1 / 19

Emergent Social Phenomena and Social Action Selection

Emergent Social Phenomena and Social Action Selection. Alan R Wagner Georgia Institute of Technology Research Institute Aerospace, Transportation, and Advanced System Lab Atlanta USA. What does the I in HRI stand for?.

miriam
Download Presentation

Emergent Social Phenomena and Social Action Selection

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Emergent Social Phenomena and Social Action Selection Alan R Wagner Georgia Institute of Technology Research Institute Aerospace, Transportation, and Advanced System Lab Atlanta USA

  2. What does the I in HRI stand for? • Interaction: influence—verbal, physical or emotional—by one individual on another (Sears, Peplau, & Taylor, 1991).

  3. How can we represent interaction computationally? Game Theory • Strategic interaction • Assumes that each player maximizes influence • Interesting equilibrium results • Normal form game Interdependence Theory • Constructed from psychological experiments • No assumptions about player’s motives • No equilibrium results • Outcome matrix

  4. The Outcome Matrix/Normal Form Game Representation: • A finite set of N individuals • For each iN a nonempty set of actions Ai • A scalar outcome value (o) for every pair of actions (influence) • -i used to represent the robot’s partner Outcome Matrix Person 1 Fair Unfair Person 2

  5. Social Situations and Interaction • Social situation is an abstract representation of a class of interactions involving particular outcome values • An interaction is a discrete period of influence between two individuals choosing between particular actions Tom i Admit Guilt Claim innoc. ai1 ai2 Eric -i Admit Guilt a-i1 Claim innoc. a-i2

  6. The Space of Social Situations • Interdependence space (Kelley, et al. 2003) • Location in the space provides info. about the situation • The dimensions are: • Interdependence • Correspondence • Control • Symmetry – not shown

  7. Deception (joint with Ron Arkin) • Deception is commonly used by animals • The use of deception by primates may indicate Theory of Mind (Byrne & Whiten, 1990) • Signal exaggeration, signal suppression, distractions, hiding objects, etc. • May be a hallmark of social intelligence (Byrne & Whiten, 1990) Mimicry Camouflage

  8. Defining Deception • Deception: “Causing another to believe what is not true; to mislead or ensnare.” (Webster’s, 1999) • Deliberately induced misperception (McClesky, 1991) • A false communication that tends to benefit the communicator (Bond and Robinson, 1988) • We use Bond and Robinson’s definition because of its general applicability

  9. Situations Warranting Deception • Deception is a false communication that tends to benefit the communicator (Bond and Robinson, 1988) 1) Deceiver provides false communication. This implies conflict. 2) Deceiver benefits from communication. This implies dependence.

  10. Computationally Representing Deception • Deceiver and Mark choose actions • Deceiver selects actions that will influences Mark’s behavior • Play dead so that Mark will approach Induced Matrix Mark Approach Stay away Deceiver

  11. Computationally Representing Deception • Deceiver’s actions make Mark think that one action is more rewarding or less costly then it actually is True Matrix Mark Approach Stay away Deceiver -3 True value

  12. Trust • Idea: Some social situations require trust and others do not • Can we use our representation of social situations and interaction to classify situations in terms of trust? 12

  13. Defining trust • Trust is a belief, held by the trustor, that the trustee will act in a manner that mitigates the trustor’s risk in a situation in which the trustee has put its outcomes at risk. • Key point: risk 13

  14. Classifying situational trust • A trustor is the individual doing the trusting. The trustee is the individual being trusted • Let action A be the trusting action and action B be the untrusting action Trustor Action A Lean back Action B Don’t lean back Trustee Action C Catch Action D Don’t catch 14

  15. Classifying situations in terms of trust If the trustor trusts the trustee, then they risk 6 for a possible gain of 12 • If the trustor doesn’t trust the trustee, then their outcome is 6, and the trustee’s action doesn’t matter, i.e. no risk! Trustor Action A Lean back Action B Don’t lean back Trustee Action C Catch 6 12 4 6 • If the trusting action is selected, then the trustee’s choice of action determines the trustor’s outcome Action D Don’t catch 6 0 4 6 15

  16. Conditions for Situational Trust • We can thus list some conditions for trust: • It is better for trustor if the trustee acts in a trusting manner (12 > 0 ) • It is better for trustor to select the untrusting action, than for the trustor to trust and have the trustee not act in a trusting manner (from example 12 > 6 > 6 > 0 ) Trustor Action B Don’t lean back Action A Lean back Trustee Action C Catch Action D Don’t catch 16

  17. How much Trust? • Trust  risk • Risk: expectation over a loss function L • R(x) =  L(x,y)p(x) where x is predicted y is the true value Context/Situation related loss Partner Model related

  18. Thank you

  19. References • Carpin, S., Jijun Wang, Michael Lewis, Andreas Birk, Adam Jacoff: High Fidelity Tools for Rescue Robotics: Results and Perspectives. RoboCup 2005: 301-31. • Deutsch, M. (1962). Cooperation and trust: Some theoretical notes. In M. R. Jones (Ed.), Nebraska symposium on motivation (pp. 275-315). Lincoln, NB: University of Nebraska. • Gambetta, D. (1990). Can we trust trust? In D. Gambetta (Ed.), Trust, making and breaking cooperative relationships (pp. pages 213--237). Oxford England: Basil Blackwell. • Kelley, H. H., J. G. Holmes, N. L. Kerr, H. T. Reis, C. E. Rusbult, and P. A. M. V. Lange, An Atlas of Interpersonal Situations. Cambridge University Press, New York, NY, 2003. • H. H. Kelley and J. W. Thibaut, Interpersonal Relations: A Theory of Interdependence. John Wiley & Sons, New York, NY, 1978. • Luhmann, N. (1979). Trust and power.Chichester: Wiley Publishers. • Osborne, M. J., & Rubinstein, A. (1994). A course in game theory. Cambridge, MA: MIT Press. Relationship. (2000). In American heritage dictionary. • Rusbult, C. E., & Van Lange, P. A. M. (2003). Interdependence, interaction and relationship. Annual Review of Psychology, 54, 351-375. • Sears, D. O., Peplau, L. A., & Taylor, S. E. (1991). Social psychology.Englewood Cliffs, New Jersey: Prentice Hall.

More Related