Skip to main content
NCBI home page
Psychological Medicine logoLink to Psychological Medicine
. 2025 Dec 24;55:e386. doi: 10.1017/S0033291725102535

The effectiveness and neurobiological actions of memory bias modification: a randomized controlled trial

Yuko Hakamata 1,2,5,✉, Shinya Mizukami 2, Shuhei Izawa 3, Mie Matsui 4, Yoshiya Moriguchi 5, Takashi Hanakawa 6, Hiroaki Hori 5, Yusuke Inoue 7, Hirokuni Tagaya 2
PMCID: PMC13058616  PMID: 41436350

Abstract

Background

Substantial evidence supports the efficacy of cognitive bias modification (CBM) for attention and interpretation. However, CBM targeting memory bias (CBM-M) remains underexplored despite its clinical relevance. This study examines the effectiveness and neurobiological mechanisms of CBM-M.

Methods

Fifty-eight individuals with elevated anxious and depressive personality traits (>1 SD) were randomly assigned to either CBM-M or sham training (n = 29 per group) in a parallel, double-blind, randomized controlled trial. The intervention involved eight sessions over 1 month. CBM-M aimed to enhance positive autobiographical memory (AM) recall by focusing on positive and negative words, whereas sham training lacked this enhancement module. Anxiety and depressive traits and symptoms, explicit and implicit memory biases, and AM specificity were assessed. Additionally, intrinsic functional connectivity was measured via functional magnetic resonance imaging, and cortisol levels were assayed via saliva collected at 10 time points across 2 days before and after the intervention.

Results

Both groups showed reduced anxiety and depressive traits from pre- to post-intervention. Compared with sham training, CBM-M specifically reduced stress vulnerability, negative explicit memory bias, and daytime cortisol levels, with a large effect size. Improvement in memory bias correlated with stress vulnerability and cortisol reductions. CBM-M also enhanced amygdala functional connectivity with the anteromedial orbitofrontal cortex in comparison with sham training from pre- to post-intervention.

Conclusions

CBM-M reduced stress vulnerability and elicited neural changes in amygdala–anteromedial orbitofrontal cortex interactions, which were involved in social reward and AM recall. Future research should identify the most responsive populations and elucidate underlying mechanisms.

Keywords: amygdala, anxiety, autobiographical memory, cortisol, depression, ventromedial prefrontal cortex, orbitofrontal cortex, anterior cingulate cortex, hippocampus, self-referent memory bias, social reward, stress

Introduction

Cognitive bias toward negative information has been linked to the development and exacerbation of mental disorders, such as anxiety and depression (Beck & Clark, 1997; Mathews & MacLeod, 2005). This led to the development of cognitive bias modification (CBM), aimed at reducing cognitive bias to alleviate these conditions. Meta-analyses have shown the effectiveness of CBMs for attention and interpretation (ABM and CBM-I, respectively) (e.g. Fodor et al., 2020; Gober, Lazarov, & Bar-Haim, 2021; Hakamata et al., 2010; Mogoaşe, David, & Koster, 2014, but see Gober et al., 2021, for a review). However, CBM for memory bias (CBM-M), targeting the encoding and retrieval of negative information, remains underexplored despite its clinical importance (Gotlib & Joormann, 2010).

Several pioneering studies have examined the effectiveness of CBM-M to enhance positive memories. Vrijsen, Hertel, and Becker (2016) showed that a single-session CBM-M, which facilitates the recall of positive words after a negative one, maintained a higher positive mood 1 week later compared to control groups (non-training and sham training) in unselected participants. CBM-M also resulted in mood improvement in ruminative individuals (Hertel, Maydon, Cottle, & Vrijsen, 2017). A follow-up study found that CBM-M increased training-congruent positive word recall in depressive individuals; however, both groups showed mood improvement and reduced depressive symptoms with no transfer to positive autobiographical memory (AM) (Vrijsen et al., 2019). Following the protocol proposed by Vrijsen et al. (2019), another study found an increase in positive word recall but no changes in depressive symptoms or emotional AM in individuals with mild depression (Atashipour, Momeni, Dolatshahi, & Mirnaseri, 2023).

Focusing on AM recall, Bovy et al. (2022) examined the effectiveness of 6-day CBM-M training to enhance positive memory in dysphoric individuals, compared with sham training for neutral memory and non-training. No group differences were observed, but CBM-M and sham training enhanced positive mood, increased positive-valence memory recall (including AM specificity), and reduced depressive symptoms. They suggested that sham training may have an unspecific effect, and the intervention duration may have been insufficient to reveal differences. Another study without a control group also showed that CBM-M induced positive AM specificity and improved mood repair in patients with major depressive disorder despite no change in depressive symptoms (Arditte Hall, De Raedt, Timpano, & Joormann, 2018). Previous studies showing CBM-M-induced mood improvement facilitated negative word recall in controls (Hertel et al., 2017; Vrijsen et al., 2016), while those showing no effect did neutral word recall (Atashipour et al., 2023; Vrijsen et al., 2019). These findings highlight the need for longer CBM-M interventions to enhance positive AM using a control group focused on negative word recall to assess its impact in a more diverse sample.

Previous research has focused primarily on depression, where individuals with depression recall more negative than neutral/positive stimuli, especially when self-relevant (Duyser et al., 2022; Gerritsen et al., 2012; Gotlib & Joormann, 2010; Hakamata et al., 2022). Depression is also linked to reduced AM specificity, particularly for positive memories (Liu et al., 2013; Ono, Devilly, & Shum, 2016; Van Vreeswijk & De Wilde, 2004). Additionally, both depression and anxiety are associated with negative implicit memory bias (Hakamata et al., 2022; Phillips, Hine, & Thorsteinsson, 2010; Teachman et al., 2019), which can occur transdiagnostically (Duyser et al., 2020) and is linked to amygdala function (Duyser et al., 2022). Consistently, we found that negative memory bias correlates with anxious and depressive traits, explained by amygdala functional connectivity (FC) with the ventromedial prefrontal cortex (vmPFC), including the subgenual anterior cingulate cortex (sgACC), and the interaction between cortisol and norepinephrine metabolites (Hakamata et al., 2022). We have also observed links between negative memory biases, amygdala FC with vmPFC/sgACC, and cortisol in independent samples (Hakamata et al., 2020, 2025). Given the stress-reducing effects of CBM (Hakamata et al., 2010; Hallion & Ruscio, 2011; Krebs et al., 2018; Mogoaşe et al., 2014), it may alter amygdala–vmPFC/sgACC FC and cortisol. Furthermore, examining the effects of anxiety and depression separately could reveal individual characteristics sensitive to CBM-M effects. However, no previous study has comprehensively explored the neurobiological actions and effectiveness of CBM-M.

This study thus investigated the effectiveness of CBM-M in enhancing positive AM recall through negative and positive word memorization over eight sessions in 1 month through a double-blind, randomized controlled trial (RCT) in vulnerable individuals exhibiting high (>1 SD) anxious/depressive personality traits, compared with sham training without the positive AM enhancement module. Individuals not on psychiatric treatment were included to avoid confounding effects from pharmacotherapy and neurocognitive deficits (Semkovska et al., 2019). Outcome measures included anxiety and depressive traits and symptoms, explicit and implicit memory biases, AM specificity, and cortisol levels. Additionally, we explored the neurobiological effects of CBM-M with pre- and post-intervention functional magnetic resonance imaging (fMRI) scans, focusing on amygdala FC. We also preliminarily explored individual personality trait profiles regarding anxiety and depression sensitive to CBM-M effects.

Methods

Participants

This study adhered to the Declaration of Helsinki and national ethical guidelines and was approved by the Kitasato University Medical Ethics Organization (C17-126). The protocol was preregistered in the University Hospital Medical Information Network database (No. 000029031) and was reported according to CONSORT guidelines. Using G*Power 3.1 software (Faul, Erdfelder, Lang, & Buchner, 2007), the sample size was calculated as n = 58 to detect a time×group interaction effect at α = 0.05 (f = 0.22, 1-β = 0.90, correlation among repeated measures 0.5), based on our previous RCT with the same design (Hakamata et al., 2018). The effect size (f = 0.22) was determined from our previous study on anxiety/depressive traits and memory bias (Hakamata et al., 2022) (see the Supplementary Material for additional details).

Inclusion criteria were: [1] Participants with ≥1 SD score on anxious/depressive personality traits of Neuroticism, assessed by the NEO Personality Inventory-Revised (Costa & McCrae, 1992), a major risk factor for depressive and anxiety disorders (Kotov, Gamez, Schmidt, & Watson, 2010; Struijs et al., 2021); [2] Age 18–59 years; [3] No current Axis-I psychiatric disorders or substance abuse history, as defined by DSM-IV; [4] No major medical illnesses; [5] No regular intake of psychotropics, steroids, or opioids; [6] No metal or medical implants; [7] No history of brain injury or trauma with loss of consciousness >10 min; and [8] No excessive caffeine intake (>400 mg/day) (Temple et al., 2017). Japanese participants from Tokyo and surrounding areas were recruited via local magazines and websites between July 2020 and October 2022. Recruitment was ceased upon reaching the required sample size. Of 212 initially recruited, 64 met the inclusion criteria and provided written informed consent. Of these, one participant declined participation and another lost contact after the first visit. A total of 62 participants were randomly assigned (Figure 1).

Figure 1.

Figure 1.

Open in a new tab

CONSORT flow diagram. One participant in CBM-M could not participate in the post-intervention assessment because of the funeral of a first-degree relative. Note: CBM-M, cognitive bias modification for memory.

The current sample was independent of our previous studies (Hakamata et al., 2020, 2021, 2022), except for one that shared 42.5% of the baseline sample (Hakamata et al., 2025). No longitudinal data following the intervention have been reported.

Study procedures

This parallel-group trial used a 1:1 allocation ratio with a double-blind design. Participants were randomly assigned to either the CBM-M (n = 31) or Sham-training (n = 31) program within 2 weeks of the baseline visit (Figure 1). Randomization was stratified by age, sex, NEO anxiety, depressive trait scores, and baseline explicit and implicit memory bias scores. H.T. adaptively finalized the assignment to ensure a balance between both groups in terms of these baseline variables (see the Supplementary Material for additional details).

The programs were web-based. Participants completed the assigned program twice a week for 1 month (eight sessions) in a quiet space at home and at a time when they could concentrate. Completion was not recorded until each participant submitted the final response to ensure their engagement with the program. Participants received an email with the program’s URL, ID, and password from an email account accessible only to an administrator who was not involved in the study. The administrator monitored progress regularly. Two weeks after completing the sessions, participants underwent post-assessment, including MRI scans and saliva collection.

Primary outcomes were NEO anxiety and depressive personality traits. Secondary outcomes included Harm Avoidance (HA) subscales from the Temperament and Character Inventory (TCI) (Cloninger, Svrakic, & Przybeck, 1993), psychological distress (including anxiety/depressive symptoms), explicit and implicit memory bias scores, AM specificity (including emotional valence), and cortisol levels. fMRI scans were conducted pre- and post-intervention to explore the action mechanisms of CBM-M.

CBM-M and sham-training programs

The CBM-M program was based on a word recall task for explicit memory bias (Friedman, Thayer, & Borkovec, 2000) and a word completion task for implicit memory bias (Warrington & Weiskrantz, 1970). It was developed online using JavaScript (https://www.ecma-international.org/) and PHP Hypertext Preprocessor (https://www.php.net/).

As in previous CBM-M programs (Arditte Hall et al., 2018; Hertel et al., 2017; Vrijsen et al., 2016), participants memorized a word list with 11 negative, 11 positive, and six neutral filler words. The same word list was used across all eight sessions, but the words were arranged randomly. Only the CBM-M group completed a module to vividly recall an event in response to a presented positive cue word (see Figure 2). Memory bias scores and mood ratings from each session were analyzed to track changes across the eight sessions, with results reported in the Supplementary Material.

Figure 2.

Figure 2.

Open in a new tab

CBM-M and sham-training programs. The CBM-M group was instructed to vividly recall a personal event that made them feel the positive word presented just before (e.g. recalling a time when they felt “competent”) with no such instruction for negative words. Sham training did not include this personal recall module. A total of 28 words were randomly presented in each session, with the first and last three words being neutral fillers to minimize primacy and recency effects. In each trial, participants first identified the color of a fixation (green or orange) to ensure focus. Subsequently, a word appeared on the screen for 8 s. Upon the disappearance of the word, participants rated its relevance to themselves on a 3-point scale (1: relevant, 2: not relevant, 3: neither). After the word presentation, participants recalled as many words as they could in a free-recall task. No time limit or prompts were provided for the recall. Participants then completed 11-word stems (one at a time) by entering the first word that came to mind. After the task, mood was rated using the Self-Assessment Manikin, on a 9-point scale from 1 (unhappy/uncomfortable) to 9 (happy/comfortable) (Bradley & Lang, 1994). Note: CBM-M, cognitive bias modification for memory.

Psychological assessment

Participants’ demographic information and neurocognitive functions were assessed using the Repeatable Battery for the Assessment of Neuropsychological Status (Matsui, Kasai, & Nagasaki, 2010; Randolph, 1998). All outcome measures demonstrated sufficient reliability and validity. Detailed descriptions of each measure are provided in the Supplementary Material.

Anxiety and depressive personality traits

Anxiety and depressive traits were assessed using the full version of the NEO-PI-R (Costa & McCrae, 1992; Shimonaka, Gondo, & Takayama, 1998). These traits were included in primary and secondary outcome analyses, with the latter examining the effects of baseline personality profiles. As mentioned earlier, the analysis of personality profiles was exploratory, as the study was not designed to specifically test a group×profile interaction. Participants were classified into three profiles: 1) anxiety-predominant (>1 SD on the anxiety trait), 2) depression-predominant (>1 SD on the depressive trait), and mixed (>1 SD on both traits). Additionally, the HA subscales of the TCI short version (Cloninger et al., 1993; Kijima et al., 1996) were assessed, which include anticipatory worry, fear of uncertainty, shyness with strangers, and fatigability and asthenia (Cloninger, 2000; Cloninger et al., 1993).

Psychological distress, including anxiety and depressive symptoms

The Hopkins Symptom Checklist (HSCL) (Derogatis et al., 1974; Nakano, 2016) was used to assess five symptom dimensions experienced in the past week: somatization, obsessive-compulsive, interpersonal sensitivity, anxiety, and depression.

Psychosocial stressor

Stressful life events were assessed using the Life Experiences Survey (Iwamitsu et al., 2008; Sarason, Johnson, & Siegel, 1978). The presence or absence of significant stressor-related negative life changes during the intervention period was considered in cortisol analyses sensitive to these effects.

Behavioral assessment

Experimental tasks were constructed using E-prime version 2.0 for Professionals (Psychology Software Tools, Inc., Pittsburgh, PA). As the explicit and implicit memory bias tasks followed the same procedures as the training, further details are provided in the Supplementary Material and in our previous study (Hakamata et al., 2022). Different word lists were used in the pre- and post-intervention assessments for the explicit memory bias task and the AM test; however, 27.3% of the words (four negative and two positive) used in the training were included in the memory bias task post-intervention.

Explicit memory bias

Explicit memory bias scores were calculated as follows: negative recall ratio (number of self-relevant negative words recalled/ total number of recalled words [22 at maximum] × 100) - positive recall ratio (number of self-relevant positive words recalled/ total number of recalled words×100) (Duyser et al., 2022; Gerritsen et al., 2012; Gotlib et al., 2004; Vrijsen et al., 2017). Positive values indicate greater negative memory bias.

Implicit memory bias

Implicit memory bias was calculated as below: (number of negative words/total number of completed words×100) - the positive word ratio (the number of positive words/total number of completed words×100) (Hakamata et al., 2022). Positive values indicate greater negative memory bias.

AM

The AM test (Williams & Broadbent, 1986) was administered in a self-written format with minimal instructions (Debeer, Hermans, & Raes, 2009). The administration and scoring procedures are detailed in the Supplementary Material and in our previous study (Hakamata et al., 2021). The specificity of recalled memories, defined as an event that occurred in a specific place and concluded within 1 day, was evaluated by two independent raters and calculated as: (number of specific memories/total valid responses [12 at maximum] × 100) (Williams & Broadbent, 1986). Valence scores were similarly calculated using the number of specific memories for each valence. We analyzed each valence score separately, as the AMT does not determine specificity scores by subtracting across valence types (Williams & Broadbent, 1986), unlike other memory bias measures (Everaert et al., 2022). A higher score indicates greater AM specificity.

Physiological assay of cortisol

Given the relevance of cortisol to stress (Wesarg-Menzel, Marheinecke, Staaks, & Engert, 2024), we measured cortisol levels from saliva collected at five time points across two consecutive weekdays: upon awakening (T1), 30 min after awakening (T2), around noon (11:30–12:30 h) (T3), late afternoon (17:30–18:30 h) (T4), and at bedtime (T5). Saliva collection occurred during the week following the pre- and post-intervention assessments. The procedures are detailed in our previous studies (Hakamata et al., 2022, 2025). Participants collected saliva in microtubes with passive drool at home. They recorded the exact collection time, sleep duration, and perceived stress on a customized web-entry form. Female participants were asked to report their menstrual cycle (see Supplementary Material for scoring details). Participants were instructed to avoid alcohol on the collection days and the night prior and to refrain from eating or drinking (except water), exercising, brushing their teeth, showering, or bathing 1 hour before each collection time point.

For the cortisol assay, slowly thawed samples were centrifuged (3000 rpm, G-force = 1710) for 10 min. Cortisol concentration was measured using an enzyme immunoassay (EIA Kit, Salimetrics LLC, USA). Inter- and intra-assay variations were below 11% and 7%, respectively.

Cortisol levels were indexed at each time point, and the area under the curve with respect to ground was calculated for total cortisol output (Pruessner, Kirschbaum, Meinlschmid, & Hellhammer, 2003). Cortisol levels were averaged across the two collection days. Indices that did not follow a normal distribution were square-root transformed.

Acquisition and preprocessing of fMRI data

As in previous studies (Hakamata et al., 2020, 2021, 2022), resting-state fMRI (rs-fMRI) was acquired using a 32-channel phased-array head coil on a 3.0 T scanner (Discovery MR750; GE Healthcare, USA). Preprocessing and FC analysis were performed using the CONN Functional Connectivity Toolbox version 21a (http://www.nitrc.org/projects/conn) with SPM12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12) operated on MATLAB 2022b (https://mathworks.com/products/matlab.html). The default band-pass filter for rs-fMRI data was applied, with a frequency range of 0.008–0.09 Hz. Scanning parameters and preprocessing details are provided in the Supplementary Material.

Data analyses

Primary outcome

As a planned analysis, we examined changes in NEO anxiety and depressive traits using a two-way repeated measures analysis of variance (rANOVA), with the intervention group (CBM-M vs. Sham training) as the between-subject factor and time (pre- vs. post-intervention) as the within-subject factor.

Secondary outcome

HA subscales, psychological distress, memory bias, and AM specificity

We conducted a multivariate analysis of covariance (MANCOVA) for the post-intervention HA subscales, HSCL symptoms, explicit and implicit memory bias scores, and AM specificity scores (including valence) as dependent variables, with group and baseline personality profiles as between-subject factors, controlling for baseline scores (O’Connell et al., 2017). We performed MANCOVA to assess the effectiveness of intercorrelated outcomes instead of the repeated measures model, which can show decreased detection power due to increased error variances by including time and interactions with multiple outcomes on different scales (Armstrong, 2017). As mentioned above, we analyzed baseline personality profiles to explore individual characteristics that were sensitive to CBM-M effects. Considering the small sample size in the analyzed groups, we used Pillai’s Trace, which is robust to the heteroscedasticity in covariance matrices (Olson, 1974; Tabachnick & Fidell, 2019).

Cortisol levels

In participants without hormonal disease or hormonal therapy, MANCOVA was conducted for total cortisol output and cortisol levels at each time point as dependent variables, with the group and the stressor (i.e. the presence/absence of negative life changes during the intervention period) as between-subject factors, controlling for pre-intervention values.

CBM-M-related changes in amygdala FC

The amygdala was anatomically defined using the FSL Harvard-Oxford atlas (http://www.cma.mgh.harvard.edu/fslatlas.html) as the seed region. While our primary focus was on the right amygdala, which has consistently been implicated in memory bias (Hakamata et al., 2020, 2022, 2025), CBM effects were explored bilaterally. As the FSL atlas does not dissect the prefrontal cortex (PFC), we used the sgACC, vmPFC, rectus, and medial and anterior orbitofrontal cortex (OFC) from the Automated Anatomical Labeling brain atlas version 3.0 (Rolls et al., 2020) (Figure 3d). We combined the bilateral anatomical masks of these regions into a single mask and used for small-volume correction (SVC) analysis.

Figure 3.

Figure 3.

Open in a new tab

CBM-M effects on fatigability, explicit memory bias, and daytime cortisol levels, and amygdala connectivity. (a) Fatigability and asthenia score. (b) Explicit memory bias score. (c) Daytime cortisol levels (Time 3). *p < .05, **p < .01. (d) The volumes of interest: subgenual anterior cingulate cortex (dark blue), ventromedial prefrontal cortex (light blue), rectus (light green), and medial and anterior orbitofrontal cortex (brown and orange, respectively). (e) Increased amygdala functional connectivity with amOFC from pre- to post-intervention in the CBM-M group compared with the sham-training group. (f) Scatterplot showing correlations of explicit memory bias with fatigability scores and daytime cortisol levels. Z scores of pre-post changes are displayed (larger values indicate a greater decrease from pre- to post-intervention). The sample size of cortisol-related analyses in (c) and (f) was 48. The effects of baseline scores, age, sex, RBANS total index score, handedness, LES balanced impact scores during the intervention period, invalid word-recall cases, sleep duration, perceived stress, and menstrual status during saliva collection were controlled for. Note: CBM-M, cognitive bias modification for memory; amOFC, anteromedial orbitofrontal cortex; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; LES, Life Experiences Survey.

We performed a two-way rANOVA for amygdala FC, with group as the between-subject factor and time as the within-subject factor. Given that relatively small effect sizes have been observed in social and affective neuroscience (Lieberman & Cunningham, 2009), we employed SVC analysis for the combined mask, applying a family-wise error (FWE)-corrected threshold of p < .05 at the peak level. For other regions, the threshold was set at an FWE-corrected p < .05 throughout the brain. For reference, the Supplementary Material reports the results of amygdala FC with the hippocampus.

Correlations between outcome changes

For the outcomes that showed a significant CBM-related change from the pre- to post-intervention time points, their correlations were explored using the difference (i.e. pre-post) scores, controlling for baseline scores. We did not use multiple testing corrections given the exploratory nature of the analysis; however, we reported the results with the corrections applied as well. Statistical analyses were performed with SPSS 27.0J (IBM, Inc., Tokyo, Japan), with a statistical threshold p < .05 (two-sided) unless otherwise stated. Additional details of the data analyses are provided in the Supplementary Material.

Results

Participant characteristics

Two participants dropped out from each group, resulting in data analysis for 58 participants: CBM-M (n = 29) and sham training (n = 29) (Figure 1). The color-identification ratio during the encoding phase was 98.7 ± 0.4%, indicating that the participants maintained proper focus on the program. No participants reported potential harm from the programs. Table 1 describes the participants’ characteristics. There were no significant baseline differences in demographics or outcome measures between groups (all p ≥ .20) (Table 2).

Table 1.

Participant baseline characteristics in CBM-M and sham-training groups

CBM-M (n = 29) Sham training (n = 29) Test statistics p-Value
Age (mean, SD) 35.8 12.1 37.0 13.8 0.36 0.72
Sex (female, %) 51.2 48.3 0.07 0.79
Handedness (EHI; mean, SD) 65.2 54.9 80.7 35.3 1.28 0.21
Years of education (mean, SD) 15.5 1.7 15.3 2.0 0.44 0.66
RBANS total index (mean, SD) 102.8 16.1 100.7 12.0 0.55 0.59
       Immediate memory 106.3 13.5 101.6 14.7 1.28 0.20
       Visuospatial construction 100.2 11.9 98.4 10.1 0.63 0.53
       Language 99.1 14.6 96.2 12.6 0.82 0.42
       Attention 99.6 12.6 102.4 11.5 0.89 0.38
       Delayed memory 104.0 15.5 103.4 10.5 0.17 0.87
Personality trait profile (%)
       Anxiety-predominant 17.2 13.8 0.14 0.93
       Depression-predominant 17.2 17.2
       Mixed 65.5 69.0
Open in a new tab

Note: χ2 test was applied to sex and personality trait profiles. All other test statistics were derived from t-tests with 56 degrees of freedom. Absolute t values are shown.

CBM-M, cognitive bias modification for memory; SD, standard deviation; EHI, Edinburgh Handedness Inventory; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status.

Table 2.

Pre- and post-intervention outcome measures in CBM-M and sham-training groups

Outcome measure Pre-intervention Group difference pre-intervention Post-intervention Overall change from pre-to post-interventiona CBM-M effects on change from pre- to post-interventionb
CBM-M (n = 29) Sham training (n = 29) t (56) p CBM-M (n = 29) Sham training(n = 29) t (57) p F η2 p
Mean SD Mean SD Mean SD Mean SD
Primary outcome
NEO-PI-R
Anxiety trait 23.6 3.4 23.4 3.9 0.22 0.83 22.0 4.3 21.9 4.3 3.9 < 0.001 0.01 0.00 0.93
Depressive trait 22.0 4.7 22.7 4.2 0.59 0.56 20.1 5.4 19.4 5.7 5.0 < 0.001 1.86 0.03 0.18
Secondary outcome
TCI
Anticipatory worry 6.3 2.9 6.4 2.8 0.18 0.85 5.9 2.8 6.2 3.0 1.2 0.3 0.06 0.00 0.80
Fear of uncertainty 5.5 1.3 6.4 2.8 1.33 0.19 5.7 1.4 5.5 1.5 1.1 0.3 2.03 0.06 0.16
Shyness 4.6 2.4 5.6 1.8 1.79 0.08 5.0 2.1 5.6 2.0 1.1 0.3 1.55 0.05 0.22
Fatigability and asthenia 4.7 2.3 4.7 2.2 0.12 0.91 4.3 2.5 5.2 2.2 0.2 0.9 6.06 0.16 0.019
Symptoms
Anxiety 12.3 4.1 12.5 4.7 0.15 0.88 10.6 2.8 10.5 5.2 3.9 < 0.001 0.06 0.00 0.81
Depression 22.3 6.9 22.3 6.1 0.00 1.00 19.6 5.6 18.8 6.8 4.9 < 0.001 0.01 0.00 0.93
Somatization 20.7 4.7 21.0 6.8 0.20 0.84 19.3 3.2 18.5 6.8 3.5 0.001 0.36 0.01 0.56
Obsessive-compulsive 19.7 6.0 20.1 6.1 0.24 0.81 16.8 4.2 16.1 6.8 5.6 < 0.001 0.00 0.00 0.96
Interpersonal sensitivity 18.6 5.5 19.3 5.5 0.55 0.58 14.9 4.5 15.2 6.9 6.7 < 0.001 0.15 0.01 0.70
Memory bias
Explicit memory bias scoresc 2.8 40.3 4.7 34.6 0.19 0.85 −16.0 23.9 −9.7 24.5 2.8 0.006 4.95 0.13 0.033
Implicit memory bias scores 8.5 23.2 8.9 24.2 0.07 0.95 −13.3 27.7 −5.1 29.3 4.4 < 0.001 3.84 0.10 0.059
Autobiographical memory (% of total valid responses)
Specific memory 48.8 23.8 50.8 23.1 0.32 0.75 46.0 21.4 50.2 25.1 0.6 0.58 0.46 0.01 0.50
Negative 16.5 11.0 16.8 12.2 0.10 0.92 13.0 10.9 15.3 10.0 1.7 0.09 0.56 0.02 0.46
Neutral 13.8 11.7 18.6 12.3 1.51 0.14 22.7 17.7 24.0 16.7 3.1 0.003 0.00 0.00 0.98
Positive 18.2 10.1 15.5 12.6 0.92 0.36 10.4 11.3 9.8 8.8 4.9 < 0.001 0.06 0.00 0.80
Cortisol (nmol/L)d
Total cortisol output (AUCg) 85.7 27.7 99.2 49.4 1.18 0.26 82.0 17.6 97.4 36.4 0.7 0.47 1.37 0.06 0.20
     Time 1 (at awakening) 8.8 2.9 8.6 6.1 0.10 0.92 8.0 3.2 8.3 3.5 0.8 0.41 0.01 0.00 0.95
     Time 2 (30 min after awakening) 13.2 5.9 14.7 6.8 0.82 0.42 12.4 4.2 13.8 4.6 1.1 0.30 0.00 0.00 0.95
     Time 3 (12:00) 4.8 2.0 5.7 3.1 1.22 0.23 4.4 1.8 6.0 2.4 0.1 0.99 10.59 0.27 0.003
     Time 4 (18:00) 3.0 1.6 3.4 2.5 0.67 0.51 3.3 1.6 3.6 1.8 0.9 0.35 0.61 0.02 0.44
     Time 5 (bedtime) 2.1 1.0 2.4 1.8 0.70 0.49 2.0 0.9 3.3 3.5 1.1 0.28 2.01 0.07 0.17
Amygdala connectivity (β values)
with amOFC −0.02 0.13 0.04 0.10 1.73 0.09 0.06 0.15 −0.04 0.11 0.1 0.90 16.8 0.24 < 0.001
Open in a new tab

Note: Effect size: η2 ≥ 0.01 (small), ≥ 0.06 (medium), ≥ 0.14 (large).

CBM-memory, cognitive bias modification for memory; SD, standard deviation; NEO-PI-R, Revised NEO Personality Inventory; HSCL, Hopkins Symptoms Checklist; AUCg, area under the curve with respect to ground; TCI, Temperament, and Character Inventory; HA, Harm Avoidance; amOFC, anteromedial orbitofrontal cortex.

a

The overall changes from pre- to post-intervention across groups were based on the results of the paired t-test. For cortisol indices, the degrees of freedom were 47.

b

Results of rANOVA and MANCOVA are reported for primary and secondary outcomes.

c

Based on self-relevant words from previous studies (e.g. Duyser et al., 2022; Gerritsen et al., 2012; Gotlib et al., 2004; Vrijsen et al., 2017). Self-relevance rating scores did not differ significantly between the CBM-M group (0.13 ± 0.30) and the sham-training group (0.18 ± 0.27) at baseline: t(56) = −0.66, p = 0.51. Note: for this statistical comparison and interpretation, the original coding values for self-relevance (1: relevant, 2: not relevant, 3: neither) were recoded as follows: 1: relevant; 0: neither; and −1: not relevant.

d

A total of 48 participants were included in the cortisol analysis (25 in the CBM-M group and 23 in the sham-training group). Ten participants were excluded due to the following reasons: Graves’ disease (n = 1), menopause (n = 4), cessation of menstruation due to hormone intake (n = 1), pre-intervention cortisol levels below the detection limit on both saliva collection days (n = 2), delivery at room temperature (n = 1), and non-submission of post-intervention saliva samples (n = 1). None of the 48 participants were using contraceptives or working shift hours. The degree of freedom for the independent t-test for baseline differences was 46. Absolute t values are shown.

Regarding behavioral data, invalid cases (see Supplementary Table S1) were controlled for as a binary variable in related analyses to prevent sample attrition. For cortisol measures, 10 participants were excluded due to conditions affecting cortisol secretion or invalid saliva samples (see notes in Table 2), leaving 48 participants. The CBM-M (n = 25) and sham-training (n = 23) groups did not differ in demographics or any outcome measures (Supplementary Table S2). Cortisol measures were square-root transformed to address non-normality (Supplementary Material).

Changes in mood and bias across sessions are reported in the Supplementary Material and in Supplementary Tables S3 and S4.

CBM-M effect on primary outcome

Two-way rANOVA revealed a significant main effect of time for anxiety and depressive traits: F1,56 = 14.97, p < .001 and F1,56 = 25.22, p < .001, respectively (Table 2; Supplementary Figure S1a,b). However, the time×group interaction was not significant: F1,56 = 0.01, p = .93 and F1,56 = 1.86, p = .18, indicating that both groups showed a reduction in anxiety and depressive traits from pre- to post-intervention.

CBM-M effect on secondary outcomes

HA subscales, psychological distress, memory bias, and AM specificity

MANCOVA showed significant effects for the intervention group (F15,19 = 2.91, η2 = 0.70, p = .015) and the group×personality profile interaction (F30,40 = 2.39, η2 = 0.64, p = .005) (Supplementary Figure S2a), despite the nonsignificant main effect of profile (F30,40 = 1.02, η2 = 0.29, p = .48). Follow-up tests revealed reductions in fatigability (F1,57 = 6.06, η2 = 0.16, p = .019; Figure 3a) and explicit memory bias (F1,57 = 4.95, η2 = 0.13, p = .033; Figure 3b) in the CBM-M group. The implicit memory bias also decreased, but this difference was not significant (F1,57 = 3.84, η2 = 0.10, p = .059) (Table 2).

Regarding the group×profile interaction, which was parsed into a single main effect, a significant group difference was found in positive specific memory scores: F1,57 = 8.05, η2 = 0.33, p = .001. Post hoc analysis revealed that anxiety-predominant individuals in the CBM-M group maintained higher positive specific memory scores compared to the sham-training group (Mean ± SD: 18.6 ± 10.8 > 3.66 ± 5.1, p = .007, respectively), while depression-predominant individuals showed a reduction (Mean ± SD: 5.0 ± 7.5 < 13.2 ± 7.8, p = .017). No differences were observed in mixed-type individuals (p = .99) (Supplementary Figure S2b), implying the beneficial effects of CBM-M on anxiety-predominant, but not depression-predominant, individuals in positive AM specificity.

Cortisol levels

MANCOVA for cortisol levels revealed a significant main effect of the intervention group (F6,23 = 2.81, η2 = 0.42, p = .033), despite the nonsignificant effects for the stressor (F6,23 = 1.11, p = .39) and group×stressor interaction (F6,23 = 0.58, p = .74). A follow-up univariate test showed reduced daytime cortisol levels at noon (T3) (F1,28 = 10.59, η2 = 0.27, p = .003; Table 2; Figure 3c) in the CBM-M group compared to the sham-training group. No significant differences were found for total cortisol output or cortisol levels at other time points (all p > .17).

CBM-M-related changes in amygdala FC

A two-way rANOVA revealed significantly increased right amygdala FC with the anteromedial OFC (amOFC) in the CBM-M group compared to the sham-training group from pre- to post-intervention (MNI coordinate: 18, 50, −14, T = 4.37, FWE-corrected p = .045, 256 mm3; Figure 3e). The right amygdala FC with the left amOFC also showed an increasing trend (MNI: −20, 52, −14, FWE-corrected p = .094, T = 4.10, 56 mm3). No significant interaction effects were found in other brain regions, including the hippocampus, with either side of the amygdala (Supplementary Table S5).

Relationships between changed outcome variables

Exploratory correlation analysis of the pre-post difference scores for fatigability, explicit memory bias, daytime (T3) cortisol levels, and amygdala–amOFC FC revealed significant correlations between explicit memory bias and fatigability (r = 0.37, p = .011) and between explicit memory bias and daytime cortisol levels (r = 0.49, p < .001). These scores still passed the threshold for significance after Bonferroni correction (i.e. p < .013; Figure 3f and Supplementary Table S6). No significant correlations were observed between the amygdala–amOFC FC and fatigability (r = −0.03, p = 0.85); memory bias (r = 0.10, p = 0.49); or cortisol levels (r = 0.19, p = 0.21) (Supplementary Table S6).

Discussion

This study found a reduction in anxiety and depressive traits from pre- to post-intervention in both groups. However, CBM-M specifically reduced fatigability, explicit memory bias, and daytime cortisol levels compared with sham training. Baseline personality profiles predicted differential responses to CBM-M in positive AM specificity. CBM-M also increased amygdala–amOFC connectivity compared with sham training. The improvement in explicit memory bias was correlated with fatigability and cortisol levels.

Contrary to expectations, anxiety and depressive traits decreased in both CBM-M and sham-training groups. A similar decrease was observed in most secondary outcomes, except for the HA subscales and cortisol levels, regardless of group (Table 2). As shown in the Supplementary Material and Supplementary Table S3, both groups shifted from negative to positive mood by the final session. Positive words were generally recalled more than negative words across groups, although the CBM-M group recalled more positive words than the sham-training group. These results align with previous studies showing increased positive mood and reduced depressive symptoms in CBM-M and sham-training groups (Bovy et al., 2022; Vrijsen et al., 2019), suggesting a potential effect of the sham training. Both groups memorized negative and positive words. Given the increase in positive, but not negative, word recall across groups, repetitive exposure to positive words may have helped the sham-training group develop a preference for them, leading to a general reduction in outcome variables.

In contrast, CBM-M significantly decreased fatigability scores, explicit negative memory bias, and daytime cortisol levels compared with sham training, with generally large effect sizes. Specifically, in the CBM-M group but not in the sham-training group, negative memory bias scores decreased with successive sessions, and this reduction correlated significantly with decreased explicit memory bias scores after the intervention (Supplementary Table S4), indicating that CBM-M ameliorated self-relevant negative memory bias. Consistent with these findings, our previous RCT observed improvements in fatigability after ABM (Hakamata et al., 2018), highlighting the sensitivity of this scale to CBM effects. Fatigability scores reflect a tendency to tire easily, requiring additional rest or slower recovery from stress, fatigue, or illness (Cloninger, 2000; Cloninger et al., 1993), thus potentially indicating high stress vulnerability. Importantly, the reduction in memory bias was associated with decreases in both daytime cortisol levels and fatigability scores. Elevated daytime cortisol levels have been linked to a family history of mood disorders (Ellenbogen et al., 2006; Ostiguy, Ellenbogen, Walker, Walker, & Hodgins, Ostiguy et al., 2011) and adverse childhood experiences (Franz et al., 2011), both of which increase the risk of developing mood disorders (Scott, Smith, & Ellis, 2010). Higher daytime cortisol levels also predict the onset of mood disorders in vulnerable individuals (Ellenbogen, Hodgins, Linnen, & Ostiguy, 2011). These findings suggest that CBM-M may have a preventive effect on stress-related disorders by mitigating stress vulnerability.

We expected an increase in positive AM specificity, given that the current CBM-M was designed to enhance the vivid recall of positive AM. However, positive AM specificity generally decreased, while neutral AM specificity increased from pre- to post-intervention across the groups despite no change in the overall specific memory score (Table 2). Previous studies have shown that elevated cortisol levels specifically impair neutral AM specificity, both following hydrocortisone administration (Wingenfeld et al., 2013) and psychosocial stress (Tollenaar, Elzinga, Spinhoven, & Everaerd, 2009). A general reduction in psychological stress (i.e. distress) might have facilitated AM recall from a more objective, neutral perspective. However, this possibility warrants further investigation.

Further, CBM-M, compared with sham training, maintained positive AM specificity in the anxiety-predominant type while decreasing it in the depression-predominant type. The mixed type generally maintained positive AM specificity, with no significant group difference observed. These profile-related findings should be interpreted cautiously given their preliminary nature; however, it should be noted that low positive AM specificity is commonly associated with depression (Liu et al., 2013; Ono et al., 2016). Depressive individuals may fail to recall positive memories in response to positive cues. This does not contradict clinical observations that depressive individuals tend to remember negative, but not positive, aspects of experienced events. Indeed, similar interventional techniques, such as using positive data logs to record positive evidence against negative self-schemas, are the most challenging to implement and require carefully guided support by clinicians because depressive individuals can easily discount and distort positive aspects of experienced events (Padesky, 1994). In addition, as the encoding of positive information is disrupted in depression (Dillon & Pizzagalli, 2018), individuals with depression may struggle to construct positive memories during retrieval (Hakamata et al., 2025). Therefore, although these findings require further investigation, CBM-M, which enhances positive AM recall, should be applied cautiously in depressive individuals without professional support, particularly in those who have persistent negative self-schemas not rooted in anxiety. For such individuals, enhanced encoding of positive stimuli without deep self-engagement might be beneficial; however, this possibility requires thorough investigation with larger samples.

The amygdala plays a critical role in emotional memory formation (Dillon & Pizzagalli, 2018; LaBar & Cabeza, 2006) and negative memory bias (Duyser et al., 2022). Compared with sham training, CBM-M increased right amygdala FC with the amPFC from pre- to post-intervention. This is consistent with previous findings in which we observed right amygdala FC with the vmPFC and sgACC in different memory bias types (Hakamata et al., 2020, 2022, 2025). The vmPFC is involved in reward processing, decision-making, emotion regulation, and social cognition (Hiser & Koenigs, 2018). The OFC generally processes rewards; however, its medial part specifically processes social reward receipt rather than its reward anticipation (Diekhof, Kaps, Falkai, & Gruber, 2012; Martins et al., 2021; Solomonov et al., 2023). Additionally, AM retrieval activates the medial PFC along with the hippocampus (Shepardson, Dahlgren, & Hamann, 2023) but engages the OFC and amygdala for the successful encoding and retrieval of emotional episodic memory (Dahlgren, Ferris, & Hamann, 2020). Patients with depression and lower positive AM specificity show stronger amygdala FC with the amPFC during positive AM retrieval, suggesting greater effortful engagement of these regions (Young, Siegle, Bodurka, & Drevets, 2016). These findings emphasize the importance of neural interactions between the amygdala and medial PFC, including the amOFC, in positive AM recall, potentially related to social reward receipt. However, the observed FC changes did not correlate with changes in outcome variables. Although the enhanced amygdala–amOFC FC presumably reflected a positive AM increase, several other pathways might have contributed to reductions in these stress-related markers. The neurobiological mechanisms underlying the CBM-M’s stress-reducing effect require further comprehensive investigation in future large-scale research.

This study had limitations that merit consideration. First, the sample size was small. We calculated the sample size based on the ‘as in GPower 3.0’ option in the G*Power software, assuming the correlation among repeated measures (i.e. pre- and post-outcome scores) to be 0.5, following our previous RCT (Hakamata et al., 2018). Although practical and usable in RCTs (Arad et al., 2023), this method was not as standard as Cohen’s recommended method, which robustly detects an effect of any designated magnitude with no assumption of correlation among repeated measures. In light of this, the current sample size was not sufficient for detecting small or medium effect sizes in the analyses of this study. Additionally, the effect size for sample size calculation (f = 0.22) was obtained from the correlation between memory bias (i.e. measured using the word recall and completion tasks employed in the current CBM-M) and anxiety/depressive traits (Hakamata et al., 2022), due to the limited research on CBM-M. Despite the comparability between the effect sizes of cognitive bias and its modification on anxiety-/depression-related measures (e.g. Bar-Haim et al., 2007; Hakamata et al., 2010; Linetzky, Pergamin-Hight, Pine, & Bar-Haim, 2015; Mogoaşe et al., 2014), applying this effect size to the assumed effect size for a repeated measures design was not a straightforward approach. These points require consideration, and the results of this study should thus be interpreted carefully. Second, the hypothesis on changes in the amygdala FC with the vmPFC/sgACC was not pre-registered. We previously observed that FC between the brain regions associated with attentional bias was altered by ABM (Hakamata et al., 2016; 2018), and this observation led us to expect that the amygdala–vmPFC/sgACC FC associated with memory bias could be altered by CBM-M. However, the hypothesis setting and SVC application of the vmPFC/sgACC should have been pre-registered, and the obtained results should be interpreted with great caution. Third, 27% of the words used in the training overlapped with those in the post-intervention experiment, including four negative and two positive words. Repetitive exposure to these words could influence word recall performance post-intervention. However, this effect is likely negligible, as negative memory bias decreased following training despite the larger number of negative words, which could have otherwise increased negative bias. Additionally, this study did not include a waiting list control group. Both groups showed reduced anxiety and depressive traits, and participants’ expectations may have influenced the results. However, CBM-M specifically improved several outcomes, including cortisol, an objective marker. Even with participants’ expectations, a key component of placebo effects (Huneke et al., 2025), it was unlikely to specifically affect the outcome differences observed in this study. Finally, although we used Pillai’s Trace, which is robust to the heteroscedasticity of covariance matrices (Olson, 1974; Tabachnick & Fidell, 2019), examining personality profile effects was preliminary and would have required a larger sample, as mentioned above. Future research with a larger sample should clearly identify individuals who benefit most from CBM-M.

Conclusion

CBM-M, which enhances positive AM recall through negative and positive word memorization, and sham training, which involves memorizing negative and neutral words without the positive AM recall module, both reduced anxiety and depressive traits from pre- to post-intervention. However, compared with sham training, CBM-M specifically mitigated stress vulnerability, reduced explicit memory bias (i.e. the tendency to preferentially recall self-relevant negative information), and lowered daytime cortisol levels. The magnitude of improvement in memory bias correlated with stress vulnerability and cortisol levels. Additionally, CBM-M increased amygdala FC with the amOFC, a region involved in social reward-associated AM retrieval. These findings provide the first evidence of its effectiveness in reducing stress and its underlying neurobiological mechanisms. CBM-M may help prevent the future development of anxiety and depressive symptoms in individuals at high risk through its stress-reducing effects. Future research should explore the specific individuals who benefit most from CBM-M and the mechanisms involved.

Supporting information

Hakamata et al. supplementary material

Hakamata et al. supplementary material

Acknowledgments

The authors would like to thank all the study participants; Ms. Yasuko Omura for her assistance with project office administration; Ms. Fuyuko Yoshida and Ms. Noriko Fukuzato for cortisol level assays; and Mr. Masato Suzuki, Mr. Takahiro Mizoguchi, Mr. Kanta Okuzumi, Ms. Ayaka Ichikawa, Ms. Miyu Onoue, Ms. Yuri Nohara, Ms. Momoka Haginoya, and Ms. Sakura Wada for their data entry and research assistance.

Supplementary material

The supplementary material for this article can be found at http://doi.org/10.1017/S0033291725102535.

Funding statement

This study was supported by a Grant-in-Aid for Scientific Research (B) (No. 18H01094 to Y.H.) from the Japanese Society for the Promotion of Science, a research grant from the Uehara Memorial Foundation (to Y.H.), and research grants from the Multilayered Stress Diseases (JPMXP1323015483), Science Tokyo (to H.H.). The funding sources did not influence any aspect of the work, including design, data collection, analysis, interpretation, writing, or submission.

Competing interests

The authors declare none.

References

  1. Arad, G., Azriel, O., Pine, D. S., Lazarov, A., Sol, O., Weiser, M., & Bar-Haim, Y. (2023). Attention bias modification treatment versus a selective serotonin reuptake inhibitor or waiting list control for social anxiety disorder: A randomized clinical trial. The American Journal of Psychiatry, 180(5), 357–366. 10.1176/appi.ajp.20220533. [DOI] [PubMed] [Google Scholar]
  2. Arditte Hall, K. A., De Raedt, R., Timpano, K. R., & Joormann, J. (2018). Positive memory enhancement training for individuals with major depressive disorder. Cognitive Behaviour Therapy, 47(2), 155–168. 10.1080/16506073.2017.1364291. [DOI] [PubMed] [Google Scholar]
  3. Armstrong, R. A. (2017). Recommendations for analysis of repeated-measures designs: Testing and correcting for sphericity and use of manova and mixed model analysis. Ophthalmic & Physiological Optics, 37(5), 585–593. 10.1111/opo.12399. [DOI] [PubMed] [Google Scholar]
  4. Atashipour, H. S., Momeni, F., Dolatshahi, B., & Mirnaseri, M. S. (2023). Effectiveness of memory bias modification in reducing depression and rumination symptoms and autobiographical memory bias: A pilot study. Frontiers in Psychology, 14, 1145259. 10.3389/fpsyg.2023.1145259. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M., & van IJzendoorn, M. (2007). Threat-related attentional bias in anxious and nonanxious individuals: A meta-analytic study. Psychological Bulletin, 133(1), 1–24. 10.1037/0033-2909.133.1.1. [DOI] [PubMed] [Google Scholar]
  6. Beck, A., & Clark, D. (1997). An information processing model of anxiety: Automatic and strategic processes. Behaviour Research and Therapy, 35(1), 49–58. 10.1016/S0005-7967(96)00069-1. [DOI] [PubMed] [Google Scholar]
  7. Bovy, L., Ikani, N., van de Kraats, L. N. M., Dresler, M., Tendolkar, I., & Vrijsen, J. N. (2022). The effects of daily autobiographical memory training on memory bias, mood and stress resilience in dysphoric individuals. Scientific Reports, 12(1), 20873. 10.1038/s41598-022-25379-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bradley, M. M., & Lang, P. J. (1994). Measuring emotion: The self-assessment manikin and the semantic differential. Journal of Behavior Therapy and Experimental Psychiatry, 25(1), 49–59. 10.1016/0005-7916(94)90063-9. [DOI] [PubMed] [Google Scholar]
  9. Cloninger, C. R. (2000). A practical way to diagnosis personality disorder: A proposal. Journal of Personality Disorders, 14(2), 99–108. 10.1521/pedi.2000.14.2.99. [DOI] [PubMed] [Google Scholar]
  10. Cloninger, C. R., Svrakic, D., & Przybeck, T. (1993). A psychobiological model of temperament and character. Archives of General Psychiatry, 50(12), 975–990. 10.1001/archpsyc.1993.01820240059008. [DOI] [PubMed] [Google Scholar]
  11. Costa, P., & McCrae, R. (1992). Revised NEO personality inventory and five-factor inventory professional manual. Psychological Assessment Resources. [Google Scholar]
  12. Dahlgren, K., Ferris, C., & Hamann, S. (2020). Neural correlates of successful emotional episodic encoding and retrieval: An SDM meta-analysis of neuroimaging studies. Neuropsychologia, 143, 107495. 10.1016/j.neuropsychologia.2020.107495. [DOI] [PubMed] [Google Scholar]
  13. Debeer, E., Hermans, D., & Raes, F. (2009). Associations between components of rumination and autobiographical memory specificity as measured by a minimal instructions autobiographical memory test. Memory, 17(8), 892–903. 10.1080/09658210903376243. [DOI] [PubMed] [Google Scholar]
  14. Derogatis, L. R., Lipman, R. S., Rickels, K., Uhlenhuth, E. H., & Covi, L. (1974). The Hopkins symptom checklist (HSCL): A self-report symptom inventory. Behavioral Science, 19(1), 1–15. 10.1002/bs.3830190102. [DOI] [PubMed] [Google Scholar]
  15. Diekhof, E. K., Kaps, L., Falkai, P., & Gruber, O. (2012). The role of the human ventral striatum and the medial orbitofrontal cortex in the representation of reward magnitude - an activation likelihood estimation meta-analysis of neuroimaging studies of passive reward expectancy and outcome processing. Neuropsychologia, 50(7), 1252–1266. 10.1016/j.neuropsychologia.2012.02.007. [DOI] [PubMed] [Google Scholar]
  16. Dillon, D. G., & Pizzagalli, D. A. (2018). Mechanisms of memory disruption in depression. Trends in Neurosciences, 41(3), 137–149. 10.1016/j.tins.2017.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Duyser, F. A., van Eijndhoven, P. F. P., Bergman, M. A., Collard, R. M., Schene, A. H., Tendolkar, I., & Vrijsen, J. N. (2020). Negative memory bias as a transdiagnostic cognitive marker for depression symptom severity. Journal of Affective Disorders, 274, 1165–1172. 10.1016/j.jad.2020.05.156. [DOI] [PubMed] [Google Scholar]
  18. Duyser, F. A., Vrijsen, J. N., van Oort, J., Collard, R. M., Schene, A. H., Tendolkar, I., & van Eijndhoven, P. F. (2022). Amygdala sensitivity for negative information as a neural marker for negative memory bias across psychiatric diagnoses. Psychiatry Research: Neuroimaging, 323, 111481. 10.1016/j.pscychresns.2022.111481. [DOI] [PubMed] [Google Scholar]
  19. Ellenbogen, M. A., Hodgins, S., Linnen, A. M., & Ostiguy, C. S. (2011). Elevated daytime cortisol levels: A biomarker of subsequent major affective disorder? Journal of Affective Disorders, 132(1–2), 265–269. 10.1016/j.jad.2011.01.007. [DOI] [PubMed] [Google Scholar]
  20. Ellenbogen, M. A., Hodgins, S., Walker, C. D., Couture, S., & Adam, S. (2006). Daytime cortisol and stress reactivity in the offspring of parents with bipolar disorder. Psychoneuroendocrinology, 31(10), 1164–1180. 10.1016/j.psyneuen.2006.08.004. [DOI] [PubMed] [Google Scholar]
  21. Everaert, J., Vrijsen, J., Martin-Willett, R., van de Kraats, L., & Joormann, J. (2022). A meta-analytic review of the relationship between explicit memory bias and depression: Depression features an explicit memory bias that persists beyond a depressive episode. Psychological Bulletin, 148, 435–463. [Google Scholar]
  22. Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39(2), 175–191. 10.3758/bf03193146. [DOI] [PubMed] [Google Scholar]
  23. Fodor, L. A., Georgescu, R., Cuijpers, P., Szamoskozi, Ş., David, D., Furukawa, T. A., & Cristea, I. A. (2020). Efficacy of cognitive bias modification interventions in anxiety and depressive disorders: A systematic review and network meta-analysis. Lancet Psychiatry, 7(6), 506–514. 10.1016/S2215-0366(20)30130-9. [DOI] [PubMed] [Google Scholar]
  24. Franz, C. E., O–Brien, R. C., Hauger, R. L., Mendoza, S. P., Panizzon, M. S., Prom-Wormley, E., & Kremen, W. S. (2011). Cross-sectional and 35-year longitudinal assessment of salivary cortisol and cognitive functioning: The Vietnam era twin study of aging. Psychoneuroendocrinology, 36(7), 1040–1052. 10.1016/j.psyneuen.2011.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Friedman, B., Thayer, J., & Borkovec, T. (2000). Explicit memory bias for threat words in generalized anxiety disorder. Behavior Therapy, 31, 745–746. [Google Scholar]
  26. Gerritsen, L., Rijpkema, M., van Oostrom, I., Buitelaar, J., Franke, B., Fernández, G., & Tendolkar, I. (2012). Amygdala to hippocampal volume ratio is associated with negative memory bias in healthy subjects. Psychological Medicine, 42(2), 335–343. 10.1017/S003329171100122X. [DOI] [PubMed] [Google Scholar]
  27. Gober, C. D., Lazarov, A., & Bar-Haim, Y. (2021). From cognitive targets to symptom reduction: Overview of attention and interpretation bias modification research. Evidence-Based Mental Health, 24(1), 42–46. 10.1136/ebmental-2020-300216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Gotlib, I., & Joormann, J. (2010). Cognition and depression: Current status and future directions. Annual Reviews in Clinical Psychology, 6, 285–312. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Gotlib, I. H., Kasch, K. L., Traill, S., Joormann, J., Arnow, B. A., & Johnson, S. L. (2004). Coherence and specificity of information-processing biases in depression and social phobia. Journal of Abnormal Psychology, 113(3), 386–398. 10.1037/0021-843X.113.3.386. [DOI] [PubMed] [Google Scholar]
  30. Hakamata, Y., Lissek, S., Bar-Haim, Y., Britton, J., Fox, N., Leibenluft, E., & Pine, D. (2010). Attention bias modification treatment: A meta-analysis toward the establishment of novel treatment for anxiety. Biological Psychiatry, 68(11), 982–990. 10.1016/j.biopsych.2010.07.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Hakamata, Y., Mizukami, S., Izawa, S., Hori, H., Matsui, M., Moriguchi, Y., & Tagaya, H. (2025). Contextual memory bias in emotional events: Neurobiological correlates and depression risk. Psychoneuroendocrinology, 171, 107218. 10.1016/j.psyneuen.2024.107218. [DOI] [PubMed] [Google Scholar]
  32. Hakamata, Y., Mizukami, S., Izawa, S., Moriguchi, Y., Hori, H., Kim, Y., & Tagaya, H. (2020). Basolateral amygdala connectivity with Subgenual anterior cingulate cortex represents enhanced fear-related memory encoding in anxious humans. Biological Psychiatry. Cognitive Neuroscience and Neuroimaging, 5(3), 301–310. 10.1016/j.bpsc.2019.11.008. [DOI] [PubMed] [Google Scholar]
  33. Hakamata, Y., Mizukami, S., Izawa, S., Moriguchi, Y., Hori, H., Matsumoto, N., & Tagaya, H. (2021). Childhood trauma affects autobiographical memory deficits through basal cortisol and prefrontal-extrastriate functional connectivity. Psychoneuroendocrinology, 127, 105172. 10.1016/j.psyneuen.2021.105172. [DOI] [PubMed] [Google Scholar]
  34. Hakamata, Y., Mizukami, S., Izawa, S., Okamura, H., Mihara, K., Marusak, H., & Tagaya, H. (2022). Implicit and explicit emotional memory recall in anxiety and depression: Role of basolateral amygdala and cortisol-norepinephrine interaction. Psychoneuroendocrinology, 136, 105598. 10.1016/j.psyneuen.2021.105598. [DOI] [PubMed] [Google Scholar]
  35. Hakamata, Y., Mizukami, S., Komi, S., Sato, E., Moriguchi, Y., Motomura, Y., Maruo, K., Izawa, S., Kim, Y., Hanakawa, T., Inoue, Y., Tagaya, H. (2018). Attentional bias modification alters intrinsic functional network of attentional control: A randomized controlled trial. Journal of Affective Disorders, 238, 472–481. 10.1016/j.jad.2018.06.018. [DOI] [PubMed] [Google Scholar]
  36. Hakamata, Y., Sato, E., Komi, S., Moriguchi, Y., Izawa, S., Murayama, N., Hanakawa, T., Inoue, Y., Tagaya, H. (2016). The functional activity and effective connectivity of pulvinar are modulated by individual differences in threat-related attentional bias. Scientific Reports, 6, 34777. 10.1038/srep34777. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Hallion, L. S., & Ruscio, A. M. (2011). A meta-analysis of the effect of cognitive bias modification on anxiety and depression. Psychological Bulletin, 137(6), 940–958. 10.1037/a0024355. [DOI] [PubMed] [Google Scholar]
  38. Hertel, P., Maydon, A., Cottle, J., & Vrijsen, J. (2017). Cognitive bias modification: Retrieval practice to simulate and oppose ruminative memory biases. Clinical Psychological Science, 5, 122–130. [Google Scholar]
  39. Hiser, J., & Koenigs, M. (2018). The multifaceted role of the ventromedial prefrontal cortex in emotion, decision making, social cognition, and psychopathology. Biological Psychiatry, 83(8), 638–647. 10.1016/j.biopsych.2017.10.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Huneke, N. T. M., Fusetto Veronesi, G., Garner, M., Baldwin, D. S., & Cortese, S. (2025). Expectancy effects, failure of blinding integrity, and placebo response in trials of treatments for psychiatric disorders: A narrative review. JAMA Psychiatry, 82(5), 531–538. 10.1001/jamapsychiatry.2025.0085. [DOI] [PubMed] [Google Scholar]
  41. Iwamitsu, Y., Yasuda, H., Kamiya, M., Wada, M., Nakajima, K., Ando, N., & Takemura, K. (2008). Reliability and validity of the Japanese version of the life experiences survey. The Japanese Journal of Stress Sciences, 23, 239–249. [Google Scholar]
  42. Kijima, N., Sato, R., Takeuchi, M., Ono, Y., Genichiro, K., & Kitamura, T. (1996). Cloninger’s seven factor model of temperament and character and Japanese version of temperament and character inventory (TCI). Archives of Psychiatric Diagnostics and Clinical Evaluation., 7(3), 379–399. [Google Scholar]
  43. Kotov, Gamez, W., Schmidt, F., & Watson, D. (2010). Linking “big” personality traits to anxiety, depressive, and substance use disorders: A meta-analysis. Psychological Bulletin, 136(5), 768–821. 10.1037/a0020327. [DOI] [PubMed] [Google Scholar]
  44. Krebs, G., Pile, V., Grant, S., Degli Esposti, M., Montgomery, P., & Lau, J. Y. F. (2018). Research review: Cognitive bias modification of interpretations in youth and its effect on anxiety: A meta-analysis. Journal of Child Psychology and Psychiatry, 59(8), 831–844. 10.1111/jcpp.12809. [DOI] [PubMed] [Google Scholar]
  45. LaBar, K., & Cabeza, R. (2006). Cognitive neuroscience of emotional memory. Nature Reviews. Neuroscience, 7, 54–64. [DOI] [PubMed] [Google Scholar]
  46. Lieberman, M. D., & Cunningham, W. A. (2009). Type I and type II error concerns in fMRI research: Re-balancing the scale. Social Cognitive and Affective Neuroscience, 4(4), 423–428. 10.1093/scan/nsp052. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Linetzky, M., Pergamin-Hight, L., Pine, D. S., & Bar-Haim, Y. (2015). Quantitative evaluation of the clinical efficacy of attention bias modification treatment for anxiety disorders. Depression and Anxiety, 32(6), 383–391. 10.1002/da.22344. [DOI] [PubMed] [Google Scholar]
  48. Liu, X., Li, L., Xiao, J., Yang, J., & Jiang, X. (2013). Abnormalities of autobiographical memory of patients with depressive disorders: A meta-analysis. Psychology and Psychotherapy, 86(4), 353–373. 10.1111/j.2044-8341.2012.02077.x. [DOI] [PubMed] [Google Scholar]
  49. Martins, D., Rademacher, L., Gabay, A. S., Taylor, R., Richey, J. A., Smith, D. V., & Paloyelis, Y. (2021). Mapping social reward and punishment processing in the human brain: A voxel-based meta-analysis of neuroimaging findings using the social incentive delay task. Neuroscience and Biobehavioral Reviews, 122, 1–17. 10.1016/j.neubiorev.2020.12.034. [DOI] [PubMed] [Google Scholar]
  50. Mathews, A., & MacLeod, C. (2005). Cognitive vulnerability to emotional disorders. Annual Review of Clinical Psychology, 1, 167–195. 10.1146/annurev.clinpsy.1.102803.143916. [DOI] [PubMed] [Google Scholar]
  51. Matsui, M., Kasai, Y., & Nagasaki, M. (2010). Reliability and validity of the Japanese version of the repeatable battery for the assessment of neuropsychological status. Toyama Medical Journal (in Japanese), 21, 31–36. [Google Scholar]
  52. Mogoaşe, C., David, D., & Koster, E. H. (2014). Clinical efficacy of attentional bias modification procedures: An updated meta-analysis. Journal of Clinical Psychology, 70(12), 1133–1157. 10.1002/jclp.22081. [DOI] [PubMed] [Google Scholar]
  53. Nakano, K. (2016). Stress management (2nd ed.) Kongo Press. [Google Scholar]
  54. O’Connell, N. S., Dai, L., Jiang, Y., Speiser, J. L., Ward, R., Wei, W., & Gebregziabher, M. (2017). Methods for analysis of pre-post data in clinical research: A comparison of five common methods. Journal of Biometrics & Biostatistics, 8(1), 1–8. 10.4172/2155-6180.1000334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Olson, C. (1974). Comparative robustness of six tests in multivariate analysis of variance. Journal of the American Statistical Association, 69(348), 894–908. [Google Scholar]
  56. Ono, M., Devilly, G. J., & Shum, D. H. (2016). A meta-analytic review of overgeneral memory: The role of trauma history, mood, and the presence of posttraumatic stress disorder. Psychological Trauma, 8(2), 157–164. 10.1037/tra0000027. [DOI] [PubMed] [Google Scholar]
  57. Ostiguy, C. S., Ellenbogen, M. A., Walker, C. D., Walker, E. F., & Hodgins, S. (2011). Sensitivity to stress among the offspring of parents with bipolar disorder: A study of daytime cortisol levels. Psychological Medicine, 41(11), 2447–2457. 10.1017/S0033291711000523. [DOI] [PubMed] [Google Scholar]
  58. Padesky, C. (1994). Schema change processes in cognitive therapy. Clinical Psychology & Psychotherapy, 1(5), 267–278. [Google Scholar]
  59. Phillips, W. J., Hine, D. W., & Thorsteinsson, E. B. (2010). Implicit cognition and depression: A meta-analysis. Clinical Psychology Review, 30(6), 691–709. 10.1016/j.cpr.2010.05.002. [DOI] [PubMed] [Google Scholar]
  60. Pruessner, J. C., Kirschbaum, C., Meinlschmid, G., & Hellhammer, D. H. (2003). Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent change. Psychoneuroendocrinology, 28(7), 916–931. 10.1016/s0306-4530(02)00108-7. [DOI] [PubMed] [Google Scholar]
  61. Randolph, C. (1998). RBANS manual—Repeatable battery for the assessment of neuropsychological status. Psychological Corp (Harcourt). [Google Scholar]
  62. Rolls, E. T., Huang, C. C., Lin, C. P., Feng, J., & Joliot, M. (2020). Automated anatomical labelling atlas 3. NeuroImage, 206, 116189. 10.1016/j.neuroimage.2019.116189. [DOI] [PubMed] [Google Scholar]
  63. Sarason, I. G., Johnson, J. H., & Siegel, J. M. (1978). Assessing the impact of life changes: Development of the life experiences survey. Journal of Consulting and Clinical Psychology, 46(5), 932–946. 10.1037/0022-006x.46.5.932. [DOI] [PubMed] [Google Scholar]
  64. Scott, K. M., Smith, D. R., & Ellis, P. M. (2010). Prospectively ascertained child maltreatment and its association with DSM-IV mental disorders in young adults. Archives of General Psychiatry, 67(7), 712–719. 10.1001/archgenpsychiatry.2010.71. [DOI] [PubMed] [Google Scholar]
  65. Semkovska, M., Quinlivan, L., O’Grady, T., Johnson, R., Collins, A., O’Connor, J., & Gload, T. (2019). Cognitive function following a major depressive episode: A systematic review and meta-analysis. Lancet Psychiatry, 6(10), 851–861. 10.1016/S2215-0366(19)30291-3. [DOI] [PubMed] [Google Scholar]
  66. Shepardson, S., Dahlgren, K., & Hamann, S. (2023). Neural correlates of autobiographical memory retrieval: An SDM neuroimaging meta-analysis. Cortex, 166, 59–79. 10.1016/j.cortex.2023.05.006. [DOI] [PubMed] [Google Scholar]
  67. Shimonaka, J. N., K. Gondo, Y. Takayama, M. (1998). Construction and factorial validity of the Japanese NEO-PI-R. The Japanese Journal of Personality, 6, 138–147. [Google Scholar]
  68. Solomonov, N., Victoria, L. W., Lyons, K., Phan, D. K., Alexopoulos, G. S., Gunning, F. M., & Flückiger, C. (2023). Social reward processing in depressed and healthy individuals across the lifespan: A systematic review and a preliminary coordinate-based meta-analysis of fMRI studies. Behavioural Brain Research, 454, 114632. 10.1016/j.bbr.2023.114632. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Struijs, S. Y., de Jong, P. J., Jeronimus, B. F., van der Does, W., Riese, H., & Spinhoven, P. (2021). Psychological risk factors and the course of depression and anxiety disorders: A review of 15 years NESDA research. Journal of Affective Disorders, 295, 1347–1359. 10.1016/j.jad.2021.08.086. [DOI] [PubMed] [Google Scholar]
  70. Tabachnick, B., & Fidell, L. (2019). Using multivariate statistics (7th ed.) Pearson. [Google Scholar]
  71. Teachman, B. A., Clerkin, E. M., Cunningham, W. A., Dreyer-Oren, S., & Werntz, A. (2019). Implicit cognition and psychopathology: Looking Back and looking forward. Annual Review of Clinical Psychology, 15, 123–148. 10.1146/annurev-clinpsy-050718-095718. [DOI] [PubMed] [Google Scholar]
  72. Temple, J. L., Bernard, C., Lipshultz, S. E., Czachor, J. D., Westphal, J. A., & Mestre, M. A. (2017). The safety of ingested caffeine: A comprehensive review. Frontiers in Psychiatry, 8, 80. 10.3389/fpsyt.2017.00080. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Tollenaar, M. S., Elzinga, B. M., Spinhoven, P., & Everaerd, W. (2009). Autobiographical memory after acute stress in healthy young men. Memory, 17(3), 301–310. 10.1080/09658210802665845. [DOI] [PubMed] [Google Scholar]
  74. Van Vreeswijk, M. F., & De Wilde, E. J. (2004). Autobiographical memory specificity, psychopathology, depressed mood and the use of the autobiographical memory test: A meta-analysis. Behaviour Research and Therapy, 42(6), 731–743. 10.1016/S0005-7967(03)00194-3. [DOI] [PubMed] [Google Scholar]
  75. Vrijsen, J. N., Dainer-Best, J., Witcraft, S. M., Papini, S., Hertel, P., Beevers, C. G., & Smits, J. A. J. (2019). Effect of cognitive bias modification-memory on depressive symptoms and autobiographical memory bias: Two independent studies in high-ruminating and dysphoric samples. Cognition & Emotion, 33(2), 288–304. 10.1080/02699931.2018.1450225. [DOI] [PubMed] [Google Scholar]
  76. Vrijsen, J. N., Hertel, P. T., & Becker, E. S. (2016). Practicing emotionally biased retrieval affects mood and establishes biased recall a week later. Cognitive Therapy and Research, 40(6), 764–773. 10.1007/s10608-016-9789-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Vrijsen, J. N., van Amen, C. T., Koekkoek, B., van Oostrom, I., Schene, A. H., & Tendolkar, I. (2017). Childhood trauma and negative memory bias as shared risk factors for psychopathology and comorbidity in a naturalistic psychiatric patient sample. Brain and Behavior: A Cognitive Neuroscience Perspective, 7(6), e00693. 10.1002/brb3.693. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Warrington, E. K., & Weiskrantz, L. (1970). Amnesic syndrome: Consolidation or retrieval? Nature, 228(5272), 628–630. 10.1038/228628a0. [DOI] [PubMed] [Google Scholar]
  79. Wesarg-Menzel, C., Marheinecke, R., Staaks, J., & Engert, V. (2024). Associations of diurnal cortisol parameters with cortisol stress reactivity and recovery: A systematic review and meta-analysis. Psychoneuroendocrinology, 163, 106976. 10.1016/j.psyneuen.2024.106976. [DOI] [PubMed] [Google Scholar]
  80. Williams, J. M., & Broadbent, K. (1986). Autobiographical memory in suicide attempters. Journal of Abnormal Psychology, 95(2), 144–149. 10.1037//0021-843x.95.2.144. [DOI] [PubMed] [Google Scholar]
  81. Wingenfeld, K., Driessen, M., Schlosser, N., Terfehr, K., Carvalho Fernando, S., & Wolf, O. T. (2013). Cortisol effects on autobiographic memory retrieval in PTSD: An analysis of word valence and time until retrieval. Stress, 16(5), 581–586. 10.3109/10253890.2013.796925. [DOI] [PubMed] [Google Scholar]
  82. Young, K. D., Siegle, G. J., Bodurka, J., & Drevets, W. C. (2016). Amygdala activity during autobiographical memory recall in depressed and vulnerable individuals: Association with symptom severity and autobiographical overgenerality. The American Journal of Psychiatry, 173(1), 78–89. 10.1176/appi.ajp.2015.15010119. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Hakamata et al. supplementary material

Hakamata et al. supplementary material

S0033291725102535sup001.docx (418.9KB, docx)

Articles from Psychological Medicine are provided here courtesy of Cambridge University Press

RESOURCES