Where the author or a verifier believed the claims file misread the paper, they asked, citing the paper, and a reader who had not seen the data answered from the paper alone, amending the claims file where it was wrong. The author then ran the claim as answered. Attempts before protocol pilot-11 also asked here for method details the claims file did not give.
The criterion takes the median trough over the four cathodal amplitudes, as the code does. The Results say "a median value of 1.95 ms across parameters". Does "across parameters" mean the cathodal amplitudes only, or all polarity and amplitude conditions?
Passage. an early peak immediately after stimulation, followed by a brief pause occurring at a median value of 1.95 ms across parameters and a broader, and sustained peak beginning around 2.5 ms (Figures 4D and S3B) (Results, Evoked spiking is sparse and increases in density with amplitude, paragraph 3)
Why it matters. Over the four cathodal amplitudes the median trough is 1.95 ms (reproduced); over all 12 polarity and amplitude conditions it is 1.45 ms, outside the 0.195 ms tolerance. The verifier reported that the claims file may misread the paper here.
No reader has answered it yet.
The criterion passes the claim when the fractions match "in some assignment of clusters to the labels", with no condition that a cluster labelled selective for an amplitude responds almost only at it. Is that what the paper's statement, clusters "responding almost only at one amplitude", allows? The same clause is in fig5ef-polarity-clusters.
Passage. we observed a subset of single units demonstrating amplitude selectivity, responding maximally, and almost exclusively, at specific intermediate amplitudes ... (19.2% for 5 uA, 15.2% for 25 uA, and 7.1% for 100 uA; Figures 5A-5C, iv-vi) (Results, Stimulation amplitude and polarity tune unit-specific recruitment, paragraph 1)
Why it matters. The only matching assignment for fig5bc labels as 100 uA-selective a cluster whose mean nDRP peaks at 50 uA (0.651 against 0.343 at 100 uA) and as 25 uA-selective one with 0.27 at 5 uA and 0.47 at 25 uA; with labels required to follow the clusters' peaks, no assignment matches and the claim would not be reproduced. For fig5ef the matching assignment labels as bipolar a cluster whose anodal response is 0.79 of its bipolar one; by the half-maximum rule bipolar is 0.205 against 0.277.
No reader has answered it yet.
The criterion lists peaks "with prominence 0.2, width at least 4 ms" and the centre of mass over +-6 ms. These are the values in the code's configuration, but the code converts them to 5 ms bins with round(), so the width becomes one bin and is then not applied (the code applies it only above one bin). Is the width filter part of the method to test, or should the criterion follow what the code does?
Passage. This was confirmed statistically by the absence of a relationship between distance from stimulation and time to peak inhibition (Figure 6F, ii; Pearson r = 0.09) or time to peak rebound excitation (Figure 6F, iii; Pearson r = -0.03) (Results, Single pulse electrical stimulation leads to extensive circuit response, paragraph 3); the Methods give no peak rule.
Why it matters. With the code's rule neither correlation is significant (r = -0.001, p = 0.987; r = 0.047, p = 0.244), the outcome recorded. With a 4 ms (0.8 bin) minimum width both are (r = 0.131, p = 0.001; r = -0.115, p = 0.004), which would make the claim not reproduced.
No reader has answered it yet.
The criteria of fig4f-responders-within-ep, fig4i-proportion-within-ep and fig5i-fs-within-ep name the code's EP_distances.pkl (evoked-potential boundaries computed from the raw AP band, which the dandiset does not hold) as an outside input for the author to judge, and the claims depend on fig2g-extent-increases, which is blocked. They are tested here with that file. Should these claims be marked as using outside inputs (`with_inputs`), so that they are counted apart from the claims on the dataset?
Passage. a responder is within when its signed distance (norm_distance) lies between the mean superficial and mean deep extents of the evoked potential at that amplitude (fig2g-extent-increases; the code reads them from EP_distances.pkl, an outside input for the author to judge) (claims file, criterion of fig4f-responders-within-ep)
Why it matters. The three outcomes (not reproduced, reproduced, not reproduced) rest on boundaries from the paper's code, not the dandiset. As the claims file stands they are counted with the claims on the dataset.
No reader has answered it yet.
The criterion asks that the region, amplitude and interaction terms have p < 0.001 under every reading, including reading (c), the Methods' 500 to 1000 ms window. The claims file also says the paper's printed values are reached only with the code's 0 to 500 ms window. Is the test under every reading what the claim intends, or did the Methods mis-state the window, so that the claim should be tested under the code's window alone?
Passage. the response firing rate was computed from binned spike counts in a 500–1000 ms post-stimulus window using 5 ms bins (STAR Methods, Modulation index calculation); the criterion of fig6c-modulation-cortex reads (a) code, 0 to 500 ms, 10 ms bins, (b) legend, 0 to 500 ms, 5 ms bins, (c) Methods, 500 to 1000 ms, 5 ms bins, and asks for p < 0.001 "under every reading" (claims file).
Why it matters. Under reading (a) the values and statistics match the paper exactly. Under reading (c) the region term has p = 0.457 and the amplitude term p = 1.06e-03, so the claim is not reproduced as the criterion stands. The verifier (round 3) reported that the outcome turns on a window that does not produce the printed numbers.
No reader has answered it yet.