SkillCraft: Information Access Strategies


Attention is a key part of human cognition, and as such, attention to our environment allows us to sort out what information is important and what is not.  By sorting this information into the appropriate categories we are able to become more effective at learning and decision-making within similar environments. Over the past few years video games have gained popularity as a spectator event with eSports (electronic sports) attracting numerous professional gamers and a constantly expanding global community. Yet, despite the increasing popularity and easily obtainable data, little research has been done to explore how experts and novices differ in their interaction within a complex Graphical User Interface (GUI) environment.

In natural environments, people use eye-movements to access information necessary for coordinating their motor-movements. Studies have shown that eye-gaze immediately precedes grasp (Hayhoe & Ballard, 2005); when you are making a peanut butter sandwich for instance, you fixate the knife before you grab it, the peanut butter before you get some on the knife, and the bread in the exact spot on which you will begin to spread.

Theorists typically ask the following sorts of questions; why is one person better than another at certain tasks? What does a professional do that a novice does not? What sets an expert apart from the rest of his colleagues? Most generally, expertise researchers are interested in examining how people learn and utilize acquired skills. Our goal is to understand how people learn what information is relevant to their task, and how they come to access this information efficiently.

So, how does expertise affect these information access strategies? A study of professional athletes’ demonstrated the idea that experts use a “just-in-time” strategy, where information that is required to do well in their respective sports is accessed at the time it necessary. Such strategies also appear to be employed in everyday tasks such as making tea or a peanut butter sandwich (Hayhoe & Ballard, 2005). It seems that experts merely use the same strategies in their specific tasks to perform better. Through the acquisition of a “just-in-time” strategy experts are able to look at what they needed to, when they needed to, faster than novices (Land & McLeod, 2000).

In the same vein, studies which compared expert surgeons to novices showed unique gaze patterns and speed differences between the two groups (Law et al, 2004). Three different gaze patterns were found in a virtual surgery experiment in which surgeons focused their attention on their target, switched their attention between the target and their surgical instrument or followed their surgical instrument with their eyes. Experts tended to focus on their targets instead of on extraneous information, a finding which is replicated in studies with aerial pilots (Schriver et al., 2008).  Pilots have also shown that experts make more necessary actions and focus on important information for the task rather than unimportant information. Their expertise had reinforced specific patterns of attention allowing them to avoid extraneous information (Schriver et al., 2008).

StarCraft 2 (SC2) is a Real-Time Strategy (RTS) video game that is played at extraordinary speeds. All matches start with each player choosing one of three species and controlling one base and 6 worker units. Each of the three species control different buildings and units with unique abilities and upgrades, creating a complex interaction within the games environment. Basically, the three species units play out a dynamic game of rock-paper-scissors. Players must command their units, prepare defences, build base expansions to gather more resources while build bigger armies to outmanoeuvre their enemies in what amounts to a high speed game of chess.

The SC2 GUI contains a top-down view over a maps terrain, where each player has their vision limited to only his or her own units and buildings. Incorporating this into the GUI creates a level of uncertainty for players regarding their opponents strategy and position, forcing them to perform regular reconnaissance, which is referred to as scouting. Top players issue hundreds of actions per minute navigating around the terrain at prodigious speeds and maintaining their forces. Optimized attention to the environment allows for proper navigation and the coordination of global and local information for real-time decision making. It is these differences of attention and actions that mark the gap between an amateur and a pro-gamer.

It appeared to us that RTS games provide a unique opportunity to better understand the cognitive processing involved in dynamic real-time resource management scenarios. Current interfaces for emergency management information systems (including both those in use and those under development) are similar to the  StarCraft 2 GUI. One key problem with expertise research is the availability of an adequate number of expert and novice subjects. Theorists often rely on a small handful of experts for their studies. We hope to circumvent this problem by acquiring experts from the Internet. Our design will allow us to recruit participants via the Internet, and we can examine their replay-files without bringing hard-to-reach experts into the lab.

The SkillCraft research project focuses on how experts and novices differ on such variables as actions-per-minute, on-screen camera movement patterns, players interactions with the interface, how players establish and use shortcuts to maximize their efficiency, and how players obtain the information necessary to inform their future decisions. Our data are ideal for confirming the presence of attentional strategies, such as just-in-time allocation, suggested in other research. Further, for the first time in any study, we will have sufficient data across a wide span of expertise to pinpoint where on this continuum these strategies emerge. We will also be able to detect interactions between measures, such as if efficient attentional allocation requires a certain level of manual keyboard efficiency to first develop.

This study will be the first very large scale research on expertise using modern video games and we are confident that the scientific study of StarCraft 2 expertise will resonate within the communities of players. With their support we hope to be able to find out just what makes experts excel, and establish how someone might be able to learn these same skills on their own. We plan to publicize our findings through normal scientific conferences and journals, as well, our results will be available online for the amazing community that StarCraft 2 players have created.

To get more information about this project, please feel free to contact with any questions you may have or if you are interested in collaborating with us.

You can visit the project site Here.



Hayhoe, M. & Ballard, D.H. (2005). Eye movements in natural behavior. Trends in Cognitive Sciences, 9, 188-193.

Land, M. F., & McLeod, P. (2000). From eye movements to actions: how batsmen hit the ball. Nature neuroscience, 3(12), 1340-5. doi: 10.1038/81887.

Law, B., Atkins, M. S., Kirkpatrick, A. E., Lomax, A. J., & Mackenzie C. L. (2004). Eye gaze patterns differentiate novice and experts in a virtual laparoscopic surgery training environment. Proceedings of the Eye tracking research & applications symposium on Eye tracking research & applications – ETRAʼ2004, 1(212), 41-48. New York, New York, USA: ACM Press. Doi: 10.1145/968363.968370.

Schriver, A. T., Morrow, D. G., Wickens, C. D., & Talleur, D. A. (2008). Expertise Differences in Attentional Strategies Related to Pilot Decision Making. Human Factors: The Journal of the Human Factors and Ergonomics Society, 50(6), 864-878. doi:  10.1518/001872008X374974.