The unequal variance signal-detection model of recognition memory: Investigating the encoding variability hypothesis

Abstract

<jats:p> Despite the unequal variance signal-detection (UVSD) model’s prominence as a model of recognition memory, a psychological explanation for the unequal variance assumption has yet to be verified. According to the encoding variability hypothesis, old item memory strength variance (σ<jats:sub>o</jats:sub>) is greater than that of new items because items are incremented by variable, rather than fixed, amounts of strength at encoding. Conditions that increase encoding variability should therefore result in greater estimates of σ<jats:sub>o</jats:sub>. We conducted three experiments to test this prediction. In Experiment 1, encoding variability was manipulated by presenting items for a fixed or variable (normally distributed) duration at study. In Experiment 2, we used an attentional manipulation whereby participants studied items while performing an auditory one-back task in which distractors were presented at fixed or variable intervals. In Experiment 3, participants studied stimuli with either high or low variance in word frequency. Across experiments, estimates of σ<jats:sub>o</jats:sub> were unaffected by our attempts to manipulate encoding variability, even though the manipulations weakly affected subsequent recognition. Instead, estimates of σ<jats:sub>o</jats:sub> tended to be positively correlated with estimates of the mean difference in strength between new and studied items ( d), as might be expected if σ<jats:sub>o</jats:sub> generally scales with d. Our results show that it is surprisingly hard to successfully manipulate encoding variability, and they provide a signpost for others seeking to test the encoding variability hypothesis. </jats:p>

Informative and uninformative prestimulus cues at encoding benefit familiarity and source memory

“…Second, we cannot rule out the possibility that the null findings for subjective measures of recollection resulted from mismodelling of the receiver operator characteristic curves. There has been substantial debate as to what signal detection model provides the best characterization of recognition memory (e.g., Jang et al, 2012;Klauer & Kellen, 2015;Spanton & Berry, 2020;Starns et al, 2012; for reviews of signal detection models of memory, see Wixted, 2007;Yonelinas & Parks, 2007). The present study does not, and cannot, distinguish between these models of memory because it was not designed to do so.…”

Does variability in recognition memory scale with mean memory strength or encoding variability in the UVSD model?

“…In the following experiment, we manipulate both encoding variability and memory strength at two levels each (high, low) in a 2 × 2 factorial design. Strength will be manipulated using a one-back digit judgement task identical to that in the 'fixed' condition in Experiment 2 of Spanton and Berry (2020). This task will be present as a simultaneous distraction in low strength condition study phases, and absent in high strength conditions.…”

“…To achieve this, a successful manipulation of encoding variability during the study phase is needed. Despite previous efforts by Spanton and Berry (2020) to add substantial Gaussian variability to individual encoding variables, the resultant effects upon old item variance were weak. This may be because even without experimental manipulation, there are already a very large number of encoding variables that sum to determine levels of added strength in any condition.…”

“…For example, if word frequency and strength are negatively related whereas concreteness and strength are positively related, then any amount of added strength that a word may receive for having low word frequency may be balanced by a decrement in strength if that word also happens to have low concreteness. Furthermore, there is likely to be a negative correlation between an item's baseline strength value and the increment of added strength it receives during study, which, in conjunction with the aforementioned factors, makes it difficult to establish a strong experimental manipulation of encoding variability (Spanton & Berry, 2020).…”

ROC asymmetry is not diagnostic of unequal residual variance in Gaussian signal detection theory

“…Recent attempts to manipulate encoding variability, however, have failed to show effects on zROC slopes (e.g., Spanton & Berry, 2020). This raises the question whether the variance inequality is really due to unequal residual variance.…”