Best Pest Method

The best-PEST method (from english best p arameters e stimation for s equential t esting , dt. Best parameter estimation for a sequential testing ) is an adaptive method of psychophysics for determining the threshold of perception of a subject to a stimulus .

background

Psychophysical experiments examine the Wahrnehmbarkeitsschwellen using experiments in which subjects have to decide after the presentation of a stimulus, whether the stimulus ( stimulus have seen) or not. The psychometric function shows the course of perceptibility as a function of stimulus strength.

Stimulus configurations such as stimulus parameters (e.g. luminance , volume) and stimulus arrangement are fixed at the beginning of an experiment. However, depending on the presentation of the stimuli during the experiment, habituation effects can occur which lead to systematic errors in the result.

The Best-PEST method offers an adaptive parameterization of stimuli, which leads to an increase in accuracy and an elimination of systematic errors (bias), and also reduces the number of stimulus presentations (and thus the duration of the experiment). The stimulus parameters used depend on the stimuli already tested and the observer's previous responses.

initialization

For the method to run, it is initially assumed that the psychometric function is a density function .

The limits of the range are chosen so that the two external borders secure positive or negative response of the subject guarantee: . The range of values ​​is also quantized . ${\displaystyle \theta \in \{\theta _{\text{min}},\theta _{\text{max}}\}}$

procedure

Initialization:

• ${\displaystyle L(\theta )=1}$or (for all ) ${\displaystyle logL(\theta )=0}$ ${\displaystyle \theta }$
• update ( , answer negative) ${\displaystyle L,\theta _{\text{min}}}$
• update ( , answer positive) ${\displaystyle L,\theta _{\text{max}}}$
• Choice with ${\displaystyle x_{\text{next}}=\theta }$ ${\displaystyle L(\theta )\rightarrow max}$
• or ${\displaystyle x_{\text{next}}=rand(\theta _{\text{min}},\theta _{\text{max}})}$

Test, repeat:

• Answer z when reading out (observer) ${\displaystyle x_{\text{next}}}$
• update ( ) ${\displaystyle L,x_{\text{next}},z}$
• Choice with ${\displaystyle x_{\text{next}}=\theta }$ ${\displaystyle L(\theta )\rightarrow max}$

swell

• Harris R. Lieberman, Alex P. Pentland: Microcomputer-based estimation of psychophysical thresholds: The Best PEST . In: Behavior Research Methods & Instrumentation . tape 14 , no. 1 , January 1982, p. 21-25 , doi : 10.3758 / BF03202110 .