Writing to your past-self can make you feel better

The study examines whether and how compassionate letter writing to oneself can improve mood and what factors modulate its effects. The context is the robust tendency toward rosy retrospection and fading affect bias, where pleasant aspects of the past are recalled more readily and unpleasant affect fades faster. Positive autobiographical memory retrieval can act as a reward and buffer stress, particularly when memories are socially connected. Prior work shows that empathy and reframing can make negative memories less negative and positive memories more positive, but many individuals are reluctant to disclose private information, making self-compassionate, private interventions appealing. Writing to one’s future-self has been linked to improved goal pursuit but may induce anxiety about current insufficiencies. The authors hypothesize that (a) writing a compassionate letter to one’s past-self will improve mood more than writing to one’s future-self, (b) individual time attitudes toward past, present, and future (ATI-TA) will moderate mood changes, and (c) focusing on socially shared past experiences will be more beneficial than focusing on experiences had alone.

The paper reviews evidence on rosy retrospection and fading affect bias (e.g., Mitchell et al., 1997; Walker et al., 1997, 2003; Skowronski et al., 2014) showing positive distortions in memory across cultures. Positive autobiographical memories can repair mood and buffer stress (Joormann & Siemer, 2004; Speer & Delgado, 2017, 2020), with socially connected memories particularly effective in dampening cortisol responses. Memory reframing and empathetic listening can alter the affective tone of memories (Speer et al., 2021; Pasupathi & Oldroyd, 2015; Sugimori et al., 2020). Self-compassion interventions, including letter writing, have demonstrated benefits across outcomes such as body satisfaction, shame reduction, smoking reduction, and eating disorders (Kelly et al., 2010; Stern & Engeln, 2018; Kelly & Waring, 2018; Swee et al., 2023; Kotera & Van Gordon, 2021). Writing to future-self can promote future-oriented behaviors and mental health under certain contexts (van Gelder et al., 2013; Rutchick et al., 2018; Chishima et al., 2021; Schippers et al., 2023), but may also increase anxiety by highlighting present shortcomings. Social vs. self-focused autobiographical themes differentially relate to self-worth (Pillemer et al., 2007, 2013). These literatures motivate testing whether compassionate letters to past-self improve mood and whether time attitudes and social focus moderate effects.

Design: Independent measures study with two laboratory experiments on healthy Japanese undergraduates. Experiment 1 used a within-subjects crossover element (each participant wrote both past- and future-self letters one week apart, order counterbalanced) with analyses focused on within-person changes by addressee. Experiment 2 compared between-group instructions focusing writing on a self-alone experience versus a socially shared experience. Study period: May–August 2022 (Experiment 1); September–October 2022 (Experiment 2). Participants: Recruited at Waseda University; inclusion: Japanese nationality, raised in Japan, healthy, no lifetime psychedelic medication; exclusion: currently seeking treatment for mood/mental disorder. Informed consent obtained; ethics approval (Waseda Ethics Review No. 2021-163). Measures: - Demographics: age, sex. - POMS2 short (Japanese; 35 items; subscales: Anger-Hostility [AH], Confusion-Bewilderment [CB], Depression-Dejection [DD], Fatigue-Inertia [FI], Tension-Anxiety [TA], Vigor-Activity [VA]). Items scored 0–4. - Total Mood Disturbance (TMD) = TA + FI + CB + AH + DD – VA; higher scores indicate more negative mood. - Adolescent Time Inventory–Time Attitudes (ATI-TA; Japanese): six subscales—Past Positive/Negative, Present Positive/Negative, Future Positive/Negative—rated 1–5. Net scores also computed as positive minus negative per time frame. Procedures: Experiment 1 (n=150; 54 males, 94 females, 2 no-response; mean age 20.86, SD 2.04): Participants first completed POMS2 short. Randomly assigned to order conditions with balanced sex: (a) past-self → future-self (n=75; 28 male, 47 female; mean age 20.81, SD 2.14), (b) future-self → past-self (n=75; 28 male, 47 female; mean age 20.93, SD 1.95). Task: write a compassionate letter either to self 5 years ago (past-self) or to self 5 years in the future (future-self) as the person who understands them best; typical duration 5–10 min; letters collected. Immediately after writing, POMS2 short repeated. One week later participants returned, completed POMS2, wrote the other-letter type under same instructions, and repeated POMS2. Finally, completed ATI-TA. The 5-year interval was chosen to map onto distinct life stages. Experiment 2 (n=62; not in Exp 1; two groups of 31 each; both groups 11 males, 20 females): All wrote a compassionate letter to their past-self (5 years ago). Random assignment to focus: self-alone group (focus on something experienced alone; mean age 20.77, SD 2.01) or social connection group (focus on something experienced with someone else; mean age 20.13, SD 1.77). POMS2 short before and after writing. Sample size planning: G*Power assuming effect size 0.3, α=0.05: Experiment 1 power 0.95 → target per group 56; Experiment 2 power 0.80 → target per group 32. Data analysis: - Compute change scores: [after letter POMS2/TMD] – [before letter]. Negative TMD change indicates reduced negative mood (improved mood). - Experiment 1: Two-factor ANOVA tested order (future→past vs past→future) and gender, and addressee (past vs future); main analyses used ANOVA on [after–before] for TMD and POMS2 subscales with within-subject factor addressee (past vs future) and subscale (TA, FI, CB, AH, DD, VA). Chi-square tests compared counts of participants categorized as more positive/no change/more negative after each letter type. ATI-TA analyses: ANOVAs with between factor mood change group (more positive, no change, more negative) and within factors time (past, present, future) and mood score valence (positive, negative); also ANOVAs on ATI-TA net scores (positive–negative). Text analysis with KH Coder quantified frequency of words describing others in past-self letters; hypothesis testing for difference in proportions between mood-change groups. - Experiment 2: ANOVA on TMD change with between factor focus (self-alone vs social connection); chi-square for mood-change categories; 2×6 ANOVA on POMS2 subscale change; t test comparisons for subscales where applicable. Significance threshold p<0.05. Software: JASP; KH Coder for text analysis.

Experiment 1 (past-self vs future-self letters): - No main effects of order or gender on TMD change; significant main effect of addressee (past vs future) (F(1,148)=60.21, MSE=8.51, η²=0.64, p<0.01). - Mood-change categories (Table 1): Past-self letters: 88 more positive, 18 no change, 44 more negative; Future-self letters: 44 more positive, 17 no change, 89 more negative; χ²(2)=29.921, p<0.01. - Mean change: Past-self letters produced negative TMD change (improved mood); Future-self letters produced positive TMD change (worsened mood). - POMS2 subscales: Changes were significantly lower (improved) for past-self than future-self on TA (F(1,149)=80.19, p<0.01), DD (F(1,149)=118.92, p<0.01), and CB (F(1,149)=5.40, p<0.05). VA increased more for past-self (F(1,149)=5.87, p<0.05). No significant differences for AH (F(1,149)=1.48, ns) or FI (F(1,149)=1.62, ns). Reduced TA and DD and increased VA primarily drove mood improvement after writing to past-self. - ATI-TA (past-self letters): ANOVA showed main effect of mood-change group (F(2,147)=4.78, MSE=0.36, p<0.05; no change < more negative) and main effect of mood-score valence (F(1,147)=67.15, MSE=2.69, p<0.01; negative < positive), with a significant interaction (F(2,147)=5.38, MSE=2.69, p<0.01). Positive scores did not differ across mood-change groups (F(2,294)=2.48, p=0.06). Negative scores differed (F(2,294)=8.12, p<0.01; no change < more positive; no change < more negative), suggesting individuals with more negative time attitudes were more emotionally affected by writing to past-self. - ATI-TA net score (past-self letters): Main effect of mood-change group significant (F(2,147)=28.99, MSE=5.39, p<0.01; no change > more negative; no change > more positive). Main effect of time not significant (F(2,294)=1.94, p=0.90); interaction not significant (F(4,294)=3.81, p=1.76). Participants whose mood worsened after writing to past-self tended to have near-zero past-attitude net scores, though not statistically significant across time frames. - ATI-TA (future-self letters): Main effect of mood-change group not significant (F(2,147)=0.14, ns). Main effect of mood-score valence significant (F(1,147)=250.39, MSE=3.02, p<0.01; negative < positive). Interaction significant (F(2,147)=0.13, MSE=3.02, p=0.04), indicating time attitudes relate to mood responses to future-self writing. - ATI-TA net score (future-self letters): No significant effects (mood-change group: F(2,147)=0.04, ns; time: F(2,294)=0.29, ns; interaction: F(4,294)=1.41, ns). - Text analysis (past-self letters): Words describing others were more frequent in letters after which mood improved (63 total; 0.72 per person) than in those after which mood worsened (21 total; 0.48 per person). Difference in proportions significant (α=0.05, g=0.0996, p=0.0125). Experiment 2 (focus: self-alone vs social connection): - TMD change showed a significant main effect of focus (F(1,60)=4.81, MSE=10.51, f=0.28, p<0.05). Mood-change category distribution favored social connection (χ²(2)=13.549, p<0.01). Counts (Table 2): Self-alone: 15 more positive, 5 no change, 11 more negative; Social connection: 28 more positive, 2 no change, 1 more negative. - POMS2 subscales: Interaction not significant (F(5,300)=1.36, MSE=0.78, η²=0.15, ns). Main effect of focus not significant across subscales (F(1,60)=2.42, MSE=1.69, η²=0.20, ns). Main effect of subscale significant (F(1,60)=4.13, MSE=0.78, η²=0.26, p<0.01) with VA > FI, VA > DD, VA > TA. DD decreased significantly more in the social connection group than the self-alone group (t test, p<0.05). Overall, focusing on socially shared experiences produced larger mood improvements.

The findings support the hypothesis that writing compassionate letters to one’s past-self improves mood, while writing to one’s future-self can transiently increase negative mood, potentially by inducing future-oriented anxiety. The subscale profile (reduced tension-anxiety and depression-dejection, increased vigor-activity) explains the TMD improvement with past-focused writing. Time attitudes moderated effects: participants with more negative attitudes toward past, present, and future were more susceptible to emotional change when writing to past-self, whereas those with less negative time attitudes showed little change. The social dimension of recalled memories emerged as a key mechanism: both text analytics and an experiment manipulating focus showed that incorporating social connections enhances mood benefits, notably reducing depression-dejection. These results align with literature indicating social components of positive memories buffer stress and enhance well-being. Conversely, future-self writing may be useful for motivating self-improvement behaviors despite a short-term mood decrease. The study elucidates how target (past vs future) and content focus (social vs self-alone) interact with individual time attitudes to shape the efficacy of self-compassionate writing.

Writing a compassionate letter to one’s past-self transiently elevates mood, especially when reflecting on experiences shared with others (friends, family, acquaintances). Individuals with neutral or positive time attitudes may show minimal mood change, while those with more negative time attitudes are more affected. Writing to one’s future-self can transiently worsen mood, which may nonetheless serve motivational purposes for self-improvement. These insights can inform the design of self-compassionate interventions for everyday use and clinical contexts, emphasizing past-focused and socially connected memories. Future research should test broader age ranges and clinical populations, evaluate repeated-use protocols, and examine behavioral outcomes of future-self writing.

- Sample characteristics: Healthy Japanese undergraduates; results may not generalize to other ages, cultures, or clinical populations. - Sample size: Experiment 2 groups were slightly smaller than the planned 32 per group (actual n=31), potentially reducing power. - Potential confounders: Uncontrolled current life circumstances, mental health status changes, and external events could have influenced mood changes. - Self-report and transient outcomes: POMS2 reflects short-term mood; longer-term effects were not assessed. - Feasibility for clinical populations: Individuals with depression/anxiety may find self-compassion challenging, potentially requiring tailored support or professional guidance.