Towards Modeling the Interplay of Personality, Motivation,Emotion, and Mood in Social Agents

Extended Abstract

Maayan ShvoUniversity of Toronto

Toronto, Canadamaayanshvo@cs.toronto.edu

Jakob BuhmannDisney Research

Zurich, Switzerlandjakob.buhmann@disneyresearch.com

Mubbasir KapadiaRutgers University

Piscataway, New Jerseymubbasir.kapadia@rutgers.edu

ABSTRACTCreating social agents that interact in believable ways is a challeng-ing task. The agent’s emotional state must be faithfully modeledand should influence its behavior. In this paper, we introduce a com-putational model of affect which incorporates an empirically-basedinterplay between its various affective components - personality,motivation, emotion, and mood. These affective components aswell as the relations between them capture a number of importantmechanisms that are observable in human beings (e.g., motivationdriven planning, emotional reactions, or coping) and influence theagent’s decisionmaking. Further, these mechanisms, reflected in theagent’s behavior, are integral to human-human interaction and aretherefore likely to contribute to improved human-agent interaction.In a preliminary evaluation of our approach, we demonstrate theimpact of the various components in the model and their interac-tion with one another on the agent’s decision making and behavior,by showing that the agent displays disparate behavior with andwithout the inclusion of specific components in our model.

KEYWORDSEmotion; Mood; Personality; Motivation; Social AgentsACM Reference Format:Maayan Shvo, Jakob Buhmann, and Mubbasir Kapadia . 2019. TowardsModeling the Interplay of Personality, Motivation, Emotion, and Mood inSocial Agents. In Proc. of the 18th International Conference on AutonomousAgents and Multiagent Systems (AAMAS 2019), Montreal, Canada, May 13–17,2019, IFAAMAS, 3 pages.

1 INTRODUCTIONA long-standing goal for Artificial Intelligence research has been thecreation of emotionally complex, believable, and engaging agentsand as these agents are given increasingly sensitive and impactfulroles in society, it is important to create social agents that benefitthose they interact with by utilizing human emotion. Specifically, itis challenging to create a social agent which is able to meaningfullyinteract with a human user via a believable mechanism that mapsthe agent’s affective state into behavioral and cognitive changes. Tofacilitate the creation of such agents, the underlying computationalmodel of affect must incorporate an interplay between its variousaffective components, as well as handle emotions in a way that isnot foreign to a human user.

Proc. of the 18th International Conference on Autonomous Agents and Multiagent Systems(AAMAS 2019), N. Agmon, M. E. Taylor, E. Elkind, M. Veloso (eds.), May 13–17, 2019,Montreal, Canada. © 2019 International Foundation for Autonomous Agents andMultiagent Systems (www.ifaamas.org). All rights reserved.

Various computational models of affect have been proposed inthe past, differing in the affective components they consist of and,crucially, the interaction between their various components (e.g.,[1, 3, 5–7]). However, previous work has struggled to capture anumber of important mechanisms which influence behavior, eitherdue to the insufficient modeling of each individual component ofaffect, or the lack of coupling between the components. E.g., whilesome previous work has incorporated the influence personality hason an individual’s emotion and mood [5], it does not incorporatea motivational profile for the agent, nor does it include an affect-consequent model which draws a correspondence between the typeof emotion experienced by the agent, and its motivations.

Our ModelIn this work, we propose a computational model of affect whichincludes an empirically-based, interplay between the various com-ponents driving the agent’s behavior, namely personality, motiva-tion, emotion, and mood. Our proposed framework consists of thefollowing components: a personality model, based on the Five Fac-tor Model (FFM) [9]; a motivational model, based on Reiss’ theoryof 16 basic desires [14]; an emotional model, based on appraisaltheory and the PAD (Pleasure, Arousal and Dominance) theory [11],where emotion and mood are differentiated and represent differenttemporal resolutions of the agent’s affect. Crucially, we model theinterplay between the aforementioned components (see Figure 1).We model the influence of personality on the emotional stateof the agent based on research that has shown that individualswith different personality traits experience emotions differently [9].Additionally, the model accounts for an impact of an agent’s per-sonality on its motivation which reflects a correlation betweenan individual’s personality traits and the importance they assignto each of Reiss’ 16 basic desires [13]. Following empirical findingsthat mood affects the intensity of an experienced emotion [12], ourmodel incorporates an interaction between mood and emotions.Further, the agent’s active emotions change its mood such thatit faithfully represents this empirically-based definition of mood:“an average of a person’s emotional states across a representativevariety of life situations” [10]. Lastly, we model the influence ofemotion on motivation via an affect-consequent model whichis grounded in therapeutic psychological literature [8] and whichallows emotions to influence the agent’s motivation (which, in turn,drive its decision making). Our proposed model addresses theseinteractions and thereby allows to model agents with rich and be-lievable behavior. The conceptual framework described in this papercan be used as the basis to drive other application-specific agent

Extended Abstract AAMAS 2019, May 13-17, 2019, Montréal, Canada

2195

P<latexit sha1_base64="DcZa5CvdCJgSgSXDHRpbMxTORTo=">AAAB23icbU/LSsNAFL1TX7W+qi7dBIvgqiQi6LLoxmUF+4A0lMn0tg6dR8hMhBKyciduu9UP8Uv8G5OahW09cOFwzrlwThgJbqzrfpPKxubW9k51t7a3f3B4VD8+6RqdxAw7TAsd90NqUHCFHcutwH4UI5WhwF44vS/83gvGhmv1ZGcRBpJOFB9zRm0u+QNJ7TOjIm1nw3rDbboLOOvEK0kDSrSH9a/BSLNEorJMUGN8z41skNLYciYwqw0SgxFlUzpBP6eKSjRBuqicORe5MnLGOs5PWWeh/v1IqTRmJsM8WVQ0q14h/uf5iR3fBilXUWJRsaUWaZG2WguT1fK13uq2ddK9anpu03u8brTuyt1VOINzuAQPbqAFD9CGDjDQMIcP+CQBeSVv5P03WiHlzyksgcx/AFaFiOQ=</latexit><latexit sha1_base64="DcZa5CvdCJgSgSXDHRpbMxTORTo=">AAAB23icbU/LSsNAFL1TX7W+qi7dBIvgqiQi6LLoxmUF+4A0lMn0tg6dR8hMhBKyciduu9UP8Uv8G5OahW09cOFwzrlwThgJbqzrfpPKxubW9k51t7a3f3B4VD8+6RqdxAw7TAsd90NqUHCFHcutwH4UI5WhwF44vS/83gvGhmv1ZGcRBpJOFB9zRm0u+QNJ7TOjIm1nw3rDbboLOOvEK0kDSrSH9a/BSLNEorJMUGN8z41skNLYciYwqw0SgxFlUzpBP6eKSjRBuqicORe5MnLGOs5PWWeh/v1IqTRmJsM8WVQ0q14h/uf5iR3fBilXUWJRsaUWaZG2WguT1fK13uq2ddK9anpu03u8brTuyt1VOINzuAQPbqAFD9CGDjDQMIcP+CQBeSVv5P03WiHlzyksgcx/AFaFiOQ=</latexit><latexit sha1_base64="DcZa5CvdCJgSgSXDHRpbMxTORTo=">AAAB23icbU/LSsNAFL1TX7W+qi7dBIvgqiQi6LLoxmUF+4A0lMn0tg6dR8hMhBKyciduu9UP8Uv8G5OahW09cOFwzrlwThgJbqzrfpPKxubW9k51t7a3f3B4VD8+6RqdxAw7TAsd90NqUHCFHcutwH4UI5WhwF44vS/83gvGhmv1ZGcRBpJOFB9zRm0u+QNJ7TOjIm1nw3rDbboLOOvEK0kDSrSH9a/BSLNEorJMUGN8z41skNLYciYwqw0SgxFlUzpBP6eKSjRBuqicORe5MnLGOs5PWWeh/v1IqTRmJsM8WVQ0q14h/uf5iR3fBilXUWJRsaUWaZG2WguT1fK13uq2ddK9anpu03u8brTuyt1VOINzuAQPbqAFD9CGDjDQMIcP+CQBeSVv5P03WiHlzyksgcx/AFaFiOQ=</latexit><latexit sha1_base64="DcZa5CvdCJgSgSXDHRpbMxTORTo=">AAAB23icbU/LSsNAFL1TX7W+qi7dBIvgqiQi6LLoxmUF+4A0lMn0tg6dR8hMhBKyciduu9UP8Uv8G5OahW09cOFwzrlwThgJbqzrfpPKxubW9k51t7a3f3B4VD8+6RqdxAw7TAsd90NqUHCFHcutwH4UI5WhwF44vS/83gvGhmv1ZGcRBpJOFB9zRm0u+QNJ7TOjIm1nw3rDbboLOOvEK0kDSrSH9a/BSLNEorJMUGN8z41skNLYciYwqw0SgxFlUzpBP6eKSjRBuqicORe5MnLGOs5PWWeh/v1IqTRmJsM8WVQ0q14h/uf5iR3fBilXUWJRsaUWaZG2WguT1fK13uq2ddK9anpu03u8brTuyt1VOINzuAQPbqAFD9CGDjDQMIcP+CQBeSVv5P03WiHlzyksgcx/AFaFiOQ=</latexit>

Mt<latexit sha1_base64="x5Gakw5Bg1G0plfb5o8Bsm9+bJc=">AAAB3XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiFy9CBdMW2hA229e6dLMbsxuhhBy9idde9Xf4S/w3JjUH2zrwYJiZBzNBJLg2tv1NKmvrG5tb1e3azu7e/kH98KijVRIzdJkSKu4FVKPgEl3DjcBeFCMNA4HdYHJb+N0XjDVX8tFMI/RCOpZ8xBk1ueQNQmqeGBXpfeYbv96wm/Yc1ipxStKAEm2//jUYKpaEKA0TVOu+Y0fGS2lsOBOY1QaJxoiyCR1jP6eShqi9dF46s85yZWiNVJyfNNZc/fuR0lDraRjkyaKkXvYK8T+vn5jRtZdyGSUGJVtokRZpo5TQWS1f6yxvWyWdi6ZjN52Hy0brptxdhRM4hXNw4ApacAdtcIHBM8zgAz6JT17JG3n/jVZI+XMMCyCzH9I6icg=</latexit><latexit sha1_base64="x5Gakw5Bg1G0plfb5o8Bsm9+bJc=">AAAB3XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiFy9CBdMW2hA229e6dLMbsxuhhBy9idde9Xf4S/w3JjUH2zrwYJiZBzNBJLg2tv1NKmvrG5tb1e3azu7e/kH98KijVRIzdJkSKu4FVKPgEl3DjcBeFCMNA4HdYHJb+N0XjDVX8tFMI/RCOpZ8xBk1ueQNQmqeGBXpfeYbv96wm/Yc1ipxStKAEm2//jUYKpaEKA0TVOu+Y0fGS2lsOBOY1QaJxoiyCR1jP6eShqi9dF46s85yZWiNVJyfNNZc/fuR0lDraRjkyaKkXvYK8T+vn5jRtZdyGSUGJVtokRZpo5TQWS1f6yxvWyWdi6ZjN52Hy0brptxdhRM4hXNw4ApacAdtcIHBM8zgAz6JT17JG3n/jVZI+XMMCyCzH9I6icg=</latexit><latexit sha1_base64="x5Gakw5Bg1G0plfb5o8Bsm9+bJc=">AAAB3XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiFy9CBdMW2hA229e6dLMbsxuhhBy9idde9Xf4S/w3JjUH2zrwYJiZBzNBJLg2tv1NKmvrG5tb1e3azu7e/kH98KijVRIzdJkSKu4FVKPgEl3DjcBeFCMNA4HdYHJb+N0XjDVX8tFMI/RCOpZ8xBk1ueQNQmqeGBXpfeYbv96wm/Yc1ipxStKAEm2//jUYKpaEKA0TVOu+Y0fGS2lsOBOY1QaJxoiyCR1jP6eShqi9dF46s85yZWiNVJyfNNZc/fuR0lDraRjkyaKkXvYK8T+vn5jRtZdyGSUGJVtokRZpo5TQWS1f6yxvWyWdi6ZjN52Hy0brptxdhRM4hXNw4ApacAdtcIHBM8zgAz6JT17JG3n/jVZI+XMMCyCzH9I6icg=</latexit><latexit sha1_base64="x5Gakw5Bg1G0plfb5o8Bsm9+bJc=">AAAB3XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiFy9CBdMW2hA229e6dLMbsxuhhBy9idde9Xf4S/w3JjUH2zrwYJiZBzNBJLg2tv1NKmvrG5tb1e3azu7e/kH98KijVRIzdJkSKu4FVKPgEl3DjcBeFCMNA4HdYHJb+N0XjDVX8tFMI/RCOpZ8xBk1ueQNQmqeGBXpfeYbv96wm/Yc1ipxStKAEm2//jUYKpaEKA0TVOu+Y0fGS2lsOBOY1QaJxoiyCR1jP6eShqi9dF46s85yZWiNVJyfNNZc/fuR0lDraRjkyaKkXvYK8T+vn5jRtZdyGSUGJVtokRZpo5TQWS1f6yxvWyWdi6ZjN52Hy0brptxdhRM4hXNw4ApacAdtcIHBM8zgAz6JT17JG3n/jVZI+XMMCyCzH9I6icg=</latexit>

~M c

~Md

~ws

⇡A⇤(s)<latexit sha1_base64="H50BzhxSlp4jmkPS9hZrwBwQavk=">AAAB3HicbU9LS8NAGPy2vmp9VT16CRaheiiJCHqsevFYwT6gSctmu61L90V2I5SQmzfx2qv+D3+J/8a05mBbBz4YZuaDmVBzZqzrfqPC2vrG5lZxu7Szu7d/UD48ahkVR4Q2ieIq6oTYUM4kbVpmOe3oiGIRctoOx/czv/1CI8OUfLITTQOBR5INGcE2k3xfs35y27tIq+a8X664NXcOZ5V4OalAjka//OUPFIkFlZZwbEzXc7UNEhxZRjhNS35sqMZkjEe0m1GJBTVBMu+cOmeZMnCGKspOWmeu/v1IsDBmIsIsKbB9NsveTPzP68Z2eBMkTOrYUkkWWiSztFWKm7SUrfWWt62S1mXNc2ve41WlfpfvLsIJnEIVPLiGOjxAA5pAQMMUPuAT9dArekPvv9ECyn+OYQFo+gPt4IiJ</latexit><latexit sha1_base64="H50BzhxSlp4jmkPS9hZrwBwQavk=">AAAB3HicbU9LS8NAGPy2vmp9VT16CRaheiiJCHqsevFYwT6gSctmu61L90V2I5SQmzfx2qv+D3+J/8a05mBbBz4YZuaDmVBzZqzrfqPC2vrG5lZxu7Szu7d/UD48ahkVR4Q2ieIq6oTYUM4kbVpmOe3oiGIRctoOx/czv/1CI8OUfLITTQOBR5INGcE2k3xfs35y27tIq+a8X664NXcOZ5V4OalAjka//OUPFIkFlZZwbEzXc7UNEhxZRjhNS35sqMZkjEe0m1GJBTVBMu+cOmeZMnCGKspOWmeu/v1IsDBmIsIsKbB9NsveTPzP68Z2eBMkTOrYUkkWWiSztFWKm7SUrfWWt62S1mXNc2ve41WlfpfvLsIJnEIVPLiGOjxAA5pAQMMUPuAT9dArekPvv9ECyn+OYQFo+gPt4IiJ</latexit><latexit sha1_base64="H50BzhxSlp4jmkPS9hZrwBwQavk=">AAAB3HicbU9LS8NAGPy2vmp9VT16CRaheiiJCHqsevFYwT6gSctmu61L90V2I5SQmzfx2qv+D3+J/8a05mBbBz4YZuaDmVBzZqzrfqPC2vrG5lZxu7Szu7d/UD48ahkVR4Q2ieIq6oTYUM4kbVpmOe3oiGIRctoOx/czv/1CI8OUfLITTQOBR5INGcE2k3xfs35y27tIq+a8X664NXcOZ5V4OalAjka//OUPFIkFlZZwbEzXc7UNEhxZRjhNS35sqMZkjEe0m1GJBTVBMu+cOmeZMnCGKspOWmeu/v1IsDBmIsIsKbB9NsveTPzP68Z2eBMkTOrYUkkWWiSztFWKm7SUrfWWt62S1mXNc2ve41WlfpfvLsIJnEIVPLiGOjxAA5pAQMMUPuAT9dArekPvv9ECyn+OYQFo+gPt4IiJ</latexit><latexit sha1_base64="H50BzhxSlp4jmkPS9hZrwBwQavk=">AAAB3HicbU9LS8NAGPy2vmp9VT16CRaheiiJCHqsevFYwT6gSctmu61L90V2I5SQmzfx2qv+D3+J/8a05mBbBz4YZuaDmVBzZqzrfqPC2vrG5lZxu7Szu7d/UD48ahkVR4Q2ieIq6oTYUM4kbVpmOe3oiGIRctoOx/czv/1CI8OUfLITTQOBR5INGcE2k3xfs35y27tIq+a8X664NXcOZ5V4OalAjka//OUPFIkFlZZwbEzXc7UNEhxZRjhNS35sqMZkjEe0m1GJBTVBMu+cOmeZMnCGKspOWmeu/v1IsDBmIsIsKbB9NsveTPzP68Z2eBMkTOrYUkkWWiSztFWKm7SUrfWWt62S1mXNc2ve41WlfpfvLsIJnEIVPLiGOjxAA5pAQMMUPuAT9dArekPvv9ECyn+OYQFo+gPt4IiJ</latexit>

⇡GP(gEi, s)

<latexit sha1_base64="nspJEonfTk/sQUstCH2DSAXs92M=">AAAB9nicbU/LSsNAFL2pr1pfUZduglWoICURQZdFEV1WsA9oQphMp3VwMhMyE2kZ8gN+hDtx2636Jf6NSc3Cth64cO4558I9QcSoVLb9bZSWlldW18rrlY3Nre0dc3evLUUSY9LCgom4GyBJGOWkpahipBvFBIUBI53g6Tr3O88kllTwBzWOiBeiIacDipHKJN88ciPqa1eRkdK3zTStDbMtROoRI6ZvUp+mp/LEN6t23Z7CWiROQapQoOmbX25f4CQkXGGGpOw5dqQ8jWJFMSNpxU0kiRB+QkPSyyhHIZGenrZJreNM6VsDEWfDlTVV/15oFEo5DoMsmT8q571c/M/rJWpw6WnKo0QRjme+0HlaCcFkWsnaOvPdFkn7rO7Ydef+vNq4KnqX4QAOoQYOXEAD7qAJLcDwAhP4gE9jZLwab8b7b7RkFDf7MANj8gMWj5O2</latexit><latexit sha1_base64="nspJEonfTk/sQUstCH2DSAXs92M=">AAAB9nicbU/LSsNAFL2pr1pfUZduglWoICURQZdFEV1WsA9oQphMp3VwMhMyE2kZ8gN+hDtx2636Jf6NSc3Cth64cO4558I9QcSoVLb9bZSWlldW18rrlY3Nre0dc3evLUUSY9LCgom4GyBJGOWkpahipBvFBIUBI53g6Tr3O88kllTwBzWOiBeiIacDipHKJN88ciPqa1eRkdK3zTStDbMtROoRI6ZvUp+mp/LEN6t23Z7CWiROQapQoOmbX25f4CQkXGGGpOw5dqQ8jWJFMSNpxU0kiRB+QkPSyyhHIZGenrZJreNM6VsDEWfDlTVV/15oFEo5DoMsmT8q571c/M/rJWpw6WnKo0QRjme+0HlaCcFkWsnaOvPdFkn7rO7Ydef+vNq4KnqX4QAOoQYOXEAD7qAJLcDwAhP4gE9jZLwab8b7b7RkFDf7MANj8gMWj5O2</latexit><latexit sha1_base64="nspJEonfTk/sQUstCH2DSAXs92M=">AAAB9nicbU/LSsNAFL2pr1pfUZduglWoICURQZdFEV1WsA9oQphMp3VwMhMyE2kZ8gN+hDtx2636Jf6NSc3Cth64cO4558I9QcSoVLb9bZSWlldW18rrlY3Nre0dc3evLUUSY9LCgom4GyBJGOWkpahipBvFBIUBI53g6Tr3O88kllTwBzWOiBeiIacDipHKJN88ciPqa1eRkdK3zTStDbMtROoRI6ZvUp+mp/LEN6t23Z7CWiROQapQoOmbX25f4CQkXGGGpOw5dqQ8jWJFMSNpxU0kiRB+QkPSyyhHIZGenrZJreNM6VsDEWfDlTVV/15oFEo5DoMsmT8q571c/M/rJWpw6WnKo0QRjme+0HlaCcFkWsnaOvPdFkn7rO7Ydef+vNq4KnqX4QAOoQYOXEAD7qAJLcDwAhP4gE9jZLwab8b7b7RkFDf7MANj8gMWj5O2</latexit><latexit sha1_base64="nspJEonfTk/sQUstCH2DSAXs92M=">AAAB9nicbU/LSsNAFL2pr1pfUZduglWoICURQZdFEV1WsA9oQphMp3VwMhMyE2kZ8gN+hDtx2636Jf6NSc3Cth64cO4558I9QcSoVLb9bZSWlldW18rrlY3Nre0dc3evLUUSY9LCgom4GyBJGOWkpahipBvFBIUBI53g6Tr3O88kllTwBzWOiBeiIacDipHKJN88ciPqa1eRkdK3zTStDbMtROoRI6ZvUp+mp/LEN6t23Z7CWiROQapQoOmbX25f4CQkXGGGpOw5dqQ8jWJFMSNpxU0kiRB+QkPSyyhHIZGenrZJreNM6VsDEWfDlTVV/15oFEo5DoMsmT8q571c/M/rJWpw6WnKo0QRjme+0HlaCcFkWsnaOvPdFkn7rO7Ydef+vNq4KnqX4QAOoQYOXEAD7qAJLcDwAhP4gE9jZLwab8b7b7RkFDf7MANj8gMWj5O2</latexit>

gEi<latexit sha1_base64="OMpubhCH62bMn3WfPWtky8D45cE=">AAAB43icbU/LSsNAFL3xWesr1aWbYBFclUQEXRZFcFnBPqANYTK9rUMnMyEzUcqQL3AnbrvVj/BL/BuTmoVtPXDhcM65cE4Yc6a0635ba+sbm1vblZ3q7t7+waFdO+oomSYU21RymfRCopAzgW3NNMdenCCJQo7dcHJb+N1nTBST4lFPY/QjMhZsxCjRuRTYtXFgBhHRT5Rwc5cFLAvsuttw53BWiVeSOpRoBfbXYChpGqHQlBOl+p4ba9+QRDPKMasOUoUxoRMyxn5OBYlQ+WZePXPOcmXojGSSn9DOXP37YUik1DQK82TRUi17hfif10/16No3TMSpRkEXWpgiraXkKqvma73lbaukc9Hw3Ib3cFlv3pS7K3ACp3AOHlxBE+6hBW2g8AIz+IBPC61X6816/42uWeXPMSzAmv0AU+2LzA==</latexit><latexit sha1_base64="OMpubhCH62bMn3WfPWtky8D45cE=">AAAB43icbU/LSsNAFL3xWesr1aWbYBFclUQEXRZFcFnBPqANYTK9rUMnMyEzUcqQL3AnbrvVj/BL/BuTmoVtPXDhcM65cE4Yc6a0635ba+sbm1vblZ3q7t7+waFdO+oomSYU21RymfRCopAzgW3NNMdenCCJQo7dcHJb+N1nTBST4lFPY/QjMhZsxCjRuRTYtXFgBhHRT5Rwc5cFLAvsuttw53BWiVeSOpRoBfbXYChpGqHQlBOl+p4ba9+QRDPKMasOUoUxoRMyxn5OBYlQ+WZePXPOcmXojGSSn9DOXP37YUik1DQK82TRUi17hfif10/16No3TMSpRkEXWpgiraXkKqvma73lbaukc9Hw3Ib3cFlv3pS7K3ACp3AOHlxBE+6hBW2g8AIz+IBPC61X6816/42uWeXPMSzAmv0AU+2LzA==</latexit><latexit sha1_base64="OMpubhCH62bMn3WfPWtky8D45cE=">AAAB43icbU/LSsNAFL3xWesr1aWbYBFclUQEXRZFcFnBPqANYTK9rUMnMyEzUcqQL3AnbrvVj/BL/BuTmoVtPXDhcM65cE4Yc6a0635ba+sbm1vblZ3q7t7+waFdO+oomSYU21RymfRCopAzgW3NNMdenCCJQo7dcHJb+N1nTBST4lFPY/QjMhZsxCjRuRTYtXFgBhHRT5Rwc5cFLAvsuttw53BWiVeSOpRoBfbXYChpGqHQlBOl+p4ba9+QRDPKMasOUoUxoRMyxn5OBYlQ+WZePXPOcmXojGSSn9DOXP37YUik1DQK82TRUi17hfif10/16No3TMSpRkEXWpgiraXkKqvma73lbaukc9Hw3Ib3cFlv3pS7K3ACp3AOHlxBE+6hBW2g8AIz+IBPC61X6816/42uWeXPMSzAmv0AU+2LzA==</latexit><latexit sha1_base64="OMpubhCH62bMn3WfPWtky8D45cE=">AAAB43icbU/LSsNAFL3xWesr1aWbYBFclUQEXRZFcFnBPqANYTK9rUMnMyEzUcqQL3AnbrvVj/BL/BuTmoVtPXDhcM65cE4Yc6a0635ba+sbm1vblZ3q7t7+waFdO+oomSYU21RymfRCopAzgW3NNMdenCCJQo7dcHJb+N1nTBST4lFPY/QjMhZsxCjRuRTYtXFgBhHRT5Rwc5cFLAvsuttw53BWiVeSOpRoBfbXYChpGqHQlBOl+p4ba9+QRDPKMasOUoUxoRMyxn5OBYlQ+WZePXPOcmXojGSSn9DOXP37YUik1DQK82TRUi17hfif10/16No3TMSpRkEXWpgiraXkKqvma73lbaukc9Hw3Ib3cFlv3pS7K3ACp3AOHlxBE+6hBW2g8AIz+IBPC61X6816/42uWeXPMSzAmv0AU+2LzA==</latexit> ⌧E

<latexit sha1_base64="hbMVNvU+XDvOIlwpA6AFhvp2pH4=">AAAB5HicbU/LSsNAFL3xWesr1qWbYBFclUQEXRZFcFnBPqAJYTKd1KHzCJmJWIb8gTtx263+g1/i35jULGzrgQuHc86Fc6KEUaVd99taW9/Y3Nqu7dR39/YPDu2jRk/JLMWkiyWT6SBCijAqSFdTzcggSQniESP9aHJb+v1nkioqxaOeJiTgaCxoTDHShRTaDV+jLDQ+R/oJI2bu8jy0m27LncNZJV5FmlChE9pf/kjijBOhMUNKDT030YFBqaaYkbzuZ4okCE/QmAwLKhAnKjDz7rlzVigjJ5ZpcUI7c/Xvh0FcqSmPimTZUS17pfifN8x0fB0YKpJME4EXWpgyraVkKq8Xa73lbaukd9Hy3Jb3cNls31S7a3ACp3AOHlxBG+6hA13A8AIz+IBPK7ZerTfr/Te6ZlU/x7AAa/YDKfOMTQ==</latexit><latexit sha1_base64="hbMVNvU+XDvOIlwpA6AFhvp2pH4=">AAAB5HicbU/LSsNAFL3xWesr1qWbYBFclUQEXRZFcFnBPqAJYTKd1KHzCJmJWIb8gTtx263+g1/i35jULGzrgQuHc86Fc6KEUaVd99taW9/Y3Nqu7dR39/YPDu2jRk/JLMWkiyWT6SBCijAqSFdTzcggSQniESP9aHJb+v1nkioqxaOeJiTgaCxoTDHShRTaDV+jLDQ+R/oJI2bu8jy0m27LncNZJV5FmlChE9pf/kjijBOhMUNKDT030YFBqaaYkbzuZ4okCE/QmAwLKhAnKjDz7rlzVigjJ5ZpcUI7c/Xvh0FcqSmPimTZUS17pfifN8x0fB0YKpJME4EXWpgyraVkKq8Xa73lbaukd9Hy3Jb3cNls31S7a3ACp3AOHlxBG+6hA13A8AIz+IBPK7ZerTfr/Te6ZlU/x7AAa/YDKfOMTQ==</latexit><latexit sha1_base64="hbMVNvU+XDvOIlwpA6AFhvp2pH4=">AAAB5HicbU/LSsNAFL3xWesr1qWbYBFclUQEXRZFcFnBPqAJYTKd1KHzCJmJWIb8gTtx263+g1/i35jULGzrgQuHc86Fc6KEUaVd99taW9/Y3Nqu7dR39/YPDu2jRk/JLMWkiyWT6SBCijAqSFdTzcggSQniESP9aHJb+v1nkioqxaOeJiTgaCxoTDHShRTaDV+jLDQ+R/oJI2bu8jy0m27LncNZJV5FmlChE9pf/kjijBOhMUNKDT030YFBqaaYkbzuZ4okCE/QmAwLKhAnKjDz7rlzVigjJ5ZpcUI7c/Xvh0FcqSmPimTZUS17pfifN8x0fB0YKpJME4EXWpgyraVkKq8Xa73lbaukd9Hy3Jb3cNls31S7a3ACp3AOHlxBG+6hA13A8AIz+IBPK7ZerTfr/Te6ZlU/x7AAa/YDKfOMTQ==</latexit><latexit sha1_base64="hbMVNvU+XDvOIlwpA6AFhvp2pH4=">AAAB5HicbU/LSsNAFL3xWesr1qWbYBFclUQEXRZFcFnBPqAJYTKd1KHzCJmJWIb8gTtx263+g1/i35jULGzrgQuHc86Fc6KEUaVd99taW9/Y3Nqu7dR39/YPDu2jRk/JLMWkiyWT6SBCijAqSFdTzcggSQniESP9aHJb+v1nkioqxaOeJiTgaCxoTDHShRTaDV+jLDQ+R/oJI2bu8jy0m27LncNZJV5FmlChE9pf/kjijBOhMUNKDT030YFBqaaYkbzuZ4okCE/QmAwLKhAnKjDz7rlzVigjJ5ZpcUI7c/Xvh0FcqSmPimTZUS17pfifN8x0fB0YKpJME4EXWpgyraVkKq8Xa73lbaukd9Hy3Jb3cNls31S7a3ACp3AOHlxBG+6hA13A8AIz+IBPK7ZerTfr/Te6ZlU/x7AAa/YDKfOMTQ==</latexit>

ACTION SCHEDULER

S<latexit sha1_base64="0vgCJLpTcayoKb51uKJgzf7AH/o=">AAAB0XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiF48tNW2hDWWzfa1L9yNkN0IJBfHaq/4af4n/xjTmYFsHHgwz82AmjAQ31nW/SWlre2d3r7xfOTg8Oj6pnp51jE5ihj7TQse9kBoUXKFvuRXYi2KkMhTYDaePS7/7irHhWj3bWYSBpBPFx5xRm0mt9rBac+tuDmeTeAWpQYHmsPo1GGmWSFSWCWpM33MjG6Q0tpwJnFcGicGIsimdYD+jiko0QZoXnTtXmTJyxjrOTlknV/9+pFQaM5NhlpTUvph1byn+5/UTO74PUq6ixKJiKy3SZdpqLcy8kq311rdtks5N3XPrXuu21ngodpfhAi7hGjy4gwY8QRN8YICwgA/4JG0yI2/k/TdaIsXPOayALH4A3lSEVQ==</latexit><latexit sha1_base64="0vgCJLpTcayoKb51uKJgzf7AH/o=">AAAB0XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiF48tNW2hDWWzfa1L9yNkN0IJBfHaq/4af4n/xjTmYFsHHgwz82AmjAQ31nW/SWlre2d3r7xfOTg8Oj6pnp51jE5ihj7TQse9kBoUXKFvuRXYi2KkMhTYDaePS7/7irHhWj3bWYSBpBPFx5xRm0mt9rBac+tuDmeTeAWpQYHmsPo1GGmWSFSWCWpM33MjG6Q0tpwJnFcGicGIsimdYD+jiko0QZoXnTtXmTJyxjrOTlknV/9+pFQaM5NhlpTUvph1byn+5/UTO74PUq6ixKJiKy3SZdpqLcy8kq311rdtks5N3XPrXuu21ngodpfhAi7hGjy4gwY8QRN8YICwgA/4JG0yI2/k/TdaIsXPOayALH4A3lSEVQ==</latexit><latexit sha1_base64="0vgCJLpTcayoKb51uKJgzf7AH/o=">AAAB0XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiF48tNW2hDWWzfa1L9yNkN0IJBfHaq/4af4n/xjTmYFsHHgwz82AmjAQ31nW/SWlre2d3r7xfOTg8Oj6pnp51jE5ihj7TQse9kBoUXKFvuRXYi2KkMhTYDaePS7/7irHhWj3bWYSBpBPFx5xRm0mt9rBac+tuDmeTeAWpQYHmsPo1GGmWSFSWCWpM33MjG6Q0tpwJnFcGicGIsimdYD+jiko0QZoXnTtXmTJyxjrOTlknV/9+pFQaM5NhlpTUvph1byn+5/UTO74PUq6ixKJiKy3SZdpqLcy8kq311rdtks5N3XPrXuu21ngodpfhAi7hGjy4gwY8QRN8YICwgA/4JG0yI2/k/TdaIsXPOayALH4A3lSEVQ==</latexit><latexit sha1_base64="0vgCJLpTcayoKb51uKJgzf7AH/o=">AAAB0XicbU9NS8NAFHxbv2r9qnr0EiyCp5KIoMeiF48tNW2hDWWzfa1L9yNkN0IJBfHaq/4af4n/xjTmYFsHHgwz82AmjAQ31nW/SWlre2d3r7xfOTg8Oj6pnp51jE5ihj7TQse9kBoUXKFvuRXYi2KkMhTYDaePS7/7irHhWj3bWYSBpBPFx5xRm0mt9rBac+tuDmeTeAWpQYHmsPo1GGmWSFSWCWpM33MjG6Q0tpwJnFcGicGIsimdYD+jiko0QZoXnTtXmTJyxjrOTlknV/9+pFQaM5NhlpTUvph1byn+5/UTO74PUq6ixKJiKy3SZdpqLcy8kq311rdtks5N3XPrXuu21ngodpfhAi7hGjy4gwY8QRN8YICwgA/4JG0yI2/k/TdaIsXPOayALH4A3lSEVQ==</latexit>

hallo

hallo zurück

Current Mood

Current Motivation

Target Motivation

Motivation Weight Vector

Active Emotions

Behavior Manager

Emotional Reaction Goal

Heuristic-based Forward Planner

EmotionalReaction Planner

Hardwired Reactions

ACTION

SCHEDULER

State

Motivation

Personality

Mood

Emotion

INPUT MANAGER

D E L I B E R A T I V E M O D U L E

Figure 1: Overview of themodel of affect, including its inter-actions and influence on the agents behavior generation.

architectures including story-driven agents [2] or emotion-awareembodied conversational agents [15].

Motivation-driven Planning. Human decisionmaking, specificallyour goal-oriented action selection, is affected by many differentfactors, with our intrinsic motivations and desires being perhapsprime among them [14]. To facilitate the idea that an agent strivesto fulfill its basic desires and motivations we use a forward plannerwith an A∗ heuristic search algorithm to produce a plan πA∗ . Theheuristic is a weighted Euclidean distance in the motivation space,where the weighting is influenced by the agent’s personality andemotions, and the distance is computed based on a state-dependentcurrent motivation and an agent specific target motivation.

Acting on Emotions. Emotions play the important function ofprompting behavior in the form of reactive action urges [8]. Forexample, if the agent is sufficiently angry (defined by a thresholdon the intensity of the emotion), it may be prompted to throw anobject, or yell, at a nearby agent. Such behavior will be consideredan action urge associated with the agent’s experienced anger. Toenable the agent to express its experienced emotions, we implementan emotionally reactive layer within the deliberative module whichcan trigger an appropriate emotional reaction, based on the emotionthat requires expression and the state in which the agent is situated.

Emotion and Motivation. Following the work on DBT [8], weallow the agent’s emotional state to influence its behavior by draw-ing a correspondence between what is implicitly communicated tothe agent by its emotions, and the 16 basic desires which drive theagent’s decision making and behavior. This influence is achievedby changing the weighting that is given to each motivation. Sincethis weighting is directly used in the heuristic of the A∗ planner,the resulting goal driven behavior may reflect the current emotion.Finally, this relation also allows an agent to address the negativeemotion as the motivations corresponding to the emotion, impor-tantly correspond to the underlying emotional need of the emotion.

2 MODEL ASSESSMENTThe objective of our preliminary evaluation is: (1) to showcase howdifferent behavior can result from slightly different agent configura-tions, thus highlighting the benefits of our proposed model; and (2)to demonstrate the impact of the various components in the modeland their interaction with one another on the agent’s behavior.

To perform the evaluation, we introduce a scenario that is inde-pendently acted out by various agents. We selectively enable dif-ferent interactions between the components in each of the agents’

models, such that on one end there are no interactions at all, andon the other we include all interactions present in our model. Thescenario is as follows: the agent is very hopeful about passing anexam it is about to take, however, it fails miserably which naturallyleads to a strong feeling of disappointment. The agent then pro-ceeds to work on a puzzle and is told that it can cheat and completeit in a fraction of the time it would normally take. We implementedthis scenario in the Racket programming language [4]. The state ofthe agent’s world, as well as its emotional state, are symbolicallyrepresented and are updated in every iteration of the system’s loop.In total, the scenario’s domain included roughly 100 actions, 60state-motivation rules, and 10 objects and the planner was able tocompute plans in less than 100ms of planning time.

From this preliminary evaluation, we can report on a number ofinteresting behavioral manifestations of our model. For instance,when simulating an agent high on the openness and agreeablenessuniversal personality traits, the agent did not cheat when solvingthe puzzle, as it especially values the honor and idealism motiva-tional dimensions. To contrast, an otherwise identical agent, exceptfor its low openness and agreeableness, chooses to cheat since themotivational gain of succeeding may outweigh the negative impactof cheating. Next, we contrast an agent whose personality inhibitsits experienced disappointment after failing the exam, thus prevent-ing an emotional reaction, with an agent whose personality doesnot inhibit this reaction. Simply by assigning the latter agent a neu-rotic personality, it experiences negative emotions more intensely,leading to an increased number of emotional reactions on average.Finally, in an agent with the emotion-motivation relation enabled,the disappointment following the exam influences the agent’s moti-vations and decision making. The agent is thus able to address thenegative emotion as the social-contact motivational dimension isgiven additional weight. In our simulation, this leads to the agentreaching out to a fellow agent to share its disappointment.

3 CONCLUSIONWe set out to create a computational model of affect, underlying anemotionally complex and believable agent. Our model incorporatesa rich, empirically-based, interplay between the different compo-nents driving the agent’s behavior, namely personality, motivation,mood, and emotion. The relations between the different compo-nents, specifically the way in which they influence one another,are grounded in psychological theory, with the hope of enabling arelatable and emotionally meaningful interaction with the humanuser. Our preliminary evaluation shows that different agents, whichdiffer in their personality profile and in the degree of interactionbetween the affective components in their model, behave very dif-ferently. Rather than having the agent’s behavior be driven solelyby its motivations, the agent’s emotional state influences its deci-sion making via an affect-consequent model. In the future, we wishto conduct a study to evaluate the nature of interactions betweenour agent and a human.

ACKNOWLEDGMENTSKapadia has been funded in part by NSF IIS-1703883, NSF S&AS-1723869, and DARPA SocialSim-W911NF-17-C-0098.

Extended Abstract AAMAS 2019, May 13-17, 2019, Montréal, Canada

2196

REFERENCES[1] Ruth S Aylett, Sandy Louchart, Joao Dias, Ana Paiva, and Marco Vala. 2005.

FearNot!–An Experiment in Emergent Narrative. In International Workshop onIntelligent Virtual Agents. Springer, 305–316.

[2] Jessica Falk, Steven Poulakos, Mubbasir Kapadia, and Robert W Sumner. 2018.PICA: Proactive Intelligent Conversational Agent for Interactive Narratives. InProceedings of the 18th International Conference on Intelligent Virtual Agents. ACM,141–146.

[3] Filipa Ferrada and Luis M Camarinha-Matos. 2017. A system dynamics andagent-based approach to model emotions in collaborative networks. In DoctoralConference on Computing, Electrical and Industrial Systems. Springer, 29–43.

[4] Matthew Flatt. 2010. PLT. Reference: Racket. Technical Report. Technical ReportPLT-TR-2010-1, PLT Inc., 2010. http://racket-lang. org/tr1.

[5] Patrick Gebhard. 2005. ALMA: a layered model of affect. In Proceedings of thefourth international joint conference on Autonomous agents and multiagent systems.ACM, 29–36.

[6] Patrick Gebhard, Tanja Schneeberger, Tobias Baur, and Elisabeth André. 2018.MARSSI: Model of Appraisal, Regulation, and Social Signal Interpretation. InProceedings of the 17th International Conference on Autonomous Agents and Multi-Agent Systems. International Foundation for Autonomous Agents and MultiagentSystems, 497–506.

[7] Jonathan Gratch. 2000. Emile: Marshalling passions in training and education. InProceedings of the fourth international conference on Autonomous agents. ACM,325–332.

[8] Marsha Linehan. 2014. DBT Skills Training Manual. Guilford Publications.[9] Robert R McCrae and Paul T Costa. 2003. Personality in adulthood: A five-factor

theory perspective. Guilford Press.[10] Albert Mehrabian. 1978. Measures of individual differences in temperament.

Educational and Psychological Measurement 38, 4 (1978), 1105–1117.[11] Albert Mehrabian. 1996. Pleasure-arousal-dominance: A general framework

for describing and measuring individual differences in temperament. CurrentPsychology 14, 4 (1996), 261–292.

[12] Roland Neumann, Beate Seibt, and Fritz Strack. 2001. The influence of moodon the intensity of emotional responses: Disentangling feeling and knowing.Cognition & Emotion 15, 6 (2001), 725–747.

[13] Kenneth R Olson and Dale A Weber. 2004. Relations between big five traits andfundamental motives. Psychological Reports 95, 3 (2004), 795–802.

[14] Steven Reiss. 2004. Multifaceted nature of intrinsic motivation: The theory of 16basic desires. Review of general psychology 8, 3 (2004), 179.

[15] Samuel S Sohn, Xun Zhang, Fernando Geraci, and Mubbasir Kapadia. 2018. AnEmotionally Aware Embodied Conversational Agent. In Proceedings of the 17thInternational Conference on Autonomous Agents and MultiAgent Systems. Interna-tional Foundation for Autonomous Agents and Multiagent Systems, 2250–2252.

Extended Abstract AAMAS 2019, May 13-17, 2019, Montréal, Canada

2197