Improving mental health by training the suppression of unwanted thoughts

Abstract

Anxiety, posttraumatic stress, and depression markedly increased worldwide during the COVID-19 pandemic. People with these conditions experience distressing intrusive thoughts, yet conventional therapies often urge them to avoid suppressing their thoughts because intrusions might rebound in intensity and frequency, worsening the disorders. In contrast, we hypothesized that training thought suppression would improve mental health. One hundred and twenty adults from 16 countries underwent 3 days of online training to suppress either fearful or neutral thoughts. No paradoxical increases in fears occurred. Instead, suppression reduced memory for suppressed fears and rendered them less vivid and anxiety provoking. After training, participants reported less anxiety, negative affect, and depression with the latter benefit persisting at 3 months. Participants high in trait anxiety and pandemic-related posttraumatic stress gained the largest and most durable mental health benefits. These findings challenge century-old wisdom that suppressing thoughts is maladaptive, offering an accessible approach to improving mental health.

Fig. 1.. Experimental design and procedure.

(A) One hundred and twenty participants from five continents participated in online suppression training and a follow-up assessment 3 months later during the COVID-19 pandemic. (B) We randomly assigned participants to suppress their fears (primary intervention; Suppress-Negative, N = 61) or neutral events (control; Suppress-Neutral, N = 59), with half of each group assigned to also imagine positive or neutral future events. (C) Participants first generated 20 fears, 36 neutral events, and 20 hopes (over 2 days) each with a cue, a key detail, a short tag line, and a brief description with more details (images in figure are only for illustration purposes and are all from free-to-use sources: The negative photo is by M. Majnun on Unsplash; the neutral photo is by K. Grabowska on Pexels, and the positive photo is by I. Lesyk on Pexels). They then rated event characteristics and had their mental health assessed; three days of retrieval suppression practice ensued, and each session was composed of 12 No-Imagine and 12 Imagine repetitions in response to No-Imagine and Imagine cues, respectively. No-Imagine cues (appearing in red) required participants to attend to the cue while suppressing retrieval of any imagery or thoughts; Imagine cues (appearing in green) required participants to imagine the event. Immediately after the final training session, we tested memory and affect for generated events and assessed mental health; we repeated these assessments after 3 months. Analyses of event measures focused on changes in memory and affect for each event after training compared to before (post versus pre or follow-up versus pre), as did measures of mental health. Event analyses permit assessment of the effects of imagination (Imagine row) or suppression (No-Imagine row).

Fig. 2.. Changes in fear vividness and emotion after suppression reveal benefits, not ironic rebound.

(A) Suppressing fears reduced their vividness on an immediate test for nearly all participants, irrespective of the initial emotional intensity of those fears or participants’ posttraumatic stress status; horizontal bars are individual participants’ average change in vividness [post–pre; percentage of maximum point (POMP) scores]. Participants are sorted vertically by the average emotional intensity of fears before training from least (bottom bars, light orange background) to most intense (top bars, darker red background); yellow highlights on bars indicate participants with probable PTSD. Across negative and neutral events, vividness reductions were significantly greater for No-Imagine than for Baseline items and did not vary reliably by event valence (mean changes indicated by solid black versus dotted vertical bars; see text for statistics) and are greater for participants whose events are initially more intense (r = 0.28, P = 0.027). (B) Suppressing feared events reduced their affective tone for most participants, irrespective of initial emotional intensity (as assessed on a separate nine-point scale; see the Supplementary Materials) and PTSD status; affect reductions were greater for No-Imagine than for Baseline items (solid black versus dotted vertical bars). (C) In the Suppress-Negative group, the more that suppression training reduced fear vividness, the greater the reduction in perceived fear intensity. (D) Suppressing future events rarely increased vividness or affect after training compared to before training, and the frequency of increases never exceeded those observed for Baseline events that were never suppressed, irrespective of event valence. The relative risk of an adverse event (a suppression-related increase in their fears’ vividness or affective intensity) was near to 1.0 (no increase in risk) for both the Suppress-Negative and Suppress-Neutral conditions [confidence intervals (in brackets) spanning 1.0 reflect nonsignificant changes in risk].

Fig. 3.. Suppressing fears both improves and protects mental health.

(A) Training participants to suppress fears (upper half, red label) significantly reduced worry, depression, and negative affect and increased well-being after training (right-hand bar within each panel) compared to before training (left-hand bar within each panel); training at suppressing neutral events (lower half, blue label) reduced worry and negative affect. Individual participants are indicated by dots connected by white lines; boxes reflect interquartile range, and lines reflect median scores. (B) A PCA on change scores (post − pre) for our four negative and two positive measures yielded a latent variable that revealed more mental health improvement (more negative scores) in the Suppress-Negative than the Suppress-Neutral group; an analysis contrasting participants who imagined positive future events versus neutral events revealed no reliable difference in mental health changes on this latent variable. (C) Suppressing fears generally reduced the chances of an adverse event (a worsening of mental health; i.e., post > pre for negative indices; post < pre for positive indices) compared to suppressing neutral events on nearly every mental health index, significantly so for depression and well-being, which exhibited 57% and 44% decreases in the chances of an adverse event, respectively [confidence intervals (in brackets) spanning 1.0 reflect no change in relative risk; significant reductions are indicated in boldface]. ns, not significant. *P < 0.05; ***P < 0.001.

Fig. 4.. Mental health benefits of suppression training in symptomatic participants.

(A) Higher pretraining trait anxiety scores [Trait portion of the State-Trait Anxiety Inventory Form Y-1 (STAI-Trait)] predicted greater mental health benefits for participants in the Suppress-Negative group on our PCA-derived mental health change latent variable. (B) Trait anxiety did not predict mental health changes in the Suppress-Neutral group. (C) Participants with likely pandemic-related PTSD [Impact of Events Scale–Revised (IES); IES score > 32] showed mental health improvement on both negative mental health indices (state anxiety, depression, worry, negative affect) and positive indices (well-being and positive affect) after training to suppress fears (Suppress-Negative group) compared to participants trained to suppress neutral events (Suppress-Neutral group). Left: POMP scores averaged over the negative scales (top) and positive scales (bottom). Right: Individual participants with likely PTSD in the Suppress-Negative group (left, red) and the Suppress-Neutral group (right, blue), sorted by IES score (least to greatest, from bottom to top) to illustrate that benefits occurred irrespective of PTSD symptom severity. (D) The relative risk of adverse events (a worsening of mental health after training) was never greater after suppressing fears compared to suppressing neutral events whether considering all subjects or subgroups with likely PTSD (IES, >32), substantial anxiety (STAI-Trait, >45), or depression [Beck’s Depression Inventory II (BDI), >19] [confidence intervals (in brackets) that include 1 indicate no increased risk]. *P < 0.05 and †P ~ 0.05.

Fig. 5.. Depression, anxiety, or posttraumatic stress symptoms before training predict mental health benefits of fear suppression immediately after training and after 3 months.

(A) Greater pretraining trait anxiety (maroon bars), pandemic-related posttraumatic stress (yellow bars), and depressive symptoms (dark blue bars) predicted larger improvements in positive mental health indices (left half of the panel) and negative mental health indices (right half of the panel) on our immediate assessment after training (solid line above bars) and 3 months later (dashed line above bars) but only when people were trained to suppress fears (Suppress-Negative) and not neutral events (Suppress-Neutral). Correlations (r values) are plotted on the y axis; significance is indicated by asterisks (*P < 0.05, **P < 0.01, and ***P < 0.001). Positive mental health is a composite of well-being and positive affect (averaged POMP scores); Negative mental health is a composite of the average POMP scores of the negative indices. (B) Subgroups with higher depression symptoms, likely PTSD, or anxiety showed significant improvements on negative mental health indices (darker shades, extending downward) and positive indices (lighter shades, extending upward) during the immediate test (post − pre; bars below solid lines) and the 3-month follow-up (follow-up − pre; bars below dotted lines). Improvements occurred only after suppressing fears (Suppress-Negative), not after suppressing neutral events (Suppress-Neutral). (C) A latent growth curve modeling (LGCM) analysis across our three time points (Pre, Post, and Follow-up) revealed greater improvement in depression after suppressing fears than after suppressing neutral events during training (y axis values are predicated values of depressive symptoms based on the LGCM model).

Fig. 6.. Participants spontaneously suppressed fears over the 3-month delay, concluding that it improved their mood and anxiety.

(A) Most participants in the Suppress-Negative group reported using suppression over the delay both for trained (left side) and novel fears (right side), irrespective of whether pretraining indices suggested good mental health or pandemic-related PTSD, anxiety, or depression. Values reflect the percentage of people claiming to use suppression either Never (pink), Sometimes (light blue), or Often (green). (B) Pretraining mental health symptom severity (x axis, composite based on the average of trait anxiety, BDI, and IES) predicted the frequency with which participants reported using suppression during the delay (left); suppression use predicted increases in mental health benefits (right). (C) Healthy and symptomatic participants reported improved mood and anxiety that they directly attributed to suppression, with the largest improvements in highly symptomatic participants.