Abstract
Hong and Thoroughgood (2025) argue that witnessing a coworker express gratitude to a mutual supervisor triggers “social comparison” processes (Festinger, 1954) that elicit envy and undermining among observers with lower leader-member exchange social comparison (LMXSC). Across two studies, they find support for this moderated mediation pathway. Their central claim is that gratitude expressions serve as “a salient source of upward social comparison information” (p. 5), prompting lower LMXSC observers to contrast themselves unfavorably with the grateful coworker (Buunk & Gibbons, 2007). Their framework offers a theoretically plausible account: in hierarchical workplace contexts, observing coworker advantages could reasonably trigger evaluative comparison processes, and individual differences in perceived relative standing (LMXSC) could plausibly moderate such reactions.
We propose that witnessing gratitude may activate an additional mechanism alongside social comparison: exclusion signaling. While Hong and Thoroughgood’s comparison framework provides valuable insights into how observers process witnessed gratitude, their evidence_comprising manipulation of information exposure and moderation by LMXSC_cannot uniquely adjudicate between comparison processes and exclusion-based interpretations. Both mechanisms could produce the observed pattern of results. We develop the exclusion account as a credible alternative interpretation, not to replace the comparison framework but to highlight the need for future research with direct process measurement to disentangle these pathways. Three features of Hong and Thoroughgood’s design and findings raise questions about mechanism identification and suggest exclusion processes warrant explicit theoretical incorporation and empirical testing.
Temporal Ordering: When Does LMXSC Develop Relative to Gratitude Events?
Hong and Thoroughgood position LMXSC as a moderator that shapes how observers interpret gratitude expressions_with a “perceptual filter through which observers perceive their capability to obtain similar benefits from a mutual supervisor” (p. 7). This framing suggests witnessing gratitude initiates the comparison producing distress. However, their own description introduces a temporal ambiguity. They note that lower LMXSC observers “already believe” such benefits reflect “their coworker’s closer relationship with the supervisor” (p. 8). If observers already hold beliefs about relational disadvantage before witnessing gratitude, this raises questions about what psychological process the gratitude observation triggers.
Under a comparison interpretation, witnessing gratitude would initiate new evaluative processing about relative standing. Under an exclusion interpretation, witnessing gratitude would remind observers of pre-existing relational deficits--functioning as what Williams (2007) terms an “ostracism cue” that activates existing belonging concerns rather than generating new comparative evaluations. Both interpretations remain consistent with Hong and Thoroughgood’s findings, but they suggest different underlying processes and different intervention targets. The comparison account suggests managing how advantages are communicated; the exclusion account suggests addressing observers’ baseline relationship quality with supervisors (Graen & Uhl-Bien, 1995).
Process Evidence: What Psychological Mechanisms Does Information Exposure Activate?
Hong and Thoroughgood (2025) manipulate whether observers learn about gratitude expressions but do not measure the cognitive or affective processes this information triggers. Their manipulation contrasts learning that a supervisor-employee exchange occurred versus not learning about such an exchange. Without process measures confirming whether observers engage in comparative evaluation versus experience exclusion-based distress, attributing effects to “comparison” reflects a theoretical inference rather than direct evidence (Mussweiler, 2003; Wood, 1996).
Consider what information gratitude expressions convey. When observers overhear a coworker thanking their supervisor, they learn two things simultaneously: (1) the coworker received something valuable, and (2) the observer was not part of that exchange. Hong and Thoroughgood’s comparison framework emphasizes the first element_learning about others’ advantages triggers self-evaluation (Festinger, 1954). An exclusion framework would emphasize the second element_learning that a valued interaction occurred without you signals relational marginalization (Williams, 2007), regardless of whether the included party is objectively ‘better’ or ‘worse’ than you. Research on workplace ostracism demonstrates that exclusion from even trivial interactions produces distress (Ferris et al., 2008; Robinson et al., 2013), suggesting pain may stem from being left out rather than unfavorable comparison.
Adjudicating between these interpretations requires measuring whether participants focus attention on relative standing (“My coworker has a better relationship than I do”) versus inclusion/exclusion (“I was left out of this supervisor interaction”). Both Hong and Thoroughgood’s proposed mechanism and the exclusion alternative remain plausible given current evidence; what is needed is research specifically designed to distinguish them.
Boundary Conditions: What Do Study Differences Reveal About Mechanisms?
Perhaps the most intriguing pattern in Hong and Thoroughgood’s findings involves differential effects across their studies. In Study 1 (experimental vignettes), even high-LMXSC observers experienced increased envy when witnessing gratitude: participants in the higher LMXSC condition reported significantly more envy in the gratitude expression condition (M = 2.19) than in the control condition (M = 1.59), F (1, 209) = 4.14, p < .05, η2ρ = .02. Yet in Study 2 (field setting), this high-LMXSC effect disappeared entirely. These differences suggest important boundary conditions that warrant further investigation.
One possibility consistent with exclusion processes: real workplace relationships may buffer against exclusion signals in ways experimentally manipulated LMXSC cannot. When people have genuine, established supervisor relationships (Study 2), witnessing a coworker’s gratitude may not signal exclusion because they have accumulated relational capital that one missed exchange cannot threaten (Baldwin, 1992; Murray et al., 2006). When LMXSC is artificially induced through vignettes (Study 1), even “high LMXSC” observers lack this buffering reservoir. This pattern aligns with research on temporal construal showing that immediate, concrete relational experiences produce different psychological responses than abstract, future-oriented representations (Fouladi, 2026; Safizadeh et al., 2026). However, alternative explanations remain equally plausible: differences in measurement timing, context-specific norms about gratitude expression, or demand characteristics in experimental versus field settings could also explain divergent patterns.
Rather than treating these differences as definitive confirmation of one mechanism over another, they are best understood as suggestive boundary conditions motivating explicit tests of when and why gratitude observations trigger different responses. The key theoretical question becomes: Under what conditions does witnessing gratitude primarily activate comparison versus exclusion processes, and do these conditions map onto individual differences (LMXSC), contextual factors (experimental vs. field), relationship history (established vs. new), or some combination?
Disentangling Mechanisms: Empirical Pathways Forward
Hong and Thoroughgood (2025) make an important contribution by documenting third-party reactions to gratitude expressions and demonstrating that LMXSC moderates these reactions. Their social comparison framework provides one theoretically coherent account of these patterns. The exclusion signaling account we develop here is not intended to replace this framework but to clarify that multiple mechanisms could produce the observed effects, and that distinguishing them requires explicit measurement and experimental design.
Future research should incorporate process measures to test discriminating predictions:
Prediction 1 (Cognitive Focus)
If social comparison drives envy, thought-listing or cognitive process measures should reveal focus on relative standing and unfavorable self-evaluation. If exclusion drives envy, process measures should reveal focus on being left out and relational marginalization. Measuring both types of cognitions would clarify their relative importance (Williams, 2007).
Prediction 2 (Temporal Dynamics)
If LMXSC reflects a stable comparison tendency, witnessing gratitude should not update LMXSC perceptions. If LMXSC reflects accumulated relational injury, gratitude observations might update these perceptions by providing new data points about relative treatment. Longitudinal designs tracking LMXSC before and after gratitude events would test this distinction.
Prediction 3 (Intervention Effectiveness)
If comparison is primary, interventions providing information about equal treatment across employees should reduce envy. If exclusion is primary, interventions emphasizing inclusion cues (e.g., supervisors acknowledging all team members before gratitude expressions) should be more effective. Experimentally manipulating intervention type while measuring process variables would clarify mechanism-intervention alignment.
Practical implications depend on which mechanism operates in a given context. If gratitude primarily triggers comparison, Hong and Thoroughgood’s recommendations_making gratitude more private or promoting peer gratitude to dilute comparisons_follow naturally. If gratitude primarily signals exclusion, different interventions become relevant: addressing observers’ baseline LMX quality, implementing inclusive recognition practices where supervisors acknowledge all team members, or creating explicit norms about how gratitude relates to broader relationship patterns rather than isolated transactions. Until process evidence clarifies which mechanism dominates, organizations may benefit from interventions that address both pathways.
Hong and Thoroughgood have surfaced a crucial empirical pattern deserving sustained attention: witnessed gratitude can trigger unintended negative reactions among observers, particularly those with lower LMXSC. Our commentary aims to sharpen the next wave of research by highlighting that gratitude expressions may transmit not only evaluative information about relative standing but also relational information about inclusion and belonging. Designing studies that can empirically distinguish these pathways_through direct process measurement, experimental manipulation of mechanism-relevant factors, and context-sensitive theory building_will enable the field to develop more precise understanding of when and why gratitude helps versus harms workplace relationships. Similarly, relational dynamics in organizational settings may be frequently misattributed to one psychological process when they actually reflect another_a pattern evident across various domains where outcomes attributed to stable individual differences may actually reflect situational or relational processes (Fouladi, 2026).
