Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Mediated Emotion and Mood Regulation

Abstract. Mediated emotion and mood regulation concerns the intentional selection and encountered consequences of mediated experiences that shape emotion, arousal, and mood. This fundamental critical narrative monograph reconstructs the scientific development of mood-management theory, prospective mood adjustment, and process models of emotion regulation; examines music, sad media, narratives, digital games, social connection, comparison, threatening news, and algorithmically arranged exposure; and evaluates experiments, longitudinal studies, experience sampling, randomized platform deactivation, and digital-trace measurement. The synthesis distinguishes anticipated emotional benefits from observed outcomes, immediate affect from longer-term functioning, selection effects from causal intervention effects, and population averages from heterogeneous responses. It analyzes adolescents, cultural and accessibility contexts, ethical platform design, emotional privacy, and adaptive human–artificial interaction. The evidence supports a context-dependent account in which regulatory goal, content, control, and social meaning jointly shape experience. Remaining problems include reciprocal selection, mechanism identification, exposure measurement, replication, and the temporal persistence of emotional change. This is a critical narrative research synthesis, not an independently conducted systematic review, meta-analysis, original experiment, or clinical treatment guideline.

Keywords: mediated emotion; mood regulation; media psychology; mood management theory; mood adjustment; emotion regulation; selective exposure; music and emotion; sad music; media entertainment; digital games; psychological recovery; social comparison; doomscrolling; news exposure; social media; digital well-being; arousal; affective forecasting; ecological momentary assessment; experience sampling; randomized deactivation; algorithmic ranking; causal inference; Artificial Era

The Scientific Object of Mediated Emotion and Mood Regulation

Mediated emotion and mood regulation is the selection, use, interpretation, or alteration of mediated experiences in ways intended to influence affective states or that systematically change their trajectory. Mediated experiences include music, filmed narratives, television, networked social interaction, news, games, streaming, and algorithmically organized feeds. The governing scientific question concerns the relation between a person's emotional goal, the media situation encountered, the mechanisms engaged, and the affective response observed across time. A media session may provide deliberate relief, emotional validation, preparation for a demanding event, social connection, or an unanticipated increase in distress. The same format can support very different outcomes because content, context, user intention, and available alternatives vary. Gross (1998) Zillmann (2000) Valkenburg et al. (2016)

The present research object is narrower than the total science of emotion regulation and more specific than the association between technology use and well-being. Emotion regulation includes interpersonal, cognitive, behavioral, and physiological strategies beyond media; media research includes learning, persuasion, identity, and political communication beyond affect. This dossier examines the meeting point: when a mediated environment becomes a selected situation, attentional target, interpretive stimulus, interpersonal resource, or feedback system for affect. Distinct outcomes include momentary pleasure, anxiety, perceived energy, tension, sadness, emotional clarity, recovery from fatigue, and the persistence or return of distress. Depression diagnoses, general subjective life satisfaction, and objective task performance require their own measures and cannot be inferred from immediate affect alone. Gross (2015) Reinecke (2009) Allcott et al. (2026)

The core questions are why people choose particular media under particular affective conditions; which design and content features make a media episode emotionally consequential; whether a perceived benefit persists beyond the session; how platform recommendation systems change access to the chosen content; and which research designs can identify causal effects. A defensible answer distinguishes user-selected exposure from randomly assigned material, subjective anticipated benefit from observed change, and an average population coefficient from variable individual responses. The present work is a critical narrative synthesis of foundational theory, experimental studies, experience-sampling research, qualitative research, randomized digital interventions, and evidence reviews. It reports published findings and proposes testable integrative research questions without presenting an original systematic review or new experimental data. Knobloch (2003) Beyens et al. (2020) Parry et al. (2021)

Affect, Emotion, Mood, and Regulatory Goals

Emotion generally refers to relatively organized responses to events appraised as significant, expressed through experiential, cognitive, behavioral, and physiological components. Mood refers to a more diffuse affective condition that can persist across situations without a single immediately identifiable eliciting object. Arousal and valence are analytically useful dimensions: a person may feel unpleasantly activated, pleasantly activated, calmly content, or fatigued and dysphoric. Media influence those dimensions through different processes, so a single pleasantness rating obscures whether the person wanted stimulation, calm, companionship, or recognition of grief. Gross's process account explains why regulation can act before, during, and after an emotion episode; Thayer and colleagues' empirical work documents multiple everyday attempts to change bad mood, energy, and tension. Gross (1998) Thayer et al. (1994)

A regulatory goal is the affective outcome that makes a behavioral choice intelligible. In one circumstance it may be hedonic relief; in another it may be emotional maintenance, acceptance, activation, performance preparation, interpersonal attunement, or deliberate exposure to difficult material for information. The distinction between wanted and experienced states matters: a listener can select sad music to sustain a reflective atmosphere, then report consolation rather than simple elevation in valence. Likewise, someone seeking reassurance may open a social platform and encounter social comparison that worsens the anticipated state. Researchers should therefore measure initial mood, desired state, anticipated benefit, media choice, immediately experienced affect, and delayed outcomes separately. The goals of regulation can change during the episode, especially when a future task becomes salient. Knobloch (2003) Van den Tol & Edwards (2013) Sagioglou & Greitemeyer (2014)

Self-reported emotion, mood, and mental health occupy different levels of analysis. A laboratory affect change following a movie clip establishes a response to that clip under experimental conditions; it does not establish sustained psychological improvement. A change in a composite happiness–depression–anxiety index in a deactivation trial concerns the whole intervention contrast over its assessment window, including displacement into alternative activities. Daily reports capture ecologically meaningful affect while inheriting recall, selection, and compliance problems. When researchers report regulation success they must state the target, baseline, timescale, comparison, and the difference between explicit intention and measured outcome. Otherwise a successful short distraction may be misdescribed as long-term coping, or a difficult but meaningful experience as affective failure. Allcott et al. (2026) Kross et al. (2013) Gross (2015)

Historical Formation from Entertainment Choice to Digital Affect

The contemporary study of mediated affect grew from several intersecting traditions: communication research on selective exposure and entertainment, psychological work on everyday mood change, and empirical examination of music and emotionally evocative narratives. Dolf Zillmann's 1988 formulation of mood management through communication choices treated audiences as active selectors of experiences expected to shift affect. This perspective transformed entertainment preference from a descriptive taste variable into a hypothesized regulatory behavior with observable selection and state-change outcomes. Its enduring contribution is the proposition that affective states and anticipated consequences influence which media encounters become attractive. The proposition can be tested by manipulating starting mood, observing real choices, and measuring later affect, rather than inferring motivation retrospectively from generic frequency of exposure. Zillmann (1988) Zillmann (2000)

By the late 1990s and early 2000s, regulation researchers gave this account a broader process vocabulary. Gross described situation selection, situation modification, attentional deployment, cognitive change, and response modulation as points of intervention in emotion generation. Zillmann subsequently clarified that informational usefulness can compete with or complement hedonic selection: people choose news, educational material, and unsettling documentaries for reasons that cannot be reduced to immediate pleasure. Knobloch's 2003 experiment on music choice showed that selection can also prepare mood for an anticipated activity, thereby extending the account beyond repair of an unpleasant present state. These developments establish the theoretical continuity of classical media choice and later multidimensional emotion-regulation research. Gross (1998) Zillmann (2000) Knobloch (2003)

Networked platforms altered the empirical setting by making selection continuous and partially delegated. In broadcast media, the available menu and timing were relatively constrained; in contemporary feeds, prior behavior helps shape the next exposure, while interpersonal feedback, alerts, and personalized ranking modify affective opportunity. Music libraries and streaming recommendations permit rapid shifts between emotional functions; social feeds can simultaneously provide affiliation and upward comparison. The evolution is therefore a change in the causal organization of exposure as well as in the amount of media available. Modern research must observe how a person's choice interacts with algorithmic presentation and changing mood, rather than equating an item seen with an item actively sought. Valkenburg et al. (2016) Verduyn et al. (2015) Beyens et al. (2020)

Mood Management and Selective Exposure Theory

Mood management theory proposes that media selection often serves a hedonic aim, tending toward experiences expected to reduce unpleasant states or sustain pleasant ones. For explanatory analysis, the model requires identifiable properties of content such as excitatory potential, absorbing capacity, and hedonic tone, together with a person's initial state and behavioral alternatives. A listener seeking high energy after fatigue may select an activating piece; someone already agitated may prefer a calmer stimulus. The theory is strongest when these terms are operationalized, media alternatives are experimentally controlled, and actual selection is recorded. A statement that people choose media because of mood, without observing either initial mood or available alternatives, is a circular reconstruction rather than an independent test. Zillmann (1988) Zillmann (2000)

The influential 2000 discussion by Zillmann explicitly brings counterhedonic selection and informational utility into view. People can deliberately consume unpleasant content to understand a threat, fulfill a social obligation, preserve moral attention, or prepare for a challenging conversation. Such behavior fits a multi-goal account in which informational, social, and affective rewards jointly influence choice. Emotional consequences must be distinguished from intentions: a distressing news report may be instrumentally useful while also intensifying anxiety, whereas a comforting fiction may produce immediate pleasure without improving a consequential decision. Competing explanations can be evaluated by measuring the importance of information, social identity, hedonic expectation, and perceived task relevance before exposure. Zillmann (2000) Holman et al. (2014)

Selective exposure has three stages that are often collapsed in observational analysis. First, anticipated consequences influence the option a person seeks; second, the distribution system determines what options and prompts become visible; third, the selected content produces a response that may change subsequent selection. At each stage prior emotion and third variables such as stress, work demands, and interpersonal context can influence both exposure and outcome. A valid experiment can manipulate one stage while holding others relatively stable, but broad platform interventions change several stages at once. Their effects therefore estimate the result of altering the overall exposure ecology rather than the intrinsic emotional quality of one type of message. Knobloch (2003) Allcott et al. (2020) Allcott et al. (2026)

Mood Adjustment and Prospective Emotional Preparation

Mood adjustment extends the analysis of entertainment selection by asking what state a person expects to need later. Knobloch's 2003 study experimentally combined an induced starting mood with anticipation of activities differing in activation and task orientation. During a waiting period, participants could select popular music; the program logged their choices, and the music alternatives had been pretested for energy and joyfulness. Patterns of choice supported an account in which early selection reflected current mood management while later selection reflected preparation for the expected activity. The experiment demonstrates why immediate maximization of positive affect is an incomplete behavioral prediction. A person may prefer the emotional readiness most suitable for a later objective. Knobloch (2003)

Prospective regulation is particularly relevant to media experiences used before sleep, study, exercise, difficult communication, or a demanding shift. Its scientific study requires specifying both the anticipated activity and the functional state thought to support it. An energetic soundtrack may raise activation in a way useful before physical activity while interfering with relaxation; reflective music may help someone organize feelings while leaving valence unchanged. Regulatory success is therefore conditional on the goal rather than definable by the largest increase in self-reported happiness. Experimental paradigms should include delayed task performance and physiological or behavioral indicators where they are valid for the target construct. A change in arousal alone cannot establish improved attention, work quality, or sleep. Knobloch (2003) Thayer et al. (1994) Gross (2015)

Prospective regulation also reveals the difference between experienced and remembered effectiveness. Participants may remember having chosen suitable material, while independent records indicate that the chosen stimulus had little measurable effect or that the effect dissipated. A future-focused intention can be reinterpreted after a favorable outcome, raising post-hoc attribution problems. Investigators can limit this ambiguity by recording intention before exposure, using pretested options, measuring independent outcomes, and specifying the relevant delay. The research lesson is broader than music: feeds, news briefings, games, and conversational systems may be used to create readiness or resilience, and evaluations of those uses must specify the counterfactual response under another available choice. Gross (1998) Parry et al. (2021)

Emotion Regulation as a Process across Media Encounters

The process model of emotion regulation offers a structured translation between psychology and media choice. Situation selection includes opening a comforting playlist or avoiding threatening coverage. Situation modification includes changing a feed preference, muting a contact, or seeking a different conversational setting. Attention deployment concerns shifting concentration toward a story, puzzle, or humorous clip. Cognitive change includes interpreting an event through a narrative or discussing it with another person. Response modulation can involve relaxation, expressive communication, or bodily arousal during an interactive experience. These processes can combine within one session and may occur without explicit awareness of a regulatory goal. A study that measures only use duration cannot determine which operation is responsible for an effect. Gross (1998) Gross (2015)

The temporal ordering of regulation should be visible in the research design. A person enters an episode with a baseline state; makes or receives a media selection; attends to some portion of the stimulus; interprets its significance; reacts physiologically and experientially; and may continue, switch, share, or withdraw. These phases offer distinct intervention points and measurable outcomes. For example, making a distracting video available changes the opportunity set, while changing its emotional content targets stimulus properties; hiding reaction counts affects interpersonal meaning rather than basic access. A single platform-disable intervention changes all of these simultaneously. Mechanistic inference becomes stronger when experiments vary one factor at a time or use factorial designs capable of detecting interactions. Gross (2015) Verduyn et al. (2015) Allcott et al. (2020)

Strategy effectiveness is context dependent and must be evaluated across immediate and longer timescales. Attention shifting can relieve an acute unpleasant state but also postpone necessary problem solving; immersion in a narrative can provide emotional space while inviting avoidance of an urgent interpersonal obligation. Reappraisal and social connection may support flexible adaptation when the content and social context are suitable. The task is to identify when a strategy performs its intended function and at what opportunity cost. Gross's framework supports a functional analysis of what media does for a person in a particular episode, while mood-management research explains why that episode is chosen. The synthesis links deliberate selection, automatic engagement, emotional learning, and later behavioral consequences. Gross (2015) Nolen-Hoeksema et al. (2008) Zillmann (2000)

Research Scope, Evidence Hierarchy, and Inference

Evidence on mediated affect comes from different research traditions that answer different questions. Controlled media-selection experiments can identify the consequences of assigned affective conditions for choice within a designed menu. Random assignment to alternative content can estimate short-run stimulus effects. Ecological momentary assessment can describe exposure–emotion sequences in daily life. Panel designs can evaluate within-person changes over days or months. Platform deactivation trials can identify overall intervention effects under concrete social and technological conditions. Qualitative interviews can clarify self-understood goals and meanings that quantitative measures might overlook. These designs contribute complementary evidence and should retain their methodological identities during synthesis. Knobloch (2003) Kross et al. (2013) Allcott et al. (2020) Saarikallio & Erkkilä (2007)

The strongest causal conclusion belongs to a randomized contrast actually implemented and measured, rather than to the general statement that media causes an emotional condition. A participant assigned to limit several applications for three weeks is exposed to a bundle of altered interactions, displaced activities, and expectations. Such findings establish the effect of that program in its studied population more directly than they reveal the unique mechanism of any single platform feature. Conversely, an experience-sampling association between use and later mood cannot eliminate all time-varying confounders or selection into use. Individual differences and cross-platform variation further restrict generalization. Hunt et al. (2018) Kross et al. (2013) Beyens et al. (2020)

This narrative synthesis selects foundational conceptual texts, research that directly operationalizes media-affect processes, and consequential counterevidence from modern digital studies. It has no registered search protocol, independent risk-of-bias scoring of every eligible article, or new pooled estimate; it must be read as interpretive scientific integration. Bibliographic records are linked to persistent identifiers wherever publisher or indexing records permit verification. Specific numerical claims are limited to results confirmed in primary abstracts or original study pages. The boundaries are important: a review can be useful for mapping hypotheses and identifying experimental designs while remaining unable to claim exhaustive coverage of every media platform or population as of the present date. Valkenburg et al. (2016) Parry et al. (2021) Fassi et al. (2024; corrected 2026)

Music as an Everyday Technology of Emotional Regulation

Music is an exceptionally informative case of mediated regulation because listeners can select it deliberately, adjust it continuously, and experience it while doing other activities. Its temporal structure, acoustic properties, learned associations, and autobiographical meanings give it multiple pathways into emotion. Saarikallio and Erkkilä's 2007 grounded-theory study used group interviews and follow-up forms with eight adolescents to construct a model of personally meaningful mood-regulation goals pursued through musical activity. The small qualitative sample was suitable for theory development, not for population prevalence estimation. Later uses-and-gratifications studies by Lonsdale and North showed that mood regulation was a prominent self-reported reason for listening, supporting the behavioral importance of this research object without establishing that every instance of listening produces improvement. Saarikallio & Erkkilä (2007) Lonsdale & North (2011)

The mechanisms of music-induced emotion include bodily responses to acoustic events, learned emotional associations, imagery, autobiographical memory, affective contagion, and expectations about musical development. Juslin and Västfjäll's theoretical synthesis specifically warned against reducing these diverse mechanisms to a single cognitive appraisal process. A listener may select a song because it recalls a supportive relationship, because its rhythm supports activation, or because its expressive quality matches a difficult state. These distinct pathways predict different temporal profiles and varying sensitivity to individual history. Strong designs combine pretested stimulus features with listener reports, physiological measures where interpretable, and repeated observation of contextual goals. Juslin & Västfjäll (2008) Juslin et al. (2008)

Music also illustrates the difference between solitary and socially embedded regulation. A track played through headphones can create a protected attentional situation; shared listening can facilitate belonging and negotiation of a collective atmosphere. The choice between nostalgic, tranquil, sad, energetic, or angry material may reflect desired congruence rather than simple mood repair. Meaning matters because the same musical passage can be soothing for one listener and distressing for another. Research should therefore distinguish selection motive, perceived fit, induced emotion, and subsequent functioning. It should also test whether repeated reliance on a narrow musical repertoire supports flexible coping or preserves rumination, instead of treating any listener-preferred selection as intrinsically adaptive. Saarikallio (2011) Van den Tol & Edwards (2013) Nolen-Hoeksema et al. (2008)

Sad Media, Emotional Congruence, and Meaning

The consumption of sad music and emotionally difficult narratives challenges a purely positive-valence account of media choice. Van den Tol and Edwards investigated why people report choosing music they themselves regard as sad during negative circumstances. Reported motives and consequences show that sadness-compatible listening can serve reflection, emotional companionship, or the processing of loss. Such findings are compatible with regulatory goals that concern understanding and emotional fit. They also leave open whether an individual episode alleviates, sustains, or intensifies distress. A coherent measurement design therefore asks whether the listener sought comfort, catharsis, memory, validation, or a change in energy and whether those functions were achieved after exposure. Van den Tol & Edwards (2013) Gross (2015)

Affective congruence can increase the subjective credibility or significance of a mediated experience. People who feel sad may find a cheerful stimulus alienating while a sorrowful narrative provides recognition of their present condition. This is a hypothesis about appraisal and fit that requires direct testing rather than a universal rule. In situations of active rumination, emotionally congruent material can also become an occasion for repetitive negative thinking. Nolen-Hoeksema, Wisco, and Lyubomirsky reviewed evidence that rumination can prolong or intensify negative thought and impede problem solving; the review does not imply that listening to sad music is equivalent to rumination. Differentiating reflective engagement from repetitive unproductive processing is essential. Nolen-Hoeksema et al. (2008) Van den Tol & Edwards (2013)

Emotionally challenging media may offer enduring significance even when immediate affect worsens. News about collective suffering can inform protective choices, fiction can invite empathy and moral imagination, and commemoration can maintain social memory. These functions should be evaluated alongside immediate hedonic consequences rather than converted into them. A longitudinal study might measure meaning and social response hours after exposure, while a tightly controlled experiment measures the immediate increase in sadness; both results can be accurate at their respective levels. The scientific task is to specify competing goals and tradeoffs in advance. Tests that compare enjoyable, meaningful, information-rich, and emotionally neutral materials can clarify the relative contributions of content tone and perceived usefulness. Zillmann (2000) Holman et al. (2024) Gross (2015)

Narratives, Film, and Emotional Transportation

Narrative media can regulate affect through attention absorption, emotional simulation, reinterpretation of personal experience, and temporary reorientation toward another person's situation. The critical mechanism is often an interaction between the story and the viewer's own needs: a comedy may interrupt repetitive negative thinking, while a serious drama can support perspective taking or deliberate grief. Researchers must distinguish emotional reactions elicited by narrative content from strategies that users intend to implement with that content. Random assignment to a film clip estimates an immediate stimulus effect; choosing that clip from a menu also introduces motives and expectations into the exposure pathway. Both processes are legitimate objects of research, but they require different causal designs. Gross (1998) Zillmann (1988) Valkenburg et al. (2016)

The absorbing quality of stories may make them useful for attentional deployment, yet their consequences depend on how the viewer interprets the experience. The same threatening narrative can increase vigilance in one viewer and provide manageable emotional mastery in another. Cultural knowledge, previous exposure, personal relevance, and existing psychological distress may moderate these responses. This variability makes genre an imperfect proxy for emotional dose. A researcher who labels every comedy as positive and every tragedy as negative imposes a stimulus category on experiences that can be emotionally mixed. Pretesting perceived emotional tone in the target population and measuring state changes directly offers a stronger basis for generalization. Valkenburg et al. (2016) Gross (2015)

Narrative regulation becomes especially complicated when recommendation systems learn from prior viewing patterns. The next offered episode may be selected by an algorithm optimized for engagement rather than by a person seeking recovery, emotional understanding, or sleep. Repeated autoplay reduces natural stopping points, which can transform a brief chosen distraction into extended exposure. The important empirical comparison is between intentionally selected content under controllable stopping conditions and content encountered through a persistent recommendation sequence. Measurements should include the user's initial goal, actual control over continuation, affect at several moments, and subsequent sleep or next-day functioning. The proposed design is a research implication from process and media-effects theories, not a reported established treatment effect. Gross (2015) Valkenburg et al. (2016) Parry et al. (2021)

Digital Games, Recovery, and Interactive Agency

Interactive games differ from passive narrative exposure because players encounter goals, feedback, uncertainty, competence demands, and sometimes live social interaction. Leonard Reinecke's 2009 online survey of 1,614 participants examined whether games were used to recover from stress and strain, relating recovery experiences and post-stress play in a structural equation model. The findings supported the relevance of recovery as a motive and experience for gaming, while the cross-sectional online design cannot alone establish that gaming causes recovery. A player's sense of accomplishment, manageable challenge, or social affiliation may help explain why a particular game feels restorative; excessive difficulty or hostile interactions may instead create additional stress. Reinecke (2009)

Games also illustrate a fundamental measurement distinction between participation time and experienced quality. Johannes, Vuorre, and Przybylski used objective industry-provided play data and reported a positive association between video game play and well-being in a limited game sample. Their observational design did not establish that every additional hour improves psychological health, and game mechanics as well as player motives may differ widely. For mediated regulation, records of session length should be combined with momentary affect, intended regulation goal, game genre, competitive or cooperative context, and the opportunity cost of play. The behavior of someone seeking quiet recovery through a low-pressure puzzle cannot be equated to that of someone engaged in a stressful competitive encounter solely because both are gaming. Johannes et al. (2021) Reinecke (2009)

Interactive recovery can be examined through experiments that randomly assign people after a standardized stressor to gameplay, another leisure activity, rest, or a non-media control. A useful design measures tension before the stressor, immediately afterward, during recovery, and after a delay, while monitoring fatigue and task reengagement. This permits direct comparison of changes over the same elapsed time and reduces the risk of attributing natural recovery to the game. It also permits assessment of a possible short-lived advantage that disappears later. The choice to play rather than rest should be measured separately from the experimentally assigned activity. Such designs expose the contrast between regulatory intention, immediate gratification, restoration of functioning, and longer-term habit. Reinecke (2009) Gross (2015)

Social Connection, Feedback, and Interpersonal Regulation

Networked media provide emotionally relevant social responses: a supportive message, acknowledgment of a difficult experience, invitations to belong, and rapid communication during uncertainty. The regulation target in such cases is often social rather than sensory. Users may seek reassurance, coordinated interpretation, distraction through conversation, or companionship. The same interface can also display exclusion, conflict, competitive status cues, and unanswered messages. Research should therefore classify interaction quality, partner relationship, reciprocal communication, and the perceived meaning of feedback. A generic count of notifications or time spent on a platform offers little information about the emotional function of an individual exchange. Valkenburg et al. (2016) Nesi & Prinstein (2015)

Experiments on passive exposure and longitudinal analyses of social-media use provide evidence about the importance of mode of engagement. Verduyn and colleagues' 2015 work reported experimental and longitudinal evidence that passive Facebook consumption undermined affective well-being, although the generalization to all passive viewing and all contemporary platforms requires additional tests. The underlying distinction concerns whether a user participates in meaningful interaction, views others without reciprocal connection, or engages in social comparison. Researchers should resist the simplifying inference that all active participation is beneficial: active harassment, compulsive reassurance seeking, and conflict may each carry costs. The relevant variable is the social and psychological process during a particular encounter. Verduyn et al. (2015) Nesi & Prinstein (2015)

Beyens and colleagues' intensive longitudinal analysis in adolescents underscored heterogeneity in social-media effects: average estimates can conceal people whose subsequent emotional responses differ in direction or magnitude. Individual susceptibility may depend on relationship context, previous emotional state, offline support, developmental circumstances, and platform-specific features. To study mediated co-regulation, dyadic designs should measure both parties' behavior and affect, while experience sampling should record who initiated interaction, what support was requested, and how it was received. Such work helps distinguish the effect of receiving a message from the preexisting distress that led someone to seek it. Emotional regulation through communication is thus a coordinated process with reciprocal selection and response. Beyens et al. (2020) Kross et al. (2013)

Social Comparison, Envy, and Emotional Spillover

Social comparison is one pathway through which a media episode affects mood without explicitly emotional content. Viewing curated achievements, appearance representations, popularity indicators, or accounts of exclusion may alter self-evaluation, envy, or belonging. Nesi and Prinstein's adolescent study associated technology-based comparison and feedback seeking with depressive symptoms and identified moderation by individual and social factors. The two-wave questionnaire design is informative about correlates and risk patterns but does not identify a randomized causal effect of comparison itself. Experimental research can strengthen inference by manipulating visible comparison cues, holding content constant, and measuring changes in state affect or perceived relative standing. Nesi & Prinstein (2015)

Tandoc, Ferrucci, and Duffy's survey of college students examined Facebook surveillance use, envy, and depression, highlighting envy as a proposed mediating pathway. A statistical mediation pattern in cross-sectional observations remains vulnerable to confounding and temporal ambiguity; it should not be described as a proven causal mechanism. Nevertheless, the model points toward testable components: exposure to idealized peer information, interpretive comparison, felt envy, and subsequent mood. An experiment manipulating the content of comparison while observing attention and emotional outcomes would provide more direct leverage. Longitudinal measurement is needed to determine whether repeated experiences accumulate, habituate, or interact with existing self-esteem and social resources. Tandoc et al. (2015) Nesi & Prinstein (2015)

Affective spillover occurs when emotion generated during a media encounter carries into later activity. After upward comparison, a person may interpret unrelated events more negatively; after reassuring contact, the same person may approach a demanding task with greater confidence. The spillover hypothesis must be assessed by repeated measurement because retrospective impressions can misattribute emotion to salient media. The design also needs a non-media comparison and careful timing: an unpleasant workday may simultaneously increase browsing and lower subsequent mood. Emotional contamination from news, entertainment, and social exchanges can be short-lived or sustained, and no single platform-level average should be assigned to the entire set of psychological mechanisms. Kross et al. (2013) Beyens et al. (2020) Gross (1998)

Affective Forecasting and the Prediction Error of Media Use

Media selection often depends on predicted emotion. A person opens a feed because it is expected to relieve boredom, provide encouragement, or restore a sense of control. Yet the anticipated reward can differ from the reported post-session experience. Sagioglou and Greitemeyer's 2014 work examined Facebook's emotional consequences and reported evidence that people expected emotional benefit even when Facebook activity was associated with decreased mood, with perceived wasted time identified as a possible explanation. The study provides a focused test of forecasting and attribution, not proof that any specific mechanism operates identically across later platforms. Sagioglou & Greitemeyer (2014)

Predicted affect and experienced affect should be measured independently and in temporal order. An ecological study could ask participants before an episode how much better they expect to feel, collect app use from logs, and ask afterward about valence, arousal, regret, social satisfaction, and fatigue. Repeating the assessment across contexts would reveal whether prediction errors concentrate in certain states such as boredom, sadness, or social exclusion. It could also identify learning: repeated disappointing experiences might reduce future use, or reinforcement features might maintain it despite poor immediate forecasts. These alternatives require longitudinal evidence; the observation that users continue to engage is insufficient to identify the function of the engagement. Sagioglou & Greitemeyer (2014) Parry et al. (2021)

A perceived waste-of-time explanation should be distinguished from hedonic properties of actual content. A session might include enjoyable clips followed by retrospective regret about lost time; another might be unpleasant during viewing and also judged futile afterward. Such cases require event-level measures rather than a single global rating. Researchers should compare pre-session intention, in-session experience, post-session evaluation, and later functioning. Media use with mixed outcomes is central to mediated regulation because the act of choosing can be rational under uncertain information even when an episode disappoints. Scientific explanation becomes strongest when it models the discrepancy instead of using continued engagement as proof of pleasure or using negative end-state as proof of initial irrationality. Sagioglou & Greitemeyer (2014) Zillmann (2000)

News Exposure, Threat Monitoring, and Public Crisis

News is frequently selected for informational orientation during uncertainty, but its affective consequences depend on repetition, vividness, personal proximity, and perceived ability to act. The mood-management tradition already recognized that audiences may seek distressing information when information has instrumental value. Contemporary crisis coverage makes the regulatory tradeoff especially visible: repeated exposure may maintain vigilance and the feeling of being informed while also prolonging arousal. In their study of reactions following the 2013 Boston Marathon bombings, Holman, Garfin, and Silver reported an association between extensive media exposure and acute stress, comparing exposure to direct experience. Because the study relied on observational self-reports rather than randomized trauma-related media exposure, it cannot settle the causal direction for every respondent. Zillmann (2000) Holman et al. (2014)

Regulatory interpretations of crisis news require an explicit distinction between informational need and compulsive repetition. An initial briefing may allow practical planning, while repeated graphic images and speculative updates may offer diminishing information. The magnitude of distress can itself predict frequency of checking, creating a reciprocal relation in which anxiety increases exposure and exposure may further sustain anxiety. Holman and colleagues' later discussion of graphic media imagery emphasized that what audiences see, and how exposure is repeated, matters for psychological understanding. The relevant scientific exposure is therefore not the label news but features such as threat salience, vividness, repetition, controllability, and available actions. Holman et al. (2024) Holman et al. (2014)

An ethical experimental design would never require vulnerable participants to view genuinely traumatic material merely to test a theory. Researchers can use previously validated mild affective stimuli, natural exposure variation, or consenting participants' own records to examine how checking and mood evolve. Real-world panel studies can measure contemporaneous stressors, direct threat, social support, political context, and offline activity to reduce plausible confounding. The purpose is to identify the circumstances in which information seeking supplies orientation and those in which repeated exposure sustains distress. The outcome may include immediate valence, sleep disruption, threat perception, practical decision quality, and persistence of distress, each measured on a defined schedule. Gross (2015) Holman et al. (2014) Holman et al. (2024)

Doomscrolling, Rumination, and Repeated Checking

Doomscrolling describes a recurring behavioral pattern of continuing to examine threatening or negatively valenced information, often with difficulty disengaging despite experienced distress. It is best treated as a descriptive behavior requiring operational specification rather than as an automatically diagnosable disorder or universally defined psychological mechanism. Proposed components include uncertainty monitoring, vigilance, attentional capture, negative reinforcement, and the perceived possibility that the next update will resolve ambiguity. Research should measure continuous exposure, the proportion of threatening material, scrolling continuation after expressed intention to stop, subjective control, and post-session affect. This framework joins media selection and emotion regulation without assuming that all sustained news consumption is maladaptive. Zillmann (2000) Gross (2015) Holman et al. (2014)

Rumination is a distinct construct involving repetitive attention to distress and its possible causes or consequences. Nolen-Hoeksema, Wisco, and Lyubomirsky's research synthesis shows why repetitive negative processing can maintain dysphoria and interfere with effective responses. An unpleasant news session may trigger rumination, but observing long reading time alone does not demonstrate rumination: a person could be gathering actionable information, performing professional work, or making sense of unfamiliar events. A robust mediation hypothesis would independently measure repetitive thought between exposure and later affect while accounting for baseline distress. Without that separation, the explanation risks defining the proposed mediator by the very outcome it is intended to explain. Nolen-Hoeksema et al. (2008) Holman et al. (2014)

Repeated checking may be maintained by variable informational reward: most refreshes add little, but an occasional consequential update encourages further monitoring. That account can be modeled prospectively by tracking prediction, uncertainty, update novelty, and emotional state around each check. Algorithmic interfaces complicate this pattern by highlighting salient content, enabling notifications, and removing natural stopping points. Researchers could compare fixed-interval summaries with continuous feeds while holding total useful information approximately stable. The contrast would test whether temporal structure and availability of stopping cues influence affect beyond the informational substance. Such an experiment is a proposed research design rather than a reported established intervention in this dossier. Gross (2015) Parry et al. (2021) Valkenburg et al. (2016)

Platform Affordances and Algorithmically Organized Affect

Digital platforms structure emotional exposure through ranking, recommendation, notifications, social feedback, visibility of audience reactions, content persistence, and ease of repetition. These are affordances: features that shape the opportunities and constraints of interaction. Their effects depend on user goals, social context, and actual implemented design. A ranked feed may surface emotionally activating material because earlier interactions predicted engagement; a personal playlist may offer user-controlled selection and repeated access to calming music. The scientific challenge is to identify whether the exposure was chosen, served, or jointly produced by earlier behavior and present ranking rules. A simple correlation between time on a platform and mood leaves these mechanisms unobserved. Valkenburg et al. (2016) Parry et al. (2021)

Media effects are dynamic because content and preferences influence each other. A person in a difficult mood may engage with certain posts, thereby altering future recommendations. The resulting feed may then change later emotional experience, which in turn affects further interactions. This feedback problem requires temporally ordered data and often experimental perturbations of the system. Randomizing feed properties, recommendation intensity, or content composition can help isolate distinct pathways, provided that the intervention is reproducible and does not alter unmeasured dimensions unpredictably. When platforms do not disclose ranking rules, independent observation has limited ability to identify the mechanism that created each exposure. Valkenburg et al. (2016) Beyens et al. (2020) Allcott et al. (2020)

Personalization may also make average treatment effects poorly transportable. An interface that improves one individual's sense of connection may direct another toward threatening comparisons or unwanted conflict. Researchers should collect baseline susceptibility, relevant interpersonal circumstances, preexisting preferences, and the material actually encountered. The evaluand is not merely the technology's availability; it is the specific experienced relationship between affordance, exposure, and state transition. This approach supports testable ethical questions about control, override mechanisms, and whether recommendation systems allow a person to pursue declared emotional goals. Claims of beneficial emotional optimization by artificial ranking require randomized evidence and transparent measurement rather than the observation of high engagement. Beyens et al. (2020) Gross (2015) Parry et al. (2021)

Notifications, Interruptions, and Arousal Transitions

Alerts and interruptions introduce unchosen transitions into an ongoing activity. A message preview may offer reassurance or companionship, but it may also demand work, expose social rejection, or provoke anticipatory anxiety. Their regulatory significance arises from timing, social meaning, and disruption of attention. A user opening a device to calm down can encounter a notification that redirects awareness toward an unresolved obligation. The affective outcome may depend less on total notifications than on how they align with the person's emotional goal, current task, and expectation of availability. Theory therefore predicts heterogeneous effects from identical alert counts. Gross (2015) Valkenburg et al. (2016)

Notification intervention research must measure both exposure and what participants do in its place. Disabling alerts can alter spontaneous checking, interpersonal responsiveness, subjective control, and the distribution of work interruptions, sometimes in conflicting directions. An experimental comparison should specify whether badges, sound, vibration, push previews, and message delivery are suppressed individually or together. Participants' compliance and compensatory opening of applications matter because nominal assignment to notifications off does not guarantee reduced exposure. Longitudinal self-report should be cross-checked with logs when ethically available, since reports of digital behavior often differ from independently recorded duration and frequency. Parry et al. (2021) Gross (2015)

Emotional arousal following an interruption can be measured with momentary ratings, task recovery intervals, and physiological signals, but physiological activation is not equivalent to distress. A desirable message can raise heart rate or alertness while improving mood; an unpleasant message can produce prolonged unease without a distinctive autonomic pattern. Interpretation requires converging measures and clear operational definitions. If the question concerns restoration of concentration, assess subsequent performance; if it concerns emotional relief, assess valence and tension; if it concerns connection, include relational measures. A multidimensional battery prevents the researcher from presenting changes in a single signal as proof of broad emotional regulation. Thayer et al. (1994) Gross (2015)

Adolescent Development and Differential Susceptibility

Adolescence is a period when peer feedback, identity development, social comparison, and changing autonomy can make mediated experiences emotionally salient. Yet adolescents are heterogeneous and media interactions differ across relationships, platforms, and individual circumstances. Beyens and colleagues' 2020 study applied intensive longitudinal methods to examine individual differences in social-media effects on well-being, finding that the size and direction of responses varied among adolescents. This finding supports designs estimating person-specific effects rather than assuming that a population mean describes every young person's emotional experience. The study does not demonstrate that media use is uniformly harmful or beneficial, nor does it identify all mechanisms behind individual variation. Beyens et al. (2020)

Nesi and Prinstein studied adolescents' technology-based comparison and feedback seeking in relation to depressive symptoms. Their findings direct attention to social processes rather than aggregate device time. Vulnerability may be associated with peer status, interpersonal stress, prior symptoms, and the meaning attached to social interactions. Such moderators should be measured before exposure where possible, while experiments should be designed to respect participants' developmental capacity and safety. An investigation of mediated regulation must therefore differentiate intended support seeking from comparison, conflict, and monitoring, even when these occur within the same application and time window. Nesi & Prinstein (2015) Fassi et al. (2024; corrected 2026)

For adolescent research, outcomes should distinguish immediate affect, stable emotional patterns, functioning, and clinically assessed symptoms. The systematic-review literature on social-media use and internalizing symptoms underscores heterogeneity in definitions, sampling, and analytical designs. Observational associations with symptoms cannot be translated directly into an intervention effect or a diagnostic conclusion. Longitudinal studies should consider development, school and family demands, existing treatment or support, and access inequalities. Consent, privacy protection, and minimizing exposure to upsetting materials are substantive methodological requirements. The question worth answering is which mediated situations promote or hinder developmentally appropriate regulation for whom, under what circumstances, and over what time horizon. Fassi et al. (2024; corrected 2026) Orben & Przybylski (2019) Beyens et al. (2020)

Cultural Context, Social Inequality, and Accessibility

Emotional meanings of media are culturally and socially organized. The same news representation, musical genre, social interaction, or public ritual can have different implications where norms of expression, institutions, language, and shared histories differ. Economic resources also determine access to uninterrupted leisure, private listening environments, reliable information, and meaningful social support. A study of self-selected emotional regulation among university students with unrestricted internet access cannot automatically represent workers, older adults, rural users, or people living under threats and material insecurity. Population description is therefore part of the scientific claim, not an optional supplement. Valkenburg et al. (2016) Gross (2015)

Digital inequality affects both exposure and control. Users with limited accessibility options may find content harder to filter, while those facing harassment or discrimination may experience platform interaction as an unavoidable source of social threat. Language and disability accommodations alter how emotion-relevant information becomes available and how easily a person can stop, mute, or reinterpret a stimulus. Researchers should assess whether the tools used to measure affect are equivalent across language and ability groups. A questionnaire designed for one cultural setting may capture different expressions of tension, meaning, or social obligation elsewhere. The interpretation of average change requires evidence of measurement validity, not merely translation of the same item labels. Gross (2015) Parry et al. (2021)

Cross-cultural designs can combine repeated affect ratings, culturally relevant stimulus validation, and qualitative interpretation of regulatory goals. Statistical comparisons should test measurement invariance when scales are used across groups; qualitative work can identify goals absent from an imported framework. Media-based regulation may serve social solidarity during collective crises as well as individual relief, and both functions warrant study. The evidence base remains uneven where platform data, reliable internet, research infrastructure, or independent access to commercial algorithms are restricted. Scientific synthesis should describe these limits explicitly and prioritize designs that include underrepresented populations without treating a single nationally dominant platform as the default human media environment. Valkenburg et al. (2016) Holman et al. (2014)

Affective Outcomes Across Minutes, Days, and Years

The meaning of an affective effect changes with the observation window. A humorous clip may generate immediate positive affect, a conversation may offer reassurance for several hours, and repeated stressful content may be associated with lingering tension or disrupted sleep. These outcomes are related but cannot be reduced to one score. Laboratory studies are well suited to immediate changes under controlled exposure; experience-sampling work observes naturally occurring state transitions; longitudinal panels examine persistence and bidirectionality. Different estimates can therefore coexist without contradiction. A study reporting improved mood after a brief intervention and another reporting lower life satisfaction in frequent users may be testing different processes with different populations. Gross (1998) Kross et al. (2013) Allcott et al. (2020)

Allcott and colleagues' randomized Facebook deactivation experiment investigated several weeks of removal from a major social platform, measuring outcomes that included subjective well-being, activities, and knowledge. Later deactivation work on Facebook and Instagram examined emotional-state outcomes under a different platform and historical context. Such interventions provide important causal evidence about particular access changes, but the measured effect combines lost content, lost contact, substituted activities, altered expectations, and changes in routines. The effect cannot be attributed to emotional toxicity inherent in every media interaction. This distinction is essential when evaluating claims that a period of less use should reliably translate into long-term clinical benefit. Allcott et al. (2020) Allcott et al. (2026)

A complete temporal assessment should therefore include before-session intention, immediate experience, later same-day state, next-day functioning, and repeated longer-term observations. It should also record what a participant did instead of engaging with the target medium. Restoration, displacement, habit, and social participation are potentially competing mediators. Models should account for natural regression toward typical mood after extreme events, time-of-day patterns, recurring stressors, and missing observations during distress. Interpretation becomes more rigorous as studies report the exact timescale of their estimand and the confidence interval around it instead of relying on the generalized assertion that a medium improves or harms mood. Beyens et al. (2020) Parry et al. (2021) Gross (2015)

Laboratory Experiments in Media-Based Regulation

An experiment on mediated regulation distinguishes manipulation of the emotional starting state from manipulation of the subsequent media environment. Mood-induction procedures make participants enter a choice situation with different affective states; selection is then measured from an available set of clips, music, or news. Alternatively, researchers can hold initial state constant and randomize exposure to stimulus properties, including tone, arousal, or perceived social evaluation. Knobloch's music-selection experiment is an instructive example of how prospective task demands can change what looks like a hedonic preference. The outcome of a selection study is behavior under an experimentally arranged menu, whereas the outcome of a stimulus-exposure study is the reaction to an assigned encounter. Keeping the two estimands distinct prevents misplaced causal claims. Knobloch (2003) Zillmann (1988)

Stimulus equivalence is difficult. A frightening video may also be more novel or visually complex than a neutral one, while a cheerful song may be familiar or personally significant. A design intending to estimate emotional tone should pretest alternative stimuli and either balance or explicitly model such features. Repeated-measures designs can improve precision by comparing individuals with themselves, but order effects, carryover, habituation, and expectation require counterbalancing. The researcher should register a primary affective outcome and the time point of interest because highly responsive people may show an initial change followed by rapid recovery. Manipulation checks are informative but can themselves redirect emotional attention. Gross (2015) Juslin & Västfjäll (2008)

Experimental ethics constrain the intensity and duration of negative affect induction, especially for young or vulnerable participants. Researchers can examine ordinary affective contrasts rather than severe distress. Inference should preserve treatment allocation, disclose exclusion and missing-data criteria, and report uncertainty in estimates. A statistically detectable average change is not necessarily meaningful for everyday regulation, and a null average may conceal prespecified heterogeneous effects if the design has power to examine them. Repeated preregistered experiments across several media genres are a better route to generalization than strong conclusions from one emotionally distinctive clip. Beyens et al. (2020) Gross (2015)

Experience Sampling and Real-World State Transitions

Experience sampling and ecological momentary assessment repeatedly ask participants about current emotion, context, and behavior during daily life. This approach brings observations nearer to the conditions in which regulation actually occurs. Kross and colleagues' 2013 study repeatedly sampled young adults' Facebook use and subjective well-being, identifying temporal associations across observations; it provides a model of within-person sequential analysis while leaving plausible time-varying confounders. The key advance is temporal resolution: investigators can ask whether an unusually intense episode of use precedes a deviation from the same person's ordinary mood. This is more specific than comparing average screen time and average well-being across different people. Kross et al. (2013)

The timing of sampling determines what can be inferred. If participants report media use over the last hour and mood at the sampling moment, the temporal order of brief shifts during that hour may remain ambiguous. Event-contingent assessments before and after voluntary episodes give a clearer sequence but can change the behavior being studied. Passive application records can reduce recall error in exposure, but they cannot reveal whether a person felt reassured, offended, or simply bored. Combining logs with short self-reports of intention and affect is therefore especially valuable. Such studies should avoid presenting repeated correlations as independent observations; episodes are nested within people and often within days, families, and social networks. Parry et al. (2021) Beyens et al. (2020)

Models can separate between-person and within-person variation, account for autocorrelation and time-of-day patterns, and evaluate lags chosen for theoretical reasons. Emotional carryover may be brief after music but longer after interpersonal conflict, so a single arbitrary lag can miss an effect. Researchers should inspect missingness because participants may fail to report precisely when distressed or absorbed in media. Compliance incentives, sampling frequency, and the perceived invasiveness of monitoring can also change real behavior. Transparent reporting of the sampling protocol and sensitivity analyses is essential for cumulative interpretation. A person's media-regulation pattern is a distribution of episodes in situations, not a stable trait inferred from one global questionnaire. Beyens et al. (2020) Gross (2015) Parry et al. (2021)

Digital Traces and the Validity of Exposure Measures

Self-reported media time often differs from device-generated use records. Parry and colleagues' systematic review and meta-analysis examined discrepancies between logged and self-reported digital media use, showing that the measures are not interchangeable. The implication for emotion research is not that device logs provide every relevant fact, but that exposure amount and the meaning of exposure are separate domains. Logs can identify app opening, duration, switching, or some forms of interaction; users can report intended goal, interpretation, perceived control, and felt consequence. Neither source alone captures the full process. A valid design maps each theoretical variable to the method able to observe it. Parry et al. (2021)

Commercial platform traces pose additional validity problems. A counted exposure may be below the viewport or rapidly skipped, while a long session may involve conversation, passive browsing, or content production. Recommendation histories and content-level exposure are often unavailable to independent investigators. Privacy and access restrictions can introduce selection bias if only technically sophisticated or highly motivated participants provide data. Researchers should report what an API or logger actually measures, how records are linked to participants, and whether missing sessions or devices systematically differ. The unit of analysis should remain conceptually aligned with the affective hypothesis: a count of application openings cannot test whether threatening visual imagery caused heightened arousal unless actual imagery exposure is measured. Parry et al. (2021) Holman et al. (2024)

When linking affect reports to trace data, researchers face consent, security, and interpretive risks. A social message is not merely an information item: it can include information about people who did not consent to participate. Data minimization, encryption, short retention, and aggregation are therefore integral to a rigorous design. Controlled exports of metadata may answer timing questions without collecting private message text. Content coding can be performed on voluntarily donated or publicly accessible material under suitable ethics approval. Methodological validity and participant rights intersect because unusably invasive measurement can alter behavior, reduce trust, and bias participation. The objective is the smallest adequate observation that supports the causal or descriptive claim. Parry et al. (2021) Gross (2015)

Longitudinal Models, Confounding, and Causal Identification

The principal challenge in observational media research is reciprocal causation. An individual may turn to a platform because of loneliness, while experiences on that platform may later influence loneliness. Ongoing life events affect both exposure and affect, and personality or social resources may influence stable averages of each. A model predicting mood from media time should therefore specify whether its purpose is descriptive, prognostic, or causal. Adjusting for baseline distress can help but may fail to capture daily changes in stress; inappropriate conditioning on variables affected by media can also introduce bias. Within-person analysis reduces some stable confounding while leaving time-varying selection unresolved. Kross et al. (2013) Valkenburg et al. (2016)

Temporal precedence alone establishes an ordering, not a randomized counterfactual. A period of bad mood followed by app use and worse mood may reflect escalation of an independent difficulty, anticipatory stress, or the medium's effect. Directed causal assumptions should distinguish common causes of exposure and outcome from mediators such as social comparison, reassurance, and attention. Selection is particularly complex when algorithms react to prior behavior: yesterday's emotional state affects yesterday's engagement, which affects today's feed and potentially today's affect. A robust study states what intervention it attempts to emulate and which histories would need to be measured for identification. Beyens et al. (2020) Gross (2015)

Natural experiments and difference-in-differences designs can be informative when platform availability or policy changes alter exposure, but parallel-trends and exclusion assumptions require justification. For randomized interventions, intention-to-treat estimates preserve assignment, while actual-use analyses can become confounded by compliance. In both settings, researchers should report effect sizes, precision, preregistration where available, attrition, sensitivity to analysis choices, and clinically meaningful thresholds when appropriate. A media effect estimate is a property of a specified comparison under measured conditions; extrapolation to every medium, age, and historical period requires separate evidence. Allcott et al. (2020) Allcott et al. (2026) Orben & Przybylski (2019)

Randomized Platform Deactivation and Use Restriction

Randomized access experiments provide consequential evidence because they intervene on the opportunity to use a platform rather than merely observing who uses it. Allcott, Braghieri, Eichmeyer, and Gentzkow's 2020 Facebook deactivation study tested the effects of a temporary suspension of Facebook use across outcomes that included subjective well-being, information, offline activity, and post-intervention use. The estimated welfare consequence is distributed across multiple channels: reduced exposure to some content, changes in social contact, displacement toward other activities, and potentially altered beliefs about habits. A measured positive change in well-being cannot be assigned with certainty to avoiding a particular content genre unless that mechanism was independently manipulated or measured. Allcott et al. (2020)

The 2026 Facebook and Instagram deactivation work extended evidence about emotional-state changes under a defined intervention. The authors reported modest mean shifts in a constructed emotional-state index, illustrating why effect magnitude and outcome construction are indispensable parts of interpretation. Direct cross-study comparison must consider differing populations, platforms, recruitment methods, social conditions, duration, compliance, and index composition. Findings support the claim that altering access can affect average emotional experience in some settings, while they leave open which experiences within those platforms are responsible. They also should not be construed as evidence that all users benefit from permanent removal. Allcott et al. (2026)

Hunt and colleagues' 2018 randomized study restricted use of selected social-media platforms and assessed loneliness and depression among university students. The experimental contrast was a particular restriction protocol, in a particular group, over a limited duration. Its findings may inform hypotheses about routines, comparison, and substitution, but they cannot demonstrate the effect of restricting every platform across all populations and outcome definitions. Researchers should track whether participants replace restricted use with direct conversation, alternative platforms, sleep, exercise, or other activity. Comparisons of program components, follow-up after restrictions end, and preregistered heterogeneity analyses could help identify mechanisms and persistence of benefits or costs. Hunt et al. (2018) Allcott et al. (2020)

Psychometrics of Affect, Recovery, and Functioning

Mediated regulation research often uses immediate mood ratings, but these measures must be matched to the proposed process. Valence and activation can distinguish calm relief from excited enjoyment; state anxiety and tension are more specific than a global satisfaction judgment. A study of music-based recovery should measure whether the regulatory target was reduced tension, increased energy, emotional clarity, or improved readiness for work. General subjective well-being measures address a broader evaluative construct and need not respond immediately to a brief media episode. Measurement instruments should have documented reliability and responsiveness over the chosen timescale, rather than being borrowed solely because they are familiar. Thayer et al. (1994) Gross (2015)

Response shifts and expectancy effects create difficulty when people evaluate personally chosen media. Participants may rate a session favorably because they believe a familiar ritual ought to help or because they feel committed to the choice. Independent outcomes can help when conceptually appropriate: attention after a recovery period, time until sleep, decision quality, or validated indicators of physiological arousal. None is a universal gold standard for emotional quality. Physiological measures are often sensitive to movement, temperature, social context, and excitation unrelated to unpleasantness. Convergent assessment is stronger than treating a single wearable signal as a direct measurement of a subjective emotion. Juslin & Västfjäll (2008) Reinecke (2009)

Comparability across studies requires explicit reporting of scale wording, reference period, scoring, and predeclared primary outcome. A one-item feeling rating immediately after a clip and a multi-item symptom inventory evaluated over two weeks are methodologically distinct. For cross-cultural research, translation and measurement invariance deserve attention. For intensive longitudinal studies, within-person reliability and sensitivity to actual changes may matter more than conventional between-person factor structures. Researchers should also check whether baseline emotional states constrain possible change through floor or ceiling effects. These details influence whether an apparently small, large, or absent effect is interpretable. Beyens et al. (2020) Parry et al. (2021)

Replication, Heterogeneity, and Evidence Quality

The literature on mediated emotional experience combines laboratory experiments, observational panels, self-selected field samples, intervention trials, qualitative investigations, and broad evidence syntheses. A theory can be supported by some designs and remain underidentified in others. For example, a music-selection experiment can establish that anticipated activity affects choices; it cannot by itself establish daily population prevalence of the same goal. A platform deactivation experiment can estimate the welfare effects of reduced access; it cannot automatically distinguish whether reduced comparison, improved sleep, or increased offline sociality produced the result. Evidence evaluation therefore begins by aligning the research question with the intervention and measurement actually used. Knobloch (2003) Allcott et al. (2020)

Heterogeneous responses do not automatically imply that any account is correct for every subgroup. Reliable moderation requires adequate sample size, prespecified moderators, and assessment of uncertainty; selecting the most extreme observed individual patterns can overstate stable differences. Replications across countries, ages, platform configurations, and media formats are needed because interfaces and social conventions change. Small experiments can be informative about mechanisms under controlled conditions yet often lack precision for modest between-person effects. Large observational datasets gain precision while remaining susceptible to systematic measurement and confounding biases. Syntheses should acknowledge both strengths and limitations without giving a simple methodological ranking that substitutes for causal analysis. Beyens et al. (2020) Valkenburg et al. (2016) Parry et al. (2021)

Transparency is particularly important in a field where commercial systems continually change the exposure environment. Researchers should archive stimulus descriptions, preregistration, exclusion decisions, code, anonymized derived data when ethically feasible, and the versions of platforms or interfaces under study. Privacy can restrict release of interpersonal messages or granular logs, and disclosure should respect those constraints. The best cumulative research establishes which mechanism and effect persist after the historical interface changes. When full source access or individual effect verification remains incomplete, editorial records should mark the limitation and avoid claiming independently replicated scientific certainty. A critical narrative monograph can organize these tests and identify methodological questions while preserving the distinction between established result and proposed synthesis. Gross (2015) Orben & Przybylski (2019) Parry et al. (2021)

Integration of Regulatory Goals, Exposure, and Emotional Consequences

The research traditions reviewed in this dossier converge on a sequence of interacting processes: a person has an emotional state and a prospective goal; the person chooses or is presented with mediated opportunities; a particular encounter activates attention, interpretation, interpersonal meaning, and bodily response; the resulting emotional change influences subsequent selection. Mood-management theory explains one recurrent motive for choice, the process model of emotion regulation differentiates mechanisms, and media-effects research specifies how exposure and context constrain outcomes. Digital platform studies show that the opportunity set itself can adapt to prior behavior, complicating the inference from time spent to personal intention. These theoretical components define complementary measurement targets and can be tested jointly without asserting a newly discovered psychological law. Zillmann (1988) Gross (2015) Valkenburg et al. (2016)

The sequence is analytically useful because it exposes distinct reasons why one intervention may work and another fail. A person can choose a soothing stimulus and encounter an unwanted notification; a story may be absorbing but too emotionally intense for the intended purpose; a reassuring conversation may improve immediate mood while imposing later expectations of constant availability. The system's distribution rules influence which options appear, and the user's skills influence whether switching or disengagement is possible. Success should be assessed relative to the declared goal and relevant timescale. A purely hedonic measure can miss growth, empathy, or informational usefulness; a purely engagement-based measure can miss distress, perceived control, and opportunity costs. Knobloch (2003) Reinecke (2009) Holman et al. (2024)

This synthesis yields a methodological commitment: never treat platform membership, format, time, and emotional mechanism as interchangeable constructs. An observed association between frequent use and low mood may arise from selection into use, consequential exposure, displacement, or unmeasured hardship. A randomized restriction effect reveals what changes when access changes, but it does not automatically identify the component responsible. A qualitative account reveals experienced motives without estimating population effects. Scientific progress comes from combining these designs in ways that preserve their evidentiary meaning and test whether the same psychological mechanism appears under different media arrangements. Allcott et al. (2020) Parry et al. (2021) Beyens et al. (2020)

Regulatory Flexibility, Tradeoffs, and Everyday Function

Regulatory flexibility means choosing strategies suitable to circumstances and revising them when conditions change. The relevance to media use is immediate: a short diversion can serve recovery after overload, while a prolonged diversion can postpone action on the source of stress. A listener may profit from sadness-congruent music during reflection, yet the same repertoire may accompany repetitive negative thought in another context. No media choice should be classified as effective solely from its category. Research should specify the individual's goal, the actual circumstances, the available alternatives, and the extent to which other responsibilities or relationships were affected. Gross (2015) Van den Tol & Edwards (2013) Nolen-Hoeksema et al. (2008)

A rigorous evaluation considers acute relief, delayed consequences, and displacement. For a worker recovering after a demanding day, immediate relaxation may be valuable, but an additional hour of unplanned browsing might compete with sleep. For a person isolated from peers, digitally mediated conversation may be an important resource even if screen time increases. Interventions should therefore be judged by predeclared targets rather than by universal prescriptions to maximize or minimize media use. The same quantitative exposure can conceal substantially different social functions, and the same reported mood change can be accompanied by divergent functional consequences. Reinecke (2009) Beyens et al. (2020) Allcott et al. (2026)

Behavioral supports that plausibly increase flexibility include clear stopping points, voluntary preference controls, interruption settings, and transparency about content selection. Their emotional effects remain empirical hypotheses until tested. A design might assign participants to user-controlled or continuously recommended media menus, collect emotion and intention before and after use, and examine changes in duration, social contact, and subsequent functioning. The comparison should be designed so that the controllability manipulation is identifiable and not confounded by total content quality. Such research would contribute to the evidence base without equating commercial engagement optimization with therapeutic benefit. Gross (2015) Valkenburg et al. (2016) Parry et al. (2021)

Ethical Questions in Emotional Platform Design

Media platforms can infer preferences from behavior and modify exposure in ways that alter feelings. This creates a question of governance: whose goal determines the ranking, what emotional outcomes are measured, and how can a person refuse or correct the system's inference? Commercial engagement metrics can be informative about attention but cannot serve as a valid proxy for well-being or effective coping. A person might remain engaged because content is distressing, uncertain, or difficult to leave. A scientifically informed design should distinguish an expressed preference for information from a desire for prolonged exposure, and should allow users meaningful agency over content, alerts, and exit conditions. Parry et al. (2021) Zillmann (2000)

Emotional personalization can generate distributional concerns. A mechanism that increases average enjoyment may nevertheless expose a subgroup to repeated threat, unwanted comparison, or manipulative relational cues. Research should therefore examine subgroup outcomes, the degree of explainability of the intervention, and the risks of optimizing behavior with unvalidated affective indicators. Youth, people facing severe stress, and those with limited digital accessibility merit careful research safeguards and participation in design. Consent to use a service should not be equated with informed authorization for every research experiment on emotional vulnerability. Independent assessment is especially important where the same organization designs and measures the intervention. Beyens et al. (2020) Fassi et al. (2024; corrected 2026) Valkenburg et al. (2016)

Emotional privacy is an important element of scientific validity as well as ethics. Inferences about sadness, anxiety, grief, or social exclusion may be sensitive even when inferred from ordinary clicks or audio choices. Recording such states requires clear disclosure of which data are collected and for what purpose. Researchers can prefer aggregated measures, local processing, and data-minimizing study protocols when feasible. Public claims about affective safety should name the outcomes, populations, and observation windows examined. Without tested measures and meaningful controls, claims that a ranking system knows a user's emotional needs or has objectively improved mental health exceed the available evidence. Parry et al. (2021) Gross (2015)

Artificial Conversational Systems as Mediated Emotional Contexts

In the Artificial Era, conversational systems add an interactive setting to the study of mediated emotion. A person may ask a system to reframe an event, organize an overwhelming task, compose a supportive message, retrieve relevant information, or provide companionship during a difficult period. The immediate psychological phenomenon is the human user's encounter with a responsive computational interface. Its regulatory pathways may involve attentional redirection, cognitive reappraisal, perceived social presence, and restored task control. These are testable hypotheses about human responses and interface behavior. The existence of fluent conversation alone does not establish a system's subjective emotion or clinical competence. The relevant evidence must measure actual interaction and the user's state before and after it. Gross (2015) Valkenburg et al. (2016)

Adaptive dialogue differs from a fixed playlist because the system generates content conditional on what the user says. This creates opportunities for personalized language and new risks: the model can misconstrue a situation, reinforce repetitive negative thoughts, offer unwarranted certainty, or invite excessive reliance on its responses. A proper evaluation distinguishes the effect of the conversational partner's actual answers from the effect of expectations about who produced them. Randomized comparisons might hold content constant while varying disclosed source, or compare standardized supportive scripts with adaptive responses under independent assessment. Outcome measures should include perceived usefulness, mood, trust calibration, accuracy, and subsequent engagement with people and tasks outside the interface. Gross (2015) Parry et al. (2021)

The theoretical language of Artificial Agency, Artificial Identity, and Artificial Sapiens belongs to the author's broader philosophical research horizon; within this psychological dossier, measurable claims concern interaction structure and human emotional consequences. A research program could compare affective outcomes across systems with different memory, disclosure, availability, and response policies, while respecting privacy and participant safety. Hypotheses about emotional co-regulation should be operationalized through observed turn-taking, changes in reported emotion, and behavior under specified conditions. Such work extends the original media-regulation problem into adaptive conversation without importing unsupported assumptions about consciousness, diagnostic authority, or universal therapeutic benefit. Gross (2015) Valkenburg et al. (2016)

Unresolved Scientific Questions and Falsifiable Predictions

Several unresolved questions follow directly from the evidence reviewed. Does pre-session intention predict affective improvement after controlling for initial distress and actual content? Does the emotional usefulness of a medium depend more on chosen content, control over stopping, or interpersonal context? How long do experimentally induced benefits last outside the laboratory? When does information seeking become repetitive checking with high opportunity cost? Can the same content evoke predictable state changes across different users, languages, and social circumstances? Each question implies a different experimental or longitudinal comparison, and a null result for one does not settle the others. Zillmann (2000) Gross (2015) Beyens et al. (2020)

A first testable prediction is that autonomy over the menu and stopping rule will moderate emotional outcomes even when stimulus categories are held approximately constant. A second is that measured fit between pre-session regulatory goal and actual encounter will explain more within-person variance in post-session affect than daily screen time alone. A third is that algorithmically promoted threat repetition will be more closely associated with sustained arousal than an equally informative fixed-interval summary, conditional on baseline threat and relevance. A fourth is that the apparent success of an emotionally pleasant session will partly depend on the delayed opportunity cost of the displaced activity. These propositions are hypotheses derived from existing evidence, not established empirical findings of the present monograph. Knobloch (2003) Holman et al. (2014) Parry et al. (2021)

A further research issue is the stability of individual differences. Intensive longitudinal studies reveal meaningful heterogeneity, but measurement noise and changing daily circumstances can produce unstable estimates of person-specific susceptibility. Large, multisite studies with repeated interventions and cross-validation are needed before systems can claim to tailor regulation reliably. Commercial platforms complicate cumulative evidence because their interfaces change across cohorts and even within studies. Sustained research infrastructure should therefore include documented stimulus archives, reproducible treatment definitions, transparent data governance, and independent replication. Theory development will benefit from hypotheses that predict failures as well as successes of particular media-regulation strategies under prespecified contexts. Beyens et al. (2020) Parry et al. (2021) Valkenburg et al. (2016)

A Scientific Agenda for Cumulative Research

The most promising next studies connect media choice, affective goal, exposure content, social situation, and subsequent functioning in one design. A longitudinal micro-randomized experiment could vary optional media prompts at suitable moments, record real-time emotional states, and compare changes with participant-chosen alternatives. A factorial laboratory study could distinguish emotional tone, stopping control, and the presence of interpersonal feedback. A cohort panel could map how repeated episodes accumulate or dissipate across weeks while keeping track of major life stressors. Each design should publish a causal estimand, an explicit measurement map, and a plausible counterfactual. This would permit direct testing of the mechanisms that existing broad platform studies often combine. Gross (2015) Allcott et al. (2020) Beyens et al. (2020)

Evidence integration requires common but flexible reporting. Studies could report initial and desired valence and arousal; selection opportunity; exposure provenance; actual content features; duration and interruptions; immediate and delayed emotion; individual differences; and privacy-preserving metadata. These dimensions make comparisons possible while preserving the meanings of music, news, conversation, and games. The researcher must also observe whether a user sought relief, performance readiness, meaning, information, or affiliation. A standardized duration measure alone cannot represent the diversity of these intentions. Consistent reporting would clarify why apparently discordant experiments often estimate distinct psychological questions. Knobloch (2003) Saarikallio & Erkkilä (2007) Reinecke (2009)

Independent replication should prioritize mechanisms that endure beyond a particular commercial platform. Researchers might test whether user control, vivid threat exposure, supportive reciprocity, or prospective mood preparation remains consequential in music apps, video platforms, social networks, and artificial conversational interfaces. Reporting both beneficial and adverse outcomes, as well as null results, would reduce selective attention to dramatic effects. Scientific interpretation should recognize what randomized trials can identify, what observational designs can describe, and what qualitative accounts can reveal about situated meaning. In this way, mediated emotional regulation can develop as a coherent field of cumulative psychological explanation rather than as a series of technology-specific judgments. Valkenburg et al. (2016) Parry et al. (2021) Gross (2015)

Conclusion

Mediated emotion and mood regulation is a distinct psychological object formed where affective goals, communication choices, media content, social interaction, and technological conditions meet. Its historical foundations extend from mood management and everyday self-regulation to contemporary models of emotion generation and digitally organized exposure. Research on music, film, games, social communication, public-crisis news, and platform access establishes that mediated encounters can shape emotional experience, while the direction, duration, and meaning of those effects vary by context and individual. Immediate pleasure, perceived recovery, emotional meaning, informed action, and long-term well-being are separate outcomes that require separate evidence. Zillmann (1988) Gross (2015) Allcott et al. (2020)

The strongest theoretical synthesis joins the user's stated regulatory goal to the particular opportunity set, exposure process, immediate psychological mechanisms, and later affective consequences. It places voluntary choice and algorithmically arranged experience within the same causal sequence while preserving their difference in identifiability. The evidence supports both laboratory manipulations of selective exposure and randomized platform interventions, supplemented by intensive longitudinal observation of real-world variation. It also demonstrates important limits imposed by bidirectional selection, self-report discrepancies, heterogeneous effects, and rapidly evolving platforms. Conclusive inference requires specifying what was changed, compared, measured, and followed through time. Knobloch (2003) Beyens et al. (2020) Parry et al. (2021)

The historical transition From Homo to Artificial adds adaptive conversational and recommendation systems to the environments in which humans pursue emotional goals. Their psychological consequences should be investigated through concrete interaction features, human outcomes, and reproducible interventions. The resulting research program can assess the conditions under which mediated experience supports recovery, understanding, connection, and agency, as well as those in which it intensifies repetitive distress or imposes an opportunity cost. Its enduring scientific contribution will be the systematic explanation of how people seek, encounter, and regulate emotion through mediated environments, supported by careful causal reasoning, valid measurement, and cumulative independent evidence. Gross (2015) Valkenburg et al. (2016)

References

  • Allcott, H., Braghieri, L., Eichmeyer, S., & Gentzkow, M. (2020). The welfare effects of social media. American Economic Review, 110(3), 629–676. https://doi.org/10.1257/aer.20190658
  • Allcott, H., Gentzkow, M., Wittenbrink, B., Mason, W., Wilkins, A., Stroud, N. J., Tucker, J. A., & US 2020 Facebook and Instagram Election Study Collective. (2026). The effect of deactivating Facebook and Instagram on users' emotional state. American Economic Journal: Economic Policy, 18(4), 188–207. https://doi.org/10.1257/pol.20240806
  • Beyens, I., Pouwels, J. L., van Driel, I. I., Keijsers, L., & Valkenburg, P. M. (2020). The effect of social media on well-being differs from adolescent to adolescent. Scientific Reports, 10, Article 10763. https://doi.org/10.1038/s41598-020-67727-7
  • Fassi, L., Thomas, K., Parry, D. A., Leyland-Craggs, A., Ford, T. J., & Orben, A. (2024). Social media use and internalizing symptoms in clinical and community adolescent samples: A systematic review and meta-analysis. JAMA Pediatrics, 178(8), 814–822. [Corrected July 6, 2026.] https://doi.org/10.1001/jamapediatrics.2024.2078
  • Gross, J. J. (1998). The emerging field of emotion regulation: An integrative review. Review of General Psychology, 2(3), 271–299. https://doi.org/10.1037/1089-2680.2.3.271
  • Gross, J. J. (2015). Emotion regulation: Current status and future prospects. Psychological Inquiry, 26(1), 1–26. https://doi.org/10.1080/1047840X.2014.940781
  • Holman, E. A., Garfin, D. R., & Silver, R. C. (2014). Media's role in broadcasting acute stress following the Boston Marathon bombings. Proceedings of the National Academy of Sciences, 111(1), 93–98. https://doi.org/10.1073/pnas.1316265110
  • Holman, E. A., Garfin, D. R., & Silver, R. C. (2024). It matters what you see: Graphic media images of war and terror may amplify distress. Proceedings of the National Academy of Sciences, 121(29), e2318465121. https://doi.org/10.1073/pnas.2318465121
  • Hunt, M. G., Marx, R., Lipson, C., & Young, J. (2018). No more FOMO: Limiting social media decreases loneliness and depression. Journal of Social and Clinical Psychology, 37(10), 751–768. https://doi.org/10.1521/jscp.2018.37.10.751
  • Johannes, N., Vuorre, M., & Przybylski, A. K. (2021). Video game play is positively correlated with well-being. Royal Society Open Science, 8(2), Article 202049. https://doi.org/10.1098/rsos.202049
  • Juslin, P. N., & Västfjäll, D. (2008). Emotional responses to music: The need to consider underlying mechanisms. Behavioral and Brain Sciences, 31(5), 559–575. https://doi.org/10.1017/S0140525X08005293
  • Juslin, P. N., Liljeström, S., Västfjäll, D., Barradas, G., & Silva, A. (2008). An experience sampling study of emotional reactions to music: Listener, music, and situation. Emotion, 8(5), 668–683. https://doi.org/10.1037/a0013505
  • Knobloch, S. (2003). Mood adjustment via mass communication. Journal of Communication, 53(2), 233–250. https://doi.org/10.1111/j.1460-2466.2003.tb02588.x
  • Kross, E., Verduyn, P., Demiralp, E., Park, J., Lee, D. S., Lin, N., Shablack, H., Jonides, J., & Ybarra, O. (2013). Facebook use predicts declines in subjective well-being in young adults. PLOS ONE, 8(8), e69841. https://doi.org/10.1371/journal.pone.0069841
  • Lonsdale, A. J., & North, A. C. (2011). Why do we listen to music? A uses and gratifications analysis. British Journal of Psychology, 102(1), 108–134. https://doi.org/10.1348/000712610X506831
  • Nesi, J., & Prinstein, M. J. (2015). Using social media for social comparison and feedback-seeking: Gender and popularity moderate associations with depressive symptoms. Journal of Abnormal Child Psychology, 43(8), 1427–1438. https://doi.org/10.1007/s10802-015-0020-0
  • Nolen-Hoeksema, S., Wisco, B. E., & Lyubomirsky, S. (2008). Rethinking rumination. Perspectives on Psychological Science, 3(5), 400–424. https://doi.org/10.1111/j.1745-6924.2008.00088.x
  • Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3, 173–182. https://doi.org/10.1038/s41562-018-0506-1
  • Parry, D. A., Davidson, B. I., Sewall, C. J. R., Fisher, J. T., Mieczkowski, H., & Quintana, D. S. (2021). A systematic review and meta-analysis of discrepancies between logged and self-reported digital media use. Nature Human Behaviour, 5, 1535–1547. https://doi.org/10.1038/s41562-021-01117-5
  • Reinecke, L. (2009). Games and recovery: The use of video and computer games to recuperate from stress and strain. Journal of Media Psychology, 21(3), 126–142. https://doi.org/10.1027/1864-1105.21.3.126
  • Saarikallio, S., & Erkkilä, J. (2007). The role of music in adolescents' mood regulation. Psychology of Music, 35(1), 88–109. https://doi.org/10.1177/0305735607068889
  • Saarikallio, S. (2011). Music as emotional self-regulation throughout adulthood. Psychology of Music, 39(3), 307–327. https://doi.org/10.1177/0305735610374894
  • Sagioglou, C., & Greitemeyer, T. (2014). Facebook's emotional consequences: Why Facebook causes a decrease in mood and why people still use it. Computers in Human Behavior, 35, 359–363. https://doi.org/10.1016/j.chb.2014.03.003
  • Tandoc, E. C., Jr., Ferrucci, P., & Duffy, M. (2015). Facebook use, envy, and depression among college students: Is facebooking depressing? Computers in Human Behavior, 43, 139–146. https://doi.org/10.1016/j.chb.2014.10.053
  • Thayer, R. E., Newman, J. R., & McClain, T. M. (1994). Self-regulation of mood: Strategies for changing a bad mood, raising energy, and reducing tension. Journal of Personality and Social Psychology, 67(5), 910–925. https://doi.org/10.1037/0022-3514.67.5.910
  • Valkenburg, P. M., Peter, J., & Walther, J. B. (2016). Media effects: Theory and research. Annual Review of Psychology, 67, 315–338. https://doi.org/10.1146/annurev-psych-122414-033608
  • Van den Tol, A. J. M., & Edwards, J. (2013). Exploring a rationale for choosing to listen to sad music when feeling sad. Psychology of Music, 41(4), 440–465. https://doi.org/10.1177/0305735611430433
  • Verduyn, P., Lee, D. S., Park, J., Shablack, H., Orvell, A., Bayer, J., Ybarra, O., Jonides, J., & Kross, E. (2015). Passive Facebook usage undermines affective well-being: Experimental and longitudinal evidence. Journal of Experimental Psychology: General, 144(2), 480–488. https://doi.org/10.1037/xge0000057
  • Zillmann, D. (1988). Mood management through communication choices. American Behavioral Scientist, 31(3), 327–340. https://doi.org/10.1177/000276488031003005
  • Zillmann, D. (2000). Mood management in the context of selective exposure theory. Communication Yearbook, 23(1), 103–123. https://doi.org/10.1080/23808985.2000.11678971