Top-down versus bottom-up processes in the formation of positive and negative retrospective affect

ABSTRACT

On the basis of two large scale diary studies (n = 2022, n = 762) We study differences in the effects of dispositions and situations in the formation of positive and negative retrospective affect (retrospective-PA and retrospective-NA, respectively), the affect associated with extended (e.g. daily) experiences, as opposed to very short (episodic) experiences. We suggest that the differences between retrospective-PA and retrospective-NA is due to the fact that positive retrospective evaluation (i.e. the evaluation of positive retrospective affect) involves primarily top-down processing, in which people resort to their dispositions in making these evaluations, whereas negative retrospective evaluation (the evaluation of negative retrospective affect) is primarily based on the cumulative affects of individual experiences.

KEYWORDS:

• Top-down versus bottom-up processes
• positive and negative affect
• affective traits
• retrospective affect
• substitution

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data are available at http://midus.wisc.edu/.

Notes

1 The study was conceptualised by the first author, but data analysis and report writing was done by all the three authors. The study was not pre-registered.

2 This question is a mixture of what Schimmack et al. (2000) called frequency (since it depends on the number of times the emotion was felt during the day) and duration (the amount of time it was felt once it was experienced). We preferred to use the term frequency because this is the common term used in the MIDUS literature to refer to this (and other similar) variables.

3 We note that our hypothesis about the level-1 predictors is that the relationship between negative events and daily-NA is stronger than the relationship between positive events on daily-PA. To be most general, we do not distinguish between the level-1 and level-2 effects of daily events: we analyze the variance in the daily affects disregarding the sources of this variance (whether it is associated with level-1 or level-2 variables). However, in the appendix we report estimates of “pure” level-1 effects based on analyses in which the positive and negative daily events were standardised within subjects. It is clear that the results of these analyses are very similar to the results reported in the text. Finally, no additional analysis was necessary regarding the level-2 predictors (trait-PA and trait-NA), since, by design, their effect concern only between person effects.

4 We also note that the distinction between experienced affect and reported affect is relevant and to the distinction between remembered utility (Schreiber & Kahneman, 2000), which corresponds to Russel’s reported affect and experienced utility (Kahneman et al., 1997), which corresponds to Russel’s core affect.

5 Kahneman et al.’s (2004) day reconstruction method in which subjects report about meaningful affective experiences may overcome this problem. However, this method involves retrospective affective evaluation, and thus may introduce substitution biases associated with this type of affective evaluations.

6 We note that Neubauer et al. (2020) focused on the effects of the intensity of positive and negative experiences, rather than the number of these experience, and therefore although they did not model the effects of affective traits, they could model effects of experiences’ intensity.

7 In this respect substitution is relevant not only to the evaluation of one’s affect, but also to the evaluation of other concepts such as well-being (Kahneman & Krueger, 2006) or life satisfaction (Kahneman & Riis, 2005).

Additional information

Funding

This work was supported by The Henry Crown Institute of Business Research in Israel.