In this section, we consider, in abstraction, the interactions between a video game environment and a game player in subsection 4.2.1. We conceptually examine how a player makes decisions in games and how game responses in subsection 4.2.2 and subsection 4.2.3, respectively. The relationship between the decisions and responses establishes a foundation of our player model framework.
4.2.1 Game States and Interaction Flows
We design our player model based on the interaction model between a video game player and a video game environment. We consider the video game environment as a set of game states being processed interactively by a three-action loop. The three fundamental actions in the loop are:
1. Input: A video game receives commands from a game player to manipulate game states via game inputs. A game input consists of acommandand optional command arguments. The command argument identifies how to perform or the degree to operate the given command.
2. Update: The video game then updates its states according to either the external inputs from the player or its internal mechanisms from the game rules. The modification of a game state can be done through a decision-making system in the game, or a pseudo-random number generator, or the mixture of both.
We use a pseudo-random number generator to construct a set of random numbers that is controllable to a certain degree via a random seed. These pseudo-random numbers retain the indeterministic behaviors when replaying the games.
Making decisions in games is a large subject of both conventional practices and on-going researches. Besides the random decisions generated from pseudo-random numbers, an overview of other decision-making techniques in video games are briefly presented in subsection 2.2.1.
3. Output: A video game responds to its game player with updated game states through game outputs. For a typical video game, the termdisplay can be used interchangeably due to the visual nature of the media.
As for the game player’s side, the same procedures are followed in the interactive manner. That means the player receives responses through game outputs, makes or updates a decision, converts the decision into game command(s), then send the command(s) back as game input.
Decision Making System interface
Video game environment Update
new states commands game data
Figure 4.1: A model of interactions between a game environment and a game player.
The entire loop of the interactions between a video game player and a video game environment is presented in Fig. 4.1. This loop also represents the flow of game states that connects the two sides of decision makings. The loop iteratively and interactively continues until the game ends.
4.2.2 Hierarchy of Game Decisions
In a video game playing, we model a player’s decisions in a top-down approach, from a conceptual idea down to concrete game commands. This hierarchy of the player’s decisions consists of:
1. Decision: According to current game states, a target decision is initiated from a strategic plan to achieve the game objectives.
2. Action Plan: A series of actions to accomplish the target decision is then developed. For other games, besides turn-based games, the plan also includes a timing to execute each action sequentially.
3. Game Commands and Arguments: Game commands, along with their corresponding command arguments, are constructed to control the game states according to each action plan.
Once game commands are realized, a player inputs them to the game environment at a particular time. (This is also where the challenges of action games reside.) This triggers a new loop of interaction flows as described in previous subsection.
4.2.3 Levels of Game Information
After updating game states according to the player’s inputs and the game mechanics, a video game sends out game data that can be classified intogame states and game information. Game information is the direct response from the game sent to its player, in correspondence to the player’s decision hierarchy described in the previous subsection.
We classified game information with a bottom-up approach, from command feed-back up to decision calls. The levels of game information, therefore, consist of:
1. Command Feedback is a quick validation made to game command inputs or their arguments. It includes, but not limited to, a notice informing a newly given command is invalid, a hint showing additional arguments are required, or a signal warning the time is almost over for the next command. It is a low-level, command-related game information.
2. Action Resultis when the game states are updated in response to an issued action command. This indicates a progress towards or a recess away from a player’s target decision. It is a middle-level, action-related game information.
The player should assess the action result and determine the next command to issue appropriately. Depending on this information, the subsequent commands are either a follow-up in the action plan or a new action due to unexpected changes of game states.
3. Decision Call is a request for a new game decision which can be either a direct or an indirect call. It is a top-level, decision-oriented information.
One example of an indirect call is a critical warning . This call requires an immediate resolution from the player. Without a suitable reaction, the player’s game states may become worse, and the player may even lose the
game. The critical warning indirectly forces the player to re-evaluate his/her current action and decision.
Commands & Arguments Decision Making
Command Feedback Action Result Decision
Figure 4.2: Association between hierarchy of game decision and levels of game in-formation.
We illustrate the relationship between the hierarchical levels of game decisions and their corresponding levels of game information in Fig. 4.2. We use this relation-ship as the underlying mechanism for our game player model, which is explained in the following section.