Simple reaction time is described by Pieron's Law which is what?
Mean Reaction Time is a power function of stimulus intensity, and says that MRT decreases with stimulus intensity. (MRT = r0 + kI-^B)
What is the distribution of RT?
With trial-to-trial variability, the distributions are unimodal and positive skewed (tail to the right)
Decision-making is only one part of RT. What is the stages model of RT?
A set of mental operations, performed in sequence which transforms stimulus into response
S -> Stimulus Encoding -> Stimulus Identification -> Response Selection -> R
Simple cognitive RT tasks are composed of a set of stages. Suppose 2 tasks differ by the presence or absence of a stage. You can estimate the duration of stage III just by the difference in MRT between task 1 and task 2.
One task required Detection, another Detection, Identification and Response Selection and the third required Detection and Identification. He thus argued he was creating his method (ie. T(Identify) = T(C) - T(A) = 36ms)
Changed stage durations. If we say go from bright to dim, RT is slowed. Then say alter similarity (make more so) at identification stage, slowing RT again. measure performance under 4 conditions. Do these things at once, which should slow RT by their 2 separate sums. Plotted cell means should be in parallel, this is an additive effect, saying ie. going bright to dim is the same regardless of stimulus similarity and likewise with similarity. This is a stage effect.
When is additivity expected?
Only if variables affect different stages. Non-parallelism is a sign of non-additivity (there is an interaction). (ie. Theios, 1973: varied number of alternatives and type of response, these are affecting the one stage (naming/button press)).
Shwartz, Pomeranz & Egeth (1978)'s experiment involving making a response to 2 arrows on a CRT screen, manipulated similarity, intensity and Stimulus-Response Compatibility (ie. responses that make more sense). What were their results?
They found that these 3 factors are mutually additive and that there is evidence for 3 distinct stages (encoding, identification, response selection)
What are Sequential-Sampling Models of Decision-Making?
They propose we: take multiple observations of a stimulus, accumulate observations over time until there is enough evidence to make a decision.
RT reflects the time taken to accumulate the criterion quantity of evidence. RT variability reflects variability in the accumulation process
We draw repeated random samples from one or another distribution depending on the stimulus presented. We sample from one or another distribution on each trial (assume stimulus 1) and accumulate a running total of observations. Each stimulus produces a distribution of sensory effect which varies moment-to-moment (like SDT). Walk varies with particular sequence of observations.
What are the properties of the Random Walk? (Xn = Z1 + Z2 + ... Zn)
- Mean and variance grow proportionately to the no. of observations - Signal-to-noise ratio (ratio of mean to SD) grows proportionately to square root of no. of observations - Signal-to-noise ratio is the basis for d': sensitivity increases in proportion to the square root of the number of observations
Because the mean of the walk drifts upward when stimulus 1 is presented, and downward when 2 is presented. This interpreted is the accumulated evidence up to that time. Hence evidence increases with time.
We are going to set 2 response criteria, positive at a1, negative at -a2 and continue accumulating evidence until our evidence total is greater than a1 (say stimulus 1) or less than -a2 (say stimulus 2). Hence response criterion is the time taken to accumulate a criterion quantity of evidence which will be both on the stimulus and the settings of the criteria. Time taken to hit a path (reach criterion) is variable (gives MRT). Some walks terminate at wrong boundary (error)
A continuous process that holds that the brain doesn't tick in discrete time steps, it allows accumulation to occur continuously in time (time steps are very small). Has the same kinds of properties as random walk
Predicts the shapes of RT distributions found experimentally
- Task Difficulty Effects: there is a faster accumulation rate for easy stimuli (the mean is larger)
- Speed-Accuracy Tradeoff: fix a narrow criterion setting for speed stress, fix a wide criterion setting for accuracy stress
If you set the R1 criterion closer to the starting point, R1 will be faster but less accurate
Want to see the other 17 Flashcards in Lecture 11 - RT - The Speed of Information Processing?JOIN TODAY FOR FREE!