Amplification of Positivity Treatment for Anxiety and Depression: A Randomized Experimental Therapeutics Trial Targeting Social Reward Sensitivity to Enhance Social Connectedness

BACKGROUND
Social disconnection is common and impairing in anxiety and depressive disorders and does not respond sufficiently to available treatments. The positive valence system supports social bond formation and maintenance yet is often hyporesponsive in those with anxiety or depression. We conducted an experimental therapeutics trial testing the hypothesis that targeting positive valence processes through cognitive and behavioral strategies would enhance responsivity to social rewards, a core mechanism underlying social connectedness.


METHODS
N=68 adults endorsing clinically elevated anxiety and/or depression with social impairment were randomized 1:1:1 to 5- (n=23) or 10-sessions (n=22) of Amplification of Positivity (AMP) treatment or waitlist (n=23). Pre- to post-treatment change in striatal activity (primary outcome) during social reward anticipation measured using functional magnetic resonance imaging; reactivity to a social affiliation task (secondary); and self-reported social connectedness (exploratory) were examined. Primary analyses compared AMP (doses combined) vs. waitlist. A second aim compared doses.


RESULTS
AMP engaged the hypothesized treatment target - leading to greater striatal activation during anticipation of social rewards versus waitlist (d=1.01 [95% CI 0.42, 1.61]; largest striatal volume). AMP yielded larger improvements on positive affect and approach behavior during the affiliation task (but not other outcomes), and social connectedness. Larger striatal and social connectedness increases were observed for 5- vs. 10-session AMP (d range=0.08-1.03).


CONCLUSIONS
Teaching people with anxiety or depression strategies to increase positive thoughts, behaviors, and emotions enhances activity in brain regions governing social reward processing and promotes social connectedness. Social reward sensitivity may be a transdiagnostic target for remediating social disconnection.

salience or "wanting") (19).The striatum is centrally involved in this circuit; it reliably engages across various social cues and contexts [e.g., the opportunity to engage in self-disclosure (20); sharing experiences with others (21); and anticipating viewing a smiling face (18)], suggesting its central role in supporting social connection (8).
Hyporesponsivity of the positive valence system characterizes depression (22) and some forms of anxiety (23) [most notably social anxiety disorder (24) and posttraumatic stress disorder (25)] as evidenced by low positive affect (24), diminished approach motivation and behavior (26,27), and reduced activation in mesolimbic circuits during reward processing (25,28), including anticipation of social rewards (29,30).People who are diagnosed with these conditions also experience social anhedonia-loss of interest in pursuing or pleasure in response to social activities-which is associated with greater social impairments (31) even controlling for anxiety and depression severity (32).Notably, research in individuals with anxiety and depression has demonstrated that positive valence processes (e.g., approach motivation, positive affect) are robustly associated with social connectedness beyond any effects of negative valence processes (e.g., avoidance motivation, negative affect) (3,14).Therefore, diminished sensitivity to social rewards may be a transdiagnostic target for improving social connectedness in anxiety and depression.Although it is not the only candidate mechanism underpinning social disconnection (cf.heightened social threat reactivity and avoidance) (3,33), and not everyone who experiences anxiety or depression is characterized by diminished sensitivity to social rewards, it is an underexplored target with the potential to address an unmet treatment need.
First-line treatments do not sufficiently repair positive valence deficits in individuals with anxiety or depression (34)(35)(36), which may explain in part why social impairments persist following established treatments (4,5).Amplification of positivity (AMP) was developed to address this gap (37).It comprises cognitive and behavioral strategies (i.e., positive activity interventions) (38)(39)(40) that target positive valence thoughts, behaviors, and emotions through repeated practices, including noticing and amplifying responsivity to positive events (e.g., savoring, reminiscing, disclosing positive events to others), promoting the experience and expression of gratitude, and engaging in acts of kindness.These strategies are not routinely part of existing evidence-based treatments for anxiety or depression.In nonclinical samples, they have been shown to engage processes that are believed to facilitate social connections.For example, engaging in kind acts (charitable donation) (41) and savoring positive memories (42) activate the striatum.Gratitude has positive dyadic effects: its experience induces social approach behaviors toward one's benefactor (43), while its expression elicits approach behaviors from the benefactor toward the expresser (44).Initial evidence from studies of individuals seeking treatment for anxiety or depression has revealed large increases in positive affect (37) and social connectedness (45) following 10 sessions of AMP compared to those in the waitlist (WL) condition, with sessionby-session improvements in connectedness accounted for by increases in positive affect beyond any effects of reductions in negative affect (45).The current trial builds on this work to establish whether AMP enhances sensitivity to social rewards, a mechanism hypothesized to underpin social connectedness (8).

Current Study
This study was grounded within the National Institute of Mental Health experimental therapeutics pipeline (46) in which engagement of an identified treatment target (i.e., hypothesized mechanism underlying a clinical or functional outcome of interest) must be established before pursuing tests of clinical efficacy.Therefore, we conducted a mechanism-focused, 3-arm, parallel randomized controlled trial of 2 doses of AMP (5 or 10 sessions) versus WL in patients who endorsed clinically elevated anxiety or depression with at least moderate social impairment to test the primary hypothesis that AMP (both doses combined) would be superior to WL in increasing striatal activation during social reward anticipation (the primary measure of target engagement and outcome upon which decisions to further evaluate AMP in future trials was based).Anticipatory processing of social reward cues was chosen as the primary outcome because it sets into motion the chain of events that support the pursuit of valued outcomes, including social connection.Secondary measures of target engagement were evaluated from psychophysiological, behavioral, and subjective responses that were obtained during a standardized social affiliation task (47).
A second aim was to compare effect size differences for 5-versus 10-session regimens to determine whether the treatment target could be engaged more efficiently (46).A small-to-medium (d = 0.4) effect size advantage of the 10session protocol was identified a priori as the criterion to support its future evaluation over the 5-session protocol, pending support for the primary hypothesis.Finally, although this mechanism-focused trial was not powered to test clinical efficacy, we nevertheless explored AMP effects on measures of social functioning, symptoms, and well-being to inform future trials.

Outcome Measures
Primary Outcome-Social Incentive Delay Task.The social incentive delay task (52) reliably activates the striatum (primary region of interest) when people anticipate obtaining social rewards (e.g., viewing a smiling face) (18). 1 During reward blocks, distinct cues indicated whether to anticipate social reward or a neutral outcome.Participants gained social reward if their reaction to the target was on time.The primary outcome was a change in striatal activation during social reward anticipation trials compared with implicit baseline.This contrast was used (cf.anticipation of social reward minus neutral outcomes) because the reliability of difference scores is always lower than the reliability of their individual parts (53), which has been shown to diminish the reliability of task-based functional MRI (54) [see also (55)] and hinder mechanistic research (56).See Supplemental Methods for a full description.
Secondary Outcomes-Social Affiliation Task.Participants completed an 18-minute conversation with a trained same-sex experimental assistant (confederate).Conversation partners alternated responding to questions gradually increasing in intimacy.Different questions and same-sex confederates were used at baseline and posttest.This task reliably induces connectedness between unacquainted partners (47).See Supplemental Methods for details.Secondary measures of social reward sensitivity were participant-reported positive affect following the conversation (57); positive facial expressions during the perceiver (listening) role of the task (11); affiliative (social approach) behavior (13,47); future approach motivation (desire for future interaction) (11,14,58); and respiratory sinus arrhythmia (59) reactivity (difference between the beginning and end of the task) (60,61).See the Supplement for details.

Key
Exploratory Outcomes-Social Connectedness.The primary clinical end point prespecified in this program of work was measured using the NIH Toolbox Friendship and Loneliness surveys (62) and the SCSR (50).
Other Exploratory Outcomes.Surveys assessing different facets of social functioning, positive and negative valence symptoms, functioning, and well-being were used (see Supplemental Methods).

Nonspecific Treatment Measures
Treatment credibility/expectancy, working alliance, and homework completion data were collected for AMP participants.See the Supplement.Treatment Amplification of Positivity.AMP is a manualized, clinician-delivered intervention comprising 3 core elements: 1) increasing exposure and responsiveness to positive events; 2) practicing gratitude; and 3) engaging in kind or generous acts toward others.The 5-and 10-session protocols were identical in the initial 4 treatment strategies (noticing and amplifying positive events; gratitude reflection; acts of kindness; scheduling pleasurable, engaging, and meaningful activities); the 10-session AMP included additional activities targeting the core domains (active/constructive responding; gratitude expression; make someone else happier; live this month like it is your last in your current city).See Table S1 for treatment modules by arm.Following their final in-person session, 5-session AMP participants reviewed their treatment plan with their clinician by phone (30 minutes; week 6) and received weekly e-mails to encourage their continued engagement in treatment activities (weeks 7-10).See Supplemental Methods for treatment adherence scale descriptions.
Waitlist.WL participants completed pre-and postassessments at a 10-week interval.They were offered AMP following the postassessment; however, their treatment data were not included in the analyses.

Procedure
Participants provided informed written consent prior to eligibility screening.Those who met inclusion criteria and agreed to participate completed baseline assessments comprising selfreport surveys and the social affiliation task followed by a separate functional MRI session involving the social incentive delay task.Participants were then randomized 1:1:1 to AMP (5 or 10 sessions) or WL using a randomly permuted block design.Stratification factors were sex assigned at birth and social connectedness (SCSR $ 60 vs. #59).Experimental personnel (e.g., confederates, functional MRI operators) were blinded to treatment assignment, but participants and clinicians were not.Participants completed postassessments after finishing their assigned treatment protocol or approximately 10 weeks after the baseline MRI (WL).Participants received monetary compensation for assessment sessions.Procedures were approved by the university's Human Research Protections Program.

Statistical Analyses
All randomized participants who had baseline data and at least one postbaseline measurement were included in the analysis (modified intent-to-treat).
Analysis of Primary Outcome.Voxelwise activation data within the striatum (Harvard-Oxford anatomical mask including the caudate, putamen, and nucleus accumbens) were entered into a generalized linear model (AFNI: 3dLME) (63) comparing activation across groups (AMP vs. WL) over time (pre, post) to social reward cues during anticipation (any reward vs. implicit baseline).Permutation testing within AFNI's 3dClustSim (63) was used to minimize identification of false-positive activations within the striatum mask (voxelwise a priori probability of .005with corrected clusterwise activation probability of .05).
Parameter estimates were extracted from significant clusters that emerged from the group 3 time analysis within the striatum mask for effect size computation and to visualize treatment-related effects.Data from 4 participants were removed (blinded to treatment assignment) due to poor quality (see the Supplement).
Analysis of Secondary Target Engagement Outcomes.Social affiliation task outcomes were analyzed using a linear mixed-effects model.Independent variables included treatment arm (AMP vs. WL), visit (pre, post), and treatment by visit interaction.
Analysis of Exploratory Social Connectedness, Symptom, and Functioning Outcomes.Our key exploratory outcome was social connectedness, measured using the NIH Toolbox Friendship and Loneliness scales and the SCSR.Additional social functioning, symptom, and wellbeing outcomes are presented in the Supplement.Exploratory outcomes were analyzed as described above for secondary outcomes.There was no correction for multiple comparisons for secondary or exploratory outcomes because this early-phase trial was intended to inform measurement decisions in future work.
Analyses of secondary and exploratory outcomes and effect size computations were conducted using R version 3.6.1 (https://www.r-project.org/).Baseline demographic variables that were both unbalanced at baseline (p , .10) and associated with the outcome (p , .15) of clinical interest were included in the model as covariates.Because all baseline demographic variables were balanced between the arms, they were not included in the model.Effect Size Computation.Cohen's d differences for the change score from baseline were computed for: 1) AMP (both doses combined) versus WL (primary aim), and 2) 5-versus 10session AMP (secondary aim).Effect size and 95% CIs are presented for primary and secondary outcomes and for key social connectedness outcomes.
Baseline Group Comparison.Categorical variables were evaluated using Fisher's exact tests.Continuous variables were analyzed with Wilcoxon's rank-sum tests.
Sample Size Determination.The study was a priori powered assuming 15% attrition and alpha = .05using a twosided, 2-sample t test.Therefore, we planned to enroll 71 subjects to have an evaluable sample size of 60 (40 AMP, 20 WL), which would provide 80% power to detect a standardized change between groups of 78%.

Preliminary Analyses
Participant progress throughout the trial is summarized in Figure 1 (CONSORT [Consolidated Standards for Reporting Trials] diagram).Groups did not differ on baseline characteristics (all ps ..10), except that AMP participants had higher Patient Health Questionnaire scores versus WL participants (p = .037)(see Table 1).There were 37 out of 45 (82%) AMP AMP Target Engagement RCT Biological Psychiatry March 1, 2024; 95:434-443 www.sobp.org/journalcompleters and 21 out of 23 (91%) WL completers.Postassessment data were obtained from 1 AMP participant who discontinued early; all other early discontinuations were lost to follow-up.The 5-and 10-session AMP arms did not differ significantly on treatment credibility (p = .60)or expectancy (p = .81),homework compliance from weeks 2 to 5 (p ..05), or working alliance (all p ..05).Treatment adherence was uniformly high (see the Supplement).Two adverse events-both mild (presence of suicidal ideation without intent)-occurred in the 10-session AMP group and were deemed to be unrelated to the intervention.

Primary Target Engagement Outcomes
The group 3 time linear mixed-effects analysis revealed that AMP participants displayed significantly greater pre-to posttreatment activation increases in several striatal regions, including the left nucleus accumbens, bilateral caudate, and bilateral putamen, compared with WL participants (see Table 2).Figure 2 illustrates activation differences by group within the largest cluster that emerged from this analysis (right putamen, 86 voxels, x = 25, y = 1, z = 2).These results demonstrate that AMP engaged the hypothesized treatment target, i.e., striatal activation anticipation of social rewards.

Other Exploratory Outcomes
Significantly larger AMP versus WL improvements were observed on measures of anxiety, depression, positive and negative affect, functional interference, satisfaction with social   S3 and S4).

AMP Dose Comparison
Brain activation data (% signal change) were extracted from significant group 3 time clusters within the striatum (Table 2), and pre-to posttreatment effect size differences were compared by AMP dose.The 5-versus 10-session protocol evidenced larger increases across all striatal regions (d range = 0.0821.03; 5 .10-session) (Table S5).Similar dose effects were observed across social affiliation task outcomes and key social connectedness exploratory outcomes (see Tables S6  and S7).

DISCUSSION
This mechanism-focused experimental therapeutics trial examined whether AMP increases responsivity to social reward in individuals seeking treatment for anxiety or depression.In support of our primary hypothesis, AMP increased striatal activation during social reward anticipation compared with WL, demonstrating a large group difference on average.To our knowledge, this study is the first to demonstrate psychosocial treatment enhancement of the striatum during social reward processing in people with anxiety or depression, suggesting engagement of a mechanism that supports the drive of humans to connect with others (8).Therefore, AMP may offer a new approach for remediating social disconnection in anxiety and depressive disorders and possibly other psychiatric conditions that are characterized by diminished social reward responsivity (31).AMP engaged several distinct regions within the striatum, including the left nucleus accumbens, bilateral caudate, and bilateral putamen, consistent with meta-analyses suggesting broad striatal engagement when people anticipate possible social rewards (18,64).Subdivisions within the striatum serve distinct but complementary functions in reward processing: the ventral striatum (comprising the nucleus accumbens) codes the value of stimuli, shaping reward expectations and approach motivation, whereas the dorsal striatum (comprising the caudate and putamen) is involved in action selection and motor behavior, guiding pursuit of anticipated reward outcomes (16).Functions served by each subdivision are engaged during the social incentive delay task as well as social connection opportunities more broadly.Although the experimental therapeutics framework necessitates focus on a central treatment target to inform next-step decisions about treatment evaluation (46), social reward processing is multifaceted, involving different phases (e.g., anticipation, responsiveness) and brain regions outside the striatum (e.g., anterior insula, anterior cingulate cortex) (18).Therefore, future work should examine broader neural effects of AMP across different phases of social incentive cue processing.
AMP also increased positive affect and social approach behavior during a dyadic affiliation task-factors that have been consistently linked to positive social outcomes, including partner liking and desire for future interaction-compared with the WL (medium-to-large effect size differences) (9,(13)(14)(15).However, groups did not differ on change in respiratory sinus arrhythmia reactivity (medium-sized group difference; AMP .WL), positive facial expressions, or desire for future interaction (small group differences).It is unclear whether AMP strategies impact those outcomes less directly or whether features of the paradigm (e.g., different conversation partners at pre-and postassessment) reduced sensitivity to detecting change.For example, desire for future interaction is likely influenced by multiple factors, including general motivational tendencies (e.g., approach vs. avoidance), specific qualities of the interaction partner, subjective affect following the encounter (14), and the future social context being considered (e.g., asking for advice vs. becoming friends).Nevertheless, dyadic task outcomes suggest that AMP may influence some positive valence affective and behavioral processes that have been shown to support social connections-complementing the primary neural outcomes.Psychosocial treatment development rarely evaluates the requisite dose or duration of treatment that is needed to exert its effects (46); however, doing so could help develop interventions with potential for maximal clinical impact (65).Accordingly, a second aim of this study involved comparing target engagement across 2 doses of AMP that contained the same core strategies intended to engage positive valence responses.The 5-session protocol showed larger striatal engagement, suggesting that additional treatment exercises included in the 10-session protocol were not needed to further engage the hypothesized treatment target compared with allowing time for participants to practice the core AMP skills.Parsimony may have facilitated deeper learning of core skills with greater opportunities to personalize them (65).It is also possible that the 5-session program induced time scarcity, thereby motivating participants to maximize engagement in the few treatment sessions that they had.Including a measure of target engagement immediately after the core AMP strategies were administered would be necessary to confirm this possibility.Regardless, current findings underscore the value of designing psychosocial trials with dose effects in mind (46,65).Of course, this trial is limited by the way that dose was conceptualized and by comparing only 2 dosing levels.It also remains unknown whether there are patients for whom the 5-versus 10-session doses may be optimal.
This mechanism-focused trial was not powered to test clinical efficacy; however, we evaluated efficacy to inform future trials.Consistent with previous findings (45), AMP led to large improvements in social connectedness compared with WL as reflected on measures of perceived friendship, belongingness, and loneliness.Converging with the primary striatal outcomes, 5-session AMP led to nominally larger improvements in connectedness compared with the 10session protocol.Extending the assessment battery to capture different aspects of social functioning more fully, we found that AMP produced larger increases in satisfaction with social roles and activities and similar but less robust increases in perceived emotional support and satisfaction with discretionary social activities.AMP-related improvements were also observed across measures of positive and negative affect, anxiety and depression symptoms, functional interference, and psychological well-being.Those outcomes converge with earlier findings (37) and other emerging positive affect-targeted approaches in anxiety and depressive disorder samples (66,67), further supporting the value of explicitly targeting positive valence processes in treating these conditions.These findings are encouraging because improvements in positive affect and well-being tend to be smaller and lag behind reductions in clinical symptoms following established treatments (4,5,34).Generalization of AMP effects to negative affect and symptom outcomes is consistent with work demonstrating that positive emotions mitigate negative responses to stressors that often fuel anxiety and depression (38,68,69).The current findings should be interpreted alongside several caveats.This trial was conducted in a small sample in which the primary AMP comparison was against WL.The next step will be to replicate target engagement in a larger sample compared with a control condition accounting for common therapeutic effects.Such sufficiently powered trials should also examine whether changes in responsivity to social reward covary with improvements in social connectedness and for whom AMP may be most useful.The transdiagnostic sample raises the question of whether response to AMP varies by principal diagnosis (e.g., depression vs. anxiety) or symptom dimension (e.g., anhedonia vs. anxious arousal).AMP may be especially efficacious for patients who are experiencing elevated anhedonia [e.g., primarily those diagnosed with major depression but also many patients with social anxiety disorder (70) or posttraumatic stress disorder (71)] compared with those primarily characterized by anxious arousal (e.g., panic disorder).Relatedly, although all participants reported at least moderate social functioning impairments at intake, the mechanisms that underlie those impairments likely vary across individuals.Diminished sensitivity to social rewards may be the central mechanism underpinning social disconnection in some patients, whereas heightened sensitivity to aversive social outcomes may be a more influential mechanism in others (3).Idiographic approaches that consider which target(s), for whom, and under what set of conditions ( 72) may be fruitful to address the multiply determined nature of social disconnection.
The majority female composition is consistent with the epidemiology of anxiety and depressive disorders; however, evidence pointing to sex differences in sensitivity to social reward cues (52) suggests that future research with larger samples should examine whether AMP response varies by gender identity or sex.Measures of the treatment target in the current study were limited by presenting static images of smiling faces (cf.dynamic social cues encountered in real-life) to participants and having them engage in a contrived, circumscribed social context.Measurement of social connectedness and symptom outcomes relied on self-report.It would be valuable to establish whether AMP results in changes in real-world connectedness that translate into improved quality of life and which specific facets of connectedness change or do not change (e.g., network size, participation in social activities and groups, perceived quality).

Conclusions
Social disconnection is common in anxiety and depressive disorders and does not improve sufficiently following first-line treatments.The current findings offer initial support for AMP in enhancing responses to positive valence social cues and contexts.To our knowledge, this study is the first to show psychosocial treatment enhancement of the striatum during social reward processing in people with anxiety or depression, thus suggesting engagement of a mechanism that supports the drive to connect with others.The dose effect finding suggests that core activities focused on increasing exposure and responsivity to positive events, practicing gratitude, and engaging in kind acts may be sufficient to enhance striatal sensitivity to social rewards and connectedness.Replicating these effects in larger samples and determining whether striatal engagement accounts for improvements in social connectedness is needed now.

Figure 1 .
Figure 1.CONSORT (Consolidated Standards for Reporting Trials) flow diagram summarizing participants' progress throughout the study.AMP, amplification of positivity; fMRI, functional magnetic resonance imaging; SID, social incentive delay.

Figure 2 .
Figure 2. Change in striatal activation (right putamen, x = 25, y = 1, z = 2) from pre-to posttreatment during social reward anticipation trials of the social incentive delay task.Data reflect parameter estimates extracted from the largest cluster emerging from the group 3 time linear mixed-effects model (voxelwise a priori probability of .005with corrected clusterwise activation probability of .05within the striatum mask).AMP, amplification of positivity.

Table 2 .
Regions of Interest Analysis Within the Striatum Showing Significant Group 3 Time Effects During Social Reward AnticipationResults shown from the linear mixed-effects analysis of group (AMP vs. waitlist) 3 time (pretreatment, posttreatment) on percent signal change for social reward anticipation trials of the social incentive delay task.Change reflects the difference from baseline (posttreatment activation minus pretreatment activation).Effect size (Cohen's d) was computed from the pre-to posttreatment change difference between groups (AMP .waitlist).AMP, amplification of positivity; MNI, Montreal Neurological Institute; NAc, nucleus accumbens.