Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Computational models of motivation for game-playing agents
Merrick K., Springer International Publishing, New York, NY, 2016. 213 pp. Type: Book (978-3-319334-57-8)
Date Reviewed: Jun 22 2017

One of the major dimensions in which current computer models of players in a gaming system differ from actual human players is affective aspects. Creating a computational model of motivation that can be implemented in such computer players blurs this distinction significantly. This book takes us through the various aspects of this challenge, and addresses one class of such models in detail. The applications of this next version of simulated players go well beyond games, though. It can significantly enhance the power and attractiveness of user interfaces in general. Let us open the book.

The preface itself advises that we are on a new thread of the currently hot field of artificial intelligence (AI), addressing computational motivation. It also sets the context for discussion as designing self-motivated game-playing agents. Later in the book, the same idea is used on non-playing characters (NPCs) to build a richer diversity in their behavior.

The book is divided into four parts. Part 1 sets the base for the book. Chapter 1 reviews relevant literature on motivation, identifying the key ingredients, and different classifications. It elaborates three types of incentive-based motivation--achievement, affiliation, and power--which define the scope for the book. Chapter 2 elaborates on these motivation types, attempting to provide formal computational models for them using approach-avoidance theory. Then, these are linked to goal selection strategies, specifically winner take all and probabilistic goal selection. Chapter 3 creates the framework for embedding such a motivation profile into a computer model, focusing on NPCs. Four agent architectures--rule-based agents, crowds, learning agents, and evolutionary algorithms--are considered. With that, we move to Part 2, where we start practical game scenarios.

Chapter 4 is on achievement motivation, and how to embed it in actual games using different architectures. Three games are used to study the outcome: ring toss, roulette, and the prisoner’s dilemma. The resulting implementations are compared to humans empirically. Using roulette and prisoner’s dilemma, chapter 5 creates a generic motivation profile and demonstrates its portability across games. All three profiles--achievement, power, and affiliation--are discussed, and compared with human behavior.

The next three chapters, 6, 7, and 8, form Part 3 of the book, and look at specific game scenarios for NPCs and discuss models and implementations of different motivation profiles. Chapter 6 is on enemies, 7 on pets/partners, and 8 on support characters. Many different games are used for the studies reported: paratrooper, prisoner’s dilemma, battle of the sexes, breadcrumbs, and so on. Different agent architectures are used as appropriate to the scenario. Detailed studies and analysis are provided, comparing performance and the behavioral diversity that this can introduce in games.

One challenge for such models is that the human player may be able to predict behavior after a while, unless the agents evolve over time, just as humans do. Chapter 9 of Part 4 discusses this aspect of the evolution of motivated agents. Chapter 10 concludes the book, reviewing what has been discussed and looking to the future.

Motivation is not only about prioritizing goals, but also about generating goals. The studies in the book assume a goal list is available, restrict attention to the prioritization of goals, and point out generation as the emerging challenge. The last chapter also provides a good perspective of what has been discussed in the book, against the overall scope in this area--one of the positive features of the book.

Another nice thing about the book is that it continually makes an attempt to connect the pieces, leaving the reader comfortable to navigate through the book. The author also attempts to provide a brief overview of the various games referred to in the book, the theories used, and so on, at the point of first use. This also helps the reader feel comfortable. Pointers to special and advanced topics are given with a brief overview at many places. Parts of the book are a bit too detailed and formal. Overall, this is an interesting book on an interesting topic. It would be of interest to many in the field of AI, particularly those in multiagent systems and game playing.

Reviewer:  M Sasikumar Review #: CR145367 (1709-0597)
Bookmark and Share
  Featured Reviewer  
 
Multiagent Systems (I.2.11 ... )
 
 
Games (I.2.1 ... )
 
 
Applications And Expert Systems (I.2.1 )
 
 
Learning (I.2.6 )
 
Would you recommend this review?
yes
no
Other reviews under "Multiagent Systems": Date
Engineering intelligent hybrid multi-agent systems
Khosla R., Dillon T. (ed), Kluwer Academic Publishers, Norwell, MA, 1998. Type: Book (9780792399827)
Aug 1 1998
Linguistic geometry: from search to construction
Stilman B., Kluwer Academic Publishers, Norwell, MA, 2000.  395, Type: Book (9780792377382)
Jan 1 2001
 Transactional agents: towards a robust multi-agent system
Nagi K., Springer-Verlag New York, Inc., New York, NY, 2002.  205, Type: Book (9783540430469)
Apr 13 2004
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy