Deception is a well studied problem in computer mediated communication. As computer mediated communication becomes pervasive, from the traditional office and home environments to health-care and education, the problem of deception will have increasing ramifications. This paper explores mechanisms to embed deception avoidance techniques into computer mediated communication instead of more widely studied solutions of deception detection and reducing deception using media richness.
Pro-social cues have been known to enforce positive behavior in organizational and child development contexts. We examine pro-social cues in psychology literature and develop corresponding technological solutions that can be applied to computer mediated communication. We review these solutions in the context of reducing deception and hypothesize that pro-social cues should be equally effective in reducing deceptive behavior as they are in encouraging positive behavior. We propose a technological solution of embedding these cues during profile creation and propose an experimental study where these cues can be tested in a controlled manner.
Six Myths about Ontologies: The Basics of Formal Ontology
Pro-social cues to reduce digital deception
1. Pro-social cues to reduce digital deception
Asim Kadav
Department of Computer Sciences
University of Wisconsin-Madison
2. Problem of digital deception
» Deliberate control of communication in
technology mediated message(Hancock
2007)
»Perceived to be widespread online(Caspi
& Gorsky, 2006 & Gibbs, Ellison & Heino,
2006)
» Increasing repercussions as CMC
becomes pervasive
3. Types of digital deception
» Identity based versus message based (Hancock, 2007)
» Self-presentational goals are the most important factor in
promoting and constraining deception (De Paulo, 1996)
» Identity based deception is more common online than in F2F
(Cornwell & Lundgren, 2001) and is very common (George,
2004)
How to tackle the problem of digital deception?
4. Deception Detection
» Increased public self consciousness : (Vrij, Edward, & Bull, 2001)
» Increased speech frequency, hesitation, and wait times before
speech
» Use fewer illustrations
» Linguistic attributes: Lower complexity, fewer self-references, more
negative emotion words.(Newman et.al., 2003)
» No “Pinocchio nose” indicator of deception
» Deception detection performs at chance levels (Bond & DePaulo,
2006)
» Better to focus on deception reduction
5. Deception reduction
» Existing research mostly focuses on channels for reducing
deception
» Broadly, use real world indicators to constrain deception
» Signaling theory and use of Warrants
6. Signaling theory
» Disconnect between physical and virtual self reduces cost of
deceptive behavior (Donath 1999)
» Use of signals, representing relationship between virtual and
physical as indicators of honesty
» Cost associated with signals keeps them reliable (Donath
2007)
7. Warranting theory
» Walther and Parks (2002) introduce warrant – a connection
between the physical and online self identity
» Warrants constrain occurrence of digital deception
» Presence of social network may act as a warrant to constrain
deception(Toma, Hancock & Ellison, 2008)
» Media rich with warrants contain lower levels of deception
(Warkentin, Woodworth, Hancock, & Cormier, 2010).
8. Can we do better?
» Deception detection not conclusive
» Deception reduction in channels: Persons can deceive where
there signals and warrants are not required/apparent
» Propose: Deception reduction in individuals as opposed to
channels
» Propose: Using pro-social cues to reduce deception
10. Pro-social behavior
» Prosocial behavior refers to "voluntary
actions that are intended to help or benefit
another individual or group of individuals"
(Eisenberg and Mussen, 1989)
» Examples include cooperation, sharing,
honesty, positive attitude towards others
» Motivations include generosity, morality,
altruism
Image: neurosciencenews.com
11. Why pro-social behavior?
» Humans are inherently selfish (Dawkins, 2006)
» Then, why would humans indulge in pro-social
behavior?
» Humans have evolved to create and maintain
individually profitable social relationships (Fessler
& Haley, 2003)
» Consciously or sub-consciously respond to subtle
cues to show pro-social behavior(Fessler & Haley,
2005)
12. Pro-social cues
» Pro-social cues are cues which
consciously or sub-consciously cause an
individual to behave in a positive pro-
social behavior
» Examples include presence of eyes,
empathy, rewarding behavior, human
voice, praise (rewarding behavior),
dispositional praise
13. Why humans respond to pro-social cues?
» Goal: Humans have evolved to participate in individually profitable social
relationships
» Cooperation due to reputation preservation (Haley and Fessler, 2003)
» Presence of eyes, auditory signals, faces
» Pro-social behavior due to pro-social self image (Damon and Lerner, 2006)
» Provision of attributions like dispositional praise
» Pro-social behavior due to positive reinforcement learning (Berwick, 1986)
» Empathy, rewarding behavior, praise
14. Presence of eyes
» Cues of being watched improved generosity
(Bateson, Nettle, Roberts, 2006)
» Attributed to maintaining pro-social reputation for
subsequent cooperation(Bateson, Nettle, Roberts,
2006)
» Reputation affects co-operation directly
(Panchanathan and Boyd, 2004)
15. Human voice
» Auditory presence of others increases generosity(Haley
and Fessler, 2006; Mayer, Sobko, & Mautone, 2003).
» Attributed to reputational concerns
Image: happyybrainstorm.com
Presence of friends
» Presence of faces improves generosity (Haley and Fessler, 2003)
» Display friend connections to risk one’s reputation and ensure cooperation(Donath
and Boyd 2004)
16. Provision of Attributions & Dispositional praise
» People are more likely to behave in pro-social manner when
induced to behave pro-socially with internal attributions
(Damon and Lerner, 2006)
» Like dispositional praise – “I guess you are a kind person,
who likes to help others”
» Fosters a pro-social self image that may last for weeks
(Damon and Lerner, 2006)
17. Empathy
» Emapthy, or feeling as the other feels, establishes a two-way relationship that
subsequently encourages pro-social behavior (Batson, 1991; Vaish, Carpenter, &
Tomasello, 2009; White & Mitchell, 1979 )
» Empathy is also known to be easily communicable via minimal cues (Same
references as above)
Praise and other rewarding behavior
» Rewarding behavior such as praise, improves co-operation in organizational
contexts(Gove, 2005) and in also in children(Christopher, 1989)
» Humans learn and indulge in a particular behavior if it is rewarded positively (Berwick,
1986; Damon and Lerner, 2006)
18. Can pro-social cues be used to reduce deception?
» Building a pro-social self image for honest behavior and
rewarding honesty with positive words will encourage more
honest behavior
» One will not indulge in deception and affect reputation
negatively
» Toma(2008) findings can be also be explained via
reputation preservation
19. Building our Hypotheses
» Test against “Will embedding pro-social cues reduce
deception? ”
» Apply in identity based scenarios (as opposed to message
based scenarios)
» Difficult to incorporate in message based without
distracting the user
20. Pro-social cues in identity based communication
» Embed pro-social cues during information collection/profile
building process
» Use a wizard to collect information that gives these cues to
the user in a controlled manner and also collects information
» Classify pro-social cues into
» Verbal pro-social cues: Praise, empathy, dispositional
praise
» Visual pro-social cues: Picture of eyes
21. Hypothesis I
» H1: Using a wizard with verbal pro-social cues to generate online
profile information can reduce deception as compared to profiles
generated using simple drop-down boxes and text fields.
» Empathy: “It is not easy to find a true match that appreciates your
true self, but our database consisting of thousands of members
makes it easier. Please fill in your date of birth to find your match.”
» Dispositional praise: “You’re the kind of person who appears to be
honest. Please fill in your interests”
» Praise: “Well done in filling your city correctly, please provide …
details”
23. Hypothesis II
» H2: Using a wizard in the form of a computer character that exhibits
visual pro-social cues can reduce deception as compared to profiles
generated using simple drop-down boxes and text fields..
Please provide your date of birth
24. Experimental Methodology
» Goal: Test our hypotheses in the scenarios of building online
resumes
» Participants : Undergraduates from University of Wisconsin,
18- 22 years olds, male or female
» Task: Prepare resume for summer internship for business
sales position
» Salary higher than standard to make it desirable
» Offer 10$ to participate in the study(apart from the job)
25. Experimental Methodology
» Prepare resume using one of three methods
» Traditional: Using conventional forms and drop-down
menus
» Wizard with verbal cues: Embed pro-social cues of
empathy, praise and motivation
» Wizard with visual cues: Embed pro-social cues of
presences of eyes in a computer character
26. Experimental Methodology
» Disclose real purpose of task
» To measure deception in online profile creation
» Option to discard profiles or mark deceptive fields in
resume for 10$
» Measure amount, type and extent of deception
27. Experimental Methodology
» Quantity: None, 1-2, Many
» Extent of deception: Subtle, Mild, Severe
» Type of deception: Interests, Abilities, Factual
» Analyze results to detect if
» pro-social cues reduce deception and by how much
» pro-social cues reduce severity of deception
» pro-social cues reduce particular types of deception
28. Conclusion
» We proposed using pro-social cues to reduce deception
» Identified and classified the pro-social cues
» Proposed a technology solution to embed these cues
» Described the experimental framework to test these cues
Proposing the hypothesis that pro-social cues can reduce deception and I will describe how we can test against this hypothesis empirically
Deception in cyberspace is : (i) It is intentional modification of message or identity information by the sender in technologically mediated message. (iii) It is intended to create false belief in the eyes of the recipient of the message .
73 percent of individuals in a web based survey of individuals pariticipating in discussion groups believe deception to be widespread online (Caspi & Gorsky, 2006) and a second study indicating that eight out of ten online daters have serious concerns about deception (Gibbs, Ellison & Heino, 2006).
Hancock (2007) argues that the major reason that digital deception is often expected to occur more frequently because online communication is typically “text-based interaction or virtual representations of self” (p. 291). Many people appear to believe that these virtual representations provide users with the opportunity to distance the actual self from the virtual self more easily and perhaps with fewer consequences (i.e., personal shame, or ridicule from significant others) than would occur in the Ftf world.
Patients can become more inclined to deceive behind computer mediated communication and this can lead to improper diagnosis and medication resulting in life threatening consequences.
Hancock (2007) further separates digital deception into identity-based deception, which is deception based on the sender’s personal identity (i.e., lying about one’s physical appearance, abilities or skills), and message-based deception, which is deception that takes place in the specific communication between sender and receiver(i.e., lying about one’s weekend plans.
Previous research demonstrates that identity-based deception occurs more frequently in computer-mediated communication environments than in FtF environments (Cornwell & Lundgren, 2001).
Indeed, according to George (2004), 90 percent of individuals admitted to lying at least once on a resume-like scholarship application. Although participants in this
study were asked to make their applications as competitive as possible, they were not prompted to be deceptive.
Most message based deception are harmless(Hancock 2009)
How deception can be detected
Verbal and no verbal cues that
Linguistic attributes
No clear indicator
Better to tackle the problem early on in CMC
Detecting detection may be too late
Deception reduction in CMC has looked to augment channels to reduce deception
Deception reduction in channels looks to expose real world equivalents making deceptive behavior impossible to fake
Donath (1999), argues when there is disconnect between the physical and virtual self, the cost of deceptive behavior is substantially reduced.
According to this argument, when there is a close relationship between the virtual and physical self represented by signals, the costs of deceptive behavior should increase.
Donath (2007) notes that what keeps signals reliable, or keeps people from being deceptive, is the costs associated with that behavior.
Certain signals as reliable indicators of honesty because they are simply too expensive or impossible to fake.
Like institutional email address
For example, the public and socially connected displays of personal information inherent to SNS should therefore act as a constraint to deception.
Walther and Parks introduce concept of warrants, which is connection
Once a person is aware of and able to view a communication partner’s social network, there is a warrant between the physical and online self-presentation, and deception is constrained. In this case, the social network constrains deception by 1) providing an opportunity to verify information about a communication partner with members of the network, and 2) by providing an audience to which the communication partner must account for deceptions (Walther & Parks, 2002).
Recent research in the context of online dating (Toma et al., 2008) demonstrates the constraining effects that social networks can have on deception. In Toma et al.’s (2008) study, members of online dating websites were asked to indicate areas where they lied on their online dating profile. The more friends who knew about the participants’ profile the more accurate their profile was, suggesting that the people in the participants’ personal network acted as a constraint to deception.
Recent study by Hancock found that media rich with warrants such as SNS contains lower level of deception than IM, chat
As we saw in pr
Deception detection works in near chance levels
Similarly, deception reduction in channles will not work when the communication is warrantless or signals are not apparent
Deception reduction which works early enough in communication process and is in not constrained by real world equivalence
This is the out line of my talk
Caring about the welfare and rights of others, feeling concern and empathy for them, and acting in ways that benefit others
Eisenberg, Nancy and Paul H. Mussen. The Roots of Prosocial Behavior in Children. Cambridge: Cambridge University Press, 1989. ISBN: 0-521-33771-2.
These behaviors include a broad range of activities: sharing, comforting, rescuing, and helping.
Researchers recently have found that when older adults engage in altruistic behavior and volunteering, they benefit from these activities. Helping others may reduce the output of stress hormones which improves cardiovascular health and strengthens the immune system. Sanstock, John W. A Topical Approach to Life span Development 4th Ed. New York: McGraw-Hill, 2007. Ch 15. 489-491
Fessler, D. M. T., & Haley, K. J. (2003). The strategy of affect: Emotions in human cooperation. In P.
Hammerstein (Ed.), Genetic and Cultural Evolution of Cooperation (pp. 7 – 36). Cambridge, MA7 MIT Press.
To benefit themselves consciously or sub-consciously
That indivudsual respond to subtle cues that…and these cues
Broad goal to respond to pro-social cues is same as the reason they indulge in pro-social behavior which is to …
Individuals response to specific pro-social cues can be cateogarized into reputation preservation and positive reinforcement learning.
Individuals preserve reputation to benefit themselves consciously or sub-consciously by hoping co-operation if future.
Individuals have evolved to indulge in behavior which rewards them psychologically or materistically thru empathy, praise or a gift.
The learning processes of brain causes it to indulge in more pro-social behavior if any previous pro-social behavior is rewarded even via a simple praise (Berwick, 1986)
Presence of eyes is known to trigger reputational concerns. In an experiment by
Attribute generosity to mainting pro-social reputation
Batson, C. D. (1991). The altruism question: Toward a socialpsychological answer. Hillsdale, NJ: Erlbaum
Similarly, indivduals are known to indulge in pro-social behavior due to auditory presence of others.
Batson, C. D. (19 91). The altruism question: Toward a socialpsychological answer. Hillsdale, NJ: Erlbaum
Displaying connections is a way of signalling a willingness to risk one’s reputation. In the real world, as Burt [13] has pointed out, reputation is a powerful force in groups with dense affiliations. This can be due to repeated interactions — we gather a reputation around our identity that, if good, is quite valuable and we benefit from continuing to act in ways that enhance that reputation.
Conscious cues
Empathy is an important mediator of helping behavior and other
prosocial behavior (Batson, 1991).
Batson, C. D. (1991). The altruism question: Toward a socialpsychological answer. Hillsdale, NJ: Erlbaum
Praise and recognition: the importance of social support in law enforcement, Gove 2005
Both empathy and rewarding behavior can be explained via positive reinforcement learning. Empathy rewards brains socially while praise and recogniton.
The learning processes of brain causes it to indulge in more pro-social behavior if any previous pro-social behavior is rewarded even via a simple praise (Berwick, 1986)
We argue that pro-social cues can also reduce deception.
If honest behavior is rewarded then individuals will subsequently indulge in more honest behavior
Similarly, if there are cues that cause reputational concerns individuals will not indulge in activities that can harm one’s reputation
We now look to build our hypotheses and test if embedding
We apply in identity based scenarios as opposed to message based since in a message based, user is focussed
In order to test effectiveness of pro-social cues
Classify pro-social cues in the manner in which we can deliver them in CMC medium.
Participants were instructed to tailor their resumes using their personal experiences and skills to ensure that they would be the most qualified candidate for the described job. Participants were not instructed or encouraged to lie, but rather were urged to use their own information in a way that would present themselves as the best candidate for the described position.
Subtle – leaving out pertinent information
Mild – mild exaggeration , handled all responsiblities for a position, may have handled only for a few days
Outright => Closed 100 deals
Factual ( No. of years of experience)