Related Experiment Videos
Estimation of psychometric functions from adaptive tracking procedures.
M R Leek1, T E Hanna, L Marshall
1Army Audiology and Speech Center, Walter Reed Army Medical Center, Washington, DC 20307-5001.
Perception & Psychophysics
|March 1, 1992
Summary
Computer simulations show that adaptive tracking procedures can accurately estimate psychometric functions with sufficient trials (≥200). Shorter tracks (50-100 trials) yield biased slope estimates, but corrections are provided.
Area of Science:
- Psychology
- Psychophysics
- Computational Neuroscience
Background:
- Adaptive tracking procedures are commonly used in psychophysics to estimate sensory thresholds.
- However, data collected during these adaptive tracks may not reliably represent performance across all stimulus levels.
- This limitation can impact the accuracy of psychometric functions derived from such data.
Purpose of the Study:
- To assess the reliability and accuracy of psychometric functions generated from adaptive tracking data.
- To evaluate the impact of track length and psychophysical procedure on the estimation of psychometric function parameters, particularly slope.
- To provide correction factors for biased slope estimates in shorter tracks.
Main Methods:
- Computer simulations were employed to generate adaptive tracks using known psychometric functions.
- Simulations varied psychophysical procedures, target performance levels, and psychometric function slopes.
- Estimates of psychometric function slopes from simulated track data were compared to the true generating functions.
Main Results:
- Psychometric functions reconstructed from adaptive tracks with at least 200 trials were generally accurate reflections of the true functions.
- Tracks with 50 or 100 trials resulted in biased high slope estimates across all simulated conditions.
- No significant differences in accuracy were found based on target performance level; minor differences were observed across psychophysical procedures.
Conclusions:
- For accurate estimation of both threshold and slope, adaptive tracks should include at least 100 trials, preferably using a three- or four-alternative forced-choice procedure.
- Good estimates are achievable with fewer trials or a two-alternative forced-choice procedure if appropriate bias correction factors are applied.
- The findings offer guidance for optimizing data collection and analysis in psychophysical research using adaptive tracking.