Enhancing Learning with Gamified Instruction: Overview of PING System and Gridlock

Table of Links

Abstract and I. Introduction

II. Overview of PING System and Gridlock

III. Design Insights and Educational Impact

IV. Conclusion and References

II. OVERVIEW OF PING SYSTEM AND GRIDLOCK

A. Intelligent Tutoring in the PING System

This section provides a non-technical overview of the PING system and, more specifically, the ITS we integrate for personalized learning. For more technical information, we ask that readers refer to our other recent publication [11, 12]. The PING system is designed as a modular intelligent tutoring system (ITS) that can be integrated with any serious game (SG). To make the ITS modular, the system developer first partitions the game’s educational domain into certain subject-specific sections, referred to here as content blocks. Within each content block, the student can then be tested and supported on a specific topic. For example, a game dealing with math might have a block for multiplication and a block for division.

Then, as the student interacts with game components and completes in-game content, the system receives a constant feed of that student’s performance, as well as any relevant data. For example, Gridlock collects timing information on how long students took to complete sections. In addition, Gridlock also records a score on each section, various data on mouse movement and keyboard interaction, as well as estimated emotional response through facial emotion recognition and webcam images. All of this data is collected through quizzes and mini-games, which are covered in the next section.

Then, the data are fed into a set of reinforcement learning agents, each of which is trained to support a student on a specific content block [12]. With reinforcement learning, the system learns optimal behavior by exploring choices, so the system needs some student interaction before providing acceptable student support. But after initial training, the reinforcement learning back-end allows the system to adapt to trends and changes in student behavior without any human intervention.

To learn, reinforcement learning requires feedback about the quality of the decisions made. As such, student performance is used as a “reward” for the agent. If a student performs better after receiving assistance from the system, the system receives a positive reward, encouraging the chosen assistance for future students. Likewise, if a student performs poorly after receiving some assistance, the specific assistance chosen will be less likely to appear to similar students in the future.

B. Gridlock

Gridlock is a game that has been iteratively redesigned and improved over the past several years. Through in-classroom testing, pilot evaluations, and student feedback, numerous changes have been made to improve both the usability of the game and to make the game more enjoyable and educational for students. The game places students in the position to redesign the logic controller for a traffic light, which is a typical lab assignment for courses that teach digital logic design. When the game starts, students are quizzed on their entry knowledge to establish a benchmark for future assistance. Then, students explore a 3D environment and unlock a door by completing tasks specific to the content the game intends to teach.

The tasks given to students vary between concepts. For example, one block focuses on the specific design specifications students need for the traffic light problem. In this section, students must complete a miniature design problem which tests their ability to set the order of the traffic light changes. This problem is structured more as a puzzle or a minigame, and students’ number of actions, time taken, and correctness of their submitted solution all give the system insight into the student’s knowledge of this section’s content. Then, the system can give the student feedback to bolster their content knowledge. As shown in Figure 1, students are given dialogue from the game, videos, images, and other relevant materials based on their performance.

Once students have mastered the concepts presented by the game, they are tasked with programming their final traffic light logic and submitting it into the game. To help minimize instructor time used by this assignment, the game has full functionality to let students program and debug and simulate their solutions, all within the game, as shown in Figure 2 and Figure 3. With this approach, students have a seamless experience, and the game can observe student performance as they write their solutions.