Three tests and three corrections: comment on Koen and Yonelinas (2010)

Area of Science:

• Cognitive Psychology
• Psychophysics
• Memory Research

Background:

• The slope of the z-transformed receiver-operating characteristic (zROC) in recognition memory is usually < 1.
• This is often interpreted as greater target distribution variance than lure distribution variance.
• The encoding variability hypothesis suggests differing memory strength increments during study cause this.

Purpose of the Study:

• To re-evaluate the Koen and Yonelinas (2010) study on encoding variability and zROC slope.
• To determine if experimental errors in Koen and Yonelinas's tests affected their conclusions.
• To assess the validity of the mixture-unequal-variance signal-detection (UVSD) model in light of these findings.

Main Methods:

• Analysis of the experimental design and statistical tests used by Koen and Yonelinas (2010).
• Correction of identified errors in the three statistical tests performed on mixed-strength target data.
• Comparison of corrected results against predictions of the encoding variability hypothesis and the mixture-UVSD model.

Main Results:

• The three tests performed by Koen and Yonelinas do not invalidate the encoding variability hypothesis.
• Errors were found in all three of Koen and Yonelinas's statistical tests.
• Corrected analyses show the data support the mixture-UVSD model, contradicting Koen and Yonelinas's original conclusion.

Conclusions:

• The encoding variability hypothesis remains a viable explanation for zROC slopes < 1.
• Koen and Yonelinas's (2010) conclusions were based on flawed statistical tests.
• The mixture-unequal-variance signal-detection model is supported by the re-analyzed data.