Coding Example 3: Non-decision time

This example is designed to give you a better feel for how the non-decision time in a DDM influences responding. It is written using R Markdown. Markdown is a simple formatting syntax for authoring HTML, PDF, and MS Word documents. For more details on using R Markdown see http://rmarkdown.rstudio.com.

Last parameter of the basic DDM! Woo, you’re almost there! Are you as excited as I am?!! Well, too bad, because the non-decision time is pretty much the most boring parameter of the lot. Still, it’s worth seeing how it works, and understanding its implications.

Example 3.1. Adding in the non-decision time.

Let’s start by modifying the ddm function we wrote in our previous example to include a non-decision time. This really won’t take much. It involves adding an input argument ndt to the code, and adding the value of ndt to the RT at the end (in step 6). That’s it.

ddm = function(driftrate = .08, threshold = .15, ndt = .3, noise = .1, precision = .001){
  # function ddm:
  # driftrate: average strength of the evidence
  # threshold: the choice-determining threshold
  # ndt: the non-decision time
  # noise: standard deviation of the evidence, by convention set to .1
  # precision: the size of the time-step for accumulating evidence (default = .001s)
  
  # 1. start at time t = 1
  t = 1
  
  # 2. start with evidence at 0
  accumulatedEvidence = 0 
  
  # 3. use a while loop that will break when the evidence passes +/- the threshold
  while(abs(accumulatedEvidence[t]) < threshold){
    t = t+1
    
    # 4. take a noisy sample, scaled for the temporal precision with which we are estimating
    noisySample = driftrate*precision + rnorm(1, mean = 0, sd = noise*sqrt(precision))
    
    # 5. add it to the previous evidence
    accumulatedEvidence[t] = accumulatedEvidence[t - 1] + noisySample
  }
  
  # 6. return a data structure with the choice, the RT (in s), and the observed accumulated evidence
  choiceInfo = list(choice = sign(accumulatedEvidence[t]), rt = t*precision + ndt, drift = accumulatedEvidence)
  
  }

Let’s see what this does now:

simAndPlot = function(driftrate = .08, threshold = .15, ndt = .3, nSim = 1000){
  # initialize vectors to hold the simulated choices and RTs
  Choices = RTs = vector(mode = 'numeric', length = nSim)
  
  # run the simulations and save the data
  for(sim in 1:nSim){
    temp = ddm(driftrate = driftrate, threshold = threshold, ndt = ndt)
    Choices[sim] = temp$choice
    RTs[sim] = temp$rt
  }
  
  # Plot the data!
  # 1. Get the frequencies of YES and NO responses in specific RT bins, determined by the maximum RT
  maxRT = ceiling(max(RTs))
  rtBins = seq(0,maxRT, length = 40)
  freqYes =   hist(RTs[Choices == 1], breaks = rtBins, plot = FALSE)$counts
  freqNo =   hist(RTs[Choices == -1], breaks = rtBins, plot = FALSE)$counts
  
  
  # 2. plot the frequencies
  par(mfrow = c(2,1), mar = c(0,4,5,1) + .2)
  yMax = ceiling(max(c(freqYes, freqNo)/50))*50
  hist(RTs[Choices == 1], breaks = rtBins, ylim = c(0,yMax), main = '', xaxt = 'n'
       , col = 'red')
  par(mar = c(4,4,1,1) + .2)
  hist(RTs[Choices == -1], breaks = rtBins, ylim = c(yMax,0), main = '', xlab = 'RT'
       , col = 'blue')
  
  # 3. Add descriptive text to the plot
  text(maxRT * .5, yMax * .4, paste('% YES: ', format(mean(Choices == 1), digits = 3)), adj = 0)
  text(maxRT * .5, yMax * .6, paste('Ave. YES RT: ', format(mean(RTs[Choices == 1]), digits = 3)), adj = 0)
  text(maxRT * .5, yMax * .8, paste('Ave. NO RT: ', format(mean(RTs[Choices == -1]), digits = 3)), adj = 0)
}

# plot with defaults (non-decision = .3)
simAndPlot()

# plot with a shorter non-decision time (ndt = .1)
simAndPlot(ndt = .1)

# plot with a longer non-decision time (ndt = .5)
simAndPlot(ndt = .8)

Note if and how the non-decision time affects the choices and rts.

Coding Example 3: Non-decision time

Example 3.1. Adding in the non-decision time.

Exercises/Test Yourself