Introduction What cognitive processes underlie experts` performance in their domain of expertise? This question has been addressed by many cognitive psychologists in various domains, such as chess (e.g., Gobet & Simon, 1996), or sport (e.g., Garland & Barry, 1991; Poplu, Baratgin, Mavromatis, & Ripoll, 2003). The template theory (Gobet & Simon, 1996), which unifies low-level knowledge, such as chunks, with high-level knowledge, such as schematic knowledge, provides a strong theoretical framework to account for expert behaviour. This theory proposes that, when perceiving a game pattern, experts players access chunks stored in long term memory (LTM) by filtering perceptual features through a discrimination net. The discriminated chunk will give access to semantic memory, allowing experts to understand the subtleties of the pattern and rapidly make appropriate decisions. Thus, perceptual discrimination of game patterns would be, in part, responsible for the quickness and the appropriateness of decisions that characterize expert behaviour. In sport psychology, expertise effects have been reported in experiments that used different kinds of game patterns stimuli (e.g., schematic drawing, 2-D video frame, 3-D pictures). The question that is addressed in this paper concerns the multiplicity of chunks stored in memory: Is there a chunk stored in LTM for each kind of stimuli? Alternatively, we assumed that chunks essentially encode common features about structural organization, rather than local information such as posture and physical characteristics of players. Method Thirteen expert soccer players volunteered to participate in the experiment. All of them were involved in competition at the national level in France (3rd Division). Stimuli were produced from digital photographs of real game situatins with 3-D reconstruction software. Two categories of pictures were made (realistic vs. schematic). For schematic pictures, 3-D crosses replaced players. We used repetition-priming paradigm. Priming effects were evaluated by comparing response times obtained for primed and unprimed stimuli in test phase. During the priming phase, 60 realistic pictures, called primed, were presented to each participant. Then, during the test phase, participants were confronted with 60 realistic and 60 schematic pictures. Each category included 30 primed and 30 unprimed (not seen during the priming phase) pictures. The participants were asked to make a quick and accurate decision concerning the next action to be done by the player holding the ball (i.e., shoot, pass, dribble). Results Participants` response time was longer for the realistic stimuli than for the schematic stimuli, [F (1,11) = 19.43, p < .05, Nano p 2 = .64]. The effect of the condition factor was also significant [F (1,11) = 20.41, p < .05, Nano p 2 = .65], but there was no interaction between kind of stimuli and condition [F (1,11) = 0.19, p > .05, Nano p 2 = .017] (see Figure 1). Discussion/Conclusion The discrimination process, which gives access to chunks stored in LTM, is not sensitive to modifications made between the priming and the test phases (i.e., when 3-D crosses were displayed instead of players). In other words, chunks do not encode local information, such as posture and physical characteristics. Instead, these results suggest that chunks encode commons features that define the structural organization of a game pattern.
© Copyright 2004 Book of Abstracts - 9th Annual Congress European College of Sport Science, July 3-6, 2004, Clermont-Ferrand, France. Alle Rechte vorbehalten.