Retrospective recall of affect in clinically depressed individuals and controls

Abstract

The current study examined the nature of retrospective recall of affect in depressed participants and nonclinical controls. Utilising the Experience Sampling Method, we compared average momentary affect reports to retrospective summaries of the same period of time. Both groups exhibited exaggerated retrospective summaries of positive affect (PA) and negative affect (NA). Greater bias for PA than NA was found in the controls but not the depressed. Across groups, depressed individuals exhibited more absolute inaccuracy in their recall of NA but not PA. For depressed and nonclinical individuals, retrospective summaries of PA were better predicted by an average of all weekly momentary PA than a combination of the most intense (“peak”) PA and most recent (“end”) PA experiences. For nonclinical participants, retrospective NA was better predicted by an average of all weekly momentary NA. For depressed individuals, however, no clear advantage to predicting retrospective NA by average weekly NA relative to a combination of peak NA and end NA was found. Based on our findings, clinicians should realise that depressive cognition may not lie in recall that is less positive and more negative than “reality” but in retrospective intensifications of all affect that are, compared to normal, relatively greater for negative than for positive affect.

Keywords:

• Depression
• Retrospective recall
• Memory
• Affect

Notes

¹In other investigations examining a peak effect in recalling affective experience, the most extreme affect rating on that measure (i.e., a single, discrete score) has been used to reflect peak experience (Fredrickson & Kahenman, 1993; Parkinson, Briner, Reynolds, & Totterdell, 1995). Defining “peak” as such in our analyses did not substantively change the pattern of results.

²The 48-item Short-scale Eysenck Personality Questionnaire- Revised (EPQ-R; Eysenck, Eysenck, & Barret, 1985) was also administered during the first screening visit. It was not used in the current analyses and is therefore not described further.

³The “end effect” that we propose, reflected in the correlation between “end” and retrospective ratings, also may be conceptualised as participants’ tendency to gradually settle on a rating over the course of time in ESM paradigms, and to vary their responses around this “anchor”, a phenomenon observed by other researchers (Csikszentmihalyi & Larson, 1987; Larson & Richards, 1994). Indeed, the correlations between retrospective recall and the average rating for the second half of the week were greater for both PA and NA (r=.52 and .72, respectively) than with such ratings from the first half of the week (r=.44 and .65, respectively).