Probiotics reduce negative mood over time: the value of daily self-reports in detecting effects

Improving mental health and preventing onset of clinical psychological symptoms is a pressing societal challenge. Emerging research on the microbiome-gut-brain axis shows that gut microbes influence brain development, neurochemistry, immune, endocrine and neural signaling, and behavior. While animal work demonstrates causal roles for the microbiome in anxiety- and depression-like behaviors, human trials of probiotics show mixed results, with stronger effects in clinical populations than in healthy volunteers. This study focuses on whether probiotics affect emotion regulation—a transdiagnostic construct—by using multiple complementary approaches. The authors aim to assess the impact of a multispecies probiotic on mood and emotion regulation in healthy young adults via standard questionnaires, daily mood monitoring to capture temporal change, and cognitive-emotional processing tasks, to provide a comprehensive understanding and address inconsistencies in prior research.

Prior work has established mechanistic pathways by which the gut microbiome may influence the brain, including vagal signaling, immune modulation (e.g., reduced pro-inflammatory cytokines), and endocrine interactions (e.g., effects on cortisol). Animal studies show probiotics can reduce anxiety and depressive-like behaviors, sometimes dependent on an intact vagus nerve, and transplantation of microbiota from depressed humans induces depression-like changes in rodents. Human studies report probiotics can improve symptoms of depression, anxiety and stress in some trials, but findings are inconsistent and not all preclinical probiotic effects translate to humans. Meta-analyses generally support small-to-moderate beneficial effects on stress, anxiety and depressive symptoms, with stronger evidence in clinical populations. In healthy volunteers, neuroimaging studies indicate probiotics can modulate neural signatures of emotion, sometimes only under stress, but translation to subjective mood is unclear. Standard questionnaire-based assessments often yield mixed findings. The authors propose that daily, temporally resolved self-reports and cognitive tasks may detect changes missed by pre-post questionnaires.

Design: Randomised, double-blind, placebo-controlled trial with a 28-day intervention in healthy volunteers. Sample size was planned via G*Power (power=0.9, alpha=0.05, 2 groups, 2 measurement points) based on a prior four-week probiotic study effect size (ηp²=0.115), estimating N=84 (42 per group), targeting 90 to allow for drop-out. Final sample: 88 participants (44 probiotics; 44 placebo). Ethics: Informed consent obtained; protocol approved by Leiden University Psychology Research Ethics Committee (#0311194); double-blinding maintained until after data collection. Participants: Inclusion required no antibiotics/probiotics in past 3 months, BMI 18–30, no medication use other than female hormonal contraceptives. Exclusions: high alcohol (>20 units/week), frequent drug use (>1/month), renal/hepatic/GI disease, CNS trauma/disorder, any psychological/psychiatric diagnosis (past/present), recent diet program or structural diet changes (past 3 months), hypersensitivity/allergy to nutrients (gluten, milk, soy, peanuts). Interventions: Probiotic group received 30 sachets (2 g per sachet; 2.5×10^9 CFU per gram) of Ecologic® Barrier (Winclove Probiotics B.V.), a 9-strain mixture: Bifidobacterium bifidum W23, B. lactis W51 and W52, Lactobacillus acidophilus W37, Levilactobacillus brevis W63, Lacticaseibacillus casei W56, Ligilactobacillus salivarius W24, Lactococcus lactis W19 and W58. Placebo group received 30 sachets of the carrier (maize starch and maltodextrins), matched in color, taste and smell. Participants dissolved one sachet daily in lukewarm water for 4 weeks. Compliance assessed by returned sachet counts. Assessments: Pre- and post-intervention laboratory sessions at the same time of day. Questionnaires included: STAI (state/trait anxiety), PSWQ (worry), PSS (perceived stress), LEIDS-R (depression sensitivity; subscales including risk aversion, rumination), CES-D (depression), PANAS (positive/negative affect), ERS (emotion reactivity), MAIA (interoceptive awareness; subscales incl. not-distracting, trusting), BVAQ (alexithymia; affective/cognitive), BPAQ (aggression). Bowel complaints questionnaire assessed GI symptoms; Bristol Stool Scale (BSS) recorded daily stool form. Emotional processing tasks: E-Prime software used. Emotional dot-probe task assessed attentional bias: pairs of emotional vs neutral faces (500 ms), probes appearing congruent/incongruent; 10 practice + 120 test trials using ADFES stimuli (six expressions: sadness, fear, anger, happiness, surprise, neutral). Reaction times filtered (<200 ms or >2000 ms removed). Bias score: RTneutralprobe − RTemotionalprobe per emotion. Facial expression recognition task (FERT): five basic emotions (happiness, anger, fear, sadness, disgust) with intensities 10–100% in steps of 10%; 12 practice, 204 total stimuli (102 pictures presented twice), two actors (one male, one female) plus neutral. Accuracy recorded per emotion intensity. Daily measures: Electronic daily reminders collected mood and stool characteristics via the same link each day. Mood rated on two 0–100 VAS items: “How much positive feeling do you have today?” and “How much negative feeling do you have today?” BSS captured stool type for past 24 hours. Intercorrelations with baseline questionnaires confirmed expected relationships. Statistics: R 3.2.3 used. For questionnaires and dot-probe RTs, changes (post−pre) computed per group; between-group differences tested using independent two-sample t-tests or Wilcoxon signed-rank tests when normality/homoscedasticity assumptions failed (Shapiro–Wilk, Levene’s test). FERT analyzed with linear mixed-effects models (dependent: percent accuracy; fixed: emotion, intensity, group, session, group×session; random: participant ID). Daily measures (positive mood, negative mood, BSS) analyzed via linear mixed-effects models (dependent: daily measure; fixed: group, time, group×time; random: participant ID), confirmed by bootstrapping. Exploratory responder analysis: per-participant Kendall Tau-b correlation of negative mood over time (more negative indicates greater decrease); pairwise Kendall Tau-b between this coefficient and pre-intervention questionnaire scores computed; associations with P ≤ 0.08 plotted. A linear model predicting negative mood change constructed via AIC-based variable selection; variance inflation factor checked to rule out multicollinearity. Baseline characteristics (means ± SD): Probiotics (n=44; 15 M, 29 F): age 22.3±3.1, BMI 23.4±3.2, PANAS positive 27.8±7.2, PANAS negative 14.5±5.9. Placebo (n=44; 14 M, 30 F): age 22.3±3.1, BMI 23.6±2.7, PANAS positive 29.5±7.1, PANAS negative 15.9±5.4.

• Daily mood: Probiotics reduced negative mood over time, particularly after approximately two weeks. In linear mixed-effects models, the group×time interaction significantly predicted negative mood (coefficient −0.22, SE 0.08, 95% CI −0.38 to −0.06, P=0.009). No significant effects were found for positive mood (group×time coefficient 0.02, P=0.784) or stool consistency (P=0.556). • Questionnaires: Overall, few differences in change scores between groups. PSWQ showed a significant reduction only in the placebo group (P=0.018). In MAIA, the probiotic group scored lower on the not-distracting subscale after intervention compared to little change in placebo (P=0.017). The MAIA trusting subscale showed a trend toward greater increase in the probiotic group (P=0.093). Given multiple comparisons, these differences would not remain significant after correction. No treatment effects on bowel complaints, frequency, or BSS. • Emotional processing: Dot-probe task showed no between-group differences in changes in attentional bias toward emotional faces (P>0.05). In FERT, accuracy increased with emotion intensity (P<0.001) and differed by emotion (P<0.05). There was a main effect of session (participants more accurate post-intervention; P=0.008) and a marginally significant group×session interaction (P<0.05) indicating improved recognition accuracy following probiotics. • Predictors of response: In the probiotic group, six pre-intervention questionnaire scores correlated (P≤0.08) with change in daily negative mood; no such correlations were observed in the placebo group. A linear model identified LEIDS-R risk aversion as the main predictor of greater reduction in negative mood (coefficient −0.026, SE 0.010, 95% CI −0.047 to −0.006, P=0.013).

Despite using a broad battery of validated questionnaires, the study did not detect robust probiotic-associated changes in emotion regulation via pre-post assessments in healthy volunteers. However, daily mood monitoring revealed a clear reduction in negative mood over time for the probiotic group, with divergence from placebo emerging around the two-week mark. This selective reduction in negative mood without a concomitant decrease in positive mood suggests probiotics may mitigate negative affective states rather than globally blunting emotional reactivity. The limited questionnaire-based effects may reflect that standard scales, often focused on higher arousal negative affect or summing diverse items, are less sensitive to subtle mood shifts in non-clinical samples. Emotional processing tasks showed no dot-probe differences but suggested marginal improvements in facial emotion recognition following probiotics, which, together with questionnaire patterns, raises the possibility that probiotics may influence recognition of emotions in self and others. These findings help reconcile prior mixed results in healthy populations: psychotropic effects of probiotics may be present but insufficiently captured by pre-post questionnaires, whereas temporally rich daily self-reports provide greater sensitivity. The observed ~2-week timescale for mood improvement parallels clinical timelines for antidepressants, potentially pointing to shared pathways such as anti-inflammatory effects and vagal signaling. Exploratory analyses suggest individuals with higher baseline risk aversion (LEIDS-R) derive greater benefit, implying that targeted probiotic use could help reduce risk for clinical onset among at-risk individuals. The results underscore the importance of incorporating ecological momentary assessments or frequent mood ratings to capture the dynamics of intervention effects, thereby improving validity over single pre-post comparisons prone to recall and context biases. Extending intervention duration beyond four weeks might reveal larger group differences, aligning with real-world probiotic use.

Daily self-reports demonstrated that a 4-week multispecies probiotic intervention reduced negative mood in healthy adults, with effects emerging after approximately two weeks. Standard pre-post questionnaires and an attentional bias task did not reliably detect changes, although facial emotion recognition accuracy showed marginal improvement in the probiotic group. Exploratory analyses identified higher baseline risk aversion (LEIDS-R) as a predictor of greater mood improvement, suggesting potential for targeting probiotics to individuals at elevated cognitive vulnerability to depression. These findings clarify discrepant results in prior probiotic trials and advocate including daily mood measures in intervention studies assessing mental health outcomes. Future research should examine longer intervention durations, delineate mechanisms (e.g., immune/vagal pathways), and evaluate targeted strategies for at-risk populations.

• Multiple comparisons: The few questionnaire differences (e.g., PSWQ reduction in placebo; MAIA not-distracting decrease in probiotics; trusting trend) would not survive correction for multiple testing, limiting firm conclusions from questionnaire outcomes. • Population: Healthy, young volunteers may show smaller or subtler effects than clinical samples, potentially reducing sensitivity of standard scales. • Duration: A four-week intervention may be insufficient to capture maximal effects; the divergence in negative mood late in the intervention suggests longer durations could yield stronger differences. • Measurement sensitivity: Standard questionnaires may focus on high-arousal negative affect and rely on summed scores that might not detect general mood shifts; daily mood ratings, while sensitive, are subjective and single-item measures. • Emotional processing: Dot-probe showed no effects and FERT improvements were marginal, which may limit interpretation of cognitive-emotional changes. • Mechanisms not directly assessed: The study did not include biological measures (e.g., inflammatory markers, vagal activity, microbiome composition changes) to elucidate mechanisms.