Narrow Results

We are fascinated by the idea of giving life to the inanimate. The fields of Artificial Life and Artificial Intelligence (AI) attempt to use a scientific approach to pursue this desire. The first steps on this approach hark back to Turing and his suggestion of an imitation game as an alternative answer to the question "can machines think?".¹ To test his hypothesis, Turing formulated the Turing test¹ to detect human behavior in computers. But how do humans pass such a test? What would you say if you would learn that they do not pass it well? What would it mean for our understanding of human behavior? What would it mean for our design of tests of the success of artificial life? We report below an experiment in which men consistently failed the Turing test.

This paper introduces the problem of active deception into the literature on search theory. We consider a game of ‘hide and seek’ in which a hider chooses to hide in one of many cells. Our hider also wishes to remain active which runs the risk of signaling his location. The searcher can look in any cell within the hider’s feasible operating environment but must prioritize his search. To do so, he looks for and has the option of following any indicator of the hider’s possible whereabouts. In conducting his search, however, he also risks signaling where he is about to look next. This gives the hider an opportunity to evade. Each player can employ ‘deceptive signals’, at some cost, to improve their respective chances of success. We examine the nature of the tradeoffs involved in deciding whether or not to use deceptive tactics under different assumptions about the nature of the players’ common operating environment.

This study aimed to evaluate a lie-detection system by nonlinear analysis of electrooculography (EOG) signals in the polygraph test. The physiological signals such as photoplethysmography signal, electrodermal response, respiratory changes as well as EOG signal were recorded based on a Control Question Test (CQT). Three psychophysiological signals were evaluated based on the extracted features in the seven-position numerical scoring. The dynamics of EOG signals in subjects that had a total negative score were analyzed by recurrence quantification analysis (RQA). The six values of RQA were calculated to analyze the EOG signals in relevant questions compared to other questions. A one-way ANOVA with multiple comparisons was performed to evaluate the extracted variables in different questions. Eleven subjects had a total score of $-$2 and less, so the EOG signals of these subjects were evaluated. Recurrence plots (RPs) of EOG signals showed clear differences in the two types of questions. The recurrence quantification analysis of vertical EOG signal indicated that $L_{max}$ and determinism (DET) values decreased significantly for relevant questions compared to other questions. Moreover, a significant decrease was observed in all RQA parameters except RR for the horizontal EOG signal. The differences of EOG signals in relevant questions observed using RPs and RQA were possibly related to the underlying changes in rapid eye movement due to the stress. The results of this study illustrate that the RQA technique is well suited to analyze the EOG signals in the detection of deception.

Nations are becoming increasingly sensitive about securing their borders and ensuring that those who infiltrate do not intend to do harmful things in the country. This has led border security organizations such as the Department of Homeland Security (DHS) to investigate better and faster technologies to screen border crossers (Nunamaker et al., 2011). Some of the screening technologies being investigated involve the analysis of the words spoken by border crossers during interviews. Researchers believe that by analyzing the words chosen by interviewees, algorithms can be developed to detect deception (Burgoon et al., 2014).

One of the problems with linguistics of deception research is that there are inconsistent findings regarding which cues are indicative of deception and truth-telling (e.g. Bond Jr. and Depaulo, 2006; Zhou et al., 2004). Because of this problem, linguistic analysis systems are limited in their usefulness in detecting deception. A comprehensive systematic review of the linguistics of deception needs to be performed to create a theoretical model of the constructs relating to the linguistics of deception and to better inform DHS and similar organizations on how to best implement linguistic systems on national borders.

This chapter will perform a meta-analysis on linguistic cues of deception to (1) inform linguistic theory, (2) enhance linguistic analysis systems, and (3) inform national security agencies on the linguistics of deception. Linguistic analysis is used by researchers and practitioners to find interesting patterns in communications, such as customer support chats and police investigations (Zhou et al., 2004). Linguistic analysis systems analyze textual language and synthesize words into linguistic features to represent cognitive or emotional states (Moffitt et al., 2012). Linguistic analysis systems, such as Linguistic Inquiry and Word Count (LIWC) (Tausczik and Pennebaker, 2009), General Architecture for Text Engineering (GATE) (Gaizauskas et al., 1996), Structured Programming for Linguistic Cue Extraction (SPLICE) (Moffitt et al., 2012), and Agent99 (Cao et al., 2003) are becoming increasingly popular in research to quantify linguistic features and to detect cognitive or emotional states of communicators that are present in the data. There are hundreds of different cues that these systems catalog, ranging from simple counts of pronouns to calculations of the expressiveness of the communicator. While most single studies that use these systems perform exploratory factor analyses to group the linguistic cues, these studies usually produce different groupings of cues that cause division in the sciences and unclear directions for practitioners (e.g. Bond Jr. and Depaulo, 2006; Zhou et al., 2004). Therefore, linguistic systems can be enhanced by solidifying the correct groups of cues based on a broad set of studies. The meta-analysis in this research will group linguistic features into more appropriate and empirically-driven categories of cognitive and emotional states.

Robots progressively become involved in more complex tasks autonomously interacting with people in public and/or private social spaces. The need to produce efficient but also aesthetically interesting and consequently engaging interfaces makes the topic of robots' appearance and that of the adequacy of their social behaviour a central issue in robotics. It is within this framework that the issue of “deception” comes up as an ethical issue.