Claims 47
From the paper Dopamine and glutamate regulate striatal acetylcholine in decision-making · claims file · checked 2026-10-21: ready
- ed1b-switch-after-block-transition
After a block transition, mice transiently increase their probability of switching ports, which brings them to the new high-reward-probability port.
After successive unrewarded trials resulting from reversals of the reward probability at block transitions, the mice transiently increase their probability of switching ports between trials
Results, Adaptive switching in decision-making, paragraph 2; Extended Data Fig. 1b,cHow it will be judged
From the behaviour code (block_transition_stats.ipynb), P(switch) and P(high port) by trial position from the block transition, per mouse, over the n = 8 mice of Extended Data Fig. 1. Reproduced when, in the mean over mice (the panels show mean ± s.e.m.), P(switch) over the first 10 trials after the transition exceeds the mean over the 10 trials before it, P(switch) returns towards the pre-transition level later in the block, and P(high port) rises after the transition (the exponential time constant tau_highport, fitted over block positions 0 to 30 as in the code, is positive). The per-mouse values are reported. The 10-trial span is a standard choice for a plotting range; the paper does not fix it.
- ed1hj-rflr-predicts-switching
The three-parameter RFLR model (alpha, beta, tau) predicts the switching dynamics at block transitions and the conditional switch probabilities given the past two trials.
the RFLR model accurately predicts the switching dynamics at block transitions as well as the probability of switching on the current trial, which depends on the choice and reward history of previous trials
Results, Adaptive switching in decision-making, paragraph 3; Extended Data Fig. 1f-jHow it will be judged
RFLR fitted by maximum likelihood with stochastic gradient descent (Supplementary Methods, Behavior performance analysis; bandit_models.py). The paper does not say how train and test sessions are split; the readings are (a) fit and evaluate on held-out sessions per mouse, (b) on all sessions pooled. Reproduced when, under both, the model's conditional switch probabilities for the two-trial histories of Extended Data Fig. 1j fall within the binomial standard error bars of the mice's for most histories, and the predicted P(switch) and P(high port) around block transitions follow the observed curves (correlation above 0.9 across block positions; a standard threshold for "accurately", named as such).
- fig1e-reward-outcome-post
After side-port entry, unrewarded trials show a lower mean DA and a larger delta-Ach than rewarded trials, in DA (13 mice) and Ach (14 mice) recordings.
Reward outcome greatly alters DA and Ach transients, with unrewarded trials resulting in a robust decrease in mean DA and a consistent increase in ∆Ach (Fig. 1e )
Results, Ach and DA are dynamically regulated, paragraph 2; Fig. 1d,eHow it will be judged
Shared method in the header (signal, windows W1 and W2, per-trial measures, T1 per mouse, T2 across mice), rewarded against unrewarded trials, post window, as in aa_processStats_final_Fig1_AllDA.m and _Fig1_Ach_Words.m. Reproduced when, under each window reading, mean DA is lower and delta-Ach higher in unrewarded trials in more than half of the mice with T1 p < 0.05. "Robust" and "consistent" are read as this count, a reading named here; the paper does not give one. T2 across mice is reported, as the header says, and does not decide.
- fig1e-reward-outcome-not-pre
Before side-port entry, mean DA and delta-Ach do not differ significantly between rewarded and unrewarded trials.
Notably, the changes are significant after ('post'), but not before ('pre') side entry
Results, Ach and DA are dynamically regulated, paragraph 2; Fig. 1eHow it will be judged
A claim of absence, judged on the paper's test, T1 per mouse (two-sided t-test, p < 0.05) on the pre window, readings W1 and W2 (header). Reproduced when fewer than half of the DA mice and fewer than half of the Ach mice show a significant pre difference (the count mirrors that of fig1e-reward-outcome-post; the paper gives none); not reproduced otherwise. The per-mouse differences and T2 across mice are reported either way.
- fig1e-lda-accuracy
An LDA classifier separates rewarded from unrewarded trials with accuracy above 80% when trained on post-side-entry signals and around 50% (chance) when trained on pre-side-entry signals.
the LDA classification accuracy is very high when trained on the signals post-side entry (higher than 80%), but not when trained on those pre-side entry (around 50%) (Fig. 1e )
Results, Ach and DA are dynamically regulated, paragraph 3; Fig. 1eHow it will be judged
LDA as in the header, windows W1 and W2, DA and Ach separately. The post value is a bound, met when accuracy exceeds 80%. The paper prints a percentage per mouse and does not say whether "higher than 80%" means every mouse or the mean; the readings are (a) the mean over mice, (b) every mouse. Reproduced when post accuracy exceeds 80% and pre accuracy is within the expected value under one reading, for both DA and Ach.
- ed3-not-lateralized
DA and Ach transients do not differ between trials to the port ipsilateral and contralateral to the fibre; DA shows no significant difference and Ach differences are inconsistent in direction with poor LDA accuracy.
Differences between these trials were not statistically different for DA (Extended Data Fig. 3e). Any significant differences for Ach were inconsistent in directionality, and the LDA classification accuracy was poor
Supplementary Results, Analysis of lateralization of DA and Ach signals; Extended Data Fig. 3d-fHow it will be judged
A claim of absence, judged on the paper's test, T1 per mouse at p < 0.05, ipsilateral against contralateral trials, rewarded and unrewarded separately, post window (W1, W2), as in processStats_Fig1_DA_RL_RewNR.m (right against left port). Which side is ipsilateral needs the hemisphere of each fibre. Reproduced for DA when no DA mouse shows a significant difference; for Ach when the significant differences do not share one direction across mice and the LDA accuracy is below 65% on average ("poor", a threshold named here). The effect sizes are reported either way.
- fig1f-cue-omission
In rewarded trials with the LED cues omitted, DA after side entry is larger and DA before side entry smaller than with cues, while Ach changes in the opposite direction.
the DA transient in cue-omission rewarded trials is significantly increased after side-port entry and decreased before side entry, whereas Ach transients are modulated in the opposite direction
Results, Ach and DA are dynamically regulated, paragraph 4; Fig. 1f; Extended Data Fig. 4aHow it will be judged
Header method, cue-omission session against a cued session of the same mouse, as paired in aa_processStats_Fig1_DA_Ach_LED.m (which cued session is compared is fixed there by date). The Extended Data Fig. 4a legend measures Ach as the mean minimum z-scored signal, not delta-Ach; both are shown and the legend's minimum decides. Reproduced when, under each of W1 and W2, in more than half of the 4 mice (at least 3, the count of fig1e-reward-outcome-post; the paper gives none) mean DA post is higher and pre lower without cues with T1 p < 0.05, and the Ach minimum moves the other way.
- fig1g-history-stay
In trials that repeat the previous port, lose-win trials have a larger mean DA and delta-Ach after side entry than win-win trials, and win-lose trials have a deeper DA dip and a larger delta-Ach than lose-lose trials.
Indeed, mean DA and ∆Ach signals increase post-side entry in the 'lose–win' scenario ... Conversely, for an unrewarded trial, mean DA dips more and ∆Ach rises more if this trial was preceded by a rewarded trial ('win–lose')
Results, Ach and DA are dynamically regulated, paragraph 5; Fig. 1g; Extended Data Fig. 4d,eHow it will be judged
Header method, stay trials, aA against AA and Aa against aa (the code's w_aA/w_AA and w_aa/w_Aa in aa_processStats_final_Fig1_DA_Words.m and _Fig1_Ach_Words.m), post window W1 and W2. Reproduced when, under each window reading, each of the four comparisons (two signals, two contrasts) goes in the stated direction in more than half of the mice with T1 p < 0.05 (the count of fig1e-reward-outcome-post). T2 across mice is reported and does not decide.
- fig1h-switch-no-history-post
When the mouse switches ports, the previous outcome does not change DA or Ach after side-port entry.
Notably, these effects of expectation on DA and Ach are absent if the mouse switches ports between trials ... the signals after side-port entry are similar
Results, Ach and DA are dynamically regulated, paragraph 5; Fig. 1h; Extended Data Fig. 4d,eHow it will be judged
A claim of absence, judged on the paper's test, T1 per mouse at p < 0.05, switch trials, aB against AB and Ab against ab (w_aB/w_AB, w_ab/w_Ab in the Fig. 1 scripts), post window W1 and W2. Reproduced when, for each comparison and signal, fewer than half of the mice show a significant difference; not reproduced otherwise. Effect sizes and T2 are reported.
- fig1h-switch-da-approach
In switch trials, DA during the move from centre to side port is larger when the previous trial was unrewarded.
Instead, the DA signals during the transition from centre to side port are greater when the previous trial was unrewarded (Fig. 1h )
Results, Ach and DA are dynamically regulated, paragraph 5; Fig. 1hHow it will be judged
Header method, mean DA in the pre window (W1, W2), switch trials with a previous loss (aB and ab) against a previous win (AB and Ab), as in the code's (pre Sw)aX/AX condition (FNoRewSw against FRewSw). Reproduced when, under each window reading, the previous-loss mean is higher in more than half of the 13 DA mice with T1 p < 0.05 (the count of fig1e-reward-outcome-post). T2 across mice is reported and does not decide.
- ed4bc-iti-port-entries
DA and Ach are modulated more around port entries in unrewarded trials than around the same port entries during the ITI; after side entry DA dips more and Ach rises more in unrewarded trials, with the opposite before side entry.
DA and Ach signals are modulated to a greater extent during unrewarded trials than during the ITI ... DA signals dip more while Ach signals rise more after side port entry, and the opposite changes occur prior to side port entry
Supplementary Results, ITI analysis of signals; Extended Data Fig. 4b,cHow it will be judged
ITI events as the Supplementary Methods define them, centre and side entries between one trial's side exit and the next centre entry, separated by at least 15 samples (0.81 s) from other port entries and exits; windows of 14 samples (0.756 s) before and after side entry, readings as W1 and W2 shortened to 14 samples; Extended Data Fig. 4a,c use mean DA and the mean minimum of Ach. Unrewarded trial side entries against ITI side entries. Reproduced when, under each window reading, post DA is lower and post Ach higher in unrewarded trials, and pre the reverse, each in more than half of the mice with T1 p < 0.05.
- ed5-base-glm-r2
A base GLM of behavioural events explains about a fifth of the variance of trial-associated DA and Ach signals (test R2 0.206 for each).
We find that it captures substantial variance across the trial-associated data (DA GLM R 2 = 0.206; Ach GLM R 2 = 0.206)
Results, Action-outcome history shapes DA and Ach, paragraph 1; Extended Data Fig. 5c,dHow it will be judged
GLM as in Methods, GLM. Binary event predictors at 54-ms bins; design matrix of 20 shifts forward and backward (41 coefficients per predictor); ITI data excluded, data from shortly before centre entry to after side exit kept, with overlapping boundary samples duplicated; trials split 50/50 into training and test; OLS (the Supplementary GLM section says OLS suffices); R2 on the test set averaged over runs. DA n = 8 mice, Ach n = 9 mice. Open, as the GLM code is not published. (1) The predictor set; the leave-out text names centre entry, centre exit, side entry, side exit, lick and reward; reading P1 uses these six, P2 adds the other recorded events (lick offset, LED onset and offset). (2) Pooling; G1 one model over all sessions of all mice; G2 one model per mouse, R2 averaged; G3 one per session, averaged. (3) Runs; Y = 10 or Y = 3. "Shortly before" and "after" are open; a standard choice of the 20-shift span (1.08 s) is used. Reproduced when both R2 fall within tolerance under every reading; blocked as methods_underspecified when readings disagree.
- ed5f-leave-out
Leaving out side entry or reward raises the GLM's MSE for both DA and Ach, leaving out side exit or lick raises it for Ach, and leaving out centre entry or centre exit alone does not change it.
For both DA and Ach models, the closely timed centre entry and centre exit are redundant because loss of either alone does not affect the model fit ... DA and Ach signals are more accurately reconstructed with the addition of side entry and reward predictors, and, additionally, side exit and lick enable better reconstructions of the Ach signal
Results, Action-outcome history shapes DA and Ach, paragraph 1; Extended Data Fig. 5fHow it will be judged
GLM as in ed5-base-glm-r2, with its open readings; Y = 10 runs, each with a 50/50 trial split held for all models, and within training a 10-fold group shuffle split by trial (80/20) for validation MSE (Methods, GLM). The paper reports no test; reproduced when, under each reading, the median validation MSE over runs rises when side entry or reward is left out (DA and Ach) and when side exit or lick is left out (Ach), with the 10 runs' ranges not overlapping the full model's, while leaving out centre entry or centre exit alone leaves the median within the full model's range. The range rules are a reading of "greatly increased" and "does not affect" named here; the paper gives no rule.
- ed5-history-glm-r2
Adding side entries split by the eight action-outcome histories improves the GLM, to a test R2 of 0.213 for DA and 0.214 for Ach, with a lower MSE and no overfitting.
This reflects an improvement in the ability of the history GLMs to capture the variance of the trial-associated data (DA GLM R 2 = 0.213; Ach GLM R 2 = 0.214) without overfitting
Results, Action-outcome history shapes DA and Ach, paragraph 2; Extended Data Fig. 5f-iHow it will be judged
As ed5-base-glm-r2, with side entry replaced by eight side-entry predictors, one per two-trial action-outcome combination (the paper does not list them; the reading is the eight combinations of stay or switch, previous outcome and current outcome). Reproduced when both R2 are within tolerance and the history model's test MSE is below the base model's in every run ("reduced the MSE", read as every run, a reading named here), under each reading.
- fig2a-rdah-smaller-than-dlight
DA signals recorded with rDAh and dLight1.1 in opposite hemispheres of the same mice are comparable in shape, with consistently smaller amplitudes for rDAh.
Both sensors yield comparable signals, but with consistently reduced amplitudes for rDAh
Results, Ach and DA release are anticorrelated, paragraph 1; Fig. 2a; Extended Data Fig. 6dHow it will be judged
Extended Data Fig. 6d compares the maximum or minimum z-scored signal in the indicated periods; the periods are not given in the text. The readings are the pre and post windows W1 and W2 (header), rewarded and unrewarded trials. Reproduced when, under each reading, the absolute peak (maximum post in rewarded, minimum post in unrewarded) is smaller for rDAh than for dLight1.1 in all 3 mice, and the trial-averaged traces of the two sensors correlate above 0.8 ("comparable", a threshold named here).
- ed6c-da-bilateral
DA release is highly correlated across the two hemispheres of the same mouse.
we exploited the fact that the release of DA is highly correlated across hemispheres within the same mouse
Results, Ach and DA release are anticorrelated, paragraph 1; Extended Data Fig. 6cHow it will be judged
Cross-correlation of trial-segregated dLight1.1 signals from the two hemispheres, as in the header (n = 4 mice). Reproduced when the peak normalized cross-correlation is positive and above 0.7 ("highly", a threshold named here) in every mouse. The lag of the peak is reported; the paper says nothing of it.
- fig2e-trial-and-noise-correlation
Trial-segregated DA and Ach signals are anticorrelated with an offset of about 100 ms in rewarded and unrewarded trials, and their noise components show a similar structure.
We observed that the trial-segregated DA and Ach signals are anticorrelated with a time offset of around 100 ms across rewarded and unrewarded trials ... This revealed a similar correlation structure (Fig. 2e )
Results, Ach and DA release are anticorrelated, paragraph 2; Fig. 2e; Extended Data Fig. 6fHow it will be judged
Header cross-correlation, 40 samples before and 100 after centre-port entry, rewarded and unrewarded trials separately, signal and noise, n = 6 mice. The tolerance is one 54-ms sample, the resolution of the signal; "around 100 ms" is two samples. The code's alignment to side entry and s.d.-scaled residuals are variants. Reproduced when, for both trial types, the signal and the noise cross-correlations have a negative minimum at a lag within tolerance in the mean over mice.
- fig2fg-off-diagonal
The two-dimensional covariance of DA and Ach fluctuations has a negative off-diagonal at which DA changes precede Ach changes by about 100 ms at most times, and it nearly disappears when the mouse enters the side port.
This revealed a strong time-lagged negative covariance (Fig. 2f ), which we call the off-diagonal (Fig. 2g ), showing that, at most time points, changes in DA precede changes in Ach by approximately 100 ms ... Notably, the negative, off-diagonal covariance nearly disappears when the mouse enters the side port
Results, Ach and DA release are anticorrelated, paragraph 3; Fig. 2f,gHow it will be judged
K(t1, t2) as the Supplementary Methods equation gives it, the trial mean of the product of residuals (each trial minus the trial-averaged signal) of DA at t1 and Ach at t2, per mouse and trial type, 40 samples before and 100 after centre entry; the code's division of residuals by the across-trial s.d. (anaCrossCor.m) is a variant. The off-diagonal is K(t, t + lag) at the lag of the minimum found in fig2e. Reproduced when, in the mean over the 6 mice, the off-diagonal is negative at more than half of the time points and its magnitude in the samples within one sample of side entry is below a quarter of its median magnitude ("nearly disappears", a threshold named here). The side-entry time in centre-aligned trials varies by trial; the readings are (a) side-aligned windows, (b) centre-aligned windows with the median side-entry time.
- fig2h-photometry-glm
Kernels of a GLM that predicts one neuromodulator from the other show a negative interaction at about 100 ms; photometry alone predicts weakly (R2 0.116 for DA, 0.171 for Ach), and adding the photometry kernels to the history GLMs raises R2 to 0.336 for DA and 0.388 for Ach.
Although the DA or Ach photometry signal alone is insufficient to robustly predict the other (Extended Data Fig. 6h) (DA GLM R2 = 0.116; Ach GLM R2 = 0.171), incorporation of these photometry kernels into history GLMs for DA and Ach improves model performance without overfitting (DA GLM R2 = 0.336; Ach GLM R2 = 0.388)
Supplementary Results, GLM with photometry variable; Fig. 2h; Extended Data Fig. 6hHow it will be judged
GLM as in ed5-base-glm-r2 and ed5-history-glm-r2, with all their open readings, on the 6 simultaneous rDAh and Ach3.0 mice, with the other channel's z-scored signal as a continuous predictor with the same 41 shifts. Reproduced when the four R2 are within tolerance and the mean kernel has its most negative coefficient at a shift of 54 to 162 ms with the predicted signal following the predictor, under every reading.
- ed6b-separate-hemispheres
DA and Ach recorded in separate hemispheres are also anticorrelated at a positive lag, with a smaller positive correlation at a negative lag.
Cross correlation analysis of these trial-segregated signals in which DA lags Ach reveals an anticorrelation with a positive time lag ... In addition, there is a smaller positive correlation at a negative time lag
Supplementary Results, Cross correlation analysis of DA and Ach transients recorded from independent hemispheres; Extended Data Fig. 6a,bHow it will be judged
Header cross-correlation of trial-segregated dLight1.1 and Ach3.0 signals from opposite hemispheres (n = 3 mice in the legend). Reproduced when the mean has a negative minimum at a positive lag and a positive maximum of smaller magnitude at a negative lag.
- fig3d-telc-abolishes-ach
TelC expressed in VLS CINs suppresses Ach release measured in vivo during rewarded trials.
potently inhibits the release of Ach from CINs in vitro (Fig. 3c ) and in vivo (Fig. 3d
Results, Striatal DA dynamics do not require CINs, paragraph 1; Fig. 3dHow it will be judged
The legend gives ∆F/F0 averages (n = 3 mice); the paper reports no test and does not give the baseline F0. The readings for F0 are (a) the session median of the demodulated signal, (b) a rolling 1-min baseline. Reproduced when, under both, the peak-to-trough range of the trial-averaged Ach ∆F/F0 within 2 s of side entry in TelC mice is below half that of mCherry mice for every TelC mouse ("potently", a threshold named here).
- fig3c-telc-blocks-cin-evoked-da-in-vitro
In acute striatal slices, optogenetic activation of Chrimson-expressing CINs evokes DA release (carbon-fibre amperometry) when the CINs coexpress mCherry, and this is blocked when they coexpress TelC, showing that TelC inhibits Ach release from CINs in vitro.
potently inhibits the release of Ach from CINs in vitro (Fig. 3c )
Results, Striatal DA dynamics do not require CINs, paragraph 1; Fig. 3cHow it will be judged
Methods, Amperometry recordings and Preparation of acute brain slices; 593-nm light, 5.86 mW, 2 ms, at 180-s intervals; +600 mV constant potential, current converted to DA concentration by calibration. The paper gives no n, test or window for this panel. Per sweep, the peak DA after the light pulse minus the mean over the 1 s before it (a standard choice named here), averaged per slice. Reproduced when the mCherry slices show a DA rise and the TelC slices' mean peak is below a quarter of the mCherry mean ("potently", a threshold named here). A two-sample two-sided t-test across slices is reported.
- fig3e-striatum-wide-telc
Striatum-wide TelC in CINs causes severe behavioural defects (changed mean ITI, decision time and RFLR alpha), yet DA still has opposite polarity after side entry for rewarded and unrewarded trials.
This widespread loss of Ach induced severe behavioural defects and greatly altered behaviourally evoked DA signals ... Nevertheless, DA retained its capacity to encode for reward, such that DA signals (Fig. 3e ) and their associated GLM kernels (Extended Data Fig. 7k ) maintain opposing polarity with reward outcome
Results, Striatal DA dynamics do not require CINs, paragraph 1; Fig. 3e; Extended Data Figs. 7k and 9f-iHow it will be judged
Behaviour metrics per mouse (behaviour code, block_transition_stats.ipynb) compared with the tests named. Reproduced when mean ITI and decision time differ with p < 0.05 (two-sided t-test), RFLR alpha differs with p < 0.05 (two-sided Wilcoxon rank-sum), and, in TelC mice, mean z-scored DA in the post window (W1, W2) is above zero in rewarded and below zero in unrewarded trials in the mean over mice. The direction of the ITI, decision-time and alpha changes is not stated in the text and is reported.
- fig3f-vls-telc-da-unchanged
Blocking Ach release from VLS CINs in one hemisphere does not change DA dynamics there compared with the control hemisphere, in any trial type or history up to three trials back.
VLS-specific loss of Ach release did not affect DA dynamics in trials recorded with dLight1.1 (Fig. 3f ) or with rDAh ... but we did not observe any significant changes in DA dynamics or in the underlying GLM kernels of DA for each input feature after loss of Ach
Results, Striatal DA dynamics do not require CINs, paragraphs 2-3; Fig. 3f; Extended Data Fig. 8d-kHow it will be judged
A claim of absence, judged on the paper's test, T1 per mouse at p < 0.05, TelC hemisphere against mCherry hemisphere, mean DA in the pre and post windows (W1, W2) for rewarded, unrewarded and history-parsed trials one, two and three back, as in aa_processStats_Fig3_DA_Ach_mCh_tet_FINAL.m. The paper does not say how per-mouse results are combined into "no significant changes"; the reading used here is that no comparison is significant in a consistent direction across mice (significant differences, if any, do not share a sign in more than half of the mice). Reproduced when that holds; not reproduced otherwise. All per-mouse effects are reported, and the count of mice with any significant comparison is reported as the strict reading.
- ed8g-vls-telc-ach-suppressed
In the simultaneous rDAh and Ach3.0 recordings, the TelC hemisphere has suppressed Ach release, validating the perturbation.
or with rDAh, in which we simultaneously validated the suppression of Ach release
Results, Striatal DA dynamics do not require CINs, paragraph 2; Extended Data Fig. 8gHow it will be judged
The paper does not say how the suppression was validated; Extended Data Fig. 8g shows z-scored traces, and z-scoring can hide a loss of amplitude. The readings are (a) the header's delta-Ach in the post window (W1, W2) of the z-scored Ach, TelC against mCherry hemisphere, T1 per mouse; (b) the peak-to-trough range of the trial-averaged Ach ∆F/F0 within 2 s of side entry, as in fig3d-telc-abolishes-ach. Reproduced when, under each reading, Ach is lower in the TelC hemisphere in every one of the 4 mice (with T1 p < 0.05 under (a)); the "every mouse" rule is a reading of "validated" named here.
- ed9jm-vls-telc-behaviour-normal
Mice with VLS-selective TelC in CINs show no behavioural deficits.
Mice did not exhibit behavioural deficits (Extended Data Fig. 9j–m )
Results, Striatal DA dynamics do not require CINs, paragraph 2; Extended Data Fig. 9j-mHow it will be judged
A claim of absence, judged on the paper's tests (Supplementary Methods, Behavior performance analysis), two-sided t-test for mean ITI, decision time, lose-switch, win-switch and mean left rates; Wilcoxon rank-sum for tau_highport, maximum P(switch) and RFLR coefficients; bootstrapped 95% CI of the difference (1000 resamples) for conditional switch probabilities, against controls, at p < 0.05. The legend's groups are TelC and mCherry mice; since each mouse has one hemisphere of each, the paper does not say what the control group is; the readings are (a) the cohort against the striatum-wide mCherry mice of Fig. 3e, (b) the cohort before and after expression, (c) the same mice's sessions split by the hemisphere recorded (TelC or mCherry). Each reading the data allow is run. Reproduced when no test is significant under any of them; when they disagree the claim is blocked as methods_underspecified.
- ed8f-amplitude-no-consistent-effect
Loss of local Ach release does not consistently change the amplitude of DA sensor transients (∆F/F0).
we analysed the amplitudes of DA sensor fluorescence transients (∆ F / F 0 ), but this analysis did not reveal consistent effects
Results, Striatal DA dynamics do not require CINs, paragraph 3; Extended Data Fig. 8fHow it will be judged
The paper names no test. Per mouse, the peak ∆F/F0 of the trial-averaged DA after side entry in the TelC and mCherry hemispheres (F0 readings as in fig3d-telc-abolishes-ach). Reproduced when the sign of the TelC minus mCherry difference is not the same in all 4 mice (a reading of "did not reveal consistent effects" named here).
- fig3g-dan-inhibition-lowers-da
Optogenetic inhibition of DANs with stGtACR2 during rewarded trials significantly reduces DA in the VLS.
inhibition of DAN activity robustly alters DA release, as evidenced by a significant reduction in the levels of DA after optogenetic inhibition of DANs with stGtACR2
Results, Striatal DA dynamics do not require CINs, paragraph 3; Fig. 3gHow it will be judged
Laser (463 nm, 0.7 mW, 5 s) triggered at side entry (Methods, Optogenetic manipulations); rDAh ∆F/F0 in stimulated rewarded trials of opsin mice against opsin-free mice (n = 3 each). The paper names no test for this panel; the readings are (a) T1 per mouse, stimulated against unstimulated rewarded trials, mean over 0 to 5 s after side entry; (b) a two-sample t-test across mice of the stimulated minus unstimulated difference, opsin against no opsin. Reproduced when DA is lower under stimulation in all 3 opsin mice with p < 0.05 under (a), or p < 0.05 under (b).
- ed7e-cin-activation-evokes-da
Optogenetic activation of Chrimson-expressing VLS CINs during the task evokes a small, significant DA release, without changing decision or side-port occupancy times.
We demonstrated that CIN stimulation consistently and significantly evokes DA release (Extended Data Fig. 7e). Therefore, exogenous activation of CINs is sufficient to drive DA release
Supplementary Results, Optogenetic activation of the CINs; Extended Data Fig. 7e,fHow it will be judged
590-nm laser, 15 mW, 1.5 s in 25% of trials at side entry (Methods, Optogenetic manipulations); stimulation artefact removed. Peak dLight1.1 change from a baseline over 1.5 s before the laser (the head-fixed baseline of the Methods; the task baseline is not given, and the reading is that same span), per mouse, paired two-sided t-test across the 7 mice (the legend's connected pairs). The legend does not say what the two dots of a pair are; the readings are (a) Chrimson against mCherry hemisphere, stimulated trials, (b) stimulated against unstimulated trials in the Chrimson hemisphere. Reproduced when, under each reading, p < 0.05 with the Chrimson or stimulated value higher, and decision and occupancy times show p > 0.05.
- fig4bc-da-bidirectionally-controls-ach
In head-fixed mice, optogenetic activation of DANs lowers Ach and inhibition of DANs raises it, opposite to the change in DA, and both effects are abolished by the D2R antagonist eticlopride.
These manipulations altered the levels of Ach in the direction opposite to optogenetically evoked changes in DA levels, consistent with DA inhibiting Ach release (Fig. 4b,c ). Notably, these effects are D2R-dependent as they are abolished by the administration of eticlopride
Results, DA inhibits Ach release through D2Rs, paragraph 1; Fig. 4b,cHow it will be judged
Methods, Optogenetic manipulations, 45-s ITI, 20 sweeps per session, baseline = mean over 1.5 s before the laser; n = 3 mice per opsin. The Methods give two durations for these panels, 5 s ("For optogenetic manipulation of DANs (Figs. 3g and 4b,c)") and 1.5 s (head-fixed mice on a wheel); the readings are a 1.5-s and a 5-s window from laser onset, and the laser timestamps, where the data hold them, say which was used. The paper names no test. Per mouse, the mean ∆F/F0 change over the stimulation relative to baseline. Reproduced when, with saline, DA rises and Ach falls under Chrimson and DA falls and Ach rises under stGtACR2 in all 3 mice, and with eticlopride the Ach change is less than a quarter of its saline size or of the opposite sign in all 3 mice ("abolished", a threshold named here). The DA change under eticlopride is reported; the paper says nothing of it.
- fig4ef-d2r-needed-for-cin-pause
In striatal slices, DA released by stimulating ChR2-expressing DAN terminals reduces CIN firing in wild-type mice, and this is absent in CINs of Drd2-cKO mice.
In wild-type mice, release of DA after laser stimulation of channelrhodopsin-expressing DAN terminals robustly reduced CIN firing, as measured by cell-attached recordings, but this effect was absent in CINs from Drd2-cKO mice
Results, DA inhibits Ach release through D2Rs, paragraph 2; Fig. 4d-fHow it will be judged
Cell-attached recordings, one 2-ms 473-nm pulse every 120 s (Methods); 9 WT and 9 Drd2-cKO cells. The paper names no test or windows. Per cell, firing rate in the 1 s after the pulse against the 1 s before (a standard choice of window, named as such; readings 0.5 s and 2 s are also shown). Reproduced when the rate falls in most WT cells (paired t-test across cells, p < 0.05) and not in Drd2-cKO cells (same test, p > 0.05), under each window.
- fig4g-cko-ach-suppression-lost
In Drd2-cKO mice the two dips in Ach that coincide with DA rises (moving from centre to side port, and after side entry in rewarded trials) are lost, so Ach signals are significantly larger than in Drd2 f/f and wild-type mice.
We found that the loss of D2Rs in CINs abolished both instances of Ach suppression that coincide with an increase in DA levels (Fig. 4g ). Together, these changes lead to significantly increased Ach signals (Extended Data Fig. 11d )
Results, DA inhibits Ach release through D2Rs, paragraph 3; Fig. 4g; Extended Data Fig. 11dHow it will be judged
Header measures on z-scored Ach, in the windows of aa_processStats_final_Fig4_cKO_Ach_Words.m (post +1 to +18 and +9 to +16, all trials pre -18 to -1 and -18 to -9 samples about side entry), the Extended Data Fig. 11d schematic not being legible in the text. The legend prints "Open circles denote a significant difference ... (two-sided ANOVA, open circle, P > 0.05)"; as open circles mark significance, "P > 0.05" is read as a misprint for P < 0.05. The unit of the ANOVA is open (its code is not published); the readings are (a) one value per mouse, the mean over its trials; (b) every trial pooled within genotype. Reproduced when, under each reading, the ANOVA gives p < 0.05 for delta-Ach in the pre window and in the rewarded post window, and two-sided t-tests show Drd2-cKO higher than both Drd2 f/f and WT at p < 0.05.
- fig4g-cko-da-unchanged
DA dynamics during the task do not differ significantly across WT, Drd2 f/f and Drd2-cKO mice.
these changes occurred despite no significant changes in DA dynamics during the task across the three genotypes (Fig. 4g and Extended Data Fig. 11d )
Results, DA inhibits Ach release through D2Rs, paragraph 3; Fig. 4g; Extended Data Fig. 11dHow it will be judged
A claim of absence, judged on the paper's test, a two-sided one-way ANOVA across the three genotypes at p < 0.05, on mean z-scored rDAh in the windows of fig4g-cko-ach-suppression-lost, with its unit readings (a) and (b) shown. Reproduced when no window and trial type gives p < 0.05 under reading (a), the unit the legend's mean ± s.d. by genotype implies; reading (b) is a variant. Effect sizes are reported.
- ed11e-cko-anticorrelation-reduced
In Drd2-cKO mice the anticorrelation between DA and Ach is significantly smaller in magnitude, and the off-diagonal negative covariance is reduced.
As a result, the relationship between Ach and DA signals in Drd2-cKO mice is severely disrupted (Extended Data Fig. 11a), with a significant reduction in the magnitude of their anticorrelation (Fig. 4h; Extended Data Fig. 11e) and the off-diagonal negative covariance
Supplementary Results, GLM for DA and Ach signals upon D2R deletion in CINs; Fig. 4h,i; Extended Data Fig. 11eHow it will be judged
Header cross-correlation and the off-diagonal of fig2fg-off-diagonal, per mouse; the amplitude of the minimum covariance per mouse compared between Drd2-cKO and each control group with a two-sample two-sided t-test across mice. Reproduced when the Drd2-cKO minimum is smaller in magnitude than both controls with p < 0.05, for signal or noise correlations, and the mean off-diagonal magnitude is smaller in Drd2-cKO.
- ed11-cko-glm-r2
History GLMs fit DA and Ach signals of the three genotypes with test R2 of 0.129, 0.122 and 0.155 for DA and 0.202, 0.262 and 0.214 for Ach (WT, Drd2 f/f, Drd2-cKO); side-entry Ach kernels are greatly altered in Drd2-cKO mice while DA kernels are comparable.
DA kernels are comparable across the three cohorts (Extended Data Fig. 11c). In contrast, all side-entry Ach kernels are greatly altered in Drd2-cKO mice
Supplementary Results, GLM for DA and Ach signals upon D2R deletion in CINs; Extended Data Fig. 11b,cHow it will be judged
History GLM as in ed5-history-glm-r2 with its open readings, per genotype. Reproduced when the six R2 are within tolerance under every reading, and the Drd2-cKO side-entry Ach kernels have larger coefficients than both controls at negative shifts for all histories, with the s.d. bands of runs not overlapping ("greatly altered", a threshold named here), and the DA kernels' s.d. bands overlap across genotypes ("comparable", likewise).
- ed11fh-cko-general-behaviour-comparable
General performance metrics and block-transition dynamics of Drd2-cKO mice do not differ from Drd2 f/f and WT mice.
Although general performance metrics, block transition dynamics and RFLR coefficients (Extended Data Fig. 11f–i ) are comparable between Drd2-cKO mice and both control groups
Results, DA inhibits Ach release through D2Rs, paragraph 4; Extended Data Fig. 11f-hHow it will be judged
A claim of absence, judged on the paper's tests (Supplementary Methods, Behavior performance analysis, and block_transition_stats.ipynb), two-sided t-tests on mean ITI, decision time, lose-switch, win-switch and mean left rates, and Wilcoxon rank-sum on tau_highport and maximum P(switch), Drd2-cKO against each control, at p < 0.05. Reproduced when none is significant; not reproduced otherwise.
- ed11i-cko-rflr
The Extended Data Fig. 11i legend reports significant differences in the RFLR beta and tau of Drd2-cKO mice, although the main text calls the RFLR coefficients comparable.
RFLR coefficients for mice in Fig. 4g . Bars denote mean with 95% C.I. Significance is determined using a two-sided Wilcoxon rank-sum test (* P ... β = 0.014; τ = 0.018)
Extended Data Fig. 11i legendHow it will be judged
RFLR per mouse (bandit_models.py), Wilcoxon rank-sum, Drd2-cKO against each control group. The legend does not name the control group; the readings are Drd2 f/f, WT, and both pooled. Reproduced when beta and tau differ with p < 0.05 under one reading; the main text's "comparable" (no difference) is reported against it, and the conflict between the two statements is recorded as a deviation.
- ed11j-cko-switching-impaired
Drd2-cKO mice are impaired in switching ports after some action-outcome histories compared with both Drd2 f/f and WT mice.
Drd2-cKO mice are impaired in their ability to switch selection ports across some histories when compared to both Drd2 f/f and wild-type cohorts
Results, DA inhibits Ach release through D2Rs, paragraph 4; Extended Data Fig. 11jHow it will be judged
Conditional switch probabilities for each two-trial (or three-trial, seq_nback = 3 in bandit_plots.py; the legend does not say) action-outcome history; per Supplementary Methods, 1000 resamples of each dataset, the difference of the parameter between groups, significant when its 95% CI excludes zero. The resampling unit (trials or sessions) is open; both are run. Reproduced when, under each reading, at least one history shows a lower switch probability in Drd2-cKO than both Drd2 f/f and WT, significant against both.
- fig5a-inputs-to-vls
Retrograde tracing from the VLS labels cells across many cortical regions and, in the thalamus, mainly in the parafascicular nucleus.
We find that a broad distribution of cells from multiple cortical regions send afferents into the VLS (Fig. 5a ). Meanwhile, thalamic inputs into this striatal region originate predominantly from the parafascicular nucleus
Results, The cortex and thalamus drive Ach release, paragraph 2; Fig. 5aHow it will be judged
Histology (Ctb in the VLS, n = 3 mice). The paper gives no counts. Reproduced when labelled cells are found in at least three cortical areas ("multiple", a threshold named here) and the parafascicular nucleus holds more labelled thalamic cells than any other thalamic nucleus ("predominantly"), in every mouse.
- fig5b-glutamate-not-lateralized
Striatal glutamate varies during the task and does not differ between ipsiversive and contraversive trials.
we find that striatal glutamate levels vary during the task but are not lateralized (Fig. 5b
Results, The cortex and thalamus drive Ach release, paragraph 3; Fig. 5bHow it will be judged
A claim of absence. The paper names no test for this panel; the test of its other lateralization claim (ed3-not-lateralized), T1 per mouse at p < 0.05 on the mean z-scored iGluSnFR signal in the pre and post windows (W1, W2), ipsiversive against contraversive trials, is used. Reproduced when fewer than half of the 4 mice show a significant difference.
- fig5c-glutamate-like-ach
Glutamate is suppressed before the choice and rises at side-port entry like Ach, and in rewarded trials shows a sustained rise during consumption that is absent in unrewarded trials.
Glutamate signals are suppressed before the choice and increase during side-port entry in an analogous manner to Ach (Fig. 5c ). In rewarded trials, glutamate exhibits an extra phase of sustained increase during consumption, which is absent in unrewarded trials
Results, The cortex and thalamus drive Ach release, paragraph 3; Fig. 5cHow it will be judged
Trial-averaged z-scored iGluSnFR (n = 4 mice) and Ach3.0 (n = 9 mice), aligned to side entry. The paper gives no test. Reproduced when, in the mean over mice, glutamate in the pre window is below its value at centre entry, glutamate rises within 0.5 s of side entry, the correlation of the mean glutamate and Ach traces from -1 to +1 s about side entry is positive, and mean glutamate from 1 to 3 s after side entry is higher in rewarded than unrewarded trials with T1 p < 0.05 in at least 3 of 4 mice. The time spans are standard choices named here; the paper does not fix them.
- fig5d-terminal-calcium-coincides
Calcium activity in cortical and thalamic terminals in the VLS coincides with glutamate release in rewarded and unrewarded trials.
The activities of cortical and thalamic terminals in the VLS, measured with genetically encoded calcium sensors, coincide with glutamate release across both trial types
Results, The cortex and thalamus drive Ach release, paragraph 3; Fig. 5dHow it will be judged
Trial-averaged z-scored jRCaMP1b (thalamic terminals, 6 mice) and GCaMP8 (cortical terminals, 5 mice) against iGluSnFR (4 mice), separate mice. The paper gives no test. Reproduced when, for each trial type, the mean calcium trace of each input correlates positively with the mean glutamate trace over -2 to +3 s about side entry (r above 0.5, a threshold named here).
- fig5ef-input-telc-dampens-ach
Silencing cortical or thalamic glutamatergic inputs with TelC robustly lowers Ach sensor ∆F/F0 across all trials, consistently across mice.
Loss of neurotransmission from each region robustly dampened Ach transients across all trials, as seen in the lowered ∆ F / F 0 of the Ach sensor signal ... The degree of suppression of ∆ F / F 0 was strong and consistent across mice
Results, The cortex and thalamus drive Ach release, paragraph 4; Fig. 5e,f; Extended Data Fig. 13eHow it will be judged
Ach3.0 ∆F/F0 (F0 readings as in fig3d-telc-abolishes-ach), cortex TelC 5 mice against mCherry 4, thalamus TelC 5 against mCherry 3. The paper names no test. Reproduced when, under each F0 reading, the peak-to-trough range of every TelC mouse's trial-averaged ∆F/F0 within 2 s of side entry is below the smallest range among its mCherry group, for both inputs ("strong and consistent across mice", a threshold named here).
- fig5ef-input-specific-ach
In unrewarded trials, loss of cortical but not thalamic input perturbs Ach after side-port entry, and loss of thalamic input shifts the timing of Ach transients more than loss of cortical input.
In unrewarded trials, loss of cortical but not thalamic inputs perturbs Ach transients after side-port entry, suggesting that the cortex has a specific role in driving this signal. In addition, loss of the thalamic input shifts the timing of Ach transients more than does loss of the cortical input
Results, The cortex and thalamus drive Ach release, paragraph 5; Fig. 5e,fHow it will be judged
Z-scored Ach (Fig. 5e,f, bottom), TelC against mCherry per input, unrewarded trials. The paper names no test or measure. Delta-Ach in the post window (W1, W2), the paper's Ach measure elsewhere, compared between the TelC and mCherry groups with a two-sample two-sided t-test across mice (a standard choice named here); timing as the latency of the trial-averaged Ach peak after side entry (likewise a choice named here). Reproduced when the cortex comparison gives p < 0.05 and the thalamus comparison does not, and the absolute peak-latency shift from mCherry is larger for thalamic than cortical TelC, under each window reading.
- ed13-input-telc-behaviour
Cortical or thalamic TelC impairs performance, with lower reward rate and win-switch differences, and the RFLR shows a smaller beta and a larger tau.
Consistent with this impairment, the RFLR model description revealed a reduction in β and an increase in τ
Results, The cortex and thalamus drive Ach release, paragraph 4; Extended Data Fig. 13f-mHow it will be judged
Behaviour code (block_transition_stats.ipynb, bandit_models.py) with the tests the legends name, TelC against mCherry per input. Reproduced when, for each input, beta is lower and tau higher in TelC mice with p < 0.05 (two-sided Wilcoxon rank-sum), and reward rate is lower with p < 0.05 (two-sided t-test). The win-switch direction is not stated and is reported. The legend prints p = 0.05 for cortical beta; a value at or just above 0.05 is recorded as a match to the paper.
- ed2op-mutant-sensors-flat
The DA and Ach transients depend on neuromodulator binding, being absent with the ligand-binding-site mutant sensors.
and that depended on neuromodulator binding, because they were absent in ligand-binding-site mutants of the sensors
Results, Ach and DA are dynamically regulated, paragraph 1; Extended Data Fig. 2k-pHow it will be judged
Rewarded-trial ∆F/F0 of dLight1.1 against its D103A mutant and Ach3.0 against its mutant (n = 4 mice each; F0 readings as in fig3d-telc-abolishes-ach). The paper gives no test. Reproduced when the peak-to-trough range of the trial-averaged mutant trace within 2 s of side entry is below a quarter of the wild-type sensor's in every mouse ("absent", a threshold named here).
Pipeline
- Paper
- done (1 of 1)
- Claims
- done
- Claims check
- done (ready)
- Attempt
- todo
- Questions
- none
- Verified
- none