Lying, deception and strategic omission : definition and evaluation

This thesis aims at improving the definition and evaluation of deceptive strategies that can manipulate information. Using conceptual, formal and experimental resources, I analyze three deceptive strategies, some of which are standard cases of deception, in particular lies, and others non-standard cases of deception, in particular misleading inferences and strategic omissions. Firstly, I consider definitional aspects. I deal with the definition of lying, and present new empirical data supporting the traditional account of the notion (called the ‘subjective definition’), contradicting recent claims in favour of a falsity clause (leading to an ‘objective definition’). Next, I analyze non-standard cases of deception through the categories of misleading defaults and omissions of information. I use qualitative belief revision to examine a puzzle due to R. Smullyan about the possibility of triggering a default inference to deceive an addressee by omission. Secondly, I consider evaluative aspects. I take the perspective of military intelligence data processing to offer a typology of informational messages based on the descriptive dimensions of truth (for message contents) and honesty (for message sources). I also propose a numerical procedure to evaluate these messages based on the evaluative dimensions of credibility (for truth) and reliability (for honesty). Quantitative plausibility models are used to capture degrees of prior credibility of messages, and dynamic rules are defined to update these degrees depending on the reliability of the source.