The VisionWorks^TM MethodMaker^TM Module

MethodMaker consists of a suite of psychophysical methods that not only control the presentation of stimuli, but also collect, organize, and analyze the results. MethodMaker is included in our Contrast-Sensitivity applications as well as all other psychophysical applications, including custom applications. MethodMaker is also part of our high-end development system in the form of a library of high-level C-functions. A single function call will bring up a menu of methods and the associated parameters

The Experiment Types

• Discrimination	• Absolute Threshold
• Point of Subjective Equality	• Identification
• Same/Different	• Detection

The Methods

• Method of Adjustment

• Method of Constant Stimuli (single-interval, forced-choice, yes-no)

• Staircase (forced-choice, up-down)

• Maximum Likelihood Estimation (QUEST and ZEST for forced choice and yes-no)

• Signal Detection (ROC curves from different confidence levels)

Features Common to all Methods

• Interweaving conditions (except Method of Adjustment)

• Reaction times automatically collected, accurate to less than one millisecond

• Results automatically sent to log and summary files and a special data file suitable for analysis by statistical packages

• Optional feedback

• Trials may be subject-initiated or automatic

• N-alternatives for forced choice

• Trial-by-trial results maintained

Selected Method of Constant Stimuli (MOCS) Features

• Levels may be set by user or automatically within a range

• Randomization-within-block option--define a block of trials within which each alternative gets presented the specified number of times

• Data collected and summarized separately for each level of the stimulus

• Probit analysis performed and ogive displayed on-screen

Selected Staircase Features

• Additive, multiplicative or log steps

• Number of steps back after a mistake can be specified

• Option to move a step only after n correct responses (permits Wetherill-Levitt staircase)

• May jump an extra amount after consecutive errors to avoid being stuck too low

• Large initial steps may be specified to be reduced at each reversal during initial practice period (permits quick threshold determination)

• Graph of results is displayed on-screen, shows each reversal

• Results include mean of reversals, standard deviation and standard error

Selected Quest / Zest Features

Results including threshold mean, mode, and confidence intervals are computed and printed. The user has control over the following parameters:

• Slope of assumed psychometric function

• Threshold Definition ( 51% to 99%)

• Assumed probability of a mistaken keypress (keeps maximum likelihood function from being overly affected by occasional wrong answers at high stimulus levels)

• Random jitter for selection of next value (keeps trial from being too predictable)

• Standard Deviation of prior probability density function

• Termination Rule (termination may be set to fixed number of trials or may keep going until a specified threshold confidence interval is reached)

• Initial probability-density function may be either Gaussian (as in QUEST), or hyperbolic secant (as in ZEST)

• Use of mode (as in QUEST) or Mean (as in ZEST) to determine placement of next trial

Selected Signal Detection Features

• Specify number of confidence levels and trials per level

• d' values are computed and printed for each confidence level for each stimulus level

• ROC curves printed on-screen for each stimulus level

• Feedback