Forming a new prospective memory intention can reduce prospective memory commission errors

Prospective memory (PM) refers to remembering to execute intended actions in the future. Beyond omission errors, recent work highlights aftereffects of completed intentions, whereby no-longer-relevant intentions remain active and can trigger commission errors. This study examines whether forming a new PM intention after completing a prior PM task reduces commission errors, and whether using an implementation intention (if-then) encoding for the new intention further modulates these effects. Two competing accounts make differing predictions: (a) the attentional dependence hypothesis posits that forming new PM tasks sustains a general PM task set, reallocates attention toward PM processing, and may inadvertently retrieve old intentions—predicting more commission errors; (b) the intention overwriting hypothesis posits that new intentions overwrite components of the old intention (cue, action, association), lowering its activation and reducing commission errors. The study tests these accounts by comparing a non-new PM group, a new PM group, and a new PM group with implementation intention encoding during the finished phase, with the broader aim of understanding mechanisms and potential strategies to mitigate commission errors that can have serious real-world consequences (e.g., double-dosing medication).

Prior research commonly uses event-based PM paradigms with an active phase followed by a finished phase where old PM cues reappear as ongoing stimuli, revealing aftereffects of completed intentions (e.g., Scullin et al., 2012; Möschl et al., 2020). Two theoretical perspectives are central: (1) the attentional dependence hypothesis (Walser et al., 2014; 2017), which suggests new PM intentions sustain a PM task set and can trigger accidental retrieval of old intentions—especially when old and new PM cues are similar—leading to more commission errors; (2) the intention overwriting hypothesis (Walser et al., 2012), suggesting that forming a new intention overwrites components of the old intention, decreasing its activation and spontaneous retrieval. Empirical findings are mixed and may depend on cue similarity and focality: Walser et al. (2012, 2017) found higher aftereffects or commission errors when new and old cues were similar (perceptually salient rectangles), whereas Anderson and Einstein (2017) found no increase, possibly due to semantically distinct word cues and inhibition of old cues. Implementation intentions—if-then plans often accompanied by visualization or repetition—typically strengthen cue–action links and improve PM (Chen et al., 2015; Bugg et al., 2013), but can increase commission errors during active phases by promoting spontaneous retrieval. It is unclear whether using an implementation intention to encode a new PM task in the finished phase will reduce commission errors by overwriting old intentions or increase them by reallocating attention, thus motivating the present comparisons.

Design: Single-factor between-subjects design with three groups in the finished phase: non-new PM (ongoing 1-back only), new PM (ongoing 1-back + new PM task), and implementation intention (ongoing 1-back + new PM task encoded via if-then). Participants were randomly assigned (N=213; 71 per group). Analytic plan included chi-square tests for categorical outcomes and ANOVAs for continuous measures (SPSS 20.0). Participants: 213 college students (87 males; Mage=20.17, SD=1.36, range 18–25). Recruitment continued until at least 66 participants committed commission errors (66 ultimately did). All provided informed consent; compensation 15 yuan. Ethical approval: Henan Provincial Key Laboratory of Psychology and Behavior (20210508017). Materials and tasks: Ongoing task was a 1-back letter comparison using 24 uppercase English letters excluding F and J (these keys used for responses). Stimulus letters approx. 2 cm×2 cm (1.43° visual angle). Response mapping: press J for “same” and F for “different.” PM responses used the spacebar. PM cues in active phase were letter G and R (two occurrences each). In the finished phase, old PM cues (G, R) reappeared as ongoing stimuli in all groups; new PM cues for groups with a new PM task were Q and W (four occurrences each). Same vs different trials ratio was 1:2; cues did not repeat consecutively. Stimuli were pseudo-randomized with equal letter frequencies. Procedure: Implemented in E-Prime 2.0. Participants first received 1-back instructions and completed 50 practice trials; only those with >60% practice accuracy proceeded. Trial structure: fixation “+” 500 ms; letter up to 2000 ms (terminated by response), followed by 500 ms blank. Phase 1 (PM active): instructions to press spacebar when G or R appeared while performing the 1-back. 160 ongoing trials with 4 PM-cue occurrences (G and R twice each). Phase 2 (PM finished): all participants were told the Phase 1 PM task was completed and no longer needed. - Non-new PM group: 1-back only; old cues (G, R) reappeared as ongoing stimuli. - New PM group: in addition to 1-back, press spacebar for new PM cues Q or W; old cues still appeared as ongoing stimuli. - Implementation intention group: same new PM task as the new PM group, but encoded with if-then instructions: “If Q or W appears, then press the spacebar directly without executing the letter comparison task,” and participants repeated the plan aloud three times. Finished phase contained 320 ongoing trials with 8 old-cue occurrences (4 G, 4 R) and 8 new-cue occurrences (4 Q, 4 W for groups with new PM). At the end, participants described both old and new PM tasks to verify memory for instructions. Measures: - Commission errors: pressing spacebar to old (completed) PM cues during the finished phase. - Commission error rate (proportion of commission errors among old-cue trials), analyzed for participants who committed at least one commission error. - Ongoing task performance: accuracy and reaction times (RTs). - New PM task performance in groups with new PM: accuracy and RTs. Statistical analyses: chi-square tests for proportions (at least one commission error; number with only one error), one-way ANOVAs for commission error rate (among error-makers), ongoing task metrics, RT to old PM cues, and new PM performance (between new PM vs implementation intention).

- Proportion with ≥1 commission error: Non-new PM 31/71; New PM 20/71; Implementation intention 15/71. Overall chi-square: χ²(2, N=213)=8.83, p<0.05, φ=0.21. Pairwise (Bonferroni): Non-new vs New, ns (p>0.05); New vs Implementation intention, ns (p>0.05); Non-new vs Implementation intention: χ²(1, N=142)=8.36, p<0.01, φ=0.24. - Among participants who committed at least one commission error, proportion with exactly one error did not differ across groups (p>0.05). Within those who erred, 13% (non-new), 20% (new PM), and 40% (implementation intention) made exactly one error. - Commission error rate (among error-makers): ANOVA F(2,63)=8.47, p=0.001, ηp²=0.21. Means (M±SD) from Table 1: Non-new PM 0.36±0.15; New PM 0.25±0.09; Implementation intention 0.23±0.10. Pairwise: New PM < Non-new PM (p<0.01); Implementation intention < Non-new PM (p<0.001). - RT to old PM cues in finished phase: no significant group differences (p>0.05). - Ongoing task performance (finished phase): accuracy and RT showed no significant differences among groups (ps>0.05). - New PM task performance (between New PM and Implementation intention groups): Accuracy higher with implementation intention, F(1,142)=23.64, p<0.001, ηp²=0.14; RT to new PM cues did not differ (p>0.05).

The pattern of results supports the intention overwriting hypothesis: forming a new PM intention reduced commission error rates relative to performing only the ongoing task, and strengthening the new intention with implementation intention encoding further reduced commission errors. These effects suggest that new intentions can overwrite or interfere with the representation of completed intentions (cue, response, and their association), decreasing spontaneous retrieval of old intentions. The absence of group differences in ongoing task accuracy/RT and RT to old cues suggests that attention allocation to the ongoing task was not detectably altered by new PM tasks under the present conditions. Differences from prior studies that reported increased aftereffects under new PM conditions (e.g., Walser et al., 2012; 2017) likely reflect cue similarity and focality: here, old and new cues were distinct focal letters (less confusable), whereas prior work used perceptually similar non-focal cues, which may increase accidental retrieval of old intentions via a shared task set. Additional mechanisms discussed include interference and retrieval-induced forgetting: retrieving new intentions may suppress or accelerate forgetting of old intention contents, particularly when new intentions are more recently encoded and thus more activated. Overall, findings indicate that forming and strengthening a new intention can mitigate persistence of completed intentions without measurable costs to ongoing task performance in this setup, clarifying conditions under which commission errors can be reduced.

Forming a new PM intention in the finished phase reduced commission error rates compared with continuing the ongoing task alone, and encoding the new intention via implementation intentions further reduced errors and improved new PM accuracy. These results favor the intention overwriting account over attentional dependence when old and new cues are distinct and focal. Practically, adopting encoding strategies such as if-then plans when creating new intentions may help minimize commission errors from completed intentions. Future research should test different cue focalities and similarities, introduce delays between active and finished phases (as in real life), vary ongoing task demands to better index attention allocation, and examine boundary conditions where overwriting may fail or yield null effects.

- Attention measures: No differences in ongoing task performance or RT to old cues were observed, but the 1-back task may have been too simple and PM cues were focal, possibly masking attention reallocations. - Cue properties: Results may depend on the distinctiveness/focality of old vs new cues; higher similarity may produce different outcomes (e.g., increased errors). - Design specifics: No manipulation of delays between active and finished phases; real-life PM often involves time gaps that reduce intention activation and could alter overwriting effects. - Statistical reporting: While the overall chi-square indicated differences across groups, pairwise differences were limited; floor effects may constrain detection of differences in some conditions. - Generalizability: Young adult sample; findings may not generalize to other age groups or clinical populations.