David Traum USC Institute for Creative Technologies William Swartout USC Institute for Creative Technologies Jonathan Gratch USC Institute for Creative.

Slides:



Advertisements
Similar presentations
What can we do in the Classroom?
Advertisements

Workpackage 2: Norms
Rationale To encourage all students to take a full part in the life of our school, college, workplace or wider community. To provide opportunities to enable.
JSIMS 28-Jan-99 1 JOINT SIMULATION SYSTEM Modeling Command and Control (C2) with Collaborative Planning Agents Randall Hill and Jonathan Gratch University.
Social Influence and Persuasion
Argumentation-based negotiation Rahwan, Ramchurn, Jennings, McBurney, Parsons and Sonenberg, 2004 Presented by Jean-Paul Calbimonte.
Paul Brinkhurst The Centre for Literacy Summer Institute 2012 Workplace Literacy & Essential Skills: Shaping a New Learning Culture June 27-29, 2012 Montreal,
Theoretical Approaches Gestalt Group Therapy  Gestalt is a German word for whole or configuration and Gestalt therapy is concerned with the foreground.
OASIS Reference Model for Service Oriented Architecture 1.0
U1, Speech in the interface:2. Dialogue Management1 Module u1: Speech in the Interface 2: Dialogue Management Jacques Terken HG room 2:40 tel. (247) 5254.
12 Entrepreneurship Managing New Ventures for Growth.
Putting It all Together Facilitating Learning and Project Groups.
Proposal Presentations for United States and International Audiences The Cain Project in Engineering and Professional Communication ENGINEERING SERIES.
Coaching Workshop.
GOALS & GOAL ORIENTATION. Needs Drive Human Behavior  Murray  Maslow.
Applying Research on Cultural Differences in Negotiation to a Negotiation Simulation Phani Radhakrishnan PhD Cross Cultural Differences in Org Behaviour.
The Entertainment Industry Meets Simulation: The Challenge of the Holodeck Bill Swartout Institute for Creative Technologies University of Southern California.
1. Human – the end-user of a program – the others in the organization Computer – the machine the program runs on – often split between clients & servers.
Path-Goal Theory Chapter 7.
Managing Social Influences through Argumentation-Based Negotiation Present by Yi Luo.
Design Science Method By Temtim Assefa.
Team Formation between Heterogeneous Actors Arlette van Wissen Virginia Dignum Kobi Gal Bart Kamphorst.
Issues in Dialogue Management for Multiparty Dialogue David Traum University of Southern California Institute for Creative Technologies (ICT)
McGraw-Hill/Irwin © 2005 The McGraw-Hill Companies, Inc. All rights reserved ChapterChapter 10 Networking and Negotiating.
Networking and Negotiating
Lecturer: Gareth Jones Class 8: Persuasive Messages.
 Organizing and Presenting a Persuasive Message.
Introducing Your name goes here Your Position goes here
Chapter 16 – Controlling the Organization
1 Contact: From Evaluation to Research Description of a continuum in the field of Global Education.
Conflict and Negotiation. Sub-topics 1. Conflict 2. Conflict – traditional, interactional, focused on solution 3. Process of conflict 4. Negotiation 5.
Human Resources Training and Individual Development Adult Learning February 2 nd, 2004.
It’s Not Just About the Horses: How to Bring Out the Best In the People You Work With John J. Martin Dina Parrello.
Coaching Youth Livestock Projects Kyle Merten – Extension Associate.
Leadership Chapter 7 – Path-Goal Theory.  Path-Goal Theory Perspective  Conditions of Leadership Motivation  Leader Behaviors & Subordinate Characteristics.
Issues in Multiparty Dialogues Ronak Patel. Current Trend  Only two-party case (a person and a Dialog system  Multi party (more than two persons Ex.
Towards a Theoretical Framework for the Integration of Dialogue Models into Human-Agent Interaction John R. Lee Assistive Intelligence Inc. Andrew B. Williams.
ENTERFACE 08 Project 1 “MultiParty Communication with a Tour Guide ECA” Mid-term presentation August 19th, 2008.
A Common Ground for Virtual Humans: Using an Ontology in a Natural Language Oriented Virtual Human Architecture Arno Hartholt (ICT), Thomas Russ (ISI),
Sophie Makris  What is a team?  A group of people pooling their skills, talents, and knowledge, with mutual support and resources, to provide.
Negotiation and Mediation Presented by Ms. Asha Menon Additional District Judge & Member Secretary, Delhi Legal Services Authority Business Session –II.
Good Agricultural Practices Teaching Adult Learners.
Cognitive Behavioral Interventions. SOCIAL SKILLS TRAINING: TWO TYPES OF INTERPERSONAL COMPETENCE Cognitive Competence  Knowledge about relationships.
Central Core CD Unit B 2-5 Employability in Agriculture/Horticulture Industry.
Directing Definition of directing: Directing is the fourth element of the management process. It refers to a continuous task of making contacts with subordinates,
TEAM, ORGANIZATIONAL, AND INTERNATIONAL CULTURE Chapter 14.
© 2006 Prentice Hall Leadership in Organizations 6-1 Chapter 6 Power and Influence.
The Unit Counseling Program A company level leader must understand: ­ The effects of the unit counseling program on the organization ­ How to assess the.
Negotiation Cultural Analysis Framework Salacuse 1991.
SPEECH ACTS Saying as Doing See R. Nofsinger, Everyday Conversation, Sage, 1991.
Lecture “6” Manage Project Team
Agent-Based Dialogue Management Discourse & Dialogue CMSC November 10, 2006.
WP6 Emotion in Interaction Embodied Conversational Agents WP6 core task: describe an interactive ECA system with capabilities beyond those of present day.
Beginning Social Communication High School: Lesson Five.
Design Evaluation Overview Introduction Model for Interface Design Evaluation Types of Evaluation –Conceptual Design –Usability –Learning Outcome.
Edit the text with your own short phrase. The animation is already done for you; just copy and paste the slide into your existing presentation.
 In Ned law are a company that provides strategic consulting and management, composed of a team of high academic and social esteem, focused on optimization,
Lecturer: Gareth Jones Class 12: Persuasive Messages.
Chapter 9 Negotiation “You often get not what you deserve, but what you negotiate.” ~ John Marrioti.
Impact-Oriented Project Planning
Managing Team Conflict Standards 8.23
By Prof. Fernando R. Pedrosa,Ph.D. Prof.-n-charge
‘Can’t we all just get along?’: Useful Conflict Management Skills
Chapter Fourteen The Persuasive Speech.
Leadership Chapter 7 – Path-Goal Theory Northouse, 4th edition.
Chapter 16 Persuasive Speaking.
2.Personality And Attitude
Presentation transcript:

David Traum USC Institute for Creative Technologies William Swartout USC Institute for Creative Technologies Jonathan Gratch USC Institute for Creative Technologies Stacy Marsella USC Information Sciences Institute Fight, Flight, or Negotiate: Believable Strategies for Conversing under Crisis

Outline 1. Background: ICT Virtual Humans and the SASO Project 2. Adversarial Negotiation: theory overview 3. Implementation of Negotiation Strategies in Vhumans 4. Preliminary Evaluation 5. Current & Next Steps

ICT Virtual Human Project ( …) Push state of the art in integrated virtual human capabilities  Basic research  in a number of component fields  Emotion  Perception  human animation  natural language dialogue  On overlap & interaction between these areas  Theory put to practice in integrated vhumans  Vhumans employed for real tasks  Immersive training applications  Tested on target user population

Virtual Human Example Applications Mission Rehearsal Exercise (MRE) Domain:Platoon-level peacekeeping Training Activity: Decision-making & Teamwork Only team negotiation Stability and Support Operations (SASO-ST) Domain:Bi-lateral (& Multi-lateral) Negotiation Training Activity : Building Trust & Negotiation Strategies

Immersive Training Environment VR Theatre 8’ 150˚ Curved Screen, Multiple Projectors d spatialized sound

SASO-ST: Dealing With Doctors Scenario Mission: Convince Doctor to move (but don’t give op details) Gain working relationship with Doctor Doctor Perez runs NGO clinic Doctor values Neutrality No prior relationship with Doctor Recently: Rise in insurgent activity More casualties in clinic Planned operations

ICT Virtual Humans: modelling negotiation Human-like bodies embedded in virtual world Advanced Integrated AI models  Perception  Task performance/planning  Emotion  Verbal and Non-verbal Communication  Speech  Gestures New extended negotiation  Adoption of orientations and strategies  Trust model

Speech Recognition (Sonic) Semantic Parser Motion/ Gesture Scheduler (BEAT) Text to Speech (Rhetorical) Simulation Environment BDI Communication Bus Audio (Protools) Voice Input UT Projection System Speakers (10.2) Soar Planning DialogueAction Selection Perception Emotion NLU pragmatics OSS ICT Virtual Human Architecture NLG

Environment Problem-Focused Emotion-Focused Appraisal Variables Coping Strategy Action Tendencies “Affect” Physiological Response Appraisal Coping Theoretical Perspective on Emotion ( Marsella & Gratch, AAMAS 02, 03) Goals/Beliefs/ Intentions Smith and Lazarus ‘91 cognitive-motivational-emotive system

Dialogue Approach:Layered Information State Layer captures coherent aspect of communicative interaction (e.g., turn, grounding, obligations) Layer consists of  Information State components (state of interaction)  Dialogue Acts (Packages of changes to information state) Realization Rules Dialogue Acts Input Utterance Recognition Rules Update Rules Output Utterance (verbal and nonverbal) Selection Rules Info State Components Dialogue Manager Dialogue Acts

Virtual Human Dialogue Layers (Traum & Rickel AAMAS 02) Contact Attention Conversation  Participants  Turn  Initiative  Grounding  Purpose Social  Obligations-Commitments  Negotiation-Collaboration  Social Roles Individual  Perception  Rational  belief,desire, intention,..  Emotional  Coping strategies

Social Commitments (Traum & Allen94, Allwood 94, Matheson et al 00) IS  Obligations, Social Commitments to Propositions Actions  Order, Request, Suggest  Promise, Offer  Statement, Question  Accept,.. Effects are to Obligations & Commitments  Belief updates based on inference, not speech act effects

Social State ^obligation + & ;#obligations to act  ^type ^holder ^obligated-to  ^action ^deadline ^sanction ^commitment + & ;# committed to states of affairs holding  ^type ^holder ^committed-to ^proposition ^sanction ^conditional + & ;# obligation or commitment if action  ^type ^trigger ;# an action to check performance of  ^consequent ;# the resulting commitment or obligation roles  ^teammate  ^superior + & ;# agents superior to self  ^subordinate + &) ;# agents subordinate to self

Team Negotiation (Traum et al AAMAS 2003) IS: task (&CGU) annotated with negotiation objects  Components: Agent, Action, Stance, audience, reason  Stances: Committed, endorsed, mentioned, not mentioned, disparaged, rejected Action effects:  Suggestion: mentioned  command, promise, request, or acceptance: committed  Rejection: rejected  Counterproposal: disparaged 1 + endorsed 2  Justification: endorsed or disparaged (depending on direction)  Offer: mention (conditional commitment)  Retract stance Factors:  Relevant Party: Authorizing or Responsible Agent  Dialogue State: who has discussed  Plan State: how do I feel about it

Theory of non-team Negotiations Context for Negotiation Orientations  Strategies  moves

Negotiation settings Team planning  Same goals or utility  Negotiate on best means to common end Stable Institution  Fixed, respected rules, enforced (courts, etc)  Bargaining, contracts  No penalty for not negotiating Hostile  Dealing with antagonists  Coercion, threats, deception  Could degenerate to use of force Spontaneous  Uncertainty of proper model  Multiple approaches  (re-)assess situation

Game theory approach Team game  Identical utilities (goals) for team members Zero-sum  Total utility is fixed, any “win” for one side is a “loss” for another Generalized Game  payoff matrixes unconstrained  Different sorts (e.g., prisoner’s dilemma)  Search for equilibria (“win-win”)

Orientations toward negotiation (Walton & Mckersie, Sillars et al, etc) Avoidance  No benefit from negotiation  avoidable Distributive  Zero-sum (win for other is loss for me) Integrative  Open to cooperation  Possibilities of win-win All of these are subjective perceptions, whatever the real situation

Coping With Negotiation Orientations Cascaded meta-strategies  Zero-level: strategies to cope with pressure to negotiate and own orientation  1st-level: strategies to cope with orientations (and strategies) of other (given desire to negotiate)  2nd-level: strategies to cope with other’s 1st- level strategies

Zero-level Strategies (own orientation) Avoidance  Avoid  Disengage Distributive  Attack  Unreasonable demands  Trap Integrative  Bargain  Find maxima

1st level Strategies (react to Other’s Orientation): (1) Work within other’s orientation Avoid  Let avoid  Make avoidance costly Distributive  Same as zero-level Distributive strategies Integrative  Negotiate  trick

1st level Strategies (react to Other’s Orientation): (2) Change other’s orientation From avoidance to distributive/integrative  Stay engaged/on topic  Show value for negotiation From distributive to integrative  Demonstrate trustworthyness  Familiarity  Credibility  Solidarity  Show value in cooperation  Show value in negotiation goal

2nd level strategies (react to other’s attempt to change orientation) Assess motivation for other’s utterances  Is offer or assertion self-serving or against own interest?  Can claims be verified independently?  Assess sincerity Strategically adopt strategies  Which are most likely to lead to adoption of helpful strategies by other  Display orientation (whether actually held or not)

Coping With Negotiation Orientations: Initial Model in SASO-ST project Extended Virtual Humans to have orientations and strategies  Implement appropriate zero-level strategies  Vhuman must act appropriately given orientation and pressure Use system to Teach effective 1st-level strategies  Vhuman Recognize effect of 1st-level moves  Vhuman changes orientation and strategy as appropriate

Modelling Trust Represented as Variable  0 (no trust) to 1 (full trust) Initial value can be set Updated as a result of interaction  Linear combination of three components  Familiarity (observance of polite social norms)  Solidarity (same goals)  Credibility (shared beliefs) Used to update beliefs based on reports (commitments, promises, belief attribution) Used in assessing proposals

Implementing Negotiation Strategies Orientations result from appraisal of negotiation itself  Reified negotiation “task”  Interactions with goals and plans Strategies chosen as part of coping  Entry & exit conditions Strategies associated with communicative behavior  Base posture and gesture set  Choice of dialogue moves  Speech act and realization  Initiative, topic selection, and type of grounding feedback  Affective tone  Aspects of interpretation  Charitability of interpretation  Assumptions vs clarification

Avoidance Strategy Entry Conditions  Negotiation costs outweigh benefits  Avoidable Exit conditions  Not avoidable (2 subsequent mentions of plan-related tasks or state)  Move is desirable?  Trust is <0.2 Moves  Avoid (mention something other than plan-related tasks or states)  Escape (pre-closing: try to leave the conversation)

Attack Strategy Entry Conditions  Not avoidable  Move has less utility than staying Exit Conditions  Move has high utility  Trust is below 0.2 Moves  Bring up problems  Pre-condition not met  Necessary task that is not commited to  Difference in utility

Negotiate Strategy Entry Conditions  Not avoidable?  Move has more utility than staying Exit Conditions  Committed agreement  Trust is below 0.2 Moves  Try to gain commitment on tasks  Offer/suggestion, acceptance

“Failure” Strategy Entry Conditions  Trust below 0.2 Exit Conditions  None (end conversation) Moves  Reject move

“Success” Strategy Entry Conditions  Committed agreement Exit Conditions  None (end conversation) Moves  Accept move

Use of SASO-ST system Doctor scenario Large variety possible in different sessions  Results depend on  Doctor personality (configurable)  Doctor strategy selection  Trainee strategy and moves  Stochastic elements  Choice of move realization & focus  Choice of coping focus  Interpretation process & errors Success in task depends on  Building trust  Motivating doctor to change strategies & reach agreement

Evaluation with target users Validation of theory by corpus analysis of roleplay  Identification of use of hypothesized strategies  Investigation & adoption of specific strategies & moves Validation of model using WOZ interaction  Shows strategy implementation and selection is viable Validation of SASO-ST system with automated doctor  Usability  Training value  learning

Current & Next Steps More Evaluation Extensions to model  Additional strategies & tactics (e.g., rejection of empathy)  1st & 2nd-order strategies Use as learning system  Targeted instruction and training  After-action review (XAI - van Lent, Core, Lane) New Scenarios  Use of culture-specific elements