Attitudes and persuasion: cognitive dissonance, elaboration likelihood model

Intuition Beginner

Attitudes are evaluations — positive or negative feelings toward people, objects, or ideas. They shape what you buy, who you vote for, and how you treat strangers. Leon Festinger discovered that when your actions conflict with your beliefs, you feel uncomfortable tension called cognitive dissonance, and you often resolve it by changing your beliefs to match your actions rather than the other way around.

In a famous experiment, Festinger and Carlsmith paid students either $1or$20 to tell the next participant that a boring task was actually fun. The students paid $20knewtheyhadliedforgoodmoney.Butthestudentspaidonly$1 could not justify lying for such a small amount, so they convinced themselves the task really had been enjoyable. Less justification produced more attitude change.

Persuasion — the deliberate attempt to change an attitude — runs along two routes, described by the elaboration likelihood model. The central route operates when you think deeply about strong arguments, producing lasting change. The peripheral route dominates when you are distracted or unmotivated, relying on surface cues such as the speaker's attractiveness, a catchy slogan, or a celebrity endorsement — change that is real but temporary.

Visual Beginner

Figure: The four pillars of the attitudes-and-persuasion unit. Attitudes decompose into affective, behavioral, and cognitive components that frequently misalign. Cognitive dissonance is the drive to reduce inconsistency between belief and action. The elaboration likelihood model partitions persuasion into a thoughtful central route and a cue-driven peripheral route, gated by motivation and ability. The Yale communication model locates persuasion inside a source-message-channel-receiver framework.

Worked example Beginner

The $1/$20 experiment

Festinger and Carlsmith (1959) asked participants to perform two deliberately dull tasks — turning wooden pegs a quarter turn at a time and restocking a tray of spools. A researcher then asked each participant to tell the next participant (actually a confederate) that the tasks were enjoyable and fun. Some were paid $1;otherswerepaid$20. A control group did no persuading.

When later asked to rate how much they enjoyed the tasks, the $20groupratedthemasboring,matchingthecontrolgroup.The$1 group rated them as significantly more enjoyable. With $20,theliewaseasilyjustifiedbythemoney.Withonly$1, the dissonance between "I said it was fun" and "I would not lie for a dollar" was resolved by deciding the tasks really had been fun.

The result is counterintuitive: smaller rewards produced greater attitude change, because smaller rewards leave more dissonance to resolve. The same logic explains why harsh initiation strengthens group loyalty — the suffering is reinterpreted as evidence that the group was worth suffering for.

Check your understanding Beginner

Formal definition Intermediate

The vocabulary of attitudes and persuasion is standardised across the anchor texts ^{[source pending]}. These terms identify empirically dissociable mechanisms, not mere labels.

Attitude structure (ABC model). An attitude is a learned, global evaluation of a person, object, place, or issue that influences thought and action. Attitudes decompose into three components: affective (how you feel toward the object — "I am afraid of dogs"), behavioural (how you act or intend to act — "I avoid dogs"), and cognitive (what you believe about the object — "dogs are dangerous"). The three components correlate only moderately: you can believe exercise is healthy (cognitive) yet dislike running (affective) and never exercise (behavioural). The attitude-behaviour link is therefore weaker than everyday intuition suggests. Kraus's (1995) meta-analysis found that attitudes predict behaviour at roughly $r\approx .38$, with stronger prediction when the attitude is specific to the behaviour, based on direct experience, and measured at the same level of specificity as the behaviour.

Attitude measurement. The dominant self-report instrument is the Likert scale, in which respondents rate their agreement with statements (typically on a 5- or 7-point scale from "strongly disagree" to "strongly agree"), and item ratings are combined into a composite score. Semantic differential scales rate the attitude object on bipolar adjective pairs (good–bad, strong–weak, active–passive). Self-report assumes that respondents are both willing and able to report their attitudes, which fails when social desirability distorts responding or when the attitude operates outside awareness. The Implicit Association Test (IAT) measures automatic associations by comparing response latencies when a target category (e.g., flowers vs insects) shares a response key with positive vs negative words; faster pairing is interpreted as a stronger implicit association. The IAT has moderate test-retest reliability and predicts small but statistically significant amounts of behaviour, particularly spontaneous decisions, though its interpretation remains debated.

Theory of planned behaviour (Ajzen). The most influential model of the attitude-behaviour link specifies that behaviour is reached through behavioural intention, itself determined by three components:

$$\text{intention}=f(\text{attitude toward the behaviour},\text{subjective norms},\text{perceived behavioural control})$$

where attitude toward the behaviour is the evaluation of performing the specific act (not the attitude toward the object in general), subjective norms are perceived social expectations weighted by motivation to comply, and perceived behavioural control is the perceived ease or difficulty of performing the behaviour, which also has a direct path to behaviour when it reflects actual control. The model explains why general attitudes are weak predictors: a person may hold a positive attitude toward donating blood but face subjective norms against it or perceive that they cannot (needle phobia, schedule conflict), and intention translates to behaviour only when perceived control is high.

Yale communication model (Hovland). The Yale group framed persuasion as a function of four elements — source (who says it), message (what is said and how), channel (the medium), and receiver (to whom). Each element carries empirically studied variables. Source effects include credibility (expertise $\times$ trustworthiness), attractiveness, and the sleeper effect, in which a message from a low-credibility source gains persuasive impact over time as the source dissociates from the message in memory. Message effects include one-sided versus two-sided presentation (two-sided is more effective with an informed or hostile audience), fear appeals (moderate fear paired with a specific recommended action is more persuasive than very high fear, which can produce denial and avoidance), and message framing (gain-framed messages work better for prevention behaviours, loss-framed for detection behaviours). Channel effects include the modality advantage of face-to-face over mass media. Receiver effects include intelligence, self-esteem, and prior attitudes (the boomerang effect, discussed in the master tier, occurs when a strong message pushes the receiver in the opposite direction).

Key experiment Intermediate

Cognitive dissonance: Festinger's paradigms

Leon Festinger's cognitive dissonance theory (1957) proposes that holding two cognitions in a dissonant relation — belief contradicts belief, or behaviour contradicts belief — generates an aversive psychological tension, which motivates reduction by changing one cognition, adding a consonant cognition, or trivialising the conflict ^{[source pending]}. Dissonance is strongest when the cognitions are central to the self, the behaviour is freely chosen and irreversible, and the negative consequences are foreseeable. The theory has been operationalised in four main paradigms.

Paradigm	Manipulation	Finding
Induced compliance (Festinger-Carlsmith 1959)	Pay $\$1$ vs $\$20$ to say a dull task was fun	Less reward $\Rightarrow$ more attitude change (insufficient justification)
Effort justification (Aronson-Mills 1959)	Severe vs mild initiation to join a group	Severe initiation $\Rightarrow$ higher group attractiveness
Free choice (post-decision dissonance)	Choose between two similarly rated items	Chosen item rated higher, rejected item rated lower (spreading of alternatives)
Justification of effort	Effortful vs effortless path to a goal	Effortful path $\Rightarrow$ higher valuation of the goal

The unifying mechanism is insufficient justification: when external reward or threat is too small to explain a counter-attitudinal behaviour, the person supplies an internal justification by shifting the attitude. The reverse — over-justification (large external rewards undermining intrinsic motivation) — is treated separately in the motivation unit.

Rival accounts of the dissonance findings

Dissonance theory dominated the 1960s, then faced three serious competitors, each of which reframes the same data.

Self-perception theory (Bem, 1967). Bem argued that no aversive tension is needed. People infer their attitudes the same way outsiders do — by observing their own behaviour and the context in which it occurred. A participant paid $\$1$ to say the task was fun reasons, "I said it was fun for almost no money, so I must have liked it"; a participant paid $\$20$ reasons, "I said it for the money." The same prediction falls out without invoking arousal. Bem's account is parsimonious, and it handles many dissonance findings, but it struggles with findings that depend on physiological arousal.

Impression management theory. The self-report attitude change observed in dissonance experiments may reflect not genuine belief revision but the participant's attempt to appear consistent to the experimenter. On this view, the internal tension is real enough but the measured attitude shift is a strategic presentation. The theory predicts that attitude change should disappear when responses are anonymous or private — a prediction that holds in some studies but not others, leaving the account as a partial contributor rather than a full replacement.

Self-affirmation theory (Steele, 1988). Steele proposed that dissonance threatens the self-concept, and what people seek to restore is not consistency per se but global self-integrity. In a decisive test, participants given the chance to affirm an unrelated important value (e.g., "I am a compassionate person") before rating the dull task showed no dissonance-driven attitude change. Self-affirmation can eliminate the effect entirely, which dissonance-as-arousal alone cannot explain, suggesting that the meaning of the behaviour for the self is what carries the aversive charge.

The modern synthesis, developed in the master tier, is that dissonance and self-perception each handle part of the territory: dissonance for high-threat, self-relevant, counter-attitudinal behaviour; self-perception for low-threat behaviour where people simply read off their attitudes from their conduct.

The elaboration likelihood model (Petty-Cacioppo)

Petty and Cacioppo's elaboration likelihood model (ELM) is the dominant contemporary account of persuasion ^{[source pending]}. Its central claim is that persuasion runs along one of two routes, determined by the audience's motivation and ability to process the message elaboratively (i.e., to think carefully about the arguments).

	Central route	Peripheral route
Trigger	High motivation and high ability	Low motivation or low ability
Processing	Depth of argument scrutiny	Heuristic cues (source expertise, attractiveness, message length, consensus)
Quality dependence	Strong arguments persuade, weak arguments backfire	Argument quality matters little
Durability	Lasting, resistant to counter-persuasion, predicts behaviour	Temporary, vulnerable to counter-persuasion
Typical context	Expert audience, high personal relevance, low distraction	Casual audience, low involvement, distraction, time pressure

Motivation is driven by personal relevance (the message concerns the receiver directly) and need for cognition (an individual difference — the chronic enjoyment of thinking). Ability is driven by distraction, knowledge (domain expertise enables scrutiny), and message comprehensibility. When both motivation and ability are high, the receiver takes the central route; when either is low, the receiver falls back on peripheral cues. The model is a continuum rather than a strict dichotomy, and a single persuasive attempt can shift a receiver partway along it.

Exercises Intermediate

Advanced results Master

Dissonance versus self-perception: the modern resolution

The dissonance–self-perception debate consumed attitude research for two decades. The decisive synthesis came from Fazio, Zanna, and Cooper (1977), who proposed that the two theories handle different territory. Dissonance operates for behaviour that is high in threat to the self-concept — counter-attitudinal acts with foreseeable negative consequences, where the conflict generates genuine physiological arousal. Self-perception operates for behaviour that is low in threat — acts well within the person's existing attitude range, where there is no conflict and therefore no arousal, only an inference from behaviour to attitude. The two theories are not competitors but complementary accounts of different regions of a continuum.

Two lines of evidence support the partition. Misattribution of arousal (Zanna and Cooper, 1974): if participants induced to behave counter-attitudinally are given a plausible external explanation for the tension they feel (a fake pill side-effect), they no longer change their attitude, because the arousal has been relabelled and no longer requires dissonance reduction. Self-perception cannot explain this, because it never invoked arousal in the first place. Pharmacological blockade (Steele, Southwick, and Critchlow, 1981): participants who consumed alcohol before the counter-attitudinal behaviour showed no dissonance-driven attitude change, consistent with alcohol dampening the arousal that carries the dissonance. The arousal signature is the nontrivial empirical lever that distinguishes the two accounts, and it points firmly to dissonance for high-threat behaviour.

The hypocrisy paradigm and behavioural consequences

Stone and Aronson's hypocrisy paradigm (1994) asks participants first to advocate publicly a behaviour they already endorse (using condoms to prevent HIV), then to recall their own past failures to do so. The combination of public commitment and awareness of personal failure generates intense dissonance — but unlike the classic paradigm, the most available reduction route is not attitude change (the attitude was already pro-attitudinal) but behaviour change. Hypocrisy participants purchased significantly more condoms afterward. The paradigm demonstrates that dissonance can drive behaviour change as well as attitude change, and it underpins interventions in health, environmental, and ethical domains: making people publicly commit to a value, then making their past lapses salient, is a reliable route to behavioural alignment.

ELM advances: multiple roles for variables

The original ELM treated variables as either central-route arguments or peripheral-route cues. The mature model (Petty and Cacioppo, 1986 onward) holds that a single variable can play multiple roles depending on the elaboration level. Source expertise, for instance, functions as a peripheral cue under low elaboration (the receiver defers to the expert without examining the arguments), as an argument under high elaboration (the expert's credentials become a piece of evidence the receiver evaluates), as a determinant of processing extent (a credible source can increase motivation to elaborate), and as a bias on processing direction (a liked source biases interpretation toward agreement, a disliked source toward disagreement). The multiple-roles principle explains why meta-analyses of source effects yield inconsistent effect sizes: the same variable produces different effects at different elaboration levels, and averaging across levels obscures the structure.

Metacognitive model and attitude strength

Petty, Briñol, and Cacioppo's metacognitive model (2007) extends the ELM by distinguishing an attitude (the evaluation itself) from the meta-attitudinal thoughts about that attitude — how valid, how certain, how accessible the person believes the attitude to be. A person can hold an attitude with high or low attitude certainty independent of the attitude's extremity, and certainty is itself persuadable: a strong argument can increase certainty, a weak source can decrease it, even when the underlying evaluation is unchanged. Attitudes held with high certainty, high accessibility, and high ambivalence-free confidence are strong attitudes, which predict behaviour better, resist counter-persuasion, and persist over time. The practical consequence is that persuasion campaigns should target not only the direction of the attitude but the confidence with which it is held, because an attitude without conviction is behaviourally inert.

Uncertainty-identity theory

Hogg's uncertainty-identity theory (2000 onward) links persuasion and attitude change to group identification. People are motivated to reduce subjective uncertainty about themselves and their world, and identification with a group — especially a highly entitative, belief-convergent group — reduces that uncertainty by supplying a clear self-definition and a ready-made belief structure. Under uncertainty, people are more persuaded by group-endorsed positions, more resistant to counter-group messages, and more attracted to extremist groups whose rigid belief systems minimise ambiguity. The theory bridges social-identity research (covered in 29.07.01) with the persuasion literature: uncertainty raises the persuasive weight of group consensus cues, which is why periods of social upheaval produce both heightened persuasibility and heightened polarisation.

Persuasion neuroscience

Cognitive neuroscience has begun to map the neural substrates of attitude change. The ventromedial prefrontal cortex (vmPFC) encodes an integrated value signal that tracks the subjective value of attitude objects and shifts as attitudes change. Falk and colleagues (2010, 2011) demonstrated a striking finding: activity in the medial prefrontal cortex (mPFC) during exposure to a persuasive message about sunscreen use predicted subsequent behaviour change better than the participants' own stated intentions. The mPFC signal appears to index self-related processing — the degree to which the persuasive message is being integrated with the self-concept — and this self-integration is a better predictor of action than explicit agreement. The implication for persuasion research is that the central route is not merely argument scrutiny but the recruitment of self-related neural systems, which is why messages framed in terms of the receiver's identity and values are more durable than messages framed in abstract terms.

Boomerang effect and psychological reactance

Brehm's psychological reactance theory (1966) predicts that when people perceive a threat to their freedom to hold a particular attitude or perform a particular behaviour, they experience reactance — a motivational state aimed at restoring the threatened freedom. The most visible consequence is the boomerang effect: a strong, heavy-handed persuasive message pushes the receiver in the opposite direction, strengthening the original attitude. Reactance is amplified when the source is seen as coercive, when the freedom is highly valued, and when the message implies the receiver has no choice. The boomerang effect explains why prohibition-style anti-drug messages can increase drug use among high-sensation-seeking adolescents, why parental pressure can entrench a forbidden relationship (the Romeo-and-Juliet effect), and why heavy-handed public-health messaging can backfire. The practical implication is that persuasion must preserve the receiver's sense of autonomous choice — a constraint that connects persuasion research to self-determination theory in the motivation unit.

Nudging versus persuasion

Thaler and Sunstein's nudge programme (2008) distinguishes changing choice architecture (altering the environment in which decisions are made — defaults, ordering, framing of options) from changing minds (altering the attitudes and beliefs that underlie decisions). A nudge does not attempt to persuade; it exploits the biases catalogued in 29.05.02 (status quo bias, loss aversion, framing) to steer behaviour without engaging the receiver's attitudes at all. Opt-out organ donation dramatically increases donation rates without changing anyone's attitude toward donation; automatic retirement enrolment raises savings rates without persuading anyone that saving is good. The distinction matters because nudging and persuasion have different ethical profiles: persuasion engages the receiver as a reasoning agent and is therefore more transparent, while nudging can operate below awareness and is therefore more vulnerable to misuse — the domain of dark patterns.

Dark patterns in digital persuasion

Digital platforms engineer persuasion environments that exploit peripheral-route processing at scale. Variable-ratio reinforcement schedules (slot-machine design — intermittent, unpredictable rewards) are embedded in social-media notifications, email refreshes, and infinite scroll, producing the same compulsive checking behaviour that operant conditioning produces in laboratory animals. Social proof cues (like counts, view counts, trending badges) manufacture consensus without underlying consensus. Friction asymmetry (one click to engage, many clicks to disengage or delete) exploits status quo bias to retain users. Fear-of-missing-out triggers exploit loss aversion. These techniques are not new in principle — they are the peripheral cues of the ELM and the compliance principles of Cialdini, applied at scale and optimised by machine learning. What is new is the asymmetry of intent: the platform's persuasive objective (maximise engagement) is not aligned with the user's welfare, which is the structural definition of a dark pattern. The ELM framework predicts that users distracted by the design will process peripherally and therefore be more susceptible, which is exactly what engagement metrics are engineered to produce.

Cultural variation in persuasion

The ELM, the Yale model, and dissonance theory were developed and tested predominantly on American, individualist samples. Cross-cultural research shows that the mechanisms generalise but the weights do not. Collectivist cultures with interdependent self-construals respond more strongly to consensus cues (what the group believes), authority cues (what a respected figure says), and relationship-based appeals, while individualist cultures with independent self-construals respond more strongly to individual-benefit arguments and autonomy-respecting framings. Han and Shavitt (1994) found that Korean advertisements emphasised in-group benefit, harmony, and family more than equivalent American advertisements, which emphasised personal success, individuality, and self-improvement — and that these culturally matched framings were more persuasive within each culture. The practical implication is that a persuasion campaign cannot be exported across cultures by translation alone: the peripheral cues and even the central arguments that work in one self-construal context may be inert or counterproductive in another. This connects to the WEIRD critique developed in 29.07.01 and 29.12.01.

Connections Master

• Social psychology: groups, prejudice 29.07.01 is the direct prerequisite and parent unit. The compliance principles (Cialdini), social-identity processes, and conformity dynamics introduced there are specialised here into the attitude-change mechanisms that drive them. Attitude change is the individual-level engine of the group-level phenomena covered in the parent unit.

• Decision-making and judgment 29.05.02 pending connects through shared dual-process architecture. The ELM's central and peripheral routes are the persuasion-domain analogue of Kahneman's System 1 and System 2: peripheral-route processing is fast, automatic, and cue-driven; central-route processing is slow, effortful, and argument-driven. The framing effect and loss aversion catalogued there are the mechanisms by which message framing and choice architecture operate here.

• Learning and memory 29.04.01 connects through classical and operant conditioning as attitude-formation mechanisms (repeated pairing of an object with positive or negative stimuli shapes the affective component of an attitude), and through the role of memory in the sleeper effect (source-message dissociation over time).

• Motivation and emotion 29.11.01 (pending) connects through self-determination theory, which constrains persuasion: messages that threaten autonomy produce reactance, while messages that support autonomy produce internalised, durable change. The overjustification effect links extrinsic reward to attitude change here.

• Cognition and intelligence 29.05.01 connects through need for cognition (the individual difference that determines central-route motivation), working memory capacity (which determines central-route ability), and the role of prior knowledge in argument scrutiny.

• Cross-cultural psychology 29.12.01 connects through the cultural variation in persuasion documented in this unit's master tier: independent versus interdependent self-construals shift the weight of peripheral cues, and collectivist cultures respond more to consensus and authority framings.

• Neuroscience [29.02.NN] (pending) provides the substrate for persuasion neuroscience: the vmPFC value signal, the mPFC self-related processing that predicts behaviour change, and the amygdala's role in fear appeals.

• Social psychology successor 29.07.03 pending (proposed) extends attitude-change mechanisms into the interpersonal and relational domain, where the persuasion dynamics covered here operate in the context of close relationships and interpersonal influence.

Historical & philosophical context Master

The study of attitudes is nearly as old as scientific psychology itself. The term "attitude" entered social psychology through Thomas and Znaniecki's The Polish Peasant in Europe and America (1918), where it denoted a relatively stable orientation toward a social object. Gordon Allport's 1935 handbook chapter declared the attitude to be "the most distinctive and indispensable concept in contemporary American social psychology" — a status it retained through the mid-twentieth century, when the field was organised almost entirely around attitude measurement and attitude change.

The persuasion research programme proper began during the Second World War, when Carl Hovland was recruited to the US Army's Information and Education Division to study the effectiveness of military training films and propaganda. Hovland's team asked a question of immediate practical importance: do one-sided messages (presenting only the sender's case) or two-sided messages (acknowledging and rebutting the opposing case) persuade more effectively? The answer — that two-sided messages are more effective with an informed or initially hostile audience — became a foundational finding, and after the war Hovland moved to Yale, where he founded the communication-and-persuasion programme that produced the Yale communication model. The Yale programme was explicitly modelled on a stimulus-response learning framework: persuasion was treated as a learning process in which the receiver learns a new attitude, and the source-message-channel-receiver taxonomy identified the variables that governed that learning.

Leon Festinger's arrival at this landscape produced the field's most consequential theoretical innovation. Festinger had studied under Kurt Lewin at Iowa and followed Lewin to MIT, where Lewin founded the Research Center for Group Dynamics. Festinger's 1957 A Theory of Cognitive Dissonance proposed that the drive for cognitive consistency is itself a motivational state — that the perception of inconsistency among one's cognitions generates an aversive tension comparable to hunger or thirst, and that the organism is motivated to reduce it. The theory was radical because it located motivation in the structure of cognition rather than in biological drives or external reinforcement. The Festinger-Carlsmith experiment (1959) was designed to test a counterintuitive prediction: if dissonance is what drives attitude change, then reducing the external justification for a counter-attitudinal act should increase attitude change, because the person must supply internal justification to reduce the larger residual dissonance. The confirmation of this prediction — that $\$1$ produced more change than $\$20$ — is one of the most famous results in the discipline, precisely because it violates the commonsense assumption that larger rewards produce larger effects.

The dissonance programme dominated the 1960s, generating the effort-justification, free-choice, and induced-compliance paradigms catalogued above. It also generated the field's most serious theoretical challenge. Daryl Bem's self-perception theory (1967) was not merely a competing account but a methodological provocation: Bem argued that the dissonance findings could be reproduced by a model in which participants simply inferred their own attitudes from their observed behaviour, with no need for the unobservable intervening variable of "aversive arousal." Bem's challenge was a version of the behaviourist critique of intervening variables generally: if the same predictions follow from a simpler model that invokes no unobservable states, parsimony favours the simpler model. The debate was unresolved by purely behavioural data — both theories made the same predictions in the standard paradigms — and was settled only when the arousal signature was isolated through misattribution and pharmacological studies in the 1970s and 1980s, confirming that dissonance involves genuine physiological arousal for high-threat behaviour.

Richard Petty and John Cacioppo's elaboration likelihood model, developed in the early 1980s, reframed persuasion research by importing the dual-process architecture that was transforming cognitive psychology more broadly. Where the Yale model treated persuasion as a single learning process with many variables, the ELM proposed that persuasion operates through two qualitatively different processes depending on the audience's elaboration level. The model was influential because it reconciled a large body of inconsistent findings — source effects sometimes help and sometimes hurt, message length sometimes persuades and sometimes backfires — by showing that the same variable plays different roles at different elaboration levels. The ELM's multiple-roles principle became the dominant organising framework for persuasion research from the late 1980s onward.

The contemporary landscape has been reshaped by three developments. First, persuasion neuroscience (beginning with Falk and colleagues in the late 2000s) has shown that neural activity during message exposure predicts behaviour change better than self-reported intentions, raising both methodological opportunities (brain-based prediction) and conceptual questions (whether attitude change is best understood at the neural rather than the cognitive level). Second, behavioural economics and nudge (Thaler and Sunstein, 2008) have drawn a sharp line between changing minds (persuasion) and changing choice architecture (nudging), with distinct ethical profiles and distinct policy implications. Third, digital persuasion at scale has created environments — social-media feeds, recommendation systems, engagement-optimised interfaces — in which the peripheral cues of the ELM and the compliance principles of Cialdini operate continuously, below awareness, and with an intent asymmetry (the platform's interests diverge from the user's) that the original theories did not anticipate. The ELM framework remains the most useful lens for analysing these environments, because it predicts that distraction and low involvement — exactly the conditions digital interfaces engineer — route processing through the peripheral path, which is exactly where cue-based manipulation is most effective.

The philosophical weight of the persuasion literature concerns autonomy. If attitudes can be changed by manipulating peripheral cues below the threshold of awareness, and if behaviour can be steered by altering choice architecture without engaging the agent's reasoning at all, then the boundary between influence and manipulation, and between persuasion and coercion, is harder to draw than democratic theory assumes. The liberal tradition presupposes that citizens hold autonomous attitudes that they can revise through reasoning; the persuasion literature shows that attitudes are also revisable through mechanisms that bypass reasoning, and that the conditions under which people reason carefully (central route) are themselves manipulable. The ethical question — under what conditions persuasion respects autonomy, and when it violates it — is now central to research on dark patterns, platform regulation, and the ethics of behavioural public policy.

Bibliography Master

1. Festinger, L. — A Theory of Cognitive Dissonance (Stanford University Press, 1957).

2. Festinger, L. & Carlsmith, J. M. — "Cognitive consequences of forced compliance", Journal of Abnormal and Social Psychology 58, 203–210 (1959).

3. Aronson, E. & Mills, J. — "Effect of severity of initiation on liking for a group", Journal of Abnormal and Social Psychology 59, 177–181 (1959).

4. Bem, D. J. — "Self-perception: an alternative interpretation of cognitive dissonance phenomena", Psychological Review 74, 183–200 (1967).

5. Steele, C. M. — "The psychology of self-affirmation: sustaining the integrity of the self", in Berkowitz (ed.), Advances in Experimental Social Psychology Vol. 21 (Academic Press, 1988), pp. 261–302.

6. Fazio, R. H., Zanna, M. P. & Cooper, J. — "Dissonance and self-perception: an integrative view of each theory's proper domain of application", Journal of Experimental Social Psychology 13, 464–479 (1977).

7. Zanna, M. P. & Cooper, J. — "Dissonance and the pill: an attribution approach to studying the arousal properties of dissonance", Journal of Personality and Social Psychology 29, 703–709 (1974).

8. Steele, C. M., Southwick, L. & Critchlow, B. — "Dissonance and alcohol: drinking your troubles away", Journal of Personality and Social Psychology 41, 831–846 (1981).

9. Stone, J., Aronson, E., Crain, A. L., Winslow, M. P. & Fried, J. B. — "Inducing hypocrisy as a means of encouraging young adults to use condoms", Personality and Social Psychology Bulletin 20, 116–126 (1994).

10. Petty, R. E. & Cacioppo, J. T. — Communication and Persuasion: Central and Peripheral Routes to Attitude Change (Springer, 1986).

11. Petty, R. E., Briñol, P. & Cacioppo, J. T. — "The metacognitive model (MCM) of attitudes: implications for attitude measurement, change, and strength", Social Cognition 25, 657–686 (2007).

12. Ajzen, I. — "The theory of planned behavior", Organizational Behavior and Human Decision Processes 50, 179–211 (1991).

13. Kraus, S. J. — "Attitudes and the prediction of behavior: a meta-analysis of the empirical literature", Personality and Social Psychology Bulletin 21, 58–75 (1995).

14. Hovland, C. I., Janis, I. L. & Kelley, H. H. — Communication and Persuasion (Yale University Press, 1953).

15. Brehm, J. W. — A Theory of Psychological Reactance (Academic Press, 1966).

16. Hogg, M. A. — "Subjective uncertainty reduction through self-categorization: a motivational theory of uncertainty-identity theory", Psychological Review 127, 555–572 (2000 onward, consolidated 2020).

17. Falk, E. B., Berkman, E. T., Mann, T., Harrison, B. & Lieberman, M. D. — "Predicting persuasion-induced behavior change from the brain", Journal of Neuroscience 30, 8421–8424 (2010).

18. Thaler, R. H. & Sunstein, C. R. — Nudge: Improving Decisions About Health, Wealth, and Happiness (Yale University Press, 2008).

19. Han, S.-P. & Shavitt, S. — "Persuasion and culture: advertising appeals in individualistic and collectivistic societies", Journal of Experimental Social Psychology 30, 326–350 (1994).

20. Cialdini, R. B. — Influence: The Psychology of Persuasion (William Morrow, 1984; revised editions 2006, 2021).

21. Myers, D. G. & DeWall, C. N. — Psychology, 13th ed. (Worth, 2021), Ch. 11.

22. Gleitman, H., Gross, J. & Reisberg, D. — Psychology, 8th ed. (Norton, 2011), Ch. 11.