Perceptual-Cognitive Assessments in Football

Abstract

Introduction: Assessments with varying levels of perceptual information or action fidelity are commonly used in the detection and identification of talent in football. Common performance assessments can range from either highly sport-specific environments with players being immersed in a realistic environment and interacting with a football (i.e. domain-specific, high ecological validity), to players sitting in front of a computer responding to various shapes and colours with no sport-specific information presented (i.e. domain-generic, low ecological validity). Many testing batteries measure athletes with a multitude of different tests that are placed along various points on ecological validity continuum. However, very few of these assessments are sufficiently validated. For the assessments that attempt to closely replicate the perception-action coupling demands experienced in football game play, there are many conditions that must be met before it can be used in future research and practice. On the other side of the spectrum, it remains contentious whether using assessments that intentionally remove ecological validity from their environments has merit. These non-sport specific assessments attempt to measure the general cognitive abilities of athletes, and many researchers have advocated their usefulness in talent identification programs. Therefore, the collection of aims within this dissertation was three-fold: i) to investigate both the domain-generic and domain-specific perceptual-cognitive abilities of all athletes (i.e. academy to senior players) in order to understand what perceptual-cognitive abilities athletes exhibit, and what factors (i.e. environment and heritable) contributed towards their cognitive profile, ii) to track both domain-specific and domain-generic abilities longitudinally in order to understand their relationships with increased exposure to football training, and iii) to learn from the limitations of the domain-specific skills assessment and incorporate new technologies in order to gain a further insight to investigate how emerging technologies could help to develop more representative assessments. Methods: To understand the between-group differences of domain-generic and domain-specific abilities across the youth developmental period of athletes, a variety of independent studies were undertaken. First, 343 male players (age: 10.34 – 34.72 years; playing experience: 5 – 22 years) from the U12-Senior age groups of a professional German football club were recruited. Age, experience and playing position were recorded to examine which factors contributed more to the development of domain-generic abilities (Chapter 3). Players participated in four generic cognitive tasks aimed at measuring higher-level cognitive functioning: a precued choice response-time task, a stop-signal reaction-time task, a sustained attention task, and a multiple-object tracking task. Second, a new football-specific skills test was used to measure the domain-specific abilities of the athletes throughout adolescence, and the reliability and age-discriminant validity of this new domain-specific skills test was investigated (Chapter 4). Third, 304 players from the same cohort as Chapter 3 had their data analysed longitudinally to track the longitudinal development of both domain-generic (assessments from Chapter 3) and domain-specific (assessment from Chapter 4) abilities across three seasons (Chapter 5). Lastly, the final investigation of the dissertation was divided in two parts to explore how to develop more representative task designs within the football specific skills assessment used in the previous chapters. Accordingly, Chapter 6a) 85 amateur male participants (19.5 ± 5.4 years old; 13.1 ± 6.0 years playing football) completed two sessions in the skills assessment task under two different visual conditions: stroboscopic and full vision Participants were subdivided into skilled (S: top 50%) and less-skilled (LS: bottom 50%) groups using their point score from the full vision condition. Chapter 6b) Exploratory head movements of fourteen U13 and thirteen U23 high-level football players were recorded with a head worn inertial sensor in the skills assessment task, from which the count, frequency and excursion of head movements were extracted before and during ball possession investigate whether visual exploratory action is associated with passing performance. Results: Chapter 3 first demonstrated that a negatively accelerated curve generally best described the relationship between age, experience and domain-generic abilities. Age and experience only explained a very low to moderate proportion of the variance in EFs (marginal explained variance ranged between 2 and 57%). Furthermore, although Chapter 4 revealed that the new domain-specific skills test yielded acceptable test-retest reliability for the correct number of passes to a target (CV = 7.5-11.1; r = 0.48; p<0.001) and the speed at which they completed each trial (CV = 2.6-5.1; r = 0.70; p<0.001), the assessment was not able to differentiate between athletes over the age of 15. This plateau in both the developmental trajectories of domain-generic (Chapter 3) and domain-specific (Chapter 4) abilities was confirmed in the longitudinal study (Chapter 5), revealing that a performance plateau was apparent for domain-specific abilities during adolescence (i.e. 15 years old), whereas domain-generic abilities improved into young adulthood (i.e. 21 years old). Consequently, a further investigation into more representative task design had merit, where Chapter 6a) reported that restricting athletes’ visual feedback in the football skills assessment impacted time of completion per trial to in both S and LS groups equally (S: 0.21s; LS: 0.18s; p=0.543), but S athletes’ accuracy (S: 11.7%; LS: 0.4%; p<0.001) were significantly more affected compared to full vision conditions. Lastly, Chapter 6b) reported that the variables that best explained faster performance were a higher number of head turns before receiving the ball, and a lower number of head turns when in possession of the ball, which older athletes perform better than younger athletes. Discussion/conclusion: Overall, the investigation into domain-generic assessments across Chapter 3 and 5 found that athletes improve their performance during late childhood until reaching adolescent (i.e. average age of 15 years old) and was independent of how many years of experience playing football or which position they played on the field. As the developmental trajectories of high-level football players’ domain-generic abilities reflected those observed in general populations’ despite long-term exposure to football-specific training and gameplay, this questions the relationship between high-level experience’s capacity to improve domain-generic abilities and challenges the validity of including non-sport specific assessments as a measure of football performance potential in high performing athletes. Lastly, despite the best efforts to use highly technical assessments to measure football skills in Chapter 4 and 5, the assessments may have under-represented the perceptual or action components necessary to allow athletes to demonstrate their expertise. Thus, more studies that aim to improve on the task designs of assessment tools has merit, and future studies could build off the foundations from the studies within Chapter 6 [i.e. stroboscopic glasses (6a) and head movement sensors (6b)] as methods to expand on the representativeness of assessment tasks.