The breadth of animacy in memory: New evidence from prospective memory

Studies using retrospective memory tasks have revealed that animates/living beings are better remembered than are inanimates/nonliving things (the animacy effect). However, considering that memory is foremost future oriented, we hypothesized that the animacy effect would also occur in prospective memory (i.e., memory for future intentions). Using standard prospective memory (PM) procedures, we explored this hypothesis by manipulating the animacy status of the PM targets. Study 1a reports data collected from an American sample; these results were then replicated with a Portuguese sample (Study 1b). Study 2 employed a new procedure, and data were collected from a broader English-speaking sample. In these three studies, animate (vs. inanimate) targets consistently led to a better PM performance, revealing, for the first time, that the animacy advantage extends to PM. These results strengthen the adaptive approach to memory and stress the need to consider animacy as an important variable in memory studies. Supplementary Information The online version contains supplementary material available at 10.3758/s13423-023-02406-y.


Additional Information about Excluded Participants
The PM literature is not consistent regarding the inclusion/exclusion of participants who did not provide any correct PM response (e.g., Horn & Bayen, 2015, included those participants, while Gilbert, 2015, excluded them).In our study, as data were collected online, we opted to exclude those participants because missing PM responses could be due to a normal PM failure, or to other non-controlled factors (e.g., misreading instructions, PM responses not being registered due to nonstandard keyboard layouts, cf.Gilbert, 2015supplementary information).However, for each study, we conducted an additional 2 (Animacy: animates vs. inanimates) x 3 (Type of trial: baseline vs. filler vs. target) repeated measures ANOVA, including also the participants who were excluded for not performing any PM response.Across studies, the results revealed the same pattern as when excluding them (Supplementary Table S1), including the follow-up paired t-tests used to disentangle the interactions.

Exclusion of Trials from the Analysis of Performance
Following previous studies (e.g., Smith & Hunt, 2014), trials immediately after the target trials were excluded, as performance on these trials may incur an additional cost due to the PM response.Trials with missing responses were also excluded from the analyses (as in, for example, Strickland et al., 2020).In Study 1a, a total of 0.7% of the baseline, 2.0% of the filler, and 1.0% of the target trials were excluded from the analyses for those reasons.In Study 1b, these corresponded to 2.2% of the baseline, 3.3% of the filler, and 2.6% of the target trials.In Study 2, 1.4% of the baseline, 0.9% of the filler, and 1.6% of the target trials were excluded for the same motives.

Versions of the Task
In Studies 1a and 2, versions 1DB and 2DB presented the target words "bottle" and "dancer", whereas versions 1NP and 2NP presented "nurse" and "phone" as targets.In Study 1b, versions 1CJ and 2CJ presented the target words "cavalo" [horse] and "janela" [window], whereas "atleta" [athlete] and "camisa" [shirt] were presented in versions 1AC and 2AC.In the PM phase, words were presented in a fixed order to every participant but counterbalanced between versions: where an animate word was presented in Version 1, an inanimate was presented in Version 2 (and vice-versa).For each study, participants were

False Alarms
The false alarm response rate (i.e., giving a PM response to a filler trial) was negligible and did no differences between animates and inanimates were obtained across studies, as presented in Supplementary Table S2.

Analyses of the Response Times (RTs)
The baseline phase provides a "purer" measure of the participants' performance in the ongoing task only, allowing us to explore the cost/interference of performing the ongoing task with an embedded PM task (PM phase).Such cost/interference is usually indexed by the participants' response times.As data were collected online, analyses of response times assumed a secondary role in this work, as these data might vary due to several uncontrollable factors (e.g., the participants' internet speed).
Due to the slightly unbalanced number of participants allocated to each version, we explored if the variable Version of the experiment influenced our results in any way, by Study 2: lowest p = .114).This suggests that the animacy effects reported in all studies are not restricted to one specific animate / inanimate PM target, nor to a specific order of presentation of the PM targets (i.e., being presented first with an animate or an inanimate target did not influence the overall PM performance).
conducting a 2 (Animacy: animates vs. inanimates) x 3 (Type of trials: baseline vs. filler vs. target) x 4 (Version) mixed ANOVA, for each study.Across studies, neither the main effect Supplemental Material 6 of Version of the experiment, nor the interactions involving this variable, were significant (Study 1a: lowest p = .505;Study 1b: lowest p = .147; Number of False Alarm Responses Given to Animate and Inanimate Trials, 1% of the baseline and 0.4% of the filler trials in Study 1b, and 1.0% of the total trials in Study 2.Results are depicted in Supplementary Fig.S1, and the statistical analyses are presented in Supplementary TableS3.Response Times Obtained in Studies 1a (N = 132), 1b (N = 63) and 2 (N = 72).Error Bars Represent Standard Errors of the Mean Smith and Hunt, 2014)studies, RTs were trimmed separately for animate and inanimate trials (e.g., Rummell et al., 2017 used a lexical decision task, and trimmed RTs separately for words and nonwords); those trials with RTs below M-3SD or above M+3SD from each participant's mean were excluded(Matos et al., 2020;as similarly done bySmith and Hunt, 2014).RTs from target trials were not trimmed, otherwise there would be too few trials/datapoints to analyze.This trimming procedure resulted in the exclusion of 0.1% of the total trials in Study 1a, 0.