DANDI:000582

Reproducing Sargolini et al. (2006) from dandiset 000582 18

Notebook reproduce_sargolini_2006.ipynb

• ✕Not reproducedMain text, paragraph 4; Fig. 2E; table S1B

  No head-direction cells were found in layer II, whereas a large proportion of the cells in layers III to VI had directional preferences (273 of 385 cells, counting conjunctive cells).

  No head-direction cells were found in layer II (Fig. 2E). In contrast, a large proportion of the cells in layers III to VI had directional preferences (273 of 385 cells) (Fig. 2E).

  Main text, paragraph 4; Fig. 2E; table S1B

  Layer II has cells passing both head-direction criteria under all four readings (9, 7, 4 and 2 of 43), all from rat 11265 in its two 20-minute files, with U2 0.36 to 9.8 against a median of 13.1 in deeper head-direction cells. Watson's U2 alone is significant for 39 to 41 of the 43 layer II cells, so the block correlation decides. The deeper fraction, 0.671 to 0.733, and layer V (0.924 to 0.937) are within tolerance; layer III (0.527 to 0.599) is below its range under every reading and layer VI (0.713 to 0.790) above it under two of four. The claim needs no layer II cell to pass, even for a partial reproduction. The two files, of 16 March 2006, show no sign of bad tracking: median LED separation 3.7 and 3.9 cm (4th and 6th smallest of 71 two-LED files), with fewer than 0.05% of samples above 15 cm; they are the only files spanning 150 cm. Without them layer II has no passing cell (0 of 17) and the deeper fractions are unchanged, which would be partially reproduced, but that variant selects files after seeing the result and does not decide the outcome. It is also weak: the 17 layer II cells left are all from rat 11278, 11 of them in sub-11278_ses-31080502, whose LEDs are more than 15 cm apart in 62% of samples; leaving that file out too, 0 of 6 pass, so the variant rests on 6 layer II cells with plausible tracking. Leaving out the two files with LEDs more than 15 cm apart in most samples leaves 9, 7, 4 and 2 of 32 layer II cells passing, and counting each (rat, depth, hemisphere, name) once 9, 7, 4 and 2 of 26. The earlier attempt had no such claim; its deep-hd-conjunctive found 12% of layer II cells tuned with its own measure.

  How the test differed from the paper’s method

  • Two-LED files, not two-LED rats; rat 11084 has files of both kinds.
  • The block correlation passes when r > 0 with a two-sided p below the level, a standard choice.
• ✕Not reproducedMain text, paragraph 4, second sentence; Fig. 2D; fig. S4

  In head-direction cells, firing rose from a low background rate (typically below 0.5 Hz) to a high maximum rate (5 to 40 Hz) when the rat's head faced the cell's range of directions.

  In these neurons, firing increased from a low background rate (typically G 0.5 Hz) to a high maximum rate (5 to 40 Hz) whenever the rat_s head faced a certain range of directions (Fig. 2D, fig. S4, and table S1).

  Main text, paragraph 4, second sentence; Fig. 2D; fig. S4

  The median background rate is 0 Hz (lowest bin) or 0.32 to 0.50 Hz (beyond 90 degrees), below 0.55 Hz under every reading, but only 65.5% to 71.2% of head-direction cells peak between 4.5 and 44 Hz where the criterion needs 90%; the median peak is 5.9 to 6.9 Hz. Counting each (rat, depth, hemisphere, name) once, 63.6% to 70.1% peak in that range. The peak-radius reading of the answer adds readings of the pure head-direction class; the outcome is the same. The earlier attempt had no such claim.

• ✕Not reproducedMain text, paragraph 5; table S1B

  The proportion of grid cells with conjunctive head-direction tuning depended on layer (66% in layer III, 90% in layer V, 28% in layer VI, none in layer II); the difference was significant across all four layers and across layers III to VI.

  The proportion of grid cells with conjunctive properties was layer-dependent (table S1). The largest proportion was encountered in layers III and V, where 66% and 90% of the grid cells had dual response properties, respectively. In layer VI, the proportion was 28%. No conjunctive cells were observed in layer II. Differences between layers were significant (all four cell layers, c(3) 0 28.4, P G 0.001; layers III to VI, c(2) 0 6.2, P G 0.05)

  Main text, paragraph 5; table S1B

  What the paper reports

  • all four layers: chi-square test, χ2(3) = 28.4, p < 0.001
  • layers III to VI: chi-square test, χ2(2) = 6.2, p < 0.05

  Layer II has conjunctive cells under all 112 readings (1 to 8, all from rat 11265); the claim needs none even for a partial reproduction. The conjunctive fraction of grid cells is 0.86 to 0.95 in layer V (within tolerance), 0.46 to 0.59 in layer III (paper 0.66) and 0.53 to 0.74 in layer VI (paper 0.28). Four-layer chi-square p 1.9e-11 to 0.23, three-layer p 1.6e-05 to 0.78. What the answer changed: the peak-radius reading adds 48 readings (64 before); none has a layer II without conjunctive cells, and the outcome is the same. The earlier attempt called this claim reproduced. Its claim was the abstract's sentence, that grid, head-direction and conjunctive cells occur together in the deeper layers, which the data still show (table S1B, first reading, conjunctive cells 58, 25 and 30 in layers III, V and VI). The claim now carries the paper's per-layer fractions and requires no conjunctive cell in layer II, and with the paper's criteria neither holds. The layer II conjunctive cells (0.04 to 0.32 of layer II grid cells) are in the two 150-cm files of rat 11265 discussed under hd-layers. Without those files layer II has none; the four-layer chi-square has p 2.4e-07 to 0.00033 under reading (a), below 0.001 under all 112 readings, but 0.00014 to 0.054 under reading (b), below 0.001 under 8 of 112, so that variant would be partially reproduced under (a) and not reproduced under most of (b), and its readings disagree (methods_underspecified). Leaving out also the two files with LEDs more than 15 cm apart in most samples, p is 3.6e-05 to 0.034 under (a) and 0.00079 to 0.34 under (b). As there, these variants select files after seeing the result and do not decide the outcome. Without the two files with LEDs more than 15 cm apart in most samples, or counting each (rat, depth, hemisphere, name) once, layer II keeps 1 to 8 conjunctive cells.

• ◐Partially reproducedMain text, paragraph 2; Fig. 1D; table S1A

  The proportion of cells with grid structure depended on layer. Most well-separated layer II cells were grid cells (66 of 108), against a smaller proportion in layers III (78 of 223), V (17 of 110) and VI (42 of 186).

  The proportion of cells with a sinusoidal modulation was layer-dependent (Fig. 1D and table S1). Whereas most well-separated layer II cells had strongly periodic firing fields, only a smaller proportion of the deeper neurons had such characteristics

  Main text, paragraph 2; Fig. 1D; table S1A

  Under all 56 readings (O1 to O3, with a peak's centre as its maximum or centroid and its radius as that of a disc of the same area or the largest distance to one of its pixels) layer II has the largest fraction of grid cells, 0.592 to 0.718, above 0.5 and above layers III (0.529 to 0.604), V (0.297 to 0.415) and VI (0.354 to 0.485). Layer II is within tolerance under 32 of 56 readings; no deeper layer is within tolerance under any, each being well above the paper's fraction. Gridness above 0 makes 277 to 329 of 614 cells grid cells, against 203 of 627 in the paper. Per rat, layer II fractions are 0.41 to 0.90 and deeper ones 0 to 0.67; counting each (rat, depth, hemisphere, name) once keeps the order (0.680, 0.511, 0.356, 0.329). What the answer changed: the question on a peak's centre and radius was answered by naming both centres and both radii as readings. The radius as the largest distance to a pixel is new here (the attempt had used the disc of the same area only), and the central peak now takes the same centre and radius. The outcome is the same under every reading, as before (32 readings, layer II 0.620 to 0.718). Same outcome as 000582-sargolini-2006-01, for another reason. That attempt used its own grid score with a shuffled-data threshold, under which about a third of layer II cells were grid cells, so the majority part was in doubt; with the paper's criterion, gridness above 0, the majority and the order hold and the deeper layers' fractions are what falls short. Gridness above 0 is lax on these data: with each spike train shifted in time (5 shifts of 20 s or more per cell, first reading), 0.297 of the shifted maps have gridness above 0 (0.288 to 0.307 by layer), against observed 0.634, 0.559, 0.314 and 0.359; above the 95% point of the shifted gridness (0.216) the fractions are 0.408, 0.419, 0.186 and 0.197, and above the 99% point (0.362) 0.296, 0.348, 0.153 and 0.192, so under either stricter cut layer III is above layer II and the layer order the partial outcome rests on does not hold. This control is not the paper's criterion and does not decide the outcome, but it limits what the partial outcome shows. Counting each (rat, depth, hemisphere, name) once (425 of 614 cells) the order holds under all 56 readings and no reading has all four layers within tolerance (layer II 0.680 to 0.780, III 0.451 to 0.564, V 0.333 to 0.437, VI 0.316 to 0.458).

  How the test differed from the paper’s method

  • Each cell-session counts as a cell; no identifier links cells across files.
  • The dandiset holds 71 layer II cell-sessions against the paper's 108, and 15 rats against 17.
  • The pair of readings in which the peak radius is unused (circle to the outer peak's centre, central peak removed as its pixels) is run once, as both radii give the same gridness.
  • Positions are rescaled in the files; rate maps use them as deposited.
  • Positions are smoothed over the whole file, so in the 23 two-trial files the 15-point filter spans the join of the trials for 7 samples on each side, and pooled maps join two trials not centred alike; the first-trial reading avoids both.
• ✓ReproducedMain text, paragraph 2; Fig. 1, A and B; table S1A

  Grid cells with tessellating firing fields were found in all four principal cell layers of MEC (II, III, V and VI).

  Grid cells with tessellating firing fields (7) were observed in all principal cell layers (Fig. 1, A and B).

  Main text, paragraph 2; Fig. 1, A and B; table S1A

  Every layer has grid cells under all 56 readings, at least 42 in layer II, 120 in III, 35 in V and 70 in VI, from 5, 6, 3 and 3 rats under the first reading; counting each (rat, depth, hemisphere, name) once, at least 34, 60, 29 and 49. What the answer changed: the peak-radius reading adds 24 readings; the outcome holds under all of them, as it did under the earlier 32. The earlier attempt had no such claim.

  How the test differed from the paper’s method

  • As for layer2-grid.
• ✓ReproducedMain text, paragraph 4; Fig. 2F

  The entire range of head directions was represented by the preferred directions of head-direction cells in each of layers III, V and VI.

  The entire range of head directions was represented in all three layers (Fig. 2F).

  Main text, paragraph 4; Fig. 2F

  Under all four readings the preferred directions fill all twelve 30-degree sectors, and so all 60- and 90-degree sectors, in layers III (96 to 109 cells), V (73 to 74) and VI (102 to 113), and still do with each (rat, depth, hemisphere, name) counted once (52 to 89 cells per layer). The earlier attempt had no such claim.

  How the test differed from the paper’s method

  • Sectors start at 0 degrees of the camera frame, a standard choice.
• ✓ReproducedMain text, paragraph 4; Fig. 2G; fig. S5

  Preferred directions of simultaneously recorded head-direction cells in the same area were widely distributed; in most ensembles of five or more cells (8 of 12) they showed no significant clustering.

  When several head-direction cells were recorded simultaneously in the same area, their peak firing directions were widely distributed (Fig. 2G and fig. S5).

  Main text, paragraph 4; Fig. 2G; fig. S5

  What the paper reports

  • clustering of the mean vectors of 7 colocalized layer VI cells (Fig. 2G): Watson's U2 test, U2 = 0.01, critical value 0.18, n.s., p > 0.05

  Under every reading fewer than half of the ensembles of five or more simultaneous head-direction cells are clustered by Watson's one-sample U2 test, 0 to 3 of 9 to 16 ensembles of pure head-direction cells and 2 to 4 of 26 to 28 with conjunctive cells. The counts differ from the paper's 4 of 12; the clustered count is within 4 +/- 0.4 under one reading only (4 of 26, with conjunctive cells). Most ensembles are from rats 11207 (18 of 28) and 11265 (6). The ensembles share cells: 92 of their 225 cells (first reading, with conjunctive cells) match a cell of another ensemble by rat, depth, hemisphere and name, in 16 of the 28 ensembles; counting each such cell once leaves 16 to 19 ensembles, 1 to 3 of them clustered, so the outcome is the same. The peak-radius reading of the answer adds readings of the pure head-direction class (112, 64 before); the outcome is the same. The earlier attempt had no such claim.

  How the test differed from the paper’s method

  • Critical values of the one-sample U2 test by Stephens' small-sample form, which gives the paper's 0.177 to 0.182.
• ✓ReproducedMain text, paragraph 4

  Directional tuning curves were always stable across the two trials (circular correlation r = 0.84, P < 0.001).

  Directional tuning curves were always stable across trials (circular correlation, r 0 0.84, P G 0.001).

  Main text, paragraph 4

  What the paper reports

  • circular correlation, r = 0.84, p < 0.001

  For 88 to 96 head-direction and conjunctive cells in the 19 or 20 two-trial files of 3 rats, the circular correlation of preferred direction between trials is 0.78 to 0.85 and the mean correlation of the trials' tuning curves 0.77 to 0.81, all p below 0.001 and within 0.84 +/- 0.084 under both readings and all four classifications. The statement's "always" holds less well, as 7 to 14 cells have tuning curves correlated below 0.5 between trials; two cells silent in one trial are left out. The third two-LED file with questionable LED geometry, sub-11025_ses-06050501 (LEDs more than 15 cm apart in 17.8% of samples), holds no such cell under the pooled readings and 2 and 1 under the first-trial readings, whose preferred directions change by 180 and 54 degrees. Without it the tuning-curve reading holds under all four readings (0.771 to 0.815), but the circular correlation falls to 0.538 and 0.645 under the first-trial readings, outside tolerance, as the circular means of preferred directions that fill the circle move when one or two cells are left out; the readings of that variant would disagree. The criterion leaves no file out, so the variant does not decide the outcome, but the circular-correlation reading is fragile here. Counting each (rat, depth, hemisphere, name) once (72 to 78 cells), circular r 0.841 to 0.897 and mean curve r 0.775 to 0.812, reproduced under both readings. The earlier attempt had no such claim.

  How the test differed from the paper’s method

  • Only the 23 twenty-minute files hold two trials, all from 3 rats (11025, 11138, 11207 among the head-direction cells).
  • Circular correlation by Jammalamadaka and SenGupta, a standard choice.
• ✓ReproducedMain text, paragraph 5; Fig. 3, A to E; fig. S7

  Grid cells and head-direction cells formed overlapping populations; the joint distribution of gridness and directional tuning was continuous, with some grid cells directionally tuned and some head-direction cells having grid structure.

  Grid cells and head-direction cells formed overlapping populations. The 2D distribution of gridness and directional tuning was continuous, such that some grid cells were directionally tuned and some head-direction cells had grid correlates (Fig. 3, A to D, and fig. S7).

  Main text, paragraph 5; Fig. 3, A to E; fig. S7

  Under all 112 readings there are 98 to 141 conjunctive cells, and among the 366 to 395 cells passing either criterion the largest gap is at most 0.209 of the gridness range and 0.126 of the log U2 range; counting each (rat, depth, hemisphere, name) once, there are 62 to 92 conjunctive cells and the largest gap is at most 0.213. What the answer changed: the peak-radius reading adds 48 readings; the largest gridness gap rises from 0.135 to 0.209 of the range, still below a quarter. The earlier attempt had no such claim.

  How the test differed from the paper’s method

  • No separate clusters is read as no gap wider than a quarter of an axis's range, the claims file's standard choice.
• ⊘BlockedMain text, paragraph 2; Fig. 1D· Methods underspecified

  Among deeper-layer cells with periodic firing, the range of gridness did not differ from that of layer II cells.

  however, the range of Bgridness[ among those cells was not different from that of the layer II cells

  Main text, paragraph 2; Fig. 1D

  The analysis ran; the criterion's two tests disagree. The Mann-Whitney test of layer II against layers III to VI finds a difference under 8 of 56 readings (p 0.009 to 0.993); the Kruskal-Wallis test across the four layers finds one under 53 of 56 (p 0.0001 to 0.159). The median gridness of grid cells is 0.42 to 0.78 in layer III, 0.23 to 0.58 in layer II, 0.28 to 0.53 in layer VI and 0.17 to 0.47 in layer V, and the layer III median is above the layer II median under all 56 readings. Pairwise Mann-Whitney tests, not the criterion's and uncorrected, find layer II different from layer III under 46 of 56 readings, layer III from V under 49 and from VI under 25, and layer II from V under 2 and from VI under none; so the Kruskal-Wallis difference comes mostly from layer III, including against layer II, which pooling layers III to VI hides from the Mann-Whitney test. The ranges are alike, from about 0 to between 1.14 and 1.49 in every layer. Counting each (rat, depth, hemisphere, name) once, the Mann-Whitney test finds a difference under 3 of 56 readings and the Kruskal-Wallis test under 10, so the tests still disagree. What the answer changed: the peak-radius reading adds 24 readings, and the tests still disagree (2 of 32 and 32 of 32 before). The earlier attempt had no such claim.

• ⊘BlockedMain text, paragraph 3; figs. S1 and S2· Methods underspecified

  All 203 grid cells had stable periodic firing patterns within and between trials.

  All 203 neurons that passed this criterion had stable periodic firing patterns both within and between trials (figs. S1 and S2).

  Main text, paragraph 3; figs. S1 and S2

  The analysis ran. No combination of readings has every grid cell stable, and the 448 combinations split between partially reproduced (84, at least 90% stable) and not reproduced (364). With map correlations within trials (odd and even minutes) and between trials, 90.2% to 96.3% of grid cells are stable; with gridness above 0 required in both parts of each trial, 49% to 65%. Between trials, 84% to 95% of the 71 to 85 grid cells in two-trial files have a significant positive map correlation and 49% to 70% have gridness above 0 in both trials. The number of grid cells, 277 to 329, is outside 203 +/- 20. Counting each (rat, depth, hemisphere, name) once (179 to 214 grid cells), 71 combinations give partially reproduced and 377 not reproduced. What the answer changed: the peak-radius reading adds combinations, which split as before (51 and 205 of 256). The earlier attempt had no such claim.

  How the test differed from the paper’s method

  • Only the 23 twenty-minute files hold two trials; between-trial stability is tested on their 71 to 85 grid cells.
• ⊘BlockedMain text, paragraph 3, last sentence; fig. S2 legend· Methods underspecifiedverifier disagreed at first

  In all layers the spatial phase of nearby grid cells was distributed (their vertices were offset), while grid orientation was consistent across simultaneously recorded cells.

  In all layers, the phase of the grid was distributed, i.e., the vertices of most nearby grid cells were offset relative to each other, but the orientation of the grids was consistent across all simultaneously recorded cells.

  Main text, paragraph 3, last sentence; fig. S2 legend

  The analysis ran; the readings disagree. The phase part holds in every layer under every reading (75% to 98.5% of simultaneous pairs displaced by more than a fifth of the spacing, 92.6% to 100% by more than 3 cm). Orientation is more alike within files than between rats in layers II and III under every reading (p below 0.001) but not in layers V (p up to 0.50 with the layer's own rats and 0.52 with all layers, holds under 42 and 41 of 56 readings) and VI (p up to 0.34 and 0.50, 35 and 23 of 56), each from three rats. Both parts hold in every layer under 90 of 224 readings (O1 to O3, the two pools of other rats' pairs and the two phase readings) and not under the other 134. These counts rest on p-values near 0.05 and have differed by a reading or two between runs of the same code on different machines (the verifier's second round quotes the counts of an earlier run); the outcome does not depend on them. Counting each (rat, depth, hemisphere, name) once, both parts hold in every layer under 66 of 224 readings. The criterion names reproduced only, so the claims file's rule applies: a claim reproduced under some readings and not others is methods_underspecified. 20 of the 487 simultaneous grid pairs of the first reading pair cells of different layers and are not counted, as the criterion judges each layer. What the answer changed: it named both pools as readings and said "every reading" covers O1 to O3; the outcome is the one the claim had before it. The previous version called the claim partially reproduced, an outcome the criterion does not name, and the verifier's first round pointed this out. The earlier attempt had no such claim.

• ⊘BlockedMain text, paragraph 4; fig. S6· Methods underspecified

  Beneath layer II, head-direction cells were usually recorded together with grid cells.

  Beneath layer II, grid cells were colocalized with head-direction cells ... Head-direction cells were usually encountered together with grid cells (fig. S6).

  Main text, paragraph 4; fig. S6

  The analysis ran; the readings disagree. By file, 75% to 91% of the 49 to 57 files with a pure head-direction cell in layers III to VI also hold a grid or conjunctive cell, and 47% to 60% a pure grid cell (above half under 83 of 112 readings). By tetrode, 36% to 48% of 107 to 128 tetrodes hold a grid or conjunctive cell and 14.5% to 19.4% a pure grid cell, below half under every reading. Counting each (rat, depth, hemisphere, name) once, by file 67% to 84% hold a grid or conjunctive cell and 30% to 48% a pure grid cell, and by tetrode both are below half, so the readings still disagree. What the answer changed: it confirmed that the head-direction cells are the pure ones of table S1B and that the cell with them is another cell, as this attempt had assumed, and named the pure grid cell as a second reading of that cell. The new reading lowers the fractions; the outcome is the same. The earlier attempt had no such claim.

• ⊘BlockedMain text, paragraph 4; fig. S4· Methods underspecified

  The breadth of directional tuning did not differ between layers III, V and VI (mean U2 18.2, 25.2 and 18.3; mean angular standard deviation 55.9, 52.2 and 58.2 degrees; P > 0.05).

  The breadth of tuning was not significantly different between layers III, V, and VI (mean U2 values, 18.2, 25.2, and 18.3, respectively; mean angular standard deviations, 55.9-, 52.2-, and 58.2-; P 9 0.05) (fig. S4).

  Main text, paragraph 4; fig. S4

  What the paper reports

  • not named, p > 0.05 (n = head-direction cells of layers III, V and VI, nine two-LED rats)

  The analysis ran; the readings disagree. On U2 no ANOVA and 10 of 116 Kruskal-Wallis tests find a difference between layers III, V and VI; on the angular standard deviation many do (angular deviation, ANOVA 35 and Kruskal-Wallis 43 of 116; circular standard deviation, 61 and 43 of 116), layer V being more sharply tuned (46.3 to 46.6 degrees against 52.6 to 55.0, with conjunctive cells). Means are within tolerance in 11 of 348 rows. Counting each (rat, depth, hemisphere, name) once, no U2 test and 1 or 2 of 116 per angular test find a difference, so the readings still disagree, though differences are much rarer. What the answer changed: it named Batschelet's angular deviation and circular standard deviation as readings, which this attempt already ran; under the claim's rule for readings they still disagree with the U2 tests. The peak-radius reading of layer2-grid adds readings of the pure head-direction class (116 per measure, 68 before). The earlier attempt had no such claim.

• ⊘BlockedMain text, paragraph 5; Fig. 3, D and E· Methods underspecified

  Directional tuning of conjunctive cells did not differ from that of pure head-direction cells (mean U2 18.0 against 19.4, P > 0.05).

  The degree of directional tuning in cells that met selection criteria for both gridness and directionality was not significantly different from that of pure headdirection cells (mean U2 values, 18.0 and 19.4, respectively; P 9 0.05) (Fig. 3, D and E).

  Main text, paragraph 5; Fig. 3, D and E

  What the paper reports

  • not named, p > 0.05 (n = 53 conjunctive and 220 head-direction cells (table S1B))

  The analysis ran; the tests disagree. The t-test finds a difference in U2 between conjunctive and pure head-direction cells under 98 of 112 readings (means 12.0 to 19.4 against 18.2 to 25.4), the Mann-Whitney test under 42 of 112 (medians 8.8 to 15.5 against 10.5 to 17.4). Means are within tolerance under 3 of 112 readings. The t-test count rests on p-values near 0.05 and has differed between runs of the same code on different machines (the verifier's second round quotes an earlier run); the numbers here are this run's. Counting each (rat, depth, hemisphere, name) once, the t-test finds a difference under 71 of 112 readings and the Mann-Whitney test under 26. What the answer changed: the peak-radius reading adds 48 readings; the tests disagree as before (49 and 14 of 64). The earlier attempt had no such claim.

• ⊘BlockedMain text, paragraph 6; Fig. 4, A and B· Methods underspecified

  Regression showed a consistent positive speed-rate relation in nearly all grid cells (141 of 150), head-direction cells (153 of 220) and conjunctive cells (45 of 53).

  Yet, regression analyses showed a consistent positive speed-rate relation in nearly all grid cells (141/150 cells, P G 0.001), headdirection cells (153/220 cells, P G 0.001) and grid ... head-direction cells (45/53 cells, P G 0.001), suggesting that a substantial proportion of the network expressed information about how fast the animal was moving (Fig. 4B).

  Main text, paragraph 6; Fig. 4, A and B

  The analysis ran. Counted as cells with a positive slope at P < 0.001, 0.129 to 0.446 of grid cells, 0.094 to 0.282 of head-direction cells and 0.081 to 0.415 of conjunctive cells pass, far below the paper under every combination. Counted by a positive slope, 0.815 to 0.886, 0.622 to 0.708 and 0.686 to 0.822 pass. Taking for each of the 448 combinations of method details (speed measure, positions, trials and cell classes) the best reading of the number, 119 give reproduced, 325 partially reproduced and 4 not reproduced; the method details disagree. Under the positive-slope count the criterion's sign test against half is applied to each class (p at most 5.5e-17, 1.5e-03 and 4.5e-05 for grid, head-direction and conjunctive cells); the head-direction count fails it in 2 of 448 rows, which are then not reproduced. Without the sign test, as in the previous version, which computed it but did not apply it, the split is 119, 327 and 2 (conjunctive 0.700 below head-direction 0.708, displacement speeds). Counting each (rat, depth, hemisphere, name) once, the split is 125, 201 and 122, so the method details still disagree. What the answer changed: the question asked whether the claim is reproduced when some combination matches or methods_underspecified when combinations disagree. The rewritten criterion takes the best of the counting and significance readings and requires the method details to agree. The outcome is the same, now because the method details disagree (before the answer, 54 reproduced and 200 partially reproduced of 256 under the positive-slope count). The earlier attempt called this claim reproduced; its claim was the abstract's "all cell types were modulated by running speed", tested as the correlation of rate with speed over 5 cm/s bins against shuffled spike trains, at the level of each cell type. The claim now carries the paper's per-cell counts and regression method, whose method details disagree.

  How the test differed from the paper’s method

  • Speeds rest on positions rescaled in the files; in the 19 circular files spanning 200 cm they are about 11% too large.
• ⊘BlockedMain text, paragraph 6; Fig. 4B· Methods underspecifiedverifier disagreed at first

  The slope of the speed-rate regression was steeper for conjunctive cells and grid cells than for head-direction cells, and did not differ between conjunctive and grid cells.

  The slope of the regression line was steeper for conjunctive cells and grid cells than for headdirection cells EF(2,422) 0 5.5, P G 0.005; conjunctive versus head-direction, P 0 0.009; grid versus head-direction, P 0 0.01; conjunctive versus grid, P 0 0.60; Tukey HSD test).

  Main text, paragraph 6; Fig. 4B

  What the paper reports

  • slope across the three cell classes: ANOVA, F(2,422) = 5.5, p < 0.005 (n = 423 cells (150 grid, 220 head-direction, 53 conjunctive))
  • conjunctive against head-direction: Tukey HSD test, p = 0.009
  • grid against head-direction: Tukey HSD test, p = 0.01
  • conjunctive against grid: Tukey HSD test, p = 0.6

  Blocked because it depends on speed, which is blocked. The analysis ran. Under all 448 combinations the ANOVA is significant (F 4.5 to 13.6, p at most 0.012) and grid cells have steeper slopes than head-direction cells (Tukey p at most 0.0084), but conjunctive cells do not differ from head-direction cells (Tukey p 0.06 to 0.99), which on its own would be partially reproduced under every reading; conjunctive against grid p 0.0004 to 0.29. Degrees of freedom F(2,444) to F(2,488). What the answers changed: speed stays blocked under its rewritten criterion, so this claim stays blocked; the peak-radius reading adds combinations (256 before), all partially reproduced on their own. The verifier would call the claim partially reproduced; it stays blocked, as the claims file lists speed under depends_on and the protocol (Author, step 1.5) says a claim that depends on a blocked claim "is blocked for the same reason", which is why blocked_by is methods_underspecified, the reason speed is blocked. The disagreement is recorded. The earlier attempt had no such claim.

• ⊘BlockedMain text, paragraph 6; Fig. 4C· Methods underspecifiedverifier disagreed at first

  The intercept of the speed-rate regression was positive in all cells but one (422 of 423), and mean rates at the lowest speeds (0 to 3 cm/s) were about 2.3 to 2.5 Hz in each cell class, so firing persisted during brief stops.

  The yintercept of the regression line was positive in 422 out of 423 cells (Fig. 4C). Average rates at the lowest velocity (0 to 3 cm/s) were 2.26 T 0.03 Hz (grid cells), 2.50 T 0.02 Hz (headdirection cells) and 2.39 T 0.05 Hz (grid ... head-direction cells).

  Main text, paragraph 6; Fig. 4C

  Blocked because it depends on speed, which is blocked. The analysis ran. The intercept is positive in 98.2% to 99.8% of cells (1 to 8 negative), within tolerance, but mean rates at 0 to 3 cm/s are 1.2 to 1.9 Hz (grid), 1.1 to 1.8 Hz (head-direction) and 0.9 to 1.6 Hz (conjunctive), more than 10% below 2.26, 2.50 and 2.39 Hz under every reading, which on its own would be not reproduced. What the answers changed: speed stays blocked under its rewritten criterion, so this claim stays blocked; the peak-radius reading adds combinations, all not reproduced on their own. The rescaling of the positions is about 11% in the 19 circular files; taking in those files the bins up to 3 cm/s times 200/180, as if scaled to the 180 cm circle, the class means are 1.22 to 1.90, 1.18 to 1.76 and 0.93 to 1.53 Hz, no row has all three within 10%, and in the 120 cells of those files the mean moves from 0.99 to 0.97 Hz (first reading), so the rescaling does not explain the low rates. The verifier would call the claim not reproduced; it stays blocked because it depends on speed, for the same reason as speed (protocol, Author, step 1.5), and the disagreement is recorded. The earlier attempt had no such claim.

  How the test differed from the paper’s method

  • The 0 to 3 cm/s bin rests on rescaled positions.