The influence of memory for impressions based on behaviours and beliefs on approach/avoidance decisions

ABSTRACT

Recent work has shown that memory for various types of information associated with social targets (impressions based on behaviours and political ideology) influences decisions to approach or avoid those same targets. The current study was intended to better understand the extent that memory for other types of details associated with targets (beliefs and behaviours) affects subsequent approach/avoidance decisions. In this investigation, participants formed impressions of social targets represented by a picture and a sentence (a belief in Experiment 1; either a belief or behaviour in Experiment 2). Later, memory for the impressions and sentences associated with each target was tested. Finally, participants decided whether they would want to approach or avoid each target. Results demonstrated that accurate memory strongly influenced approach/avoidance decisions for targets associated with different types of information (beliefs; behaviours), although there was no difference in the effect of memory between beliefs and behaviours. Overall, results suggest an important role of memory in decisions to approach or avoid others.

KEYWORDS:

• impression memory
• person memory
• approach/avoidance
• social cognition

Disclosure statement

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

Notes

1 Past research has shown differences in memory for in-group and out-group members (Bernstein et al., 2007; Sahdra & Ross, 2007; Wolff et al., 2014; Zengel et al., 2021), but evidence of memory being influenced by beliefs themselves is lacking. Strongly held personal values have been shown to be memorable (Howe & Krosnick, 2017), largely due to selective attention to information that aligns with ones’ own values, but, though related, values are not synonymous with political beliefs.

2 These four participants gave extremely few “negative” impression memory responses (zero, zero, one, and one).

3 In the self-similarity task, participants rated self-similarity and importance of the 168 beliefs from the sentences seen in earlier phases. This phase of the experiment is not the focus of this report and will not be discussed further. This phase occurred at the very end of the experiment and therefore did not influence any data reported in this paper.

4 63.3% of participants identified as liberal, 23.3% as conservative, and 13.3% as other. Altogether, 37% of participants identified their ideology strength as average, 27% as somewhat strongly, 20% as a little, 13% as very strongly, 3.3% as not at all. The average ideology strength was 3.27 (3.00 for conservatives, and 3.16 for liberals).

5 We also conducted the same analyses within trials where belief memory was correct. When beliefs were correctly remembered, there were more yes responses than baseline for targets associated with positive impressions (M = .61, SE = .03), t(29) = 3.79, p < .001, d = 0.69, 95% CI [0.05, 0.17], and fewer yes responses than baseline for targets associated with negative impressions (M = .39, SE = .03), t(29) = −3.82, p < .001, d = 0.70, 95% CI [−0.17, −0.05] (see Supplemental Figure 1). These findings further suggest that memory plays a role in subsequent AA decisions about social targets.

6 We also ran the same ANOVA for belief memory accuracy: an initial impression x belief memory accuracy ANOVA on the percentage of yes responses showed a main effect of initial impression, with more yes responses on positive impression trials (M = 0.56, SE = 0.03) than on negative impression trials (M = 0.40, SE = 0.03), F(1, 29) = 51.53, p < .001, η²p = 0.08. This main effect was qualified by an initial impression x belief memory accuracy interaction, F(1, 29) = 11.42, p = .002, η²p = 0.28. There were significantly more yes responses for positive (M = 0.61, SE = 0.03) than negative impression trials (M = 0.39, SE = 0.03) when belief memory was correct, t(29) = 6.68, p < .001, d = 1.22, 95% CI [0.15, 0.29], and more yes responses for positive (M = 0.52, SE = 0.03) than negative impression trials (M = 0.42, SE = 0.03) when belief memory was incorrect t(29) = 4.53, p < .001, d = 0.83, 95% CI [0.06, 0.15], but the difference between positive and negative was greater when belief memory was correct than when it was incorrect. The main effect of belief memory accuracy was not significant, (F = 2.47, p = .127; see Supplemental Figure 2).

7 These three participants gave extremely few “negative” impression memory responses (six, two, and two).

8 53% of participants identified as liberal, 28% as conservative, and 19% as other. The majority of participants (53%) identified their ideology strength as average, 28% as somewhat strongly, and 9% each as a little or very strongly, and 0% as not at all. The average ideology strength was 3.38 (3.22 for conservatives, and 3.47 for liberals).

9 We also conducted the same analyses within trials where sentence memory was correct (see Supplemental Figure 3). For behaviors, there were more yes responses than baseline for targets associated with positive impressions (M = .61, SE = .04), t(31) = 3.03, p = .005, d = 0.54, 95% CI [0.04, 0.18]; and fewer yes responses than baseline for targets associated with negative impressions (M = .36, SE = .03), t(31) = −4.73, p < .001, d = 0.84, 95% CI [−0.20, −0.08] when the behavior was correctly remembered. For beliefs, there were more yes responses than baseline for targets associated with positive impressions (M = .64, SE = .04), t(31) = 3.84, p < .001, d = 0.68, 95% CI [0.07, 0.22]; and fewer yes responses than baseline for targets associated with negative impressions (M = .34, SE = .03), t(31) = −5.10, p < .001, d = 0.90, 95% CI [−0.22, −0.09] when the belief was correctly remembered. These results suggest that memory for beliefs and behaviors do impact AA decisions.

10 We also ran the same ANOVA for sentence memory accuracy: an Initial Impression × Sentence Memory Accuracy × Sentence Type ANOVA on the percentage of yes responses (see Supplemental Figure 4) showed a main effect of initial impression, with more yes responses for positive (M = 0.57, SE = 0.03) than negative (M = 0.36, SE = 0.03) impressions, F(1, 31) = 47.78, p < .001, ${\eta }_p^2$ = 0.61. There was also a main effect of sentence memory accuracy, with more yes responses when sentence memory was correct (M = 0.49, SE = 0.02) than when it was incorrect (M = 0.44, SD = 0.03), F(1, 31) = 5.35, p = .028, ${\eta }_p^2$ = 0.15. These main effects were qualified by an initial impression × sentence memory accuracy interaction, F(1, 31) = 15.23, p < .001, ${\eta }_p^2$ = 0.33. There were significantly more yes responses for positive (M = 0.63, SE = 0.03) than negative impression trials (M = 0.35, SD = 0.03) when sentence memory was correct, t(31) = 6.91, p < .001, d = 1.22, 95% CI [0.19, 0.35] and for positive (M = 0.51, SD = 0.03) than negative impression trials (M = 0.38, SD = 0.03) when sentence memory was incorrect, t(31) = 4.53, p < .001, d = 0.80, 95% CI [0.08, 0.21], but the difference between positive and negative was greater when belief memory was correct than when it was incorrect. Neither the main effect of sentence type nor any of the other interactions were significant, (Fs < 1.40, ps > .246). These results provide further support for the impact of memory on AA decisions.