Academic journal article Attention, Perception and Psychophysics

High False Positive Rates in Common Sensory Threshold Tests

Academic journal article Attention, Perception and Psychophysics

High False Positive Rates in Common Sensory Threshold Tests

Article excerpt

Published online: 19 November 2014

# The Psychonomic Society, Inc. 2014

Abstract Large variability in thresholds to sensory stimuli is observed frequently even in healthy populations.Much of this variability is attributed to genetics and day-to-day fluctuation in sensitivity. However, false positives are also contributing to the variability seen in these tests. In this study, randomnumber generation was used to simulate responses in threshold methods using different "stopping rules": ascending 2- alternative forced choice (AFC) with 5 correct responses; ascending 3-AFC with 3 or 4 correct responses; staircase 2- AFC with 1 incorrect up and 2 incorrect down, as well as 1 up 4 down and 5 or 7 reversals; staircase 3-AFC with 1 up 2 down and 5 or 7 reversals. Formulas are presented for rates of false positives in the ascending methods, and curves were generated for the staircase methods. Overall, the staircase methods generally had lower false positive rates, but these methods were influenced even more by number of presentations than ascending methods. Generally, the high rates of error in all these methods should encourage researchers to conduct multiple tests per individual and/or select a method that can correct for false positives, such as fitting a logistic curve to a range of responses.

Keywords Sensory thresholds . Type I error . False positive

Introduction

Threshold testing has long been used to evaluate sensory perception in a wide variety of fields (pain research, water contamination, taste sensation, auditory acuity, offflavors, etc). Thresholds are generally grouped into categories of detection thresholds (lowest concentration of a substance/ sensation that is detectable from the background), recognition thresholds (lowest concentration at which a substance/ sensation can be identified), and discrimination thresholds (smallest difference in concentration or intensity of a substance/sensation that can be detected in a particular range). Methods have been developed to assess sensory thresholds, all of which require an individual to distinguish the stimulus from a background. Most of these threshold tests are also "forced choice," meaning that participants are required to make a choice among samples, such as choose a stimulus compared to one or more blanks or choosing a stronger stimulus; if the participant is uncertain which sample to choose, he or she must make a guess. In such cases, participants will occasionally give correct responses accidentally, leading to false positives, or lower than actual thresholds, in the dataset.

In fields of sensory research where participants may be guessing frequently, such as an anosmic person in an olfactory threshold test or when a stimulus is unfamiliar such as in fatty acid "taste" research, rates of false positives in threshold tests become particularly important in interpretation of results. This article is designed to investigate the frequencies of such false positives in sensory threshold experiments, focusing on a few primary techniques common in the field of odor and taste sensitivity research. The high rates of false positives in these methods have been acknowledged (Lawless and Heymann 1998, 2010), but are often not taken into account when analyzing final data. Typical methods for dealing with the false thresholds have included correcting for the proportion of expected "guessers," which can be done at each concentration step or across the ranges of concentrations; or fitting psychometric functions to the data, which assumes a certain rate of false positives. Experiments comparing methods of threshold testing acknowledge that multiple tests, or even multiple methods, will give the most reliable data regarding an individual's true range of sensitivity, as the variance both among and within subjects in these datasets are high (Boesveldt, de Muinck Keizer, Knol, Wolters, & Berendse, 2009; Doty, McKeown, Lee, & Shaman, 1995; Doty, Smith, McKeown, &Raj, 1994; Haehner et al. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.