Group Therapy for Major Depression: Comparing Expectation-Focused and Reward-Focused Psychotherapy Approaches

Major depressive disorder (MDD) remains a leading cause of disability and often shows incomplete response or relapse after established treatments. A promising strategy is to move beyond broad, symptom-focused interventions toward therapies that directly target the psychological and neurobiological mechanisms maintaining depression. This trial evaluates two mechanism-based, group-delivered psychotherapies that each target a core process implicated in MDD: maladaptive expectation processing and reduced reward sensitivity.

Rationale and mechanistic framework. Expectation pathway. Predictive coding accounts propose that the brain continuously updates prior beliefs in light of new information. In depression, negative expectations about the self, others, and the future often become rigid and resistant to change. Cognitive immunization-discounting or reframing disconfirming evidence-prevents adaptive belief updating, maintains symptoms, and can weaken responses to conventional cognitive-behavioral interventions. Targeting dysfunctional expectations directly-via structured behavioral experiments that produce positive expectation violations and procedures that reduce cognitive immunization-may restore flexibility in belief revision and enhance treatment effects.

Reward pathway. Depressed patients frequently show reduced reward sensitivity, including blunted anticipation of reward and impaired reinforcement learning, processes linked to mesolimbic dopaminergic circuitry (e.g., ventral striatum/nucleus accumbens). These deficits can limit the benefit of nonspecific activity scheduling, which increases exposure to potentially rewarding situations but does not specifically remediate anticipatory and learning-related components of reward processing. Interventions that systematically enhance reward sensitivity and strengthen reinforcement contingencies may therefore improve motivational drive and positive affect.

Neurobiological component. Expectation updating has been associated with activity in the dorsolateral prefrontal cortex (DLPFC) and anterior insula, whereas reward processing consistently engages the ventral striatum and insular cortex. This trial includes a functional MRI (fMRI) module to probe these mechanisms pre- and post-intervention. Participants perform established paradigms tapping threat/expectation learning, future-directed thinking, and social reward/expectation. Linking neural responses to clinical change will clarify how mechanism-based therapies exert their effects and may inform biomarkers for treatment selection and monitoring.

Study design overview. This is a randomized, controlled, superiority trial with three parallel groups: (1) EFPI, (2) REACT, and (3) waiting-list control. The trial follows a 3 × 5 repeated-measures design with planned assessments at baseline (pre-treatment), post-treatment (after 5 weeks), and 3- and 6-month follow-ups. Intervention groups complete all follow-ups; after the 6-month assessment they may initiate further treatment if desired. Waiting-list participants complete baseline and the 3-month follow-up before being ethically permitted to access other treatments. They continue to receive follow-up questionnaires at later time points, including items on additional treatments, to enable sensitivity analyses.

Randomization is performed at the individual level with a 1:1:1 allocation ratio. Allocation is managed through secure electronic data capture to ensure concealment until assignment. Given the nature of group psychotherapy, participants and therapists are not blinded; however, outcome assessors remain blind to group allocation, and standardized procedures are applied for clinician-rated measures. Because treatment is delivered in groups, potential therapist- and group-related effects are monitored and considered analytically.

Interventions (concise). EFPI and REACT are manualized group interventions delivered in 10 sessions over five weeks (two 100-minute sessions per week) by trained CBT therapists under regular supervision.

• EFPI targets maladaptive expectations and cognitive immunization through psychoeducation; identification/documentation of negative expectations; structured real-life behavioral experiments (predict-test-experience-reflect); guided cognitive integration to promote flexible belief updating; and maintenance strategies.
• REACT targets reduced reward sensitivity through systematic reinforcement procedures: identifying reward deficits and attentional biases; enhancing and prolonging positive experiences (e.g., savoring, imagery); structured engagement in meaningful rewarding activities with explicit reinforcement plans; and relapse-prevention strategies to sustain reward engagement.

Therapist training, supervision, adherence checklists, and standardized materials support fidelity. The waiting-list control receives no active group intervention during the first three months but is offered standard treatment options thereafter.

Assessments and timeline (overview). Clinical and behavioral data are collected at baseline, post-treatment, 3- and 6-month follow-ups in the intervention arms; the waiting-list control is assessed at baseline and 3 months prior to accessing further care. Digital data capture (e.g., REDCap) includes automated range and plausibility checks; follow-ups are facilitated via secure online questionnaires and structured telephone contacts to minimize attrition. In addition to clinician- and self-rated depression severity, the battery includes validated measures of expectation processing, reward sensitivity (anticipatory and learning components), and related cognitive-emotional constructs. Eligibility criteria and outcome measures are specified elsewhere in this record.

Neuroimaging component (overview). An fMRI module is acquired at baseline and post-intervention (or the matched 3-month time point for the waiting-list group) to assay neural correlates of expectation updating and reward processing. Paradigms include: (i) conditioning-based tasks probing threat expectancy and expectation violation; (ii) future-directed thinking tasks eliciting positive/negative prospective/retrospective imagery; and (iii) a social reward/expectation paradigm using structured interactions with virtual partners. Region-of-interest analyses focus on DLPFC/anterior insula (expectation) and ventral striatum/insula (reward); whole-brain and connectivity analyses complement ROI findings. Standard preprocessing (e.g., motion correction, normalization, smoothing) and task-based GLMs with multiple-comparison control are used. Neuroimaging endpoints are related to clinical trajectories to test mechanistic hypotheses.

Adherence, concomitant care, and retention. Participants confirm availability for the 5-week schedule before enrollment; attendance is monitored and automated reminders are sent. During the 5-week intervention phase, participants are asked to maintain stable psychotropic medication and to refrain from initiating additional psychotherapy; deviations are documented. After the intervention, participants may seek external psychotherapy; additional treatments at follow-up are recorded for sensitivity/covariate analyses. Retention is supported through reminders, flexible scheduling, and brief telephone assessments if full questionnaires are declined.

Sample size and operating characteristics. Planned enrollment is N = 150 (~50 per arm), allowing for expected attrition while preserving power to detect group × time effects of practical relevance in linear mixed-effects models. Power considerations were based on moderate effects from prior mechanism-focused psychotherapy literature; primary analyses will use linear mixed models.

Statistical analysis (overview). Primary clinical analyses use linear mixed-effects models with fixed effects for Group, Time, and their interaction, and random intercepts (and where appropriate, random slopes) to account for within-subject dependency and between-subject heterogeneity. Models will consider therapist/group clustering where indicated. Prespecified mediation analyses test whether changes in expectation processing and reward sensitivity account for symptom change; moderation analyses examine whether baseline profiles predict differential benefit. Missing data are handled under missing-at-random assumptions using maximum likelihood, with multiple imputation in sensitivity analyses. Neuroimaging analyses relate pre-to-post changes in activation/connectivity to clinical improvements; ROI-based hypotheses are complemented by exploratory whole-brain analyses with appropriate error control.

Safety monitoring and harms. Given the non-invasive, low-risk nature of group psychotherapy, no independent Data Safety Monitoring Board is planned. Safety is monitored by the investigative team throughout treatment and follow-up. Adverse events (including clinically significant symptom worsening or emergent suicidality) are documented and managed according to predefined procedures, with referral to appropriate care as needed and ethics notification when required. Criteria for discontinuation include persistent non-attendance or need for higher-level care during the intervention phase.

Data management and confidentiality. All data are pseudonymized (unique study IDs); identifying information is stored separately on access-restricted institutional servers. Electronic data capture uses role-based access, audit trails, and automated checks. After completion, personally identifying data are deleted; anonymized datasets are retained per policy and may be shared under GDPR-compliant agreements to support transparency and reproducibility.

Ethical conduct and dissemination. The protocol adheres to SPIRIT recommendations and has received ethics approval from the responsible institutional review board. Written informed consent is obtained from all participants prior to any study procedures. Results will be disseminated via peer-reviewed publications and scientific meetings; de-identified data and analysis code may be shared under controlled access agreements to promote open science.