Explicit Instructions Increase Cognitive Costs of Deception in Predictable Social Context

Convincing participants to deceive remains one of the biggest and most important challenges of laboratory-based deception research. The simplest and most prevalent method involves explicitly instructing participants to lie or tell the truth before presenting each task item. The usual finding of such experiments is increased cognitive load associated with deceptive responses, explained by necessity to inhibit default and automatic honest responses. However, explicit instructions are usually coupled with the absence of social context in the experimental task. Context plays a key role in social cognition by activating prior knowledge, which facilitates behaviors consistent with the latter. We hypothesized that in the presence of social context, both honest and deceptive responses can be produced on the basis of prior knowledge, without reliance on truth and without additional cognitive load during deceptive responses. In order to test the hypothesis, we have developed Speed-Dating Task (SDT), which is based on a real-life social event. In SDT, participants respond both honestly and deceptively to questions in order to appear similar to each of the dates. The dates are predictable and represent well-known categories (i.e., atheist or conservative). In one condition participants rely on explicit instructions preceding each question (external cue). In the second condition no explicit instructions are present, so the participants need to adapt based on prior knowledge about the category the dates belong to (internal cue). With internal cues, reaction times (RTs) are similar for both honest and deceptive responses. However, in the presence of external cues (EC), RTs are longer for deceptive than honest responses, suggesting that deceptive responses are associated with increased cognitive load. Compared to internal cues, deception costs were higher when EC were present. However, the effect was limited to the first part of the experiment, only partially confirming our initial hypothesis. The results suggest that the presence of social context in deception tasks might have a significant influence on cognitive processes associated with deception.