Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Chris Hazard - Measuring and Manipulating Player Biases and Trust Through Choice and Game Mechanics


Published on

Chris Hazard, CEO/Founder, Hazardous Software

This presentation was given at the 2016 Serious Play Conference, hosted by the UNC Kenan-Flagler Business School.

Information theory offers a mathematical measure of expected surprisal and a measure of ambiguity. When coupled with decision science, psychology, and a little game theory, human biases and trust can be formalized, measured and, ultimately, manipulated. This talk will provide an overview of how such techniques can be employed in the mechanics of serious games to create more believable experiences, more challenging and effective AI, and to help players understand their own biases and weaknesses in their own disciplines.

Audiences will become familiar with types of biases, limitations of human abilities to asses risk probability, and how nuanced real-world interactions can be recreated in serious game environments. This is useful for those interested in employing serious games for training and evaluation, as well as for practitioners to implement in their own games.

Published in: Technology
  • Be the first to comment

  • Be the first to like this

Chris Hazard - Measuring and Manipulating Player Biases and Trust Through Choice and Game Mechanics

  1. 1. Measuring and Manipulating Player Trust through Choice and Game Mechanics Christopher J. Hazard, PhD CEO Hazardous Software Inc.
  2. 2. ● having reason or understanding ● relating to, based on, or agreeable to reason: reasonable
  3. 3. Rational ● having reason or understanding ● relating to, based on, or agreeable to reason: reasonable -Merriam Webster
  4. 4. Humans are rational* *given limited computational bounds, unfounded beliefs of others, inaccurate capability assessments, inexplicable valuations, and some level of [im]patience
  5. 5. Machines are rational* *given limited computational bounds, unfounded beliefs of world, wrong models, inexplicable valuations, and some level of implicit [im]patience
  6. 6. from from by tinyfroglet, cc from from Seattle Weekly from
  7. 7. Reputation ● Belief about attribute ● Hindsight, capabilities, statistics ● Concern: adverse selection Trust ● Belief will not exploit ● Foresight, strategy, game theory ● Concern: moral hazard
  8. 8. +
  9. 9. Trustworthiness ~ Patience
  10. 10. Trustworthiness Isomorphic to Discount Factor ● Compare two agents interacting with third in pure moral hazard situation ● Assumptions – Consistent valuations – Quasilinearity – Trustworthiness sufficiently consistent – Individually rational ● All else equal, given definitions & assumptions, only factor that affects trustworthiness is discount factor
  11. 11. Discounting Everywhere  Stochastic search  Amortization  Bellman Equation  Reinforcement Learning  Markov Decision Processes & POMDPs  Normalize discount rate wrt time
  12. 12. Creeping Sniper's Dilemma ● Single sniper optimal strategy; slow creep out = low risk ● Multiple sniper optimal strategy ● Match quickest visible discount strategy unless too risky
  13. 13. Negotiating  Rubenstein Negotiation  v1 = (1-γ2)/(1-γ1γ2)  Inequalities if rationality not guaranteed  Player & NPC interaction inequalities  Impatience  NPC disagreements with player over choices
  14. 14. Measuring discount factor by choice
  15. 15. Combining Observations: Bayesian Inference
  16. 16. Optimal Level of Patience for Given Scenario
  17. 17. Trust Exploration ● Measure valuations, discount factor, beliefs, maxent regions ● NPCs of different trustworthiness ● Reputations Trust Exploitation ● Push player's ethics buttons: “what is your price?” ● Stability & comfort vs conflict ● Trickery
  18. 18. Psychological Heuristics of Trust Homophily Embedding Corroboration Mass Effect 3 Image from WoW Cataclysm Image from Heavenly Sword
  19. 19. Homophily Embedding Corroboration
  20. 20. Acceptance and Affirmation Antitheses (Alliterated) - Algorithm aversion Dietvorst et al., J Exp Psych 2013 + Anthropomorphization - Abeyance of absorbtion + Acceptance acquiered after asking assistance Flynn, Org Behav & Hum Dec Proc, 2003
  21. 21. Selling Trust With Nuance 1: Don't [unintentionally] Scare Players Nevermind (forthcoming) – Monitor fright Balloon Brigade NBA 2K14 - Swearing IQ, Depression, Behavior, Health, Preferences, Diet, Injuries, Friends, etc. -Newman, Jerome, & Hazard, AIPLA, 2014 Psychometrics for Predicting Behavior – Poore et al., J Cognitive Engineering, 2014.
  22. 22. Selling Trust With Nuance 2: Physiology ● More permissive on right ear than left - Marzoli & Tommasi, Sci of Nat, 2009 ● Two-streams hypothesis for vision processing ● Foveal & spatial detail vs perifoveal & temporal detail ● Mutual exclusion between physical & social reasoning – Jack et al., Neuroimage, 2012 ● Push players to practice self-control – Denson, DeWall, Finkel, Cur Dir in Psych Sci., 2012
  23. 23. Selling Trust With Nuance 3: Hypnosis & Trance ● Relaxation ● Memory ● Creativity ● Suggestability ● Awe & comfort ● Biases ● Placebo effect
  24. 24. Trust & Society ● Enforcing/sanctioning to combat lies ● Incentive compatibility & revelation principle wrt information asymmetry ● Level of trust req'd for system & efficiency ● Too trusting with homophily, embedding, corroboration? ● Common inability to play “red player”
  25. 25. Direct Applications (Conclusions)  NPC decisions: favors, purchases, alliances  Measuring player patience  Adversary willingness to look ahead related to organizational trust (e.g., big bad)  NPC subordinates following player commands based on trustworthiness (explicit or implicit)
  26. 26. Questions?