A novel bias-free approach for robust perceptual threshold estimation

Tarasi, Luca; Covelli, Margherita; Bertini, Caterina; Romei, Vincenzo · 2026 · Crossref

DOI: 10.3758/s13428-026-03038-5

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses a fundamental flaw in standard psychophysical methods for estimating perceptual thresholds: the failure to account for response bias. Traditional techniques, such as the constant stimuli and staircase methods, rely exclusively on hit rates, ignoring false alarms. According to signal detection theory, this oversight conflates genuine sensory sensitivity with liberal response criteria, leading to unreliable threshold estimates. The authors propose and validate "bias-free" versions of these methods that incorporate target-absent trials (catch trials) to calculate false alarm rates, thereby isolating sensitivity from decisional bias. The study involved 69 participants (after excluding five outliers) who performed a visual detection task involving gray circles on checkerboard backgrounds. Participants underwent threshold estimation using four methods: classic staircase, classic constant stimuli, bias-free staircase, and bias-free constant stimuli. The bias-free methods included 50% catch trials to compute balanced accuracy. The bias-free staircase used mini-blocks of 10 trials (five signal, five noise), adjusting contrast based on the average of hit rates and correct rejection rates to converge on 70% balanced accuracy. Following estimation, participants completed independent test blocks at their estimated thresholds to validate performance. The target sensitivity ($d'$) for 70% accuracy was theoretically estimated at approximately 1.46. Results demonstrated that classical methods systematically overestimated perceptual thresholds. The bias-free staircase yielded significantly lower threshold values (33/222 RGB) compared to the classic staircase (44/211 RGB) and constant stimuli methods. Crucially, only the bias-free staircase achieved the target 70% accuracy in the validation test blocks (mean accuracy 71%). In contrast, thresholds derived from classical methods resulted in significantly higher accuracy (84–85%), indicating that these methods set thresholds too high, making stimuli easier to detect than intended. Sensitivity analysis confirmed that the bias-free staircase produced a $d'$ of 1.50, aligning closely with the theoretical target, whereas classical methods yielded inflated $d'$ values (2.34–2.57). Furthermore, the bias-free staircase showed a balanced distribution of errors around the target, whereas classical methods exhibited predominantly ceiling effects, confirming that they fail to control for interindividual variability in response criteria. The significance of this work lies in the demonstration that uncontrolled decisional bias compromises the reliability of perceptual threshold estimation. The authors conclude that the bias-free staircase method provides a robust, simple, and accurate alternative that mitigates the influence of response criteria. By adopting this approach, researchers can enhance the precision of sensory measurements and reduce variability in subsequent experimental manipulations across psychology and neuroscience.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• signal detection theory