Emotional Memory and Adaptive Personalities

Auteur-es

Francis, Anthony G. Jr.; Mehta, Manish; Ram, Ashwin

Nombre Auteurs

3

Titre

Emotional Memory and Adaptive Personalities

Année de publication

2009

Référence (APA)

Francis, A. G. Jr., Mehta, M., & Ram, A. (2009). Emotional Memory and Adaptive Personalities. In Handbook of Research on Synthetic Emotions and Sociable Robotics : New Applications in Affective Computing and Artificial Intelligence (p. 391‑412). IGI Global. https://doi.org/10.4018/978-1-60566-354-8.ch020

Mots-clés

ND

URL

https://web.archive.org/web/20150905184906id_/http://www.cc.gatech.edu/faculty/ashwin/papers/er-08-10.pdf

Accessibilité de l'article

Open access

Champ

Machine Intelligence

Type contenu (théorique Applicative méthodologique)

Applicative

Méthode

We propose to create believable, engaging artificial characters capable of long-term interaction with a human user by explicitly modeling the emotional adaptation that goes on in humans and animals.
We developed two implementations of the PEPE architecture (Projet "Personal Pet", un animal de compagnie artificiel capable d'interagir avec plusieurs utilisateurs humains & Jack et Jill, deux personnages incarnés créés à la main et conçus pour jouer à un jeu de Tag).
The first was used for testing reactive control and facial recognition [Stoytchev & Tanawongsuwan 1998]. The second implementation was focused on testing the emotional long term memory system.

Cas d'usage

Project "Personal Pet" (PEPE), an artificial pet capable of interacting with several human users.
Two embodied agents (characters in a game) involved in a game of tag (wolf game).

Objectifs de l'article

To aid the authoring of adaptive agents, we present an artificial intelligence model inspired by these psychological results in which an emotion model triggers case-based emotional preference learning and behavioral adaptation guided by personality models.

Question(s) de recherche/Hypothèses/conclusion

Research question(s) : Believable agents designed for long-term interaction with human users need to adapt to them in a way which appears emotionally plausible while maintaining a consistent personality. For short-term interactions in restricted environments, scripting and state machine techniques can create agents with emotion and personality, but these methods are labor intensive, hard to extend, and brittle in new environments.

Hypothesis(es) : Fortunately, research in memory, emotion and personality in humans and animals points to a solution to this problem. [...]
We argue that robots and synthetic characters should have the same ability to interpret their interactions with us, to remember these interactions, and to recall them appropriately as a guide for future behaviors, and we present a working model of how this can be achieved.
[...] It may seem a tall order make robots have this kind of flexibility – but we argue it is possible by using emotion to trigger behavior revision guided by a personality model, and we present a working model of how it can be achieved.
[...] We argue that using explicit emotion models integrated into an agent’s memory but guided by the agent’s personality model can aid the development of agents for long term interaction

Conclusion(s) : Our tests of this model on robot pets and embodied characters show that emotional adaptation can extend the range and increase the behavioral sophistication of an agent without the need for authoring additional hand-crafted behaviors.

Cadre théorique/Auteur.es

Believable agents (Loyall, 1997)

Techniques for Creating Agent Personalities (Johnston et Thomas, 1995 ; Paiva, 2005 ; Standife,r 1995 ; Saltzman 1999 ; Schwab 2004 ; Millington 2006 ; Huebner, 2000 ; Spector, 2000 ; Ohlen et al., 2001)

Challenges in Creating Agent Personalities (Mateas et Stern, 2003 ; Reilly, 1996 ; Maxis, 2000)

Nature of Emotion (LeDoux, 1996 ; Damasio, 2000 ; Minsky, 2007 ; Simon, 1983 ; Frijda, 1987 ; Ohman et al., 2000 ; Winkielman et Berridge, 2004 ; Ruys & Stapel, 2008)

Memory and Learning (Anderson, 2000 ; Tulving & Craik, 2000 ; Purdy et al., 2001 ; Gluck, 2008 ; McGaugh, 2003 ; McGaugh, 2007 ; Haist et al., 2001 ; Bartlett, 1932)

Relationship of Memory and Emotion (McGaugh, 2003 ; Gluck et al., 2008 ; Benjamin et al., 1981)

Personality and Self-Regulation (Caprara & Cervone, 2000 ; Minsky, 2007)

Modeling Emotion in Intelligent Systems (Simon, 1983 ; Frijda, 1987 ; Velasquez, 1997 ; Velasquez, 1998 ; Ekman, 1992, Izard, 1991 ; Johnson-Laird et Oately, 1992 ; Ortony, Clore et Collins, 1988 ; Reilly, 1996 ; Elliott, 1992 ; Studdard, 1995 ; Koda, 1996 ; Karunaratne & Yan, 2001 ; Bartneck, 2002 ; Li et al., 2007)

Memory Retrieval and Machine Learning (Mitchell, 1997 ; Alpaydin, 2004 ; Kolodner, 1993 ; Kolodner, 1984 ; Sutton & Barto, 1998 ; Santamaria, 1997)

Personality Regulation and Behavior Transformation (Peot et Smith, 1992 ; Weld et. al., 1998)

Concepts clés

Emotional memory

Adaptive personality

Emotional Long Term Memory

Données collectées (type source)

Behavioral data for artificial characters.
To test the PEPE architecture and our ELTM model, we implemented a simple library of behaviors, such as wandering, approach and avoidance, which could in turn be composed into higher level behaviors such as “playing” (alternately wandering and approaching an object) and “fleeing” (backing up, executing a fast 180, and running away). The emotion model extended this with a simple set of concerns, including avoiding pain, which we derived from “kicks” to the rear sensor, and socialization, which we derived from a combination of proximity to people objects and “petting” the head sensor. The robot had several emotional states, including a neutral state, a “happy” state associated with socialization, and a “fearful” state associated with pain.

Définition des émotions

Definition of emotions

Ampleur expérimentation (volume de comptes)

ND

Technologies associées

Pet robots and embodied characters

Artificial Intelligence

Mention de l'éthique

ND

Finalité communicationnelle

Our work used emotion as a trigger for learning about the environment and agents within it, and as a trigger for behavioral change. This model made it possible for us to develop sophisticated and sometimes surprising agent behaviors in less time and with less effort than we could have done otherwise. Therefore, we conclude that emotion-driven learning and emotiondriven behavioral updates are a useful method for developing believable agents that adapt to their environments and users in a way which appears emotionally plausible while maintaining a consistent personality.

Résumé

Believable agents designed for long-term interaction with human users need to adapt to them in a way which appears emotionally plausible while maintaining a consistent personality. For short-term interactions in restricted environments, scripting and state machine techniques can create agents with emotion and personality, but these methods are labor intensive, hard to extend, and brittle in new environments. Fortunately, research in memory, emotion and personality in humans and animals points to a solution to this problem. Emotions focus an animal’s attention on things it needs to care about, and strong emotions trigger enhanced formation of memory, enabling the animal to adapt its emotional response to the objects and situations in its environment. In humans this process becomes reflective: emotional stress or frustration can trigger re-evaluating past behavior with respect to personal standards, which in turn can lead to setting new strategies or goals. To aid the authoring of adaptive agents, we present an artificial intelligence model inspired by these psychological results in which an emotion model triggers case-based emotional preference learning and behavioral adaptation guided by personality models. Our tests of this model on robot pets and embodied characters show that emotional adaptation can extend the range and increase the behavioral sophistication of an agent without the need for authoring additional hand-crafted behaviors.