Aff_Comp_lect03a_2011Aff_Comp_lect03b_2011Aff_Comp_lect03c_20119/9/20111ExperimentExperiment9/9/201129/9/20113Question: Who should we blame?Reasoning about Anger Basic Questions: – How do ordinary people (not lawyers) evaluate and form judgments about such situations?– Can computer models predict judgments?Some appraisals require social inferences (theory of mind) – e.g. Weiner; Shavere.g., Causal attribution involved in AngerAttributional InferencesBlAppraisal theory perspective on anger(Mao&Gratch2005;Oh et al 2007; Melissen)NegativeConsequenceCause ForeseenUnforeseenIntendedUnintendedVoluntaryCoercedBlame or CreditNo Blame No Blame No Blame No BlameIncreasing responsibility but not blameSuch reasoning underlies how people make sense of complex social activities involving the assignment of blame or credit9/9/20114Illustration: deriving appraisals of blame(Mao&Gratch2005;Oh et al 2007; Melissen)Scenario1. The VP British Petroleum discusses plans for new oil well2. The VP states the program will likely increase profits3. The VP states the program will likely harm the environment4. The CEO orders the program to be started5. The VP executes the new program6. The environment is harmed7Illustration: deriving appraisals of blame(Mao&Gratch2005;Oh et al 2007; Melissen)Scenario 11. The VP British Petroleum discusses plans for new oil well2. The VP states the program will likely increase profits3. The VP states the program will likely harm the environment4. The CEO orders the program to be started5. The VP executes the new program6. The environment is harmed8VP: CauseCEO: Cause + Intent + Voluntary  Blame9/9/20115Illustration: deriving appraisals of blame(Mao&Gratch2005;Oh et al 2007; Melissen)Scenario 11. The VP British Petroleum discusses plans for new oil well2. The VP states the program will likely increase profits3. The VP states the program will likely harm the environment4. The CEO orders the program to be started5. The VP executes the new program6. The environment is harmed9VP: Cause + IntentIllustration: deriving appraisals of blame(Mao&Gratch2005;Oh et al 2007; Melissen)Scenario 11. The VP British Petroleum discusses plans for new oil well2. The VP states the program will likely increase profits3. The VP states the program will likely harm the environment4. The CEO orders the program to be started5. The VP executes the new program6. The environment is harmed10VP: Cause + Intent + ¬Voluntary  ¬Blame9/9/20116Illustration: deriving appraisals of blame(Mao&Gratch2005;Oh et al 2007; Melissen)Scenario 11. The VP British Petroleum discusses plans for new oil well2. The VP states the program will likely increase profits3. The VP states the program will likely harm the environment4. The CEO orders the program to be started5. The VP executes the new program6. The environment is harmed11VP: Cause + Intent + ¬Voluntary  ¬Blame CEO: Cause + Intent + Voluntary  Blame Scenario 1E1 The vice president of Beta Corporation goes to the chairmanof the board and requests, “Can we start a new program?”Appraise RepresentationIntentional actionIllustration: deriving appraisals of blame(Mao&Gratch2005;Oh et al 2007; Melissen)E2 The vicepresident continues, “The newprogramwill help usincrease profits,E3 but according to our investigation report, it will harm to theenvironment.”E4 The chairman answers, “Start the program anyway.”E5 The vice president executes the new program.E6 However, the environment is harmed by the new program.intend(x, p, t1) ∧ do’(p, x, A) ∧ t1<t3 ∧¬(∃t2)(t1<t2<t3 ∧¬intend(x, p, t2)) ∧ execute(x, A, t3)Side effecteffect(A) ∧¬intend(x, b, t1) ∧ by(b, A, e) ∧¬(∃t2)(t1<t2<t3 ∧ intend(x, b, t2)) ∧t1<t3<t4 ∧ execute(x, A, t3) ∧occur(e, t4)CoercionCounterfactual Scenario 2E1 The vice president of Beta Corporation goes to the chairmanof the board and requests, “Can we start a new program?”E2 The vice president continues, “The new program will help usincrease profits,E3ANDaccordingtoourinvestigationreportitWON’Tharmtothe12Coercioncoerce(y, x, p, t1) ∧ do’(p, x, A) ∧e∈effect(A) ∧ t1<t2<t3 ∧etc29(x, y, e, t2)  coerce(y, x, e, t3)BLAME = YESE3ANDaccordingtoourinvestigationreport,itWON Tharmtotheenvironment.”E4 The chairman answers, “Start the program.”E5 The vice president executes the new program.E6 However, the environment is harmed by the new program.9/9/20117Empirically validated commonsense axioms allow agents to form basic inferences from observations of actions and speech Intentional action--1009080ResponsibilityBlameConfidence interval--1009080ResponsibilityBlameConfidence intervalIllustration: evaluating predictive success(Mao&Gratch2005;Oh et al 2007; Melissen)intend(x, p, t1) ∧ do’(p, x, A) ∧ t1<t3 ∧¬(∃t2)(t1<t2<t3 ∧¬intend(x, p, t2)) ∧ execute(x, A, t3) Negligence¬intend(x, p, t1) ∧ do’(p, x, A) ∧ t1<t3 ∧¬(∃t2)(t1<t2<t3 ∧ intend(x, p, t2)) ∧ execute(x, A, t3) Intentional effectintend(x, b, t1) ∧ by(b, A, e) ∧¬(∃t2)(t1<t2<t3 ∧¬intend(x, b, t2))) ∧t1<t3<t4 ∧ execute(x, A, t3) ∧ occur(e, t4)Side effect--------8070605040302010with bullets all marksmen noneScenario 1--------8070605040302010with bullets all marksmen noneScenario 1Model Predictions13Side effecte∈effect(A) ∧¬intend(x, b, t1) ∧ by(b, A, e) ∧¬(∃t2)(t1<t2<t3 ∧intend(x, b, t2)) ∧ t1<t3<t4 ∧ execute(x, A, t3) ∧ occur(e, t4) Coercioncoerce(y, x, p, t1) ∧ do’(p, x, A) ∧ primitive(A) ∧ e∈effect(A) ∧ t1<t2<t3∧ etc29(x, y, e, t2)  coerce(y, x, e, t3)...Empirically validated commonsense axioms allow agents to form basic inferences from observations of actions and speech Intentional action--1009080ResponsibilityBlameConfidence interval--1009080ResponsibilityBlameConfidence intervalIllustration: evaluating appraisal derivation(Mao&Gratch2005;Oh et al 2007; Melissen)intend(x, p, t1) ∧ do’(p, x, A) ∧ t1<t3 ∧¬(∃t2)(t1<t2<t3 ∧¬intend(x, p, t2)) ∧ execute(x, A, t3) Counterfactual scenario creation– What would have to be different in the initial scenario to change the causal attribution?–Generate these novel scenarios--------8070605040302010with bullets all marksmen noneScenario 1--------8070605040302010with bullets all marksmen noneScenario