Effects of positive and negative social feedback on motivation, evaluative learning, and socio-emotional processing

The study addresses whether more naturalistic social feedback (dynamic videos with facial expressions, gestures, and verbal content) enhances motivation, evaluative learning, and socio-emotional processing compared to static photos, and whether these effects differ with age. Prior work shows that anticipating rewards/punishments speeds response times (RTs) but has largely used simple stimuli and young samples. Aging affects learning and emotion (e.g., positivity effect, dopaminergic decline), yet effects of dynamic social feedback on these processes are unclear. The authors hypothesized that (1) positive and negative feedback anticipation would speed RTs in both age groups, more so with videos than photos; (2) evaluative learning would occur such that cues paired with positive feedback acquire positive valence (vs neutral/negative); (3) older adults would show a positivity effect and cue valence would correlate with feedback valence independent of age.

- Anticipation of rewards/punishments accelerates RTs across feedback types (text, photos, videos, money). Prior SID/MID studies: faster RTs for high vs low incentives and for positive/negative vs neutral social feedback. - Dynamic stimuli can be more engaging, but biological motion alone may not enhance motivation; social presence may be critical. - Evaluative learning: neutral stimuli acquire valence through repeated pairing with valent outcomes; prior evidence in conditioning with aversive stimuli and changes in face ratings. - Aging: dopaminergic system degeneration linked to impaired prediction error updating and probabilistic learning; however, SID-related motivational RT benefits can be preserved in older adults despite motor slowing. Socio-emotional aging shows a positivity effect (reduced negative affect, less negative ratings, attention/memory biases for positive, reduced amygdala response to negative), and emotion recognition differences depend on static vs dynamic stimuli. - Socioemotional selectivity theory (SST) posits increased prioritization of emotional goals with age; dynamic expressions may mitigate age declines in emotion recognition.

Design: Two behavioral studies comparing static photo (SID-Photo) vs dynamic video (SID-Video with audio) social feedback within a Social Incentive Delay (SID) task. Between-subjects design across studies; within-subject factor of condition (neutral, positive, negative) in each study. Participants: Combined new and previously published behavioral data. SID-Photo: 52 young (18–32; M±SD=22.8±3.1; 29 female) and 52 older adults (51–75; M±SD=64.0±6.4; 29 female). SID-Video: 49 young (18–35; M±SD=23.9±3.7; 36 female) and 55 older adults (50–84; M±SD=64.2±8.2; 31 female; MoCA mean=27.9, all ≥22). Exclusions for predefined criteria and technical issues were applied per analysis (final ns reported in Results/Stats). Tasks: - Simple RT task (50 trials) to establish baseline RTs for adaptive difficulty in SID. - SID task: Trial phases: cue (predicts neutral/positive/negative), target (respond as fast as possible), feedback. Positive condition: fast responses yield positive feedback; Negative condition: slow responses yield negative feedback; Neutral condition: always neutral feedback. Staircase adjusted difficulty to ~83% hit rate. SID-Photo used photos (10 volunteers; neutral/smiling/angry mimics and gestures) displayed 750 ms. SID-Video used videos (2 actors; neutral/positive/negative mimics, gestures, and verbal feedback) average duration ~2400 ms (SD=400 ms); 16 videos/condition (8 female, 8 male) across six blocks. - Explicit cue valence rating (SID-Video only): extra trials appended; participants rated how they felt about cues on −10 (unpleasant) to +10 (pleasant). - Single-Category Implicit Association Test (SC-IAT; SID-Video only): four blocks (positive/negative congruent vs incongruent), categorizing pleasant/unpleasant words (from BAWL-R) and the learned cues via keypress; computed D-scores (positive indicates positive association; negative indicates negative association). RT trimming (below 350 ms, above 1500 ms), error penalty +400 ms. - Feedback video evaluation (SID-Video): participants rated emotional valence (−10 to +10) and credibility (−10 unbelievable to +10 believable) of feedback videos (each twice; randomized order). Video stimulus set was pre-validated online (n=76) and selected for matched valence/credibility across actors and gestures. Stimuli: Videos recorded with one female (27) and one male (24) actor; gestures included thumbs up/horizontal/down and varied celebratory/other gestures; white t-shirts, gray background. Photos from 10 volunteers matched by expressions/gestures. Procedure: Recruitment via university/community channels. SID-Photo data collected 2016 (young) and 2019 (older); SID-Video collected Oct 2020–Jun 2021 (during COVID-19 restrictions). Standardized instructions; tasks run on similar laptops (SID-Photo with Cogent; SID-Video with MATLAB/Psychtoolbox). Analysis: RTs preprocessed to remove implausible values; condition means computed; for between-study RT comparisons, z-transform within study to account for baseline differences. Permutation-based ANOVAs and t-tests (5,000 permutations) with Bonferroni corrections; effect sizes (η²p, Cohen’s d) reported. Bayesian 2×2×3 mixed ANOVA for RTs (jamovi). Partial Spearman correlations (controlling for age) between cue valence and video valence (Bonferroni-corrected).

- Baseline simple RT: SID-Photo slower than SID-Video; young faster than older; main effect of study F(1,202)=257.9, p=0.0001, η²=0.5608; main effect of age F(1,202)=47.9, p=0.0001; no interaction (p=0.6663). - SID RTs (z-transformed): Significant main effects of age (p=0.0001; young faster) and condition (p=0.0001), no main effect of study and no interactions. Post hoc: Positive vs neutral and Negative vs neutral both faster (p=0.0001); Positive vs Negative not different. Bayesian ANOVA favored age+condition model (BF10=2.44×10^13) and provided evidence against age×condition interaction (BF ratio ≈24.15). - Explicit cue valence (SID-Video): Main effect of condition (p=0.0001); interaction age×condition (p=0.0082); no main effect of age. Ratings: positive cue > neutral and > negative (both p=0.0001); negative vs neutral n.s. Older adults showed smaller differences for Positive−Neutral (p=0.0106) and Positive−Negative (p=0.0150) than young. - Implicit cue valence (SC-IAT): Main effect of condition (p=0.0042); no age effect or interaction. D-scores: Positive cue D>0 (M≈0.15) significantly above zero (p=0.0003); Negative cue D not different from zero; indicating implicit positive association for reward-predicting cues but not negative association for punishment-predicting cues. - Video feedback valence ratings (SID-Video): Main effect of age (p=0.0003), condition (p=0.0001), and interaction (p=0.0003). Post hoc: All conditions differed (Pos>Neu>Neg; all p=0.0001). Older adults rated negative videos less negatively than young (p=0.0003); no group differences for neutral or positive videos. - Video credibility ratings: Significant age×condition interaction (p=0.0001); no main effects. Young vs older: Neutral videos rated more credible by young (p=0.0050); Positive videos more credible by older (p=0.0001); Negative condition no difference. - Evaluative learning correlation (partial, controlling age): Explicit cue valence correlated with positive video valence (r=0.4389, p<0.0001); no correlation in neutral (r=0.0596, p=0.5597) or negative (r=0.0856, p=0.4018). Implicit cue valence showed no correlations with video valence (positive: r=−0.1302, p=0.2012; negative: r_s=0.1144, p=0.2619). Overall: Anticipating social reward or punishment speeds responses across ages; dynamic videos do not confer additional RT benefit over photos. Positive evaluative learning occurs for reward-predictive cues (explicit and implicit), while punishment-predictive cues remain near neutral. Older adults exhibit a positivity bias in ratings of negative social videos.

Findings confirm that anticipating both positive and negative social feedback invigorates behavior (faster RTs) in young and older adults. Contrary to expectations, dynamic video feedback did not further accelerate RT beyond static photos, possibly due to context-dependent value normalization: within each study, relative incentive values may have been scaled to the available options, equalizing motivational impact. Evaluative learning was evident for positive cues: cues paired with positive feedback acquired positive valence (explicit ratings and SC-IAT), and explicit cue valence correlated with perceived positivity of feedback videos, indicating valence transfer. Negative cues did not show significant valence transfer, likely influenced by the low frequency of negative feedback (17%) to avoid frustrating participants. Age-related socio-emotional differences emerged: older adults rated negative videos as less negative, consistent with the positivity effect and potential reliance on emotion regulation strategies (e.g., reappraisal). Explicit (but not implicit) cue valence differences between ages suggest contributions of higher-order cognitive control and dual-process mechanisms. Together, static and dynamic social feedback similarly motivate action, while age modulates socio-emotional appraisal of negative feedback and explicit evaluative learning strength.

The study demonstrates that both static photos and dynamic videos of social feedback effectively motivate faster responding when anticipating rewards or punishments across the adult life span, with no added motivational benefit of videos. Positive evaluative learning occurs for reward-predictive cues, and explicit cue valence reflects the perceived positivity of corresponding feedback. Older adults show a positivity bias in socio-emotional processing (less negative ratings of negative videos) and smaller explicit cue valence differences, consistent with age-related emotion regulation and dual-process accounts. Future work should employ within-subject designs to directly compare feedback modalities, balance feedback frequencies (especially negative outcomes) to test evaluative learning for punishment cues, align stimulus durations across modalities, and explore increasingly naturalistic, interactive social stimuli to assess impacts on motivation and learning.

- Between-subjects comparison of photos vs videos; modality evaluated in separate samples and time periods, introducing potential context effects (e.g., SID-Video during COVID-19 restrictions) and limiting causal inference about modality differences. - Imbalance in outcome frequencies: positive feedback followed positive cues ~83%, negative feedback followed negative cues ~17%, potentially suppressing evaluative learning for negative cues. - Stimulus duration mismatch (photos 750 ms vs videos ~2400 ms), which may confound modality effects. - Actor age (24–27 years) may introduce intergroup bias in perception across participant age groups. - Baseline RT differences between studies required z-transformation; raw RT comparisons across studies not interpretable. - Reuse of previously collected data for SID-Photo; potential unmeasured differences across collection periods and experimenters. - Between- vs within-subject design trade-offs (e.g., lack of direct within-person modality comparison).