The notion that overly vehement avowals and overly emphatic behaviors betray knowledge of a disavowed reality is not new. In Hamlet, the lady's vow of fidelity to her husband is so passionate and insistent as to arouse suspicion. One possibility is that she is a pure hypocrite, attempting to deceive her audience while knowing full well that her feelings are otherwise. A less cynical observer, however, might conclude that she is only attempting to deceive herself.

For von Hippel & Trivers (VH&T), self-deception and other-deception are not mutually exclusive possibilities. Their evolutionary claim is that the former has evolved in order to facilitate the latter. As they acknowledge, this claim has received surprisingly little attention in the empirical literature (but see McKay & Dennett Reference McKay and Dennett2009), which makes the hypothesis almost entirely speculative but not for that reason any less interesting.

The aspect that we focus on here is the psychological architecture that enables self-deception. Although VH&T endorse a “non-unitary mind,” defined by separate mental processes with access to privileged information, they resist treating these processes as fully fledged subagents with distinct interests, decision roles, and modes of interaction. Consequently, their theory leaves unresolved the crucial issues of author, audience, and credibility in self-deceptive speech.

Observe, first, that for VH&T the benefits of self-deception are defined as performance enhancement: The “self-deceived deceiver” puts on a smoother show and makes fewer slips that might give the game away. What seems to be ignored in this performance-centered account is the moral dimension of deception. One may ask why psychopathy is not a universal condition if glib performance is so valuable from an evolutionary standpoint.

An alternative interpretation is available, namely, that the benefits of self-deception are realized in the internal moral economy of the self-deceiving individual: The conveniently self-deceived deceivers are absolved from the burden of dealing with unpleasant awareness of their own treachery (Elster Reference Elster1999). Like Russell's Gladstone, they have license to deceive others without any attendant loss of self-esteem.

On this interpretation, therefore, the motive to self-deceive arises from a desire to perceive oneself as a moral agent. There remains the question of whether the desire will be satisfied, whether ostensibly self-deceptive judgments and affirmations will achieve their goal (Funkhouser Reference Funkhouser2005). This issue of self-credibility can be assessed if we view self-deceptive speech as a form of self-signaling, the attempt to convince ourselves that we possess some desired underlying characteristic or trait (Mijović-Prelec & Prelec Reference Mijović-Prelec and Prelec2010). If the self-signaling attempt does succeed, and the characteristic is also socially desirable, then guilt-free deception of others may follow as a collateral benefit. However, even if it fails, and fails repeatedly, that need not remove the compulsion to self-signal. Ritualistic affirmations may remain in force, even as they fail to convince.

The prediction emerges once we conceptualize self-signaling by analogy to signaling between individuals. In theoretical biology, signaling refers to actions taken by a sender to influence the beliefs of receivers about the sender's unobservable characteristics, for example, reproductive quality (Grafen Reference Grafen1990). The sender plays offense by emitting signals that exaggerate his qualities, and the receiver plays defense by discounting or ignoring the messages altogether. The tug of war between offense and defense encourages futile but costly signaling. Even if senders with inferior characteristics do succeed in perfectly emulating the signals emitted by their superiors, the receiver, according to theory, will take this into account and will discount the value of the signal accordingly. The signaling equilibrium is a losing proposition all round; what makes it stick is the fact that failure to send the mandated signal immediately brands the deviant as undesirable.

With self-signaling, this entire dynamic is internalized, and messages conveying desired characteristics are reinterpreted as messages to oneself (Bodner & Prelec Reference Bodner, Prelec, Brocas and Carillo2003; Quattrone & Tversky Reference Quattrone and Tversky1984). The details of this approach are spelled out elsewhere (Mijović-Prelec & Prelec Reference Mijović-Prelec and Prelec2010), but the basic assumption, with respect to psychological architecture, is that there is a division of labor between a sender subsystem responsible for authoring signals, and a receiver subsystem responsible for interpreting them. It is crucial that the two subsystems cannot share information internally, but only through externalized behavior.

What determines whether attempted self-deception is successful? As in the interpersonal case, it all hinges on the credulity of the receiver. If the receiver takes the sender's signal at face value, not discounting for ulterior motives, then attempted self-deception will succeed and we have the “Gladstone” mode. However, the receiver may also discount the signal. This might occur because the receiver has some prior expectation of an ulterior sender motive, or because the deceptive sender misjudges the signal strength. Interestingly, however, discounting may not eliminate the sender's motive to self-signal, because self-serving and pessimistic statements may be discounted asymmetrically (the latter lack an obvious ulterior motive). In such cases, self-deceptive speech becomes mandatory not because it is believed but because deviating from the self-deceptive norm could lead to a catastrophic loss in self-esteem. Self-signaling can therefore lead to ritualistic expression that appears self-deceptive on the surface but that may not truly reflect what a person feels. There will be a mismatch, often noted in the psychotherapeutic literature (Shapiro Reference Shapiro1996), between beliefs-as-expressed, for example, about one's self-esteem, sexuality, future prospects, family relationships, and so forth, and beliefs as actually experienced.

If VH&T's evolutionary story is right, then individuals who cannot deceive themselves will be poor at deceiving others. This would not, however, preclude occasional dissociations between self-deception and the deception of others. Some individuals with crushing self-doubts may fail to conceal these doubts from themselves yet manage to maintain an external façade of confidence. Others, with sufficiently credulous receiver subselves, may manage to convince themselves of their self-worth; if, however, their self-aggrandizing statements ring hollow to others, they may be suspected – and accused – of protesting too much.