR cell activity often updates before behavior, paragraph 2; Figure 8eHow it will be judged
As the Methods give it (K-means clustering and distance score analysis). Per session and RR population, deconvolved activity in 10 cm bins smoothed (10 cm Gaussian), min-max scaled per cell; lvl factorized k-means, k = 2, 100 restarts; sessions accepted when the k = 2 model beats rotation shuffles on speckled holdout (50 repeats) by a Wilcoxon test at P < 0.05 (the Methods say two-sided, the code one-sided greater, kMeansDistScore.py:211-217; both shown; the paper accepted 50 of 77); cluster labels set so cluster one comes first; distance score; sigmoid L/(1 + exp(-k(x - x0))) + b by curve_fit (dogbox, p0 [max y, median x, 1, min y]), its inflection the remap trial (regression.py:172-195); the same for max-normalized binned licks and speed; sessions kept when all three fits converge. Direction, forward when the new zone is further down the track. Mixed model remap_trial ~ C(datatype, reference PV) * C(direction) + switch, random intercept per mouse, coefficient p Benjamini-Hochberg corrected (Fig8_Ext10_DistanceScore.md:759-790). The four values equal the model's coefficients (backward lick 2.125; forward 2.125 - 0.398; backward speed 4.417; forward 4.417 - 3.871), but the paper does not say how the "mean +/- s.e.m." is taken. Readings (a) the model coefficients and their combinations; (b) mean over sessions of per-session differences; (c) mean over mice of per-mouse means. Reproduced under a reading when the four values are within tolerance and the lick - PV and speed - PV coefficients are positive and the interaction negative at corrected p < 0.05.
- ed10-appearing-remap-later
Remapping of appearing cells followed changes in behavior, and RR remap trials were the most consistent; mean remap trials were 31.91 +/- 2.37 for RR, 32.86 +/- 3.91 for non-RR remapping and 43.0 +/- 7.22 for appearing cells.
By contrast, remapping of the appearing cells significantly followed changes in behavior (Extended Data Fig. 10c,d,f,g )
Results, RR cell activity often updates before behavior, paragraph 2; Extended Data Fig. 10e-gHow it will be judged
The procedure of fig8e-rr-remap-precedes-behavior for appearing and non-RR remapping populations (appearing limited to sessions where two clusters could be identified). Reproduced when for appearing cells the lick - PV and speed - PV coefficients are negative at corrected p < 0.05, the appearing - RR coefficient of the pooled model (fixed effects subpopulation, direction, their interaction and switch day; random intercept per mouse) is positive at p < 0.05, the RR remap trials have the smallest s.d. of the three populations, and the means are within tolerance.
- ed3d-single-field-increases
The fraction of cells keeping a single place field both before and after the switch increased across switch days for all subpopulations, most clearly for RR cells.
However, the fraction of cells maintaining a single place field increased with task experience (Extended Data Fig. 3d )
Results, Independent remapping of individual place fields per cell, paragraph 1; Extended Data Fig. 3dHow it will be judged
Place fields as the Methods give them (Place field definition and inclusion criteria), deconvolved activity smoothed 10 cm, regions of 20 cm or more above 20% of the trial-averaged maximum over the 30 trials before or the last 30 after, active (above mean + 1 s.d.) in at least 8 of the 30 trials; cells with significant SI or a significant field before and after. Per subpopulation (RR, TR, non-RR remapping), the pooled fraction with one field before and one after, linear regression on switch 1-7 (Ext3_FieldQuant.md). Reproduced when the slope is positive at p < 0.05 for all three subpopulations and the RR p is the smallest; partially reproduced when it is positive at p < 0.05 for RR only or for two of the three.
- ed3fg-fields-remap-independently
For RR cells with two fields before and after the switch, the offsets between the two fields were not correlated before against after, indicating independent remapping of the fields.
We found no correlation in the offset between the two fields at the population level for RR cells, indicating largely independent remapping of fields
Results, Independent remapping of individual place fields per cell, paragraph 2; Extended Data Fig. 3f,gHow it will be judged
A claim of absence. For RR cells with exactly two fields both before and after, the circular-circular correlation of the field offsets before against after, per switch day, cells pooled over mice, two-sided p without correction (legend). Reproduced when no switch day has a significant correlation of either sign at p < 0.05; not reproduced otherwise. The coefficients and p of every day are reported.
- ed3mo-appearing-late-backward
Appearing cells formed their fields on later laps and shifted backward more after the formation lap than RR, TR and non-RR remapping cells.
At the population level, appearing cells showed more backwards shifting (Extended Data Fig. 3o ) and later formation laps than the other subpopulations
Results, Independent remapping of individual place fields per cell, paragraph 2; Extended Data Fig. 3m,oHow it will be judged
Formation lap as the Methods give it (Formation lap identification and backwards shifting), within the trial-averaged field +/- 10 cm after the switch, trials whose binned activity exceeds mean + 1 s.d., the first five-trial window with at least three such trials, its first active trial; shift, the field's centre of mass on the formation lap subtracted from that of the mean of the last 30 trials (negative backward), from raw deconvolved activity. Primary fields (the field with peak activity), pooled over mice and switch days. Reproduced when appearing fields have later formation laps and more negative shifts than each other subpopulation by two-sided rank-sum tests at the legend's Bonferroni levels (0.0083 and 0.004).
- ed3p-shift-follows-reward-direction
Primary fields shifted more backward when the reward moved backward on the track than when it moved forward, in every subpopulation.
Notably, the degree of backward and forward shifting depended on whether the reward moved backward or forward on the track, respectively (Extended Data Fig. 3p )
Results, Independent remapping of individual place fields per cell, paragraph 2; Extended Data Fig. 3pHow it will be judged
Shifts of ed3mo-appearing-late-backward split by switch direction (backward when the new zone is earlier on the track). The legend's test is a two-sided permutation test on the rank-sum Z, permuting backward and forward session labels 100 times, at p < 0.01. Reproduced when for each subpopulation backward shifts are more negative and the permutation p is below 0.01; partially reproduced when that holds for some subpopulations.
Pipeline
- Paper
- done (1 of 1)
- Claims
- done
- Claims check
- done (ready)
- Attempt
- todo
- Questions
- none
- Verified
- none