The differential similarity of positive and negative information – an affect-induced processing outcome?

ABSTRACT

People judge positive information to be more alike than negative information. This good-bad asymmetry in similarity was argued to constitute a true property of the information ecology (Alves, H., Koch, A., & Unkelbach, C. (2017). Why good is more alike than bad: Processing implications. Trends in Cognitive Sciences, 21, 69–79). Alternatively, the asymmetry may constitute a processing outcome itself, namely an influence of phasic affect on information processing. Because no research has yet tested whether phasic affect influences perceived similarity among stimuli, we conducted 5 Experiments that also tested whether phasic affect can account for the higher judged similarity among positive compared to negative stimuli. In three experiments, we affectively charged pictures of different Pokemon by pairing them with monetary gains and losses (Exp. 1a, 1b) as well as positive and negative trait words (Exp. 2); yet, the evaluative charge did not differentially influence perceived similarity among the Pokemon. Experiment 3 replicated the basic similarity asymmetry among positive and negative words, and found that it was unaffected by externally induced phasic affect. Experiment 4 showed that phasic affect had no influence on perceived similarity of non-evaluative words either. We conclude that albeit a weak influence of phasic affect on perceived similarity of stimuli cannot be ruled out entirely, it can most likely not account for the typically medium to large sized asymmetry in similarity among positive and negative stimuli.

KEYWORDS:

• Valence
• affect
• similarity
• information processing

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The fact that we collected data from more participants than intended was due to procedural constraints in our lab. Sometimes participants who are recruited on campus show up at the lab later than they intended, and we then allow them to participate even if the desired sample size has already been reached.

2. Note that this analysis is factually equivalent to an ANOVA with one between and one repeated factor, but it prevents us from expressing one of the two factors (word and sound valence) as an interaction term, which may create confusion.