OSCR

Model-based analysis of stop-signal data reveals robust neural and clinical correlates of evidence accumulation but not inhibition.

Code ↔ Paper

3 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 3 matches
  1. [1] § METHODS AND MATERIALS › RDEX-ABCD model ↔ Simulation Studies/dmc/models/WALD-ABCDlg_05/dists.R, lines 595–706 · score 0.71 · trigger failure, go failure, Stop signal, exponential, accuracy, race
  2. [2] § METHODS AND MATERIALS › RDEX-ABCD model ↔ Data Preparation/ABCD_SST_subject_selection.R, lines 1–76 · score 0.66 · choice accuracy, incorrect, omissions, Go trials, stimulus, RTs
  3. [3] § METHODS AND MATERIALS › RDEX-ABCD model evaluation and estimation ↔ Modeling and Recovery/interpretation_plots.R, lines 2–40 · score 0.54 · generalized power, RDEX ABCD model, growth, linear, fit

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

R · 1,183 lines · 43 KB · no license · 1 match

  1. ################ Wald exg stop signal with truncaiton in lower bound of exg
  3. ######################## ExGaussian ----
  5. # DISTRIBTION AND RANDOM FUNCTIONS ----
  7. # Modified from gamlss.dist to make cdf in tau > 0.05 * sigma case robust,
  8. # and robust to -Inf and Inf inputs, returns NA for bad sigma or tau and
  9. # robust against small sigma cases. Also changes nu name to tau
  10. pexGAUS <- function (q, mu = 5, sigma = 1, tau = 1, lower.tail = TRUE, log.p = FALSE)
  11. {
  12. if (sigma <= 0) return(rep(NA,length(q)))
  13. if (tau <= 0) return(rep(NA,length(q)))
  15. if (sigma < 1e-4)
  16. return(pexp(q-mu,1/tau,log=log.p,lower.tail=lower.tail)) # shfited exponential
  18. ly <- length(q)
  19. sigma <- rep(sigma, length = ly)
  20. mu <- rep(mu, length = ly)
  21. tau <- rep(tau, length = ly)
  22. index <- seq(along = q)
  23. z <- q - mu - ((sigma^2)/tau)
  24. cdf <- ifelse(is.finite(q),
  25. ifelse(tau > 0.05 * sigma,
  26. pnorm((q - mu)/sigma) -
  27. exp(log(pnorm(z/sigma)) + ((mu + (sigma^2/tau))^2 - (mu^2) -
  28. 2 * q * ((sigma^2)/tau))/(2 * sigma^2)),
  29. pnorm(q, mean = mu, sd = sigma)),
  30. ifelse(q<0,0,1)
  31. )
  32. if (lower.tail == TRUE)
  33. cdf <- cdf
  34. else cdf <- 1 - cdf
  35. if (log.p == FALSE)
  36. cdf <- cdf
  37. else cdf <- log(cdf)
  38. cdf
  39. }
  41. # gamlss.dist function, but returns NA for bad sigma or tau, and
  42. # robust against small sigma cases and changed parameter name nu to tau.
  43. dexGAUS <- function (x, mu = 5, sigma = 1, tau = 1, log = FALSE)
  44. {
  45. if (sigma <= 0) return(rep(NA,length(x)))
  46. if (tau <= 0) return(rep(NA,length(x)))
  48. if (sigma < 1e-4)
  49. return(dexp(x-mu,1/tau,log=log)) # shfited exponential
  51. ly <- length(x)
  52. sigma <- rep(sigma, length = ly)
  53. mu <- rep(mu, length = ly)
  54. tau <- rep(tau, length = ly)
  55. z <- x - mu - ((sigma^2)/tau)
  56. logfy <- ifelse(tau > 0.05 * sigma, -log(tau) - (z + (sigma^2/(2 *
  57. tau)))/tau + log(pnorm(z/sigma)), dnorm(x, mean = mu, sd = sigma,
  58. log = TRUE))
  59. if (log == FALSE)
  60. fy <- exp(logfy)
  61. else fy <- logfy
  62. fy
  63. }
  65. # TRUNCATED VERSIONS ----
  67. # n=100; mu=0; sigma=.05; tau=.1; a=0; b=Inf
  68. rEXG <- function (n,mu=.5,sigma=.05,tau=.1,a=-Inf,b=Inf,max.try=1000)
  69. {
  70. out <- rnorm(n=n, mean=mu,sd=sigma) + rexp(n=n,rate=1/tau)
  71. ok <- (out > a) & (out < b)
  72. if (any(!ok)) {
  73. nneed <- sum(!ok)
  74. nget <- ceiling(1.1*nneed/max(c(mean(ok),.001)))
  75. rtok <- numeric(nneed)
  76. full <- logical(nneed)
  77. n.try=0
  78. repeat {
  79. tmp <- rnorm(n=nget, mean=mu,sd=sigma) + rexp(n=nget,rate=1/tau)
  80. okrt <- (tmp > a) & (tmp < b)
  81. if ( sum(okrt)>=nneed ) {
  82. rtok[!full] <- tmp[okrt][1:nneed]
  83. break
  84. }
  85. ngot <- sum(okrt)
  86. if (ngot>0) {
  87. rtok[!full][1:ngot] <- tmp[okrt]
  88. full[!full][1:ngot] <- TRUE
  89. }
  90. nneed <- nneed - ngot
  91. if (all(!ok))
  92. nget <- nneed*10 else
  93. nget <- ceiling(nneed*2/max(c(mean(ok),.001)))
  94. n.try <- n.try+1
  95. if (n.try==max.try) break
  96. }
  97. out[!ok] <- rtok
  98. }
  99. out
  100. }
  102. dEXG <- function (x, mu = 5, sigma = 1, tau = 1, a=-Inf, b = Inf, log = FALSE)
  104. {
  105. out <- numeric(length(x))
  106. ok <- (x>a) | (x<b)
  107. if (a == -Inf) Fa <- 0 else Fa <- pexGAUS(a,mu=mu,sigma=sigma,tau=tau)
  108. if (b == Inf) Fb <- 1 else Fb <- pexGAUS(b,mu=mu,sigma=sigma,tau=tau)
  109. out <- dexGAUS(x[ok],mu,sigma,tau,log=FALSE)/(Fb-Fa)
  110. if (log) log(out) else out
  111. }
  113. pEXG <- function (q, mu = 5, sigma = 1, tau = 1, a=-Inf, b = Inf,
  114. lower.tail = TRUE, log.p = FALSE)
  116. {
  117. out <- numeric(length(q))
  118. small <- (q<=a); big <- (q>=b); ok <- !small & !big
  119. out[big] <- 1
  120. if (a == -Inf) Fa <- 0 else Fa <- pexGAUS(a,mu=mu,sigma=sigma,tau=tau)
  121. if (b == Inf) Fb <- 1 else Fb <- pexGAUS(b,mu=mu,sigma=sigma,tau=tau)
  122. out[ok] <- (pexGAUS(q[ok],mu,sigma,tau,log.p=FALSE,lower.tail=lower.tail)-Fa)/(Fb-Fa)
  123. if (log.p) log(out) else out
  124. }
  126. # par(mfrow=c(1,2))
  127. # n=1e5;mu=.3;sigma=.2;tau=.2;a=0;b=Inf
  128. # sim <- rEXG(n=n,mu=mu,sigma=sigma,tau=tau,a=a,b=b)
  129. # if (!is.finite(b)) hi <- max(sim) else hi <- b
  130. # tmp=hist(sim,breaks=seq(a,hi,length.out=100),freq=FALSE)
  131. # x=tmp$mids; points(x,dEXG(x,mu=mu,sigma=sigma,tau=tau,a=a,b=b),pch=16,cex=.5)
  132. # plot(x,cumsum(tmp$density)/sum(tmp$density),type="l",ylab="cdf")
  133. # points(x,pEXG(x,mu=mu,sigma=sigma,tau=tau,a=a,b=b),pch=16,cex=.5)
  135. ######################## Wald ----
  137. # n-choice uniformly varying start point (0-A) Wald
  138. # race, with t0, v, A, b (boundary) parameterixaiton
  140. ### Single accumulator model ----
  142. # pigt, digt, rwaldt Copyright (C) 2013 Trisha Van Zandt distributed with:
  143. # Logan, Van Zandt, Verbruggen, and Wagenmakers (2014). On the ability to
  144. # inhibit thought and action: General and special theories of an act of control.
  145. # Psychological Review. Comments and changes added by Andrew Heathcote. Trish's
  146. # code is for k = threshold, a = half width of uniform threshold variability,
  147. # l = rate of accumulation. Note that Wald mean = k/l and shape = k^2.
  149. # Following functions use a different parameterization in terms of v=l (rate),
  150. # uniform start point variability from 0-A (A>=0), threshold b (>0) and hence
  151. # B=b-A (>=0) as a threshold gap. Hence k = b-A/2 = B + A/2 and a=A/2
  153. rWald <- function(n,B,v,A)
  154. # random function for single acumulator
  155. {
  157. rwaldt <- function(n,k,l,tiny=1e-6) {
  158. # random sample of n from a Wald (or Inverse Gaussian)
  159. # k = criterion, l = rate, assumes sigma=1 Browninan motion
  160. # about same speed as statmod rinvgauss
  162. rlevy <- function(n=1, m=0, c=1) {
  163. if (any(c<0)) stop("c must be positive")
  164. c/qnorm(1-runif(n)/2)^2+m
  165. }
  167. flag <- l>tiny
  168. x <- rep(NA,times=n)
  170. x[!flag] <- rlevy(sum(!flag),0,k[!flag]^2)
  171. mu <- k/l
  172. lambda <- k^2
  174. y <- rnorm(sum(flag))^2
  175. mu.0 <- mu[flag]
  176. lambda.0 <- lambda[flag]
  178. x.0 <- mu.0 + mu.0^2*y/(2*lambda.0) -
  179. sqrt(4*mu.0*lambda.0*y + mu.0^2*y^2)*mu.0/(2*lambda.0)
  181. z <- runif(length(x.0))
  182. test <- mu.0/(mu.0+x.0)
  183. x.0[z>test] <- mu.0[z>test]^2/x.0[z>test]
  184. x[flag] <- x.0
  185. x[x<0] <- max(x)
  186. x
  187. }
  189. # Act as if negative v never terminates, cluge to do single accumulator
  190. # case by passing negative v
  191. if (length(v)!=n) v <- rep(v,length.out=n)
  192. if (length(B)!=n) B <- rep(B,length.out=n)
  193. if (length(A)!=n) A <- rep(A,length.out=n)
  195. # Kluge to return -Inf for negative rates, so can implment one accumulator case
  196. out <- numeric(n)
  197. ok <- v>0
  198. nok <- sum(ok)
  199. bs <- B[ok]+runif(nok,0,A[ok])
  200. out[ok] <- rwaldt(nok,k=bs,l=v[ok])
  201. out[!ok] <- Inf
  202. out
  203. }
  206. dWald <- function(t,v,B,A)
  207. # density for single accumulator
  208. {
  210. digt <- function(t,k=1,l=1,a=.1,tiny=1e-10) {
  211. # pdf of inverse gaussian at t with k +/- a/2 uniform variability
  212. # returns digt.0 if a<1e-10
  214. digt.0 <- function(t,k=1,l=1) {
  215. # pdf of inverse gaussian at t with no k variability
  216. # much faster than statmod's dinvgauss funciton
  218. lambda <- k^2
  219. l0 <- l==0
  220. e <- numeric(length(t))
  221. if ( any(!l0) ) {
  222. mu <- k[!l0]/l[!l0]
  223. e[!l0] <- -(lambda[!l0]/(2*t[!l0])) * (t[!l0]^2/mu^2 - 2*t[!l0]/mu + 1)
  224. }
  225. if ( any(l0) ) e[l0] <- -.5*lambda[l0]/t[l0]
  226. x <- exp(e + .5*log(lambda) - .5*log(2*t^3*pi))
  227. x[t<=0] <- 0
  228. x
  229. }
  231. options(warn=-1)
  232. if(length(k)!=length(t)) k <- rep(k,length.out=length(t))
  233. if(length(l)!=length(t)) l <- rep(l,length.out=length(t))
  234. if(length(a)!=length(t)) a <- rep(a,length.out=length(t))
  236. tpos <- t<=0
  238. atiny <- a<=tiny & !tpos
  239. a[atiny] <- 0
  241. ltiny <- (l<=tiny) & !atiny & !tpos
  242. notltiny <- (l>tiny) & !atiny & !tpos
  243. l[l<=tiny] <- 0
  245. x <- numeric(length(t))
  247. # No threshold variability
  248. if ( any(atiny) )
  249. x[atiny] <- digt.0(t=t[atiny],k=k[atiny],l=l[atiny])
  251. # Threshold variability
  252. if ( any(!atiny) ) {
  254. if ( any(notltiny) ) { # rate non-zero
  256. sqr.t <- sqrt(t[notltiny])
  258. term.1a <- -(a[notltiny]-k[notltiny]+t[notltiny]*l[notltiny])^2/(2*t[notltiny])
  259. term.1b <- -(a[notltiny]+k[notltiny]-t[notltiny]*l[notltiny])^2/(2*t[notltiny])
  260. term.1 <- (exp(term.1a) - exp(term.1b))/sqrt(2*pi*t[notltiny])
  262. term.2a <- log(.5)+log(l[notltiny])
  263. term.2b <- 2*pnorm((-k[notltiny]+a[notltiny])/sqr.t+sqr.t*l[notltiny])-1
  264. term.2c <- 2*pnorm((k[notltiny]+a[notltiny])/sqr.t-sqr.t*l[notltiny])-1
  265. term.2d <- term.2b+term.2c
  266. term.2 <- exp(term.2a)*term.2d
  268. term.3 <- term.1+term.2
  269. term.4 <- log(term.3)-log(2)-log(a[notltiny])
  270. x[notltiny] <- exp(term.4)
  271. }
  273. if ( any(ltiny) ) { # rate zero
  274. log.t <- log(t[ltiny])
  275. term.1 <- -.5*(log(2)+log(pi)+log.t)
  276. term.2 <- (k[ltiny]-a[ltiny])^2/(2*t[ltiny])
  277. term.3 <- (k[ltiny]+a[ltiny])^2/(2*t[ltiny])
  278. term.4 <- (exp(-term.2)-exp(-term.3))
  279. term.5 <- term.1+log(term.4) - log(2) - log(a[ltiny])
  280. x[ltiny] <- exp(term.5)
  281. }
  283. }
  285. x[x<0 | is.nan(x) ] <- 0
  286. x
  287. }
  289. out <- numeric(length(t))
  290. ok <- v>0
  291. out[ok] <- digt(t[ok],k=B[ok]+A[ok]/2,l=v[ok],a=A[ok]/2)
  292. out[!ok] <- 0
  293. out
  294. }
  297. pWald <- function(t,v,B,A)
  298. # cumulative density for single accumulator
  299. {
  300. pigt <- function(t,k=1,l=1,a=.1,tiny=1e-10) {
  301. # cdf of inverse gaussian at t with k +/- a/2 uniform variability
  302. # returns pigt.0 if a<=0
  304. pigt.0 <- function(t,k=1,l=1) {
  305. # cdf of inverse gaussian at t with no k variability
  306. # much faster than statmod's pinvgauss funciton
  308. mu <- k/l
  309. lambda <- k^2
  311. e <- exp(log(2*lambda) - log(mu))
  312. add <- sqrt(lambda/t) * (1 + t/mu)
  313. sub <- sqrt(lambda/t) * (1 - t/mu)
  315. p.1 <- 1 - pnorm(add)
  316. p.2 <- 1 - pnorm(sub)
  317. x <- exp(e + log(p.1)) + p.2
  319. x[t<0] <- 0
  320. x
  321. }
  323. options(warn=-1)
  324. if(length(k)!=length(t)) k <- rep(k,length.out=length(t))
  325. if(length(l)!=length(t)) l <- rep(l,length.out=length(t))
  326. if(length(a)!=length(t)) a <- rep(a,length.out=length(t))
  328. tpos <- t<=0
  330. atiny <- a<=tiny & !tpos
  331. a[atiny] <- 0
  333. ltiny <- (l<=tiny) & !atiny & !tpos
  334. notltiny <- (l>tiny) & !atiny & !tpos
  335. l[l<=tiny] <- 0
  337. x <- numeric(length(t))
  339. # No threshold variability
  340. if ( any(atiny) )
  341. x[atiny] <- pigt.0(t[atiny],k[atiny],l[atiny])
  343. # Threshold variability
  344. if ( any(!atiny) ) {
  346. if ( any(notltiny) ) { # rate non-zero
  348. log.t <- log(t[notltiny])
  349. sqr.t <- sqrt(t[notltiny])
  351. term.1a <- .5*log.t-.5*log(2*pi)
  352. term.1b <- exp(-((k[notltiny]-a[notltiny]-t[notltiny]*l[notltiny])^2/t[notltiny])/2)
  353. term.1c <- exp(-((k[notltiny]+a[notltiny]-t[notltiny]*l[notltiny])^2/t[notltiny])/2)
  354. term.1 <- exp(term.1a)*(term.1b-term.1c)
  356. term.2a <- exp(2*l[notltiny]*(k[notltiny]-a[notltiny]) +
  357. log(pnorm(-(k[notltiny]-a[notltiny]+t[notltiny]*l[notltiny])/sqr.t)))
  358. term.2b <- exp(2*l[notltiny]*(k[notltiny]+a[notltiny]) +
  359. log(pnorm(-(k[notltiny]+a[notltiny]+t[notltiny]*l[notltiny])/sqr.t)))
  360. term.2 <- a[notltiny] + (term.2b-term.2a)/(2*l[notltiny])
  362. term.4a <- 2*pnorm((k[notltiny]+a[notltiny])/sqr.t-sqr.t*l[notltiny])-1
  363. term.4b <- 2*pnorm((k[notltiny]-a[notltiny])/sqr.t-sqr.t*l[notltiny])-1
  364. term.4c <- .5*(t[notltiny]*l[notltiny] - a[notltiny] - k[notltiny] + .5/l[notltiny])
  365. term.4d <- .5*(k[notltiny] - a[notltiny] - t[notltiny]*l[notltiny] - .5/l[notltiny])
  366. term.4 <- term.4c*term.4a + term.4d*term.4b
  368. x[notltiny] <- (term.4 + term.2 + term.1)/(2*a[notltiny])
  369. }
  371. if ( any(ltiny) ) { # rate zero
  372. sqr.t <- sqrt(t[ltiny])
  373. log.t <- log(t[ltiny])
  374. term.5a <- 2*pnorm((k[ltiny]+a[ltiny])/sqr.t)-1
  375. term.5b <- 2*pnorm(-(k[ltiny]-a[ltiny])/sqr.t)-1
  376. term.5 <- (-(k[ltiny]+a[ltiny])*term.5a - (k[ltiny]-a[ltiny])*term.5b)/(2*a[ltiny])
  378. term.6a <- -.5*(k[ltiny]+a[ltiny])^2/t[ltiny] - .5*log(2) -.5*log(pi) + .5*log.t - log(a[ltiny])
  379. term.6b <- -.5*(k[ltiny]-a[ltiny])^2/t[ltiny] - .5*log(2) -.5*log(pi) + .5*log.t - log(a[ltiny])
  380. term.6 <- 1 + exp(term.6b) - exp(term.6a)
  382. x[ltiny] <- term.5 + term.6
  383. }
  385. }
  387. x[x<0 | is.nan(x) ] <- 0
  388. x
  389. }
  391. out <- numeric(length(t))
  392. ok <- v>0
  393. out[ok] <- pigt(t[ok],k=B[ok]+A[ok]/2,l=v[ok],a=A[ok]/2)
  394. out[!ok] <- 0
  395. out
  397. }
  401. ######################## Race model ----
  403. rWaldRace <- function(n,v,B,A,t0,return.ttf=FALSE)
  404. # random function for Wald race.
  405. {
  406. B[B<0] <- 0 # Protection for negatives
  407. A[A<0] <- 0
  408. n_v <- ifelse(is.null(dim(v)), length(v), dim(v)[1])
  409. ttf <- matrix(t0 + rWald(n*n_v,B=B,v=v,A=A),nrow=n_v)
  410. if (return.ttf) return(ttf)
  411. resp <- apply(ttf, 2, which.min)
  412. data.frame(RT = ttf[cbind(resp,1:n)], R = apply(ttf, 2, which.min))
  413. }
  415. rWaldexgRace <- function(n,v,B,A,t0,gf=0,a=-Inf,b=Inf,return.ttf=FALSE)
  416. # random function for Wald race.
  417. {
  418. B[B<0] <- 0 # Protection for negatives
  419. A[A<0] <- 0
  420. n_v <- ifelse(is.null(dim(v)), length(v), dim(v)[1])
  421. ttf <- t0 + rbind(rEXG(n,mu=v[1,1],sigma=B[1,1],tau=A[1,1],a=a[1]),
  422. matrix(rWald(n*(n_v-1),B=B[-1,],v=v[-1,],A=A[-1,]),nrow=n_v-1))
  423. if (return.ttf) return(ttf)
  424. resp <- apply(ttf, 2, which.min)
  425. out <- data.frame(RT = ttf[cbind(resp,1:n)], R = apply(ttf, 2, which.min))
  427. if (gf[1] > 0) {
  428. is.gf <- as.logical(rbinom(dim(out)[1],1,gf))
  429. out$RT[is.gf] <- NA
  430. out$R[is.gf] <- 1
  431. }
  433. out
  434. }
  436. n1Wald <- function(dt,v,B,A,t0=0,gf=0)
  437. # Generates defective PDF for responses on node=1, dt (decison time) is a vector of times
  438. {
  439. B[B<0] <- 0 # Protection for negatives
  440. A[A<0] <- 0
  441. n_acc <- ifelse(is.null(dim(v)),length(v),dim(v)[1])
  442. if (is.null(dim(dt))) dt <- matrix(rep(dt,each=n_acc),nrow=n_acc)
  443. # dt <- dt-t0
  445. is.go <- !is.na(dt[1,])
  446. n.go <- sum(is.go)
  448. if (!is.matrix(v)) v <- matrix(rep(v,n.go),nrow=n_acc)
  449. if (!is.matrix(B)) B <- matrix(rep(B,n.go),nrow=n_acc)
  450. if (!is.matrix(A)) A <- matrix(rep(A,n.go),nrow=n_acc)
  452. # Winner
  453. dt[1,is.go] <- (1-gf[1])*dWald(dt[1,is.go],A=A[1,],v=v[1,],B=B[1,])
  454. if (n_acc > 1) for (i in 2:n_acc)
  455. dt[1,is.go] <- dt[1,is.go]*(1-pWald(dt[i,is.go],A=A[i,],v=v[i,],B=B[i,]))
  457. dt[1,!is.go] <- gf[1]
  459. dt[1,]
  460. }
  463. n1Waldexg <- function(dt,v,B,A,Si=1,t0=0,gf=0,minEXG=-Inf)
  464. # Generates defective PDF for responses on node=1, dt (decison time) is a vector of times
  465. {
  466. B[B<0] <- 0 # Protection for negatives
  467. A[A<0] <- 0
  468. n_acc <- ifelse(is.null(dim(v)),length(v),dim(v)[1])
  469. if (is.null(dim(dt))) dt <- matrix(rep(dt,each=n_acc),nrow=n_acc)
  470. # dt[Si,] <- dt[Si,]-t0
  471. # dt <- dt-t0
  473. is.go <- !is.na(dt[1,])
  474. n.go <- sum(is.go)
  476. if (!is.matrix(v)) v <- matrix(rep(v,n.go),nrow=n_acc)
  477. if (!is.matrix(B)) B <- matrix(rep(B,n.go),nrow=n_acc)
  478. if (!is.matrix(A)) A <- matrix(rep(A,n.go),nrow=n_acc)
  480. # Winner
  481. if (Si==1)
  482. dt[1,is.go] <- (1-gf[1])*dEXG(dt[1,is.go],mu = v[1,1],sigma = B[1,1],tau = A[1,1],a=minEXG) else
  483. dt[1,is.go] <- (1-gf[1])*dWald(dt[1,is.go],A=A[1,],v=v[1,],B=B[1,])
  484. if (n_acc > 1) for (i in 2:n_acc)
  485. if (Si==i)
  486. dt[1,is.go] <- dt[1,is.go]*(1-pEXG(q=dt[i,is.go],mu=v[i,1],sigma=B[i,1],tau=A[i,1],a=minEXG)) else
  487. dt[1,is.go] <- dt[1,is.go]*(1-pWald(dt[i,is.go],A=A[i,],v=v[i,],B=B[i,]))
  489. dt[1,!is.go] <- gf[1]
  491. dt[1,]
  492. }
  495. # NB1: no st0
  497. # NB2: TRIALS effect on B (threshold), and values < 0 set to 0
  499. rWaldss <- function (n, v, B, t0, A=0, tf=0, gf=0, ts = 0, minEXG=-Inf,
  500. SSD=Inf, TRIALS = NA, staircase=NA)
  501. # Race among length(v) accumulators (if v a vector),
  502. # or dim(v)[1] (if a matrix), first of which is a stop accumulator.
  503. # Acts the same as rWald except NA returned for RT when winner = 1.
  504. # Optional SSD argument can be used to adjust start time for first
  505. # accumulator. SSD can be a scalar or vector length n; output has an SSD column
  506. # For trials with winning first accumulator RT and R set to NA.
  507. # tf = trigger failure probability, gf = go failure probability
  508. # If any !is.na in staircase runs a staircase
  509. # t0 is a scalar with the standard interpritaiton for GO accumulators.
  510. # minEXG is lower bound on EXG wald
  511. # ts = slope of slowing (speeding if negative) over TRIALS, meanlog - ts*TRIALS
  512. # This has a linear effect on mean and sd
  514. {
  515. if ( length(SSD)==1 ) SSD <- rep(SSD,n)
  516. if ( any(is.na(SSD)) || length(SSD) != n )
  517. stop("SSD cannot have NAs and must be a scalar or same length as n")
  519. n_acc <- ifelse(is.null(dim(v)),length(v),dim(v)[1])
  521. # head start for stop accumulator relative to 0 for go accumulators
  522. t0S <- matrix(rep(c(-t0[1],rep(0,n_acc-1)),length.out=n*n_acc),nrow=n_acc)
  524. if (!is.matrix(v)) v <- matrix(rep(v,n),nrow=n_acc)
  525. if (!is.matrix(B)) B <- matrix(rep(B,n),nrow=n_acc)
  526. if (!is.matrix(A)) A <- matrix(rep(A,n),nrow=n_acc)
  528. if ( !any(is.na(TRIALS)) ) {
  529. if (length(TRIALS)!=n)
  530. stop("TRIALS must have length n")
  531. B[-1,] <- B[-1,] + rep(ts*TRIALS,each=n_acc-1)
  532. }
  534. if ( gf > 0 ) # Setup for GO failure
  535. is.gf <- as.logical(rbinom(length(SSD),1,gf)) else
  536. is.gf <- logical(length(SSD))
  538. if ( all(!is.finite(SSD)) ) { # ALL GO
  539. out <- rWaldRace(n,v=v[-1,,drop=FALSE],B=B[-1,,drop=FALSE],A=A[-1,,drop=FALSE],t0=t0)
  540. out$R <- out$R+1
  541. } else { # SOME STOP
  542. if ( any(is.na(staircase)) ) { # STOP fixed SSD
  543. # add SSD to stop accumulator
  544. t0S[1,] <- t0S[1,] + SSD
  545. out <- rWaldexgRace(n,v=v,B=B,A=A,t0=t0S,a=minEXG)
  546. if ( tf>0 ) {
  547. is.tf <- logical(length(SSD))
  548. is.tf[is.finite(SSD)][as.logical(rbinom(sum(is.finite(SSD)),1,tf))] <- TRUE
  549. if ( any(is.tf) ) {
  550. out[is.tf,] <- rWaldRace(sum(is.tf),v=v[-1,is.tf,drop=FALSE],
  551. B=B[-1,is.tf,drop=FALSE],A=A[-1,is.tf,drop=FALSE],t0=0)
  552. out[is.tf,"R"] <- out[is.tf,"R"]+1
  553. }
  554. }
  555. } else { # STOP, staircase
  556. if ( !is.numeric(staircase) | length(staircase)!=1 )
  557. stop("Staircase must be a numeric vector of length 1 specifying the step.")
  558. SSDi <- SSD[is.finite(SSD)][1] # begining SSD
  559. dt <- rWaldexgRace(n,v=v,B=B,A=A,t0=t0S,return.ttf=TRUE,a=minEXG)
  560. # Setup
  561. winner <- numeric(n)
  562. for ( i in c(1:n) ) {
  563. if ( !is.finite(SSD[i]) ) # not staircase
  564. dt[1,i] <- dt[1,i] + SSD[i] else
  565. dt[1,i] <- dt[1,i] + SSDi # staircase
  566. if ( runif(1)<tf ) # Trigger failure
  567. winner[i] <- which.min(dt[2:n_acc,i])+1 else
  568. winner[i] <- which.min(dt[,i])
  569. if (is.gf[i]) winner[i] <- 1
  570. if ( is.finite(SSD[i]) ) { # update staircase
  571. SSD[i] <- SSDi
  572. if ( winner[i]==1 )
  573. SSDi <- SSDi + staircase else
  574. SSDi <- SSDi - staircase
  575. if (SSDi<1e-10) SSDi <- 0
  576. }
  577. }
  578. out <- data.frame(RT=dt[cbind(winner,1:n)],R=winner)
  579. }
  580. out$RT <- out$RT + t0 # Add t0 for go responses
  581. }
  583. # print(out)
  585. out[out$R==1,"RT"] <- NA
  586. if ( gf > 0 ) {
  587. out$RT[is.gf] <- NA
  588. out$R[is.gf] <- 1
  589. }
  590. if ( any(is.na(TRIALS)) ) cbind.data.frame(out,SSD=SSD) else
  591. cbind.data.frame(out,SSD=SSD,TRIALS=TRIALS)
  592. }
  595. rWaldssABCD <- function (n, v, B, t0,v0,vT,vF,k=1,minEXG=-Inf,
  596. A=0, tf=0, gf=0, ts = 0,SSD=Inf, TRIALS = NA, staircase=NA)
  598. # Like rWaldss except adds change in v with SSD
  599. # Exponential from v0 to vT and vF
  601. {
  602. if ( length(SSD)==1 ) SSD <- rep(SSD,n)
  603. if ( any(is.na(SSD)) || length(SSD) != n )
  604. stop("SSD cannot have NAs and must be a scalar or same length as n")
  606. n_acc <- ifelse(is.null(dim(v)),length(v),dim(v)[1])
  608. # head start for stop accumulator relative to 0 for go accumulators
  609. t0S <- matrix(rep(c(-t0[1],rep(0,n_acc-1)),length.out=n*n_acc),nrow=n_acc)
  611. # Get index of true and false rates
  612. iT <- c(1:length(v))[v==Inf]
  613. iF <- c(1:length(v))[v==-Inf]
  614. # Get index of stop-signal trials
  615. isSS <- is.finite(SSD)
  617. if (!is.matrix(v)) v <- matrix(rep(v,n),nrow=n_acc)
  618. if (!is.matrix(B)) B <- matrix(rep(B,n),nrow=n_acc)
  619. if (!is.matrix(A)) A <- matrix(rep(A,n),nrow=n_acc)
  621. if ( any(!isSS) ) {
  622. v[iT,!isSS] <- vT
  623. v[iF,!isSS] <- vF
  624. }
  626. if ( !any(is.na(TRIALS)) ) {
  627. if (length(TRIALS)!=n)
  628. stop("TRIALS must have length n")
  629. B[-1,] <- B[-1,] + rep(ts*TRIALS,each=n_acc-1)
  630. }
  632. if ( gf > 0 ) # Setup for GO failure
  633. is.gf <- as.logical(rbinom(length(SSD),1,gf)) else
  634. is.gf <- logical(length(SSD))
  636. if ( all(!is.finite(SSD)) ) { # ALL GO
  637. out <- rWaldRace(n,v=v[-1,,drop=FALSE],B=B[-1,,drop=FALSE],A=A[-1,,drop=FALSE],t0=t0)
  638. out$R <- out$R+1
  639. } else { # SOME STOP
  640. if ( any(is.na(staircase)) ) { # STOP fixed SSD
  641. # Set v
  642. lint <- pmin(k*SSD[isSS],1)
  643. v[iT,isSS] <- v0 + (vT-v0)*lint
  644. v[iF,isSS] <- v0 + (vF-v0)*lint
  647. # add SSD to stop accumulator
  648. t0S[1,] <- t0S[1,] + SSD
  649. out <- rWaldexgRace(n,v=v,B=B,A=A,t0=t0S,a=minEXG)
  650. if ( tf>0 ) {
  651. is.tf <- logical(length(SSD))
  652. is.tf[is.finite(SSD)][as.logical(rbinom(sum(is.finite(SSD)),1,tf))] <- TRUE
  653. if ( any(is.tf) ) {
  654. out[is.tf,] <- rWaldRace(sum(is.tf),v=v[-1,is.tf,drop=FALSE],
  655. B=B[-1,is.tf,drop=FALSE],A=A[-1,is.tf,drop=FALSE],t0=0)
  656. out[is.tf,"R"] <- out[is.tf,"R"]+1
  657. }
  658. }
  659. } else { # STOP, staircase
  660. if ( !is.numeric(staircase) | length(staircase)!=1 )
  661. stop("Staircase must be a numeric vector of length 1 specifying the step.")
  662. SSDi <- SSD[is.finite(SSD)][1] # begining SSD
  663. dt <- v
  664. if ( any(!isSS) )
  665. dt[,!isSS] <- rWaldexgRace(n,v=v[,!isSS,drop=FALSE],
  666. B=B[,!isSS,drop=FALSE],A=A[,!isSS,drop=FALSE],t0=t0S[,!isSS,drop=FALSE],
  667. return.ttf=TRUE,a=minEXG)
  668. # Setup
  669. winner <- numeric(n)
  670. for ( i in c(1:n) ) {
  671. if ( !is.finite(SSD[i]) ) # not staircase
  672. dt[1,i] <- dt[1,i] + SSD[i] else { # staircase
  673. lint <- min(c(k*SSDi,1))
  674. v[iT,i] <- v0 + (vT-v0)*lint
  675. v[iF,i] <- v0 + (vF-v0)*lint
  676. dt[,i] <- rWaldexgRace(1,v=v[,i,drop=FALSE],B=B[,i,drop=FALSE],
  677. A=A[,i,drop=FALSE],t0=t0S[,i,drop=FALSE],return.ttf=TRUE,a=minEXG)
  678. dt[1,i] <- dt[1,i] + SSDi
  679. }
  680. if ( runif(1)<tf ) # Trigger failure
  681. winner[i] <- which.min(dt[2:n_acc,i])+1 else
  682. winner[i] <- which.min(dt[,i])
  683. if (is.gf[i]) winner[i] <- 1
  684. if ( is.finite(SSD[i]) ) { # update staircase
  685. SSD[i] <- SSDi
  686. if ( winner[i]==1 )
  687. SSDi <- SSDi + staircase else
  688. SSDi <- SSDi - staircase
  689. if (SSDi<1e-10) SSDi <- 0
  690. }
  691. }
  692. out <- data.frame(RT=dt[cbind(winner,1:n)],R=winner)
  693. }
  694. out$RT <- out$RT + t0 # Add t0 for go responses
  695. }
  697. # print(out)
  699. out[out$R==1,"RT"] <- NA
  700. if ( gf > 0 ) {
  701. out$RT[is.gf] <- NA
  702. out$R[is.gf] <- 1
  703. }
  704. if ( any(is.na(TRIALS)) ) cbind.data.frame(out,SSD=SSD) else
  705. cbind.data.frame(out,SSD=SSD,TRIALS=TRIALS)
  706. }
  709. n1PDF.Waldss <- function(rt,v,B,t0,A=0,tf=0,gf=0,ts=0,minEXG=-Inf,
  710. SSD=Inf,TRIALS=NA,Si)
  711. # Same as n1Wald except SSD is either a scalar or vector of length(rt)
  712. # stop accumulator must have name "NR". SSD is subtracted stop accumulator time
  713. # and dt=NA done by integration.
  714. #
  715. # tf= probabiliy of trigger failure, where
  716. # L = trigger fail & respond + trigger and respond + trigger and no-response
  717. # = tf*L(N-1)+(1-tf)[L(N)+p(S)],
  718. # L(N-1) = choice race likelihood (no stop accumulator),
  719. # L(N) = full N unit race likelihood given they did respond,
  720. # p(S) probability of stop winning
  721. #
  722. # gf = probabiliy of go failure.
  723. # On go trials: L = go fail (so no response) + go and L above
  724. # L = gf + (1-gf)*[tf*L(N-1)+(1-tf)[L(N)+p(S)]] or similarly
  725. # L = [ p(non-response) ] + [ p(response) ]
  726. # = [ gf + (1-gf)(1-tf)p(S) ] + [ (1-gf){(tf*Ln(n-1) + (1-tf)*L(N))} ]
  727. #
  728. # NB:rt is NOT decision time, but rather full RT as t0 has to be passed
  729. # in order to include properly in cases where RT is NA (i.e., sucessful stop)
  731. {
  733. stopfn <- function(t,vj,Bj,Aj,t0,SSD,Si,minEXG)
  734. {
  735. t0S <- rep(0,length(vj))
  736. t0S[-Si] <- t0S[-Si]+SSD
  737. t0S[Si] <- t0S[Si]+t0
  738. # t0S[-Si] <- t0S[-Si]-t0
  739. # t0S[Si] <- t0S[Si]-SSD
  740. dt <- matrix(rep(t,each=length(vj)),nrow=length(vj))+t0S
  741. i <- c(Si,c(1:length(vj))[-Si])
  742. n1Waldexg(dt[i,,drop=FALSE],v=vj[i],B=Bj[i],A=Aj[i],minEXG=minEXG)
  743. }
  745. # NOTE: t0 is not subtracted when making dt but passed to handle RT=NA case
  747. if ( length(SSD)==1 ) SSD <- rep(SSD,length(rt))
  748. if (length(SSD) != length(rt))
  749. stop("SSD must be a scalar or same length as rt")
  750. n_acc <- ifelse(is.null(dim(v)),length(v),dim(v)[1])
  752. rt <- matrix(rep(rt,each=n_acc),nrow=n_acc)
  753. is.stop <- is.na(rt[1,])
  755. if (!is.matrix(v)) v <- matrix(rep(v,dim(rt)[2]),nrow=n_acc)
  756. if (!is.matrix(B)) B <- matrix(rep(B,dim(rt)[2]),nrow=n_acc)
  757. if (!is.matrix(A)) A <- matrix(rep(A,dim(rt)[2]),nrow=n_acc)
  759. if ( any(is.na(TRIALS)) | ts == 0 ) {
  760. p <- SSD[is.stop]
  761. pj <- c(1:sum(is.stop))[!duplicated(p)] # index of unique SSD
  762. } else {
  763. B[-Si,] <- B[-Si,] + ts*TRIALS
  764. p <- apply(
  765. rbind(B[,is.stop,drop=FALSE],SSD[is.stop]),
  766. 2,paste,collapse="")
  767. pj <- c(1:sum(is.stop))[!duplicated(p)] # index of unique p and SSD
  768. }
  770. if ( any(!is.stop) )
  771. {
  772. rt[Si,!is.stop] <- rt[Si,!is.stop] - SSD[!is.stop]
  773. rt[-Si,!is.stop] <- rt[-Si,!is.stop]-t0
  774. if ( tf > 0 )
  775. {
  776. rt[1,!is.stop] <- (1-gf)*(
  777. tf*n1Wald(rt[-Si,!is.stop,drop=FALSE],v=v[-Si,!is.stop,drop=FALSE],
  778. A=A[-Si,!is.stop,drop=FALSE],B=B[-Si,!is.stop,drop=FALSE]) +
  779. (1-tf)*n1Waldexg(rt[,!is.stop,drop=FALSE],v=v[,!is.stop,drop=FALSE],
  780. minEXG=minEXG[1],
  781. B=B[,!is.stop,drop=FALSE],A=A[,!is.stop,drop=FALSE],Si=Si)
  782. )
  783. } else
  784. rt[1,!is.stop] <- (1-gf)*n1Waldexg(dt=rt[,!is.stop,drop=FALSE],
  785. minEXG=minEXG[1],
  786. v=v[,!is.stop,drop=FALSE],B=B[,!is.stop,drop=FALSE],
  787. A=A[,!is.stop,drop=FALSE],Si=Si)
  788. }
  790. if ( any(is.stop) ) for (j in pj) {
  791. if ( !is.finite(SSD[j]) ) tmp <- 0 else {
  792. if (minEXG==-Inf) lower <- -1 else lower <- pmin(minEXG[1]-t0,0)
  793. tmp <- my.integrate(f=stopfn,lower=minEXG[1]-t0,vj=v[,j],minEXG=minEXG[1],
  794. Bj=B[,j],Aj=A[,j],t0=t0,SSD=SSD[j],Si=Si)
  795. }
  796. rt[1,is.stop][p %in% p[j]] <- gf +(1-gf)*(1-tf)*tmp
  797. }
  798. rt[1,]
  799. }
  802. n1PDF.WaldssABCD <- function(rt,v,B,t0,v0,vT,vF,k=1,
  803. A=0,tf=0,gf=0,ts=0,minEXG=-Inf,SSD=Inf,TRIALS=NA,Si)
  805. # Same as n1PDF.Waldss adds change in v with SSD
  806. # Exponential from v0 to vT and vF
  808. {
  810. stopfn <- function(t,vj,Bj,Aj,t0,SSD,Si,minEXG)
  811. {
  812. t0S <- rep(0,length(vj))
  813. t0S[-Si] <- t0S[-Si]+SSD
  814. t0S[Si] <- t0S[Si]+t0
  815. # t0S[-Si] <- t0S[-Si]-t0
  816. # t0S[Si] <- t0S[Si]-SSD
  817. dt <- matrix(rep(t,each=length(vj)),nrow=length(vj))+t0S
  818. i <- c(Si,c(1:length(vj))[-Si])
  819. n1Waldexg(dt[i,,drop=FALSE],v=vj[i],B=Bj[i],A=Aj[i],minEXG=minEXG)
  820. }
  822. # NOTE: t0 is not subtracted when making dt but passed to handle RT=NA case
  824. if ( length(SSD)==1 ) SSD <- rep(SSD,length(rt))
  825. if (length(SSD) != length(rt))
  826. stop("SSD must be a scalar or same length as rt")
  827. n_acc <- ifelse(is.null(dim(v)),length(v),dim(v)[1])
  829. rt <- matrix(rep(rt,each=n_acc),nrow=n_acc)
  830. is.stop <- is.na(rt[1,])
  832. # Get index of true and false rates
  833. iT <- c(1:length(v))[v==Inf]
  834. iF <- c(1:length(v))[v==-Inf]
  835. # Get index of stop-signal trials
  836. isSS <- is.finite(SSD)
  838. if (!is.matrix(v)) v <- matrix(rep(v,dim(rt)[2]),nrow=n_acc)
  839. if (!is.matrix(B)) B <- matrix(rep(B,dim(rt)[2]),nrow=n_acc)
  840. if (!is.matrix(A)) A <- matrix(rep(A,dim(rt)[2]),nrow=n_acc)
  842. if ( any(!isSS) ) {
  843. v[iT,!isSS] <- vT
  844. v[iF,!isSS] <- vF
  845. }
  846. if ( any(isSS) ) {
  847. lint <- pmin(k*SSD[isSS],1)
  848. v[iT,isSS] <- v0 + (vT-v0)*lint
  849. v[iF,isSS] <- v0 + (vF-v0)*lint
  850. }
  852. if ( any(is.na(TRIALS)) | ts == 0 ) {
  853. p <- SSD[is.stop]
  854. pj <- c(1:sum(is.stop))[!duplicated(p)] # index of unique SSD
  855. } else {
  856. B[-Si,] <- B[-Si,] + ts*TRIALS
  857. p <- apply(
  858. rbind(B[,is.stop,drop=FALSE],SSD[is.stop]),
  859. 2,paste,collapse="")
  860. pj <- c(1:sum(is.stop))[!duplicated(p)] # index of unique p and SSD
  861. }
  863. if ( any(!is.stop) )
  864. {
  865. rt[Si,!is.stop] <- rt[Si,!is.stop] - SSD[!is.stop]
  866. rt[-Si,!is.stop] <- rt[-Si,!is.stop]-t0
  867. if ( tf > 0 )
  868. {
  869. rt[1,!is.stop] <- (1-gf)*(
  870. tf*n1Wald(rt[-Si,!is.stop,drop=FALSE],v=v[-Si,!is.stop,drop=FALSE],
  871. A=A[-Si,!is.stop,drop=FALSE],B=B[-Si,!is.stop,drop=FALSE]) +
  872. (1-tf)*n1Waldexg(rt[,!is.stop,drop=FALSE],v=v[,!is.stop,drop=FALSE],
  873. minEXG=minEXG[1],
  874. B=B[,!is.stop,drop=FALSE],A=A[,!is.stop,drop=FALSE],Si=Si)
  875. )
  876. } else
  877. rt[1,!is.stop] <- (1-gf)*n1Waldexg(dt=rt[,!is.stop,drop=FALSE],
  878. minEXG=minEXG[1],
  879. v=v[,!is.stop,drop=FALSE],B=B[,!is.stop,drop=FALSE],
  880. A=A[,!is.stop,drop=FALSE],Si=Si)
  881. }
  883. if ( any(is.stop) ) for (j in pj) {
  884. if ( !is.finite(SSD[j]) ) tmp <- 0 else {
  885. if (minEXG==-Inf) lower <- -1 else lower <- pmin(minEXG[1]-t0,0)
  886. tmp <- my.integrate(f=stopfn,lower=minEXG[1]-t0,vj=v[,j],minEXG=minEXG[1],
  887. Bj=B[,j],Aj=A[,j],t0=t0,SSD=SSD[j],Si=Si)
  888. }
  889. rt[1,is.stop][p %in% p[j]] <- gf +(1-gf)*(1-tf)*tmp
  890. }
  891. rt[1,]
  892. }
  895. # # VERY EXTENSIVE TESTING WITH Two different SSDs ----
  896. # plot.cell.density <- function(data.cell,C=NA,xlim=NA,limx=c(0,Inf),ymax=NA,lwd=2,
  897. # save.density=FALSE,digits=2,main="",show.mean=FALSE,
  898. # CorrectError.col=c("black","red"),CorrectError.lty=c(1,1))
  899. # # If !is.na(C) plots density for correct and error responses for a data frame
  900. # # with columns R (a factor) and RT, adding a boolean score column for R=C.
  901. # # Otherwise plots each response. Can deal with NA in the RT column, in which
  902. # # case it provides a summary of p(NA)
  903. # {
  904. # if (!is.factor(data.cell$R)) data.cell$R <- factor(data.cell$R)
  905. # if (length(C)==1) C <- rep(C,dim(data.cell)[1])
  906. # p.na <- mean(is.na(data.cell$RT))
  907. # is.in <- !is.na(data.cell$RT)
  908. # is.in[is.in] <- data.cell$RT[is.in]>limx[1] & data.cell$RT[is.in]<limx[2]
  909. # dat <- data.cell[is.in,]
  910. # if ( !any(is.na(C)) ) {
  911. # if ( is.logical(C) & length(C)==dim(data.cell)[1] )
  912. # dat$C <- C[is.in] else dat$C <- dat$R==C[is.in]
  913. # if (length(dat$RT[dat$C])>2)
  914. # dns.correct <- density(dat$RT[dat$C]) else dns.correct <- NULL
  915. # if (length(dat$RT[!dat$C])>2)
  916. # dns.error <- density(dat$RT[!dat$C]) else dns.error <- NULL
  917. # if (is.null(dns.error) & is.null(dns.correct))
  918. # stop("There are no densities to plot")
  919. # acc <- mean(dat$C)
  920. # if (!is.null(dns.correct))
  921. # dns.correct$y <- dns.correct$y*acc*(1-p.na)
  922. # if (!is.null(dns.error))
  923. # dns.error$y <- dns.error$y*(1-acc)*(1-p.na)
  924. # if (is.na(ymax)) ymax <- max(c(dns.correct$y,dns.error$y))
  925. # if (!is.null(dns.correct)) {
  926. # if (any(is.na(xlim))) plot(dns.correct,xlab="RT",ylab="density",ylim=c(0,ymax),main=main,
  927. # col=CorrectError.col[1],lty=CorrectError.lty[1],lwd=lwd) else
  928. # plot(dns.correct,xlab="RT",ylab="density",ylim=c(0,ymax),xlim=xlim,main=main,
  929. # col=CorrectError.col[1],lty=CorrectError.lty[1],lwd=lwd)
  930. # if (!is.null(dns.error)) lines(dns.error,col=CorrectError.col[2],
  931. # lty=CorrectError.lty[2],lwd=lwd)
  932. # } else {
  933. # if (any(is.na(xlim))) plot(dns.error,xlab="RT",ylab="density",ylim=c(0,ymax),main=main,
  934. # col=CorrectError.col[2],lty=CorrectError.lty[2],lwd=lwd) else
  935. # plot(dns.error,xlab="RT",ylab="density",ylim=c(0,ymax),main=main,xlim=xlim,
  936. # col=CorrectError.col[2],lty=CorrectError.lty[2],lwd=lwd)
  937. # if (!is.null(dns.correct)) lines(dns.correct,col=CorrectError.col[1],
  938. # lty=CorrectError.lty[2],lwd=lwd)
  939. # }
  940. # nams <- "Accuracy ="
  941. # ps <- round(acc,digits)
  942. # if (p.na!=0) {
  943. # nams <- c(nams,"p(NA) =")
  944. # ps <- c(ps,round(p.na,digits))
  945. # }
  946. # legend("topright",paste(nams,ps),bty="n")
  947. # legend("topright",xjust=0, inset=c(0,0.1), c("correct","error"),bty="n",
  948. # lty=CorrectError.lty, col=CorrectError.col,lwd=lwd)
  949. # if ( save.density ) list(correct=dns.correct,error=dns.error)
  950. # } else {
  951. # rs <- levels(dat$R)
  952. # dns <- vector(mode="list",length=length(rs))
  953. # names(dns) <- rs
  954. # ps <- table(dat$R)/dim(dat)[1]
  955. # for (i in rs) {
  956. # rt <- dat$RT[dat$R==i]
  957. # if (length(rt)>2) {
  958. # dns[[i]] <- density(rt)
  959. # dns[[i]]$y <- ps[i]*dns[[i]]$y*(1-p.na)
  960. # }
  961. # }
  962. # ymax <- suppressWarnings(max(unlist(lapply(dns,function(x){max(x$y)}))))
  963. # no.dns <- unlist(lapply(dns,is.null))
  964. # if (all(no.dns))
  965. # stop("There are no densities to plot!")
  966. # dns1 <- dns[!no.dns]
  967. # ltys <- c(1:length(dns1))
  968. # plot(dns1[[1]],xlab="RT",ylab="density",ylim=c(0,ymax),lty=ltys[1],
  969. # main=main)
  970. # if (length(dns1)>1) for (i in 2:length(dns1)) lines(dns1[[i]],lty=ltys[i])
  971. # nams <- paste("p(",names(dns1),") =",sep="")
  972. # ps <- round(ps[!no.dns]*(1-p.na),digits)
  973. # if ( p.na!=0 ) {
  974. # nams <- c(nams,"p(NA) =")
  975. # ps <- c(ps,round(p.na,digits))
  976. # lty <- c(ltys,NA)
  977. # }
  978. # legend("topright",paste(nams,ps),lty=ltys,bty="n",lwd=lwd)
  979. # if ( save.density ) dns
  980. # }
  981. # }
  982. #
  983. #
  984. # ########### TWO ACCUMULATOR CASE ----
  985. # {
  986. # n=1e4
  987. # # n=10
  988. # v=c(.5,1); B=c(1,1); A=c(1,1); minEXG=0
  989. # SSD = rep(c(1,10)/10,each=n/2)
  990. #
  991. # # Run one of the follwing two lines
  992. # do.trials=FALSE
  993. # # do.trials = TRUE # requires very differnet plotting check, can be SLOW!
  994. #
  995. # t0=.2
  996. #
  997. # ### RUN ONE OF THE FOLLOWING FOUR LINES
  998. # # Without trigger failure or go failure
  999. # tf=0; gf=0
  1000. # # With trigger failure, no go failure
  1001. # tf=.1;gf=0
  1002. # # Without trigger failure, with go failure
  1003. # tf=0; gf=.1
  1004. # # With trigger failure and go failure
  1005. # tf=.1;gf=.1
  1006. #
  1007. # if (do.trials) {
  1008. # ts=.5; TRIALS=log10(seq(1,10,length.out=n)) # 1..10 natural so 0-1 on log
  1009. # TRIALS <- as.vector(t(matrix(TRIALS,nrow=2))) # interleave SSDs
  1010. # # Plot slowing in GO (usually nice and linear, up to smooting overfitting)
  1011. # sim.go <- rWaldss(n=n,v=v,B=B,A=A,t0=t0,tf=tf,gf=gf,SSD=SSD,TRIALS=TRIALS,ts=ts,minEXG=minEXG)
  1012. # is.in <- !is.na(sim.go$RT) # in case go failure
  1013. # plot(smooth.spline(TRIALS[is.in],sim.go$RT[is.in]),ylab="Smooth",xlab="TRIALS",type="l")
  1014. # } else {TRIALS=NA;ts=0}
  1015. #
  1016. # # Simulate stop trials
  1017. # sim <- rWaldss(n=n,v=v,B=B,A=A,t0=t0,tf=tf,gf=gf,SSD=SSD,TRIALS=TRIALS,ts=ts,minEXG=minEXG)
  1018. # # Sucessful inhibition
  1019. # tapply(is.na(sim$RT),sim$SSD,mean)
  1020. #
  1021. #
  1022. # # Plot densities
  1023. # par(mfrow=c(1,2))
  1024. # dns1 <- plot.cell.density(sim[sim$SSD==.1,],limx=c(0,7),save.density=TRUE,main="SSD=.1")
  1025. # dns2 <- plot.cell.density(sim[sim$SSD!=.1,],limx=c(0,7),save.density=TRUE,main="SSD=1")
  1026. # x1c <- dns1$'2'$x; x2c <- dns2$'2'$x
  1027. #
  1028. # # Signal respond RT
  1029. # dat <- sim; dat <- dat[!is.na(dat$RT),]; dat$R <- factor(as.character(dat$R))
  1030. # round(tapply(dat$RT,dat[,c("R","SSD")],mean),2)
  1031. #
  1032. # if (do.trials) {
  1033. # tmp <- n1PDF.Waldss(sim$RT[!is.na(sim$RT)],v=v[2:1],B=B[2:1],A=A[2:1],t0=t0,minEXG=minEXG,
  1034. # ts=ts,TRIALS=TRIALS[!is.na(sim$RT)],SSD=SSD[!is.na(sim$RT)],Si=2,tf=tf,gf=gf)
  1035. # par(mfrow=c(1,3))
  1036. # # red=black?
  1037. # plot(x1c,dns1$'2'$y,pch=".",main="SSD=.1",ylab="Density",xlab="RT")
  1038. # lines(smooth.spline(sim$RT[!is.na(sim$RT) & SSD==.1],
  1039. # tmp[c(SSD==.1)[!is.na(sim$RT)]]),col="red")
  1040. # # red=black?
  1041. # plot(x2c,dns2$'2'$y,pch=".",main="SSD=1",ylab="Density",xlab="RT")
  1042. # lines(smooth.spline(sim$RT[!is.na(sim$RT) & SSD==1],
  1043. # tmp[c(SSD==1)[!is.na(sim$RT)]]),col="red")
  1044. # # print(tapply(is.na(sim$RT),sim$SSD,mean)) # empirical
  1045. # tmp <- n1PDF.Waldss(rep(NA,n),v=v,B=B,A=A,t0=t0,
  1046. # SSD=SSD,Si=1,tf=tf,gf=gf,ts=ts,TRIALS=TRIALS)
  1047. # # print(mean(tmp[SSD==.1]))
  1048. # # print(mean(tmp[SSD==1]))
  1049. # plot(TRIALS,tmp,pch=".",xlab="TRIALS",ylab="p(NA)",ylim=c(0,1))
  1050. # lines(smooth.spline(TRIALS[SSD==.1],as.numeric(is.na(sim$RT)[SSD==.1])),col="red")
  1051. # lines(smooth.spline(TRIALS[SSD==1],as.numeric(is.na(sim$RT)[SSD==1])),col="red")
  1052. # } else {
  1053. # # Save simulated densities
  1054. # r1 <- c(2,1)
  1055. # d.r1 <- n1PDF.Waldss(rt=c(x1c,x2c),v=v[r1],B=B[r1],A=A[r1],t0=t0,minEXG=minEXG,
  1056. # SSD=c(rep(.1,length(x1c)),rep(1,length(x2c))),Si=2,tf=tf,gf=gf)
  1057. # # Plot simulated (black) and theoretical (red) densities
  1058. # par(mfrow=c(1,2))
  1059. # # red=black?
  1060. # plot(x1c,dns1$'2'$y,type="l",main="SSD=.1",ylab="Density",xlab="RT",
  1061. # ylim=c(0,max(dns1$'2'$y)))
  1062. # lines(x1c,d.r1[1:length(x1c)],col="red")
  1063. # # red=black?
  1064. # plot(x2c,dns2$'2'$y,type="l",main="SSD=1",ylab="Density",xlab="RT",
  1065. # ylim=c(0,max(dns2$'2'$y)))
  1066. # lines(x2c,d.r1[(length(x2c)+1):(2*length(x2c))],col="red")
  1067. #
  1068. # # # p(Stop check)
  1069. # # print(tapply(is.na(sim$RT),sim$SSD,mean)) # empirical
  1070. # # print(n1PDF.Waldss(NA,v=v,A=A,B=B,t0=t0,SSD=.1,Si=1,tf=tf,gf=gf,minEXG=minEXG))
  1071. # # print(n1PDF.Waldss(NA,v=v,A=A,B=B,t0=t0,SSD=1,Si=1,tf=tf,gf=gf,minEXG=minEXG))
  1072. # }
  1073. #
  1074. # }
  1075. #
  1076. # ########### THREE ACCUMULATOR CASE ----
  1077. # {
  1078. # n=1e5
  1079. # SSDs = c(1,10)/10
  1080. # v=c(1,.75,.25); B=c(1,1,1); A=c(1,1,1); minEXG=0
  1081. # SSD = rep(SSDs,each=n/2)
  1082. #
  1083. # do.trials=FALSE
  1084. # do.trials = TRUE # requires very differnet plotting check, can be SLOW!
  1085. #
  1086. # t0=.2
  1087. # ### RUN ONE OF THE FOLLOWING FOUR LINES
  1088. # # Without trigger failure or go failure
  1089. # tf=0; gf=0
  1090. # # With trigger failure, no go failure
  1091. # tf=.1;gf=0
  1092. # # Without trigger failure, with go failure
  1093. # tf=0; gf=.1
  1094. # # With trigger failure and go failure
  1095. # tf=.1;gf=.1
  1096. #
  1097. # if (do.trials) {
  1098. # ts=.5; TRIALS=log10(seq(1,10,length.out=n)) # 1..10 natural so 0-1 on log
  1099. # TRIALS <- as.vector(t(matrix(TRIALS,nrow=2))) # interleave SSDs
  1100. # # Plot slowing in GO (usually nice and linear, up to smooting overfitting)
  1101. # sim.go <- rWaldss(n=n,v=v,B=B,A=A,t0=t0,tf=tf,gf=gf,TRIALS=TRIALS,ts=ts,minEXG=minEXG)
  1102. # is.in <- !is.na(sim.go$RT) # in case go failure
  1103. # plot(smooth.spline(TRIALS[is.in],sim.go$RT[is.in]),ylab="Smooth",xlab="TRIALS",type="l")
  1104. # } else {TRIALS=NA;ts=0}
  1105. #
  1106. # # Simulate stop trials
  1107. # par(mfrow=c(1,3)); xlim=c(0,1)
  1108. # simgo <- rWaldss(n=n,v=v,B=B,A=A,t0=t0,tf=tf,gf=gf,TRIALS=TRIALS,ts=ts,SSD=Inf,minEXG=minEXG)
  1109. # plot.cell.density(simgo,main="Go",xlim=xlim)
  1110. # sim <- rWaldss(n=n,v=v,B=B,A=A,t0=t0,tf=tf,gf=gf,TRIALS=TRIALS,ts=ts,SSD=SSD,minEXG=minEXG)
  1111. # # Sucessful inhibition
  1112. # tapply(is.na(sim$RT),sim$SSD,mean)
  1113. # dns1 <- plot.cell.density(sim[sim$SSD==SSDs[1],],limx=c(0,7),save.density=TRUE,main="SSD=.1",xlim=xlim)
  1114. # dns2 <- plot.cell.density(sim[sim$SSD!=SSDs[1],],limx=c(0,7),save.density=TRUE,main="SSD=1",xlim=xlim)
  1115. # x1c <- dns1$'2'$x; x2c <- dns2$'2'$x
  1116. # x1e <- dns1$'3'$x; x2e <- dns2$'3'$x
  1117. #
  1118. # # Signal respond RT
  1119. # dat <- sim; dat <- dat[!is.na(dat$RT),]; dat$R <- factor(as.character(dat$R))
  1120. # round(tapply(dat$RT,dat[,c("R","SSD")],mean),2)
  1121. #
  1122. # if (do.trials) {
  1123. # r1 <- c(2,1,3)
  1124. # is.in1 <- !is.na(sim$RT) & sim$R==2
  1125. # d.r1 <- n1PDF.Waldss(sim$RT[is.in1],v=v[r1],ts=ts,TRIALS=TRIALS[is.in1],
  1126. # B=B[r1],A=A[r1],t0=t0,SSD=SSD[is.in1],Si=2,tf=tf,gf=gf,minEXG=minEXG)
  1127. # r2 <- c(3,1,2)
  1128. # is.in2 <- !is.na(sim$RT) & sim$R==3
  1129. # d.r2 <- n1PDF.Waldss(sim$RT[is.in2],v=v[r2],ts=ts,TRIALS=TRIALS[is.in2],
  1130. # B=B[r2],A=A[r2],t0,SSD=SSD[is.in2],Si=2,tf=tf,gf=gf,minEXG=minEXG)
  1131. # par(mfrow=c(1,3))
  1132. # # red=black?
  1133. # plot(x1c,dns1$'2'$y,pch=".",main="SSD=.1",ylab="Density",xlab="RT",type="l")
  1134. # lines(x1e,dns1$'3'$y,lty=2)
  1135. # lines(smooth.spline(sim$RT[is.in1 & sim$SSD==.1],
  1136. # d.r1[c(sim$SSD==.1)[is.in1]]),col="red")
  1137. # lines(smooth.spline(sim$RT[is.in2 & sim$SSD==.1],d.r2[c(sim$SSD==.1)[is.in2]]),
  1138. # lty=2,col="red")
  1139. # # red=black?
  1140. # plot(x2c,dns2$'2'$y,pch=".",main="SSD=1",ylab="Density",xlab="RT",type="l")
  1141. # lines(x2e,dns2$'3'$y,lty=2)
  1142. # lines(smooth.spline(sim$RT[is.in1 & sim$SSD==1],
  1143. # d.r1[c(sim$SSD==1)[is.in1]]),col="red")
  1144. # lines(smooth.spline(sim$RT[is.in2 & sim$SSD==1],
  1145. # d.r2[c(sim$SSD==1)[is.in2]]),col="red",lty=2)
  1146. #
  1147. # # print(tapply(is.na(sim$RT),sim$SSD,mean)) # empirical
  1148. # tmp <- n1PDF.Waldss(rep(NA,n),v=v,A=A,B=B,t0=t0,SSD=SSD,Si=1,tf=tf,gf=gf,ts=ts,
  1149. # TRIALS=TRIALS,minEXG=minEXG)
  1150. # # print(mean(tmp[SSD==.1]))
  1151. # # print(mean(tmp[SSD==1]))
  1152. # plot(TRIALS,tmp,pch=".",xlab="TRIALS",ylab="p(NA)",ylim=c(0,1))
  1153. # lines(smooth.spline(TRIALS[SSD==.1],as.numeric(is.na(sim$RT)[SSD==.1])),col="red")
  1154. # lines(smooth.spline(TRIALS[SSD==1],as.numeric(is.na(sim$RT)[SSD==1])),col="red")
  1155. # } else {
  1156. # # Save simulated densities
  1157. # r1 <- c(2,1,3)
  1158. # d.r1 <- n1PDF.Waldss(rt=c(x1c,x2c),v=v[r1],B=B[r1],A=A[r1],t0=t0,minEXG=minEXG,
  1159. # SSD=c(rep(.1,length(x1c)),rep(1,length(x2c))),Si=2,tf=tf,gf=gf)
  1160. # r2 <- c(3,1,2)
  1161. # d.r2 <- n1PDF.Waldss(rt=c(x1e,x2e),v=v[r2],B=B[r2],A=A[r2],t0=t0,minEXG=minEXG,
  1162. # SSD=c(rep(.1,length(x1e)),rep(1,length(x2e))),Si=2,tf=tf,gf=gf)
  1163. # # Plot simulated (black) and theoretical (red) densities
  1164. # par(mfrow=c(1,2))
  1165. # # red=black?
  1166. # plot(x1c,dns1$'2'$y,type="l",main="SSD=.1",ylab="Density",xlab="RT",
  1167. # ylim=c(0,max(dns1$'2'$y)))
  1168. # lines(x1c,d.r1[1:length(x1c)],col="red")
  1169. # lines(x1e,dns1$'3'$y,lty=2)
  1170. # lines(x1e,d.r2[1:length(x1e)],col="red",lty=2)
  1171. # # red=black?
  1172. # plot(x2c,dns2$'2'$y,type="l",main="SSD=1",ylab="Density",xlab="RT",
  1173. # ylim=c(0,max(dns2$'2'$y)))
  1174. # lines(x2c,d.r1[(length(x2c)+1):(2*length(x2c))],col="red")
  1175. # lines(x2e,dns2$'3'$y,lty=2)
  1176. # lines(x2e,d.r2[(length(x2e)+1):(2*length(x2e))],col="red",lty=2)
  1177. #
  1178. # # # p(Stop check)
  1179. # # print(tapply(is.na(sim$RT),sim$SSD,mean)) # empirical
  1180. # # print(n1PDF.Waldss(NA,v=v,B=B,A=A,t0=t0,SSD=.1,Si=1,tf=tf,gf=gf))
  1181. # # print(n1PDF.Waldss(NA,v=v,B=B,A=A,t0=t0,SSD=1,Si=1,tf=tf,gf=gf))
  1182. # }
  1183. # }

dists.R, no license · at the source

Overview

Authors: Yihe Weng1, Rory Boyle1, Chi Tak Lee1, Declan Quinn1, Clodagh Earley1, Maike Splittgerber1, Lili Zhang2, Luisa Franzen1,3, Tobias Banaschewski4,5, Arun L. W. Bokde6, Sylvane Desrivières7, Herta Flor8,9, Antoine Grigis10, Hugh Garavan11, Penny Gowland12, Andreas Heinz13,14,15, Rüdiger Brühl16, Jean-Luc Martinot17,18, Marie-Laure Paillère Martinot17,19, Eric Artiges17,18
and 13 other authorsJane McGrath6, Frauke Nees20, Dimitri Papadopoulos Orfanos9, Luise Poustka21, Nathalie Holz4,5, Sarah Hohmann4,5,22, Michael N. Smolka15, Nilakshi Vaidya23, Gunter Schumann23,24,25, Henrik Walter26, Alexander Weigard27,28, Robert Whelan1,28, IMAGEN Consortium
ORCID iDs: Yihe Weng, Tobias Banaschewski, Arun L. W. Bokde, Sylvane Desrivières, Herta Flor, Hugh Garavan, Penny Gowland, Rüdiger Brühl, Jean-Luc Martinot, Eric Artiges, Frauke Nees, Dimitri Papadopoulos Orfanos, Nathalie Holz, Michael N. Smolka, Gunter Schumann, Henrik Walter, Alexander Weigard, Robert Whelan
28 affiliations
  1. School of Psychology and Global Brain Health Institute, Trinity College Dublin, Dublin, Ireland
  2. School of Computing, Dublin City University, Dublin, Ireland
  3. Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands
  4. Department of Child and Adolescent Psychiatry and Psychotherapy, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, Square J5, 68159 Mannheim, Germany
  5. German Center for Mental Health (DZPG), Partner Site Mannheim-Heidelberg-Ulm, Heidelberg, Germany
  6. Discipline of Psychiatry, School of Medicine and Trinity College Institute of Neuroscience, Trinity College Dublin, Dublin, Ireland
  7. Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology & Neuroscience, King’s College London, London, UK
  8. Institute of Cognitive and Clinical Neuroscience, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, Square J5, Mannheim, Germany
  9. Department of Psychology, School of Social Sciences, University of Mannheim, 68131 Mannheim, Germany
  10. NeuroSpin, CEA, Université Paris-Saclay, F-91191 Gif-sur-Yvette, France
  11. Departments of Psychiatry and Psychology, University of Vermont, 05405 Burlington, Vermont, USA
  12. Sir Peter Mansfield Imaging Centre School of Physics and Astronomy, University of Nottingham, University Park, Nottingham, UK
  13. Department of Psychiatry and Psychotherapy, University of Tübingen, Tübingen, Germany
  14. German Center for Mental Health (DZPG), Site Tübingen, Germany
  15. Department of Psychiatry and Psychotherapy, Technische Universität Dresden, Dresden, Germany
  16. Physikalisch-Technische Bundesanstalt (PTB), Braunschweig and Berlin, Berlin, Germany
  17. INSERM U1299 “Developmental Trajectories in Psychiatry & Addictology”; University Paris-Saclay, Ecole Normale Supérieure Paris-Saclay, Mathematics department, Centre Borelli CNRS UMR9010, Gif-sur-Yvette, France
  18. Department of Psychiatry, LabD-Psy, Barthélémy Durand Hospital, Etampes, France
  19. AP-HP. Sorbonne Université, Department of Child and Adolescent Psychiatry, Pitié-Salpêtrière Hospital, Paris, France
  20. Institute of Medical Psychology, Ludwig-Maximilians-Universität (LMU) in Munich, Munich, Germany
  21. Department of Child and Adolescent Psychiatry, Center for Psychosocial Medicine, University Hospital Heidelberg, Heidelberg, Germany
  22. Department of Child and Adolescent Psychiatry, Psychotherapy and Psychosomatics, University Medical Centre Hamburg-Eppendorf, Hamburg, Germany
  23. Centre for Population Neuroscience and Stratified Medicine (PONS), Department of Psychiatry and Psychotherapy, Charité Universitätsmedizin Berlin, Berlin, Germany
  24. Centre for Population Neuroscience and Stratified Medicine (PONS), Institute for Science and Technology of Brain-Inspired Intelligence and National Centre for Neurological Disorders, Huashan Hospital, Fudan University, Shanghai, China
  25. German Center for Mental Health (DZPG), Site Berlin-, Potsdam, Germany
  26. Department of Psychiatry and Psychotherapy CCM, Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Berlin, Germany
  27. Department of Psychiatry and Addiction Center, University of Michigan, 4250 Plymouth Rd, Ann Arbor, MI 48109, USA
  28. These authors contributed equally: Alexander Weigard, Robert Whelan
Institutions: Trinity College Dublin (Ireland); Dublin City University (Ireland); Maastricht University (Netherlands); Heidelberg University (Germany); Central Institute of Mental Health (Germany); Medizinische Fakultät Mannheim; Deutsches Zentrum für Psychische Gesundheit (Germany); King's College London (United Kingdom); University of Mannheim (Germany); Commissariat à l'Énergie Atomique et aux Énergies Alternatives (France); Université Paris-Saclay (France); CEA Paris-Saclay - Etablissement de Saclay (France); CEA Paris-Saclay (France); University of Vermont (United States); University of Nottingham (United Kingdom); Technische Universität Dresden (Germany); University of Tübingen (Germany); Physikalisch-Technische Bundesanstalt (Germany); École Normale Supérieure Paris-Saclay (France); Centre National de la Recherche Scientifique (France); Inserm (France); Trajectoires développementales & psychiatrie; Centre Borelli (France); Sorbonne Université (France); Assistance Publique – Hôpitaux de Paris (France); Pitié-Salpêtrière Hospital (France); Institut für Medizinische Psychologie (Germany); Ludwig-Maximilians-Universität München (Germany); University Hospital Heidelberg (Germany); University Medical Center Hamburg-Eppendorf (Germany); Charité - Universitätsmedizin Berlin (Germany); Fudan University (China); Huashan Hospital (China); Humboldt-Universität zu Berlin (Germany); University of Michigan (United States)
Journal: Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology, pages 10.1038/s41386-026-02401-6
Dates: published online 22 April 2026; in print April 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1038/s41386-026-02401-6 · PMID 42020764 · PMCID PMC13592875 · OpenAlex W7155221093
Open access: hybrid, a free copy (OpenAlex)
Status: code verified
Categories: other condition (population)
Methods: Statistics, Machine learning, Connectivity, fMRI & imaging
Topic: Functional Brain Connectivity Studies (Cognitive Neuroscience, Neuroscience), according to OpenAlex
Funding: Agence Nationale de la Recherche (French National Research Agency) (AAPG2019 - GeBra, ANR-12-SAMA-0004); NIDA NIH HHS (K23 DA051561, R01 DA049238); NIA NIH HHS (R56 AG058854); NIMH NIH HHS (R01 MH085772); NIBIB NIH HHS (U54 EB020403); Science Foundation Ireland (SFI) (16/ERCD/3797)
Citations: cited by 1 paper (Europe PMC); 71 references in the paper

Abstract

Poor inhibitory control and decision-making are often considered as risks for substance use and other adverse psychiatric outcomes. The Stop-Signal Task (SST) is a widely used protocol, from which inhibitory control is indexed by stop signal reaction time (SSRT). However, heretofore models of SSRT may be too simplistic to capture complex processes underlying task performance. In contrast, the Racing Diffusion Ex-Gaussian ABCD (RDEX-ABCD) model provides a more mechanistic framework, capturing both inhibitory control and task-general decision-making processes during the SST. Here, we applied the RDEX-ABCD model to SST data from the IMAGEN cohort (n > 1000) at ages 19 and 23, and examined model parameters in relation to substance use via Elastic Net regression. Connectome-based predictive modeling was then performed to identify brain networks predicting parameters, and the association between these networks and substance use was examined. We found that parameters indexing inhibitory control had no associations with substance use and were only weakly associated with brain connectivity. In contrast, parameters reflecting general decision-making processes – such as efficiency of evidence accumulation, decision threshold (response caution), probability of go failure – and their associated brain activity were significant predictors of cannabis and cigarette use. These findings suggested that efficiency of evidence accumulation, a neurocognitive mechanism that facilitates adaptive decision making across many contexts, emerged as a robust predictor of substance use vulnerability. Overall, general decision-making mechanisms may act as more reliable indicators of vulnerability to substance use than the conventional inhibitory control measures.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repositories

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

OSF 2h8a7

License: none: the authors keep all their rights
State: the link answers, verified on 29 September 2026
Evidence: files inventoried
Languages: R (27), Jupyter (1)
Size: 38 files, 28 scripts
Software Heritage: not checked
Found in: the text, “RDEX-ABCD model evaluation and estimation”
Holds: README, 1 notebook
Not found: license file, CITATION.cff, environment file, tests, continuous integration, documentation
Tools: tidyverse (2 files), ggplot2 (1 file), NumPy (1 file), pandas (1 file)
Availability: 1 check, the latest on 29 September 2026: the link answers (HTTP 200)
  • 29 September 2026: the link answers (HTTP 200)
28 files
At the source: osf.io/2h8a7/

OSF 6ejpd

License: none: the authors keep all their rights
State: the link answers, verified on 29 September 2026
Evidence: files inventoried
Size: 0 files, 0 scripts
Software Heritage: not checked
Found in: “DATA AVAILABILITY”
Not found: README, license file, CITATION.cff, environment file, tests, continuous integration, documentation
Availability: 1 check, the latest on 29 September 2026: the link answers (HTTP 200)
  • 29 September 2026: the link answers (HTTP 200)
At the source: osf.io/6ejpd/

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 2 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 28 scripts, each with its path and the digest of its content;
  • 3 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data availability

IMAGEN data are available from a dedicated database: https://imagen2.cea.fr. Due to participant consent restrictions, IMAGEN data cannot be made fully open access. Code for CPM analysis is available at https://osf.io/6ejpd/.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 29 September 2026: the first record

Recorded: type, language, journal, pages, dates, 33 authors, 6 funders, 62 references.

Cite

This paper

Weng, Y., Boyle, R., Lee, C. T., Quinn, D., Earley, C., Splittgerber, M., Zhang, L., Franzen, L., Banaschewski, T., Bokde, A. L. W., Desrivières, S., Flor, H., Grigis, A., Garavan, H., Gowland, P., Heinz, A., Brühl, R., Martinot, J.-L., Martinot, M.-L. P., . . . IMAGEN Consortium. (2026). Model-based analysis of stop-signal data reveals robust neural and clinical correlates of evidence accumulation but not inhibition. Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology, 10.1038/s41386-026-02401-6. https://doi.org/10.1038/s41386-026-02401-6

BibTeX

@article{weng2026model,
author = {Weng, Yihe and Boyle, Rory and Lee, Chi Tak and Quinn, Declan and Earley, Clodagh and Splittgerber, Maike and Zhang, Lili and Franzen, Luisa and Banaschewski, Tobias and Bokde, Arun L. W. and Desrivières, Sylvane and Flor, Herta and Grigis, Antoine and Garavan, Hugh and Gowland, Penny and Heinz, Andreas and Brühl, Rüdiger and Martinot, Jean-Luc and Martinot, Marie-Laure Paillère and Artiges, Eric and McGrath, Jane and Nees, Frauke and Orfanos, Dimitri Papadopoulos and Poustka, Luise and Holz, Nathalie and Hohmann, Sarah and Smolka, Michael N. and Vaidya, Nilakshi and Schumann, Gunter and Walter, Henrik and Weigard, Alexander and Whelan, Robert and {IMAGEN Consortium}},
title = {{Model-based analysis of stop-signal data reveals robust neural and clinical correlates of evidence accumulation but not inhibition}},
journal = {Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology},
year = {2026},
month = apr,
pages = {10.1038/s41386--026--02401--6},
publisher = {Nature Publishing Group},
issn = {0893-133X},
doi = {10.1038/s41386-026-02401-6},
url = {https://doi.org/10.1038/s41386-026-02401-6},
pmid = {42020764},
pmcid = {PMC13592875}
}

RIS

TY - JOUR
AU - Weng, Yihe
AU - Boyle, Rory
AU - Lee, Chi Tak
AU - Quinn, Declan
AU - Earley, Clodagh
AU - Splittgerber, Maike
AU - Zhang, Lili
AU - Franzen, Luisa
AU - Banaschewski, Tobias
AU - Bokde, Arun L. W.
AU - Desrivières, Sylvane
AU - Flor, Herta
AU - Grigis, Antoine
AU - Garavan, Hugh
AU - Gowland, Penny
AU - Heinz, Andreas
AU - Brühl, Rüdiger
AU - Martinot, Jean-Luc
AU - Martinot, Marie-Laure Paillère
AU - Artiges, Eric
AU - McGrath, Jane
AU - Nees, Frauke
AU - Orfanos, Dimitri Papadopoulos
AU - Poustka, Luise
AU - Holz, Nathalie
AU - Hohmann, Sarah
AU - Smolka, Michael N.
AU - Vaidya, Nilakshi
AU - Schumann, Gunter
AU - Walter, Henrik
AU - Weigard, Alexander
AU - Whelan, Robert
AU - IMAGEN Consortium
TI - Model-based analysis of stop-signal data reveals robust neural and clinical correlates of evidence accumulation but not inhibition
T2 - Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology
J2 - Neuropsychopharmacology
PY - 2026
DA - 2026/04/22
SP - 10.1038/s41386
EP - 026-02401-6
SN - 0893-133X
PB - Nature Publishing Group
DO - 10.1038/s41386-026-02401-6
UR - https://doi.org/10.1038/s41386-026-02401-6
LA - en
ER -

CSL-JSON

{
"id": "10.1038/s41386-026-02401-6",
"type": "article-journal",
"title": "Model-based analysis of stop-signal data reveals robust neural and clinical correlates of evidence accumulation but not inhibition",
"container-title": "Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology",
"author": [
{
"family": "Weng",
"given": "Yihe"
},
{
"family": "Boyle",
"given": "Rory"
},
{
"family": "Lee",
"given": "Chi Tak"
},
{
"family": "Quinn",
"given": "Declan"
},
{
"family": "Earley",
"given": "Clodagh"
},
{
"family": "Splittgerber",
"given": "Maike"
},
{
"family": "Zhang",
"given": "Lili"
},
{
"family": "Franzen",
"given": "Luisa"
},
{
"family": "Banaschewski",
"given": "Tobias"
},
{
"family": "Bokde",
"given": "Arun L. W."
},
{
"family": "Desrivières",
"given": "Sylvane"
},
{
"family": "Flor",
"given": "Herta"
},
{
"family": "Grigis",
"given": "Antoine"
},
{
"family": "Garavan",
"given": "Hugh"
},
{
"family": "Gowland",
"given": "Penny"
},
{
"family": "Heinz",
"given": "Andreas"
},
{
"family": "Brühl",
"given": "Rüdiger"
},
{
"family": "Martinot",
"given": "Jean-Luc"
},
{
"family": "Martinot",
"given": "Marie-Laure Paillère"
},
{
"family": "Artiges",
"given": "Eric"
},
{
"family": "McGrath",
"given": "Jane"
},
{
"family": "Nees",
"given": "Frauke"
},
{
"family": "Orfanos",
"given": "Dimitri Papadopoulos"
},
{
"family": "Poustka",
"given": "Luise"
},
{
"family": "Holz",
"given": "Nathalie"
},
{
"family": "Hohmann",
"given": "Sarah"
},
{
"family": "Smolka",
"given": "Michael N."
},
{
"family": "Vaidya",
"given": "Nilakshi"
},
{
"family": "Schumann",
"given": "Gunter"
},
{
"family": "Walter",
"given": "Henrik"
},
{
"family": "Weigard",
"given": "Alexander"
},
{
"family": "Whelan",
"given": "Robert"
},
{
"literal": "IMAGEN Consortium"
}
],
"container-title-short": "Neuropsychopharmacology",
"page": "10.1038/s41386-026-02401-6",
"DOI": "10.1038/s41386-026-02401-6",
"PMID": "42020764",
"PMCID": "PMC13592875",
"ISSN": "0893-133X",
"publisher": "Nature Publishing Group",
"URL": "https://doi.org/10.1038/s41386-026-02401-6",
"language": "en",
"issued": {
"date-parts": [
[
2026,
4,
22
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1126/sciadv.aed0772 [code]
Deep learning reveals a neurocomputational mechanism predicting depression risk in adolescents.
Journal: Science advances
In common: ggplot2, tidyverse, 16 authors
[2] doi:10.1038/s41398-026-04010-9 [code]
Bullying victimization and brain development: a longitudinal structural magnetic resonance imaging study from adolescence to early adulthood.
Journal: Translational psychiatry
In common: ggplot2, tidyverse, 16 authors
[3] doi:10.1038/s41398-026-04114-2 [code]
Plasma Glial Fibrillary Acidic Protein (GFAP) shows age-dependent associations with externalizing psychopathology and atypical brain connectivity.
Journal: Translational psychiatry
In common: ggplot2, tidyverse, pandas, 1 other tool, 1 reference, 3 authors
[4] doi:10.1523/eneuro.0370-25.2026 [code]
Pretraining for Large-Scale Functional Connectome Fingerprinting Supports Generalization and Transfer Learning in Functional Neuroimaging.
Journal: eNeuro
In common: pandas, NumPy, 5 references
[5] doi:10.1038/s41467-026-71404-0 [code]
Nonergodicity and Simpson's paradox in neurocognitive dynamics of cognitive control.
Journal: Nature communications
In common: pandas, NumPy, 4 references
[6] doi:10.1016/j.isci.2026.116034
Distinct temporal patterns of beta bursts differentiate successful and failed movement cancellation.
Journal: iScience
In common: 4 references
[7] doi:10.1111/psyp.70397 [code]
Cortical Contributions to Attentional Orienting and Response Cancellation in Action Stopping.
Journal: Psychophysiology
In common: pandas, NumPy, 3 references
[8] doi:10.1017/s0033291726104103 [code]
Linking brain structure to stress reactivity: cingulate surface area predicts acute cortisol responses.
Journal: Psychological medicine
In common: pandas, NumPy, author Henrik Walter
[9] doi:10.1093/braincomms/fcag045 [code]
Examining neuroimaging biomarkers, plasma biomarkers and cognitive functions in patients with recovered COVID-19 infection: a multicentre study using 7T MRI.
Journal: Brain communications
In common: NumPy, other condition, author Penny Gowland
[10] doi:10.1038/s41398-026-04140-0 [code]
Predicting stress response trajectories: Differential contributions of limbic and prefrontal regions to cortisol and affective responses.
Journal: Translational psychiatry
In common: tidyverse, author Henrik Walter

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.