ActAffAct

IRIS Wiki - IS Systems - ActAffAct

Availability

Documentation and the Java-based ActAffAct system itself are available online.

Technical Description

The project investigates to which extent equipping characters with a simulation of Ortony's (2003) derivation of control principles from the OCC appraisal theory of emotion (Ortony et al., 1998) is sufficient to achieve dramatic structures while interacting in a conflict prone environment.

On the one hand, the system provides an empirical demonstration of what courses of action can be engendered by the characters interactions with each other and their world --- in particular, it is also made apparent how certain timings, re-occurring sequences, and other dynamic qualities are implicitly (co-)determined by exogenous factors such as the topology of the scene and the durations of physical acts. Further, domain-independent mechanisms could be abstracted from the specific implementation. Both the mechanisms themselves and the insights gained in the process of their definition, form important knowledge regarding the (im)possibilities and costs of reuse of architectures and architectural components across different scenarios.

At the same time, the importance of deliberately omitted constructs that could exert further influences on behaviour is made evident by the observable deficiencies of generated behaviour that can be clearly diagnosed. In this way, the basis for a principled incremental method for the construction of control architectures of virtual characters is provided, where components are included based on demonstrated needs. As such, traditional issues of overdesign can be avoided, such as equipping characters with overly sophisticated cognitive capabilities that do not match their sensori-motor equipment or the actual complexity of the setting or integrating components from different theories with overlapping (redundant or contradictory) functionalities that cause haphazard behaviour. Conversely, the actual impact and relevance of specific cognitive faculties for this specific class of scenarios (i.e., characters' capabilities + environmental properties + the overall aim of staging a dramatic performance) is made clear.

Result Description (end user perspective)

The system is not geared towards users as an audience. Rather, the intended target user is the researcher employing the tool to test the results generated with different initial conditions and parameters, as well as inspecting the state of the simulated world at different points in time.

Another potential class of target users could be given with the larger community of non-technical authors of storytelling systems. Interacting and experimenting with this system may prove a useful device to improve their understanding of the capabilities, limitations, and the overall particular "shape" of the properties of generative technology. In particular, it may be used to demonstrate higher-level control in technical interactive systems (i.e., not direct control over each single movement and action, but directing of unfolding of dramatic courses of action via setting of motivations and sensitivities, and definition of physical capabilities that may be exploited for different purposes by the virtual characters, also in an autonomous fashion).

Strong Points

The main result of the system is the possibility of some level of generativity based solely on the conflicts between the characters in the set-up of a play, the possibilities offered by the physical environment, and the simulated emotions and skills involved in resolving them. Affective reactions can then be seen as the constituents of a dramatic structure. The system offers detailed inspection access to the mechanisms operating "under the hood", as well as some limited configurability at run-time.

Limitations

Since the system is intended neither as an authoring tool nor as a presentation system, the options to interact and alter behaviour are limited and quickly require direct modification of the programme base.

Main Publications

Rank S. (2004). Affective Acting: An Appraisal-based Architecture for Agents as Actors. Institut für Medizinische Kybernetik und Artificial Intelligence, Universität Wien, Diplomarbeit (M.S.Thesis).

Rank S. (2005). Toward Reusable Roleplayers Using an Appraisal-based Architecture. In Payr S. (ed.): Educational Agents and (e-)Learning, Applied Artificial Intelligence 19(3-4):313-340.

Rank S., Petta P. (2005). Appraisal for a Character-based Story-World. In Panayiotopoulos T. et al. (eds.), Intelligent Virtual Agents, 5th International Working Conference, IVA 2005, Kos, Greece, September 2005, Proceedings, LNAI 3661, Springer Berlin Heidelberg, pp.495-496.

Rank S., Petta P. (2005). Motivating Dramatic Interactions, in Agents that Want and Like: Motivational and Emotional Roots of Cognition and Action, AISB, The Society for the Study of Artificial Intelligence and the Simulation of Behaviour, University of Sussex, Falmer, Brighton, UK, EU, pp.102-107.

Rank S., Petta P. (2007). From ActAffAct to BehBehBeh: Increasing Affective Detail in a Story-World, in Cavazza M. and Donikian S.(eds.), Virtual Storytelling: Fourth International Conference (ICVS 2007), St.Malo, France, EU, December, 2007. Proceedings, LNCS 4871, Springer Berlin Heidelberg, pp.206-209.

Supporting Narrative Theories

ActAffAct is based mainly on Campbell1968/Vogler1996 (monomyth) plus Egri, and Forster (motivated causality as essential for plot).

Computational Model

A pragmatic integration of Ortony's (2003) model derived from the original OCC model into a BDI (belief-desire-intention) type plan-based practical reasoning agent framework, JAM (Huber 1999).

References

Huber M.J. (1999) JAM: a BDI-theoretic mobile agent architecture, in Etzioni O., et al.(eds.), Proceedings of the third annual conference on Autonomous Agents (Agents'99), Seattle, WA, USA, May 1-5, ACM Press, New York, NY, USA, pp.236-243.

Ortony A., Clore G.L., Collins A.: The Cognitive Structure of Emotions, Cambridge University Press, Cambridge, UK, 1988.

Ortony A. (2003) On Making Believable Emotional Agents Believable, in Trappl R., et al.(eds.), Emotions in Humans and Artifacts, MIT Press Cambridge, Massachusetts London, England, pp.189-212.