Defined four overarching research questions, which will motivate my research over the next 3+ years:
What are the cognitive correlates of playing FPS? We will assess various aspects of cognition such as processing speed, task mixing, task switching and statistical learning.
How does cognitive performance differ with FPS expertise? We will measure expertise in CS: GO by interviewing CS: GO players and conducting questionnaires with suggested measures such as total hours of playtime, weekly hours of playtime, self-rated expertise, and ranking. We will then compare cognitive performance in different expertise groups – with the expectation that more expert players will show greater cognitive performance.
What cognitive and in-game skills are important for success/expertise in CS: GO? We aim to interview professional CS: GO players, measure their cognitive skills, and observe their playing to explore the possible methods of measuring, predicting and nurturing cognition and in-game skill. Gaining an understanding of the mechanisms which are important for expertise in CS: GO may aid our understanding of how playing FPS play enhances cognition.
Can FPS training improve cognition? We aim to train novice FPS players in CS: GO for an extended period, assessing their cognitive performance and in-game performance development. These results will be compared to their baseline levels and perhaps a cognitive control sample.
NB: These data were collected during my MSc in CNHN at the University of Sheffield however, during the PhD I used advanced statistical analysis to further explore this data set before collecting data for my second study – where I will apply similar statistical analysis.
Data from a large cross-sectional study with N = 273 CS: GO players were re-analysed. Participants completed an online colour-shape task to measure processing speed, task mixing and switching ability, followed by a questionnaire to measure CS: GO expertise and proficiency. K-means cluster analysis was used to group participants into expertise clusters based on three expertise variables which were weighted equally: total hours of playtime, weekly hours of playtime and self-rated expertise. A consensus-based algorithm plot determined that four groups would best fit our data. The four groups were named Casual, Experienced, Aspiring and Semi/Professionals. As expected, most of the participants (78%) were placed in the lower two expertise groups, with a small number of participants reaching the Aspiring and Semi/Professional levels of expertise.
An ANOVA showed a significant effect of group on processing speed. On average, Semi/Professionals were 85ms faster than Casual players, and Experienced players were 48ms faster than Casual players, with pairwise t-tests revealing these differences to be significant. However, we found no effects of group on task mixing performance. Although non-significant, we observed a U-shaped relationship between task switching and expertise, with Experienced players showing the best and Semi/Professionals the worst switching performance, with a difference of over 100ms on average between these two groups.
Results suggest that CS: GO players with greater expertise possess faster-processing speed. This supports previous literature and suggests that FPS play may improve processing speed performance however, correlation does not imply causality. Mixing and switching performance were similar across different levels of CS: GO expertise, and whilst the U-shaped relationship between CS: GO expertise and switching performance was non-significant, it may suggest that with expertise gains there are strategic shifts in attention allocation. Perhaps highly expert players have reduced task mixing and switching ability because they use a more focused attentional strategy compared to less expert players who employ a more divided attentional strategy. Further research is required to explore this relationship.
Drift-Diffusion Modelling (DDM), which decomposes the observed RTs and accuracy scores into latent decision-making processes such as drift rate, is ongoing. Using an EZ-DDM the values have been computed; however, further learning is required to understand the meaning of these new values. My aim is to present them at the Confirmation Review Meeting. Furthermore, we aim to analyse learning rates by evaluating accuracy rates across blocks, with increases in accuracy expected to occur earlier in more expert players compared to less expert players. Once the analysis is completed, we aim to publish these results.
NB: Data collection is complete and preliminary analysis has begun.
A large sample of N = 250+ CS: GO players completed four online cognitive tasks, followed by a revised version of the CS: GO expertise questionnaire from Study 1. To measure processing speed, a choice RT task was used, in which participants had to quickly decide whether a shape was green or blue. Probabilistic inference was measured with two tasks. First, in a simple RT task at the beginning of the experiment, participants pressed a key as soon as a stimulus appeared on the screen. Unbeknownst to participants, this task showed the stimuli in pre-determined sequences. In the final task, participants were presented with triplets of stimuli, some of which matched the sequences of stimuli in the simple RT task, whilst others were foil triplet sequences. Accuracy in the detection of target and foil stimuli will be used as a measure of probabilistic inference.
We have currently obtained a sample of N = 590. However, some individuals completed the study several times and, therefore, the data requires extensive cleaning before any data analysis can be conducted. I plan to present preliminary findings from the processing speed task at the confirmation review meeting. We have various plans for data analysis such as k-means cluster analysis on expertise measures, DDM on RT distributions and accuracy data and learning rate speed and accuracy analysis on the statistical learning data.
Study 3: to answer the question ‘what cognitive and in-game skills are important for success/expertise in CS: GO?’, we aim to interview professional eSports CS: GO players to gain rich expert qualitative data regarding the in-game skills they deem important for success in their field. This will build upon interviews I have previously conducted during my MSc. Additionally, observation of their gameplay in comparison with novice CS: GO players will reveal important aspects of in-game skills in real-time, which do not otherwise present themselves in an interview setting. Furthermore, I plan to evaluate professional players’ cognitive performance on various tasks in a lab setting to determine the specific cognitive domains they particularly excel in compared to novice players.
Study 4: to answer the question ‘can FPS training improve cognition?’, we hope to conduct an intervention study in a large sample of non-FPS and non-expert video gamers. Ideally, a large sample of N = 100+ participants would be training at home for a period of 100+ hours, which poses challenges to the feasibility of this project. For example, at-home training requires close training monitoring, possibly via Steam, the platform CS: GO is played on.
Preview of my Study 1 Cluster Grouping
2021 - 2022
Improving my coding and analysis methods.
Presenting at my first international conference - Learning and Plasticity (LaP) in Finland.
Passing my Confirmation Review (with minor corrections).
Starting my job as a Research Coordinator on the ORA Project.
Working as a GTA on a project which made Tatool-web more accessible for novice users.
Mentoring my first Master's student.
Understanding and executing the script for the Drift-Diffusion Model (DDM).
Balancing my time between my PhD, various roles and moving house.
Achieved a Distinction on my MSc
Moved into our long-awaited first home.
Women in Cognitive Science (WiCSW)
Women in Games Careers and Networking Expo
Learning and Plasticity (LaP)
Recently, I completed the fast-DM modelling on the reaction time distribution data from Study 1. I presented these findings at Psychonomics and received helpful feedback which I will implement before writing up.
The data requires extensive cleaning before any data analysis can be conducted. We have various plans for data analysis such as k-means cluster analysis on expertise measures, DDM on RT distributions and accuracy data and learning rate speed and accuracy analysis on the statistical learning data. We may refine the questionnaire and task battery before implementing for Study 3.
Boston, MA, USA