Psychobiology and Behavioral Strategies
Visuomotor Expertise and Dimensional Complexity of Cerebral Cortical Activity TSUNG-MIN HUNG1, AMY J. HAUFLER2, LI-CHUAN LO2,4, GOTTFRIED MAYER-KRESS3, and BRADLEY D. HATFIELD2,4 Graduate Institute of Exercise and Sport Science, Taipei Physical Education College, Taipei, TAIWAN; 2Department of Kinesiology, University of Maryland, College Park, MD; 3Department of Kinesiology, The Pennsylvania State University, University Park, PA; and 4Neural and Cognitive Sciences Program, University of Maryland, College Park, MD
1
ABSTRACT
APPLIED SCIENCES
HUNG, T.-M., A. J. HAUFLER, L.-C. LO, G. MAYER-KRESS, and B. D. HATFIELD. Visuomotor Expertise and Dimensional Complexity of Cerebral Cortical Activity. Med. Sci. Sports Exerc., Vol. 40, No. 4, pp. 752–759, 2008. Purpose: This study employed the correlation dimension (D2) to examine whether visuomotor expertise was inversely related to the complexity of cerebral cortical activity. Method: Expert rifle shooters (N = 15) and novices (N = 21) completed 40 shots in the standing position during which the electroencephalogram (EEG) was recorded at 10 sites (F3, F4, C3, C4, T3, T4, P3, P4, O1, and O2) during a 5-s aiming period prior to trigger pull. D2 was derived for each trial and averaged across shots. A 2 2 5 (group cerebral hemisphere region) ANOVA was employed to contrast D2, while correlation analyses were used to determine the relationship between D2 and target shooting accuracy as well as variability of shot placement. Results: As predicted, experts exhibited lower D2 (5.02 T 0.16 vs 5.49 T 0.13, respectively) and greater accuracy of shot placement ((339.8 T 44.7 vs 90.7 T 38.9 points out of 400 possible, respectively). Experts also exhibited an inverse relationship between D2 and shooting accuracy, while, in contrast, novices revealed a positive relationship. Discussion: The results suggest that refinement and efficiency of cerebral cortical activity facilitates visuomotor performance. Lower complexity may be associated with less neuromotor ‘‘noise’’ in the brain, thus reducing interference with intended action. Key Words: BRAIN, EEG, PSYCHOMOTOR SKILL, NONLINEAR DYNAMICS
E
fficiency is a fundamental characteristic of expert motor performance in both the psychological and physiological domains. In this regard, Hatfield and Hillman (15) theorize that the human brain subscribes to the same principles as those of other biological systems when adapting to environmental challenges. For example, a fundamental characteristic of skeletal muscle is the minimization of energy expenditure when adapting to the constraints imposed by task and environmental demands. This process results in efficient movement (22). Similarly, efficiency in metabolic processes has been noted in elite distance runners who exhibit superior economy, or reduced O2 consumption (i.e., per unit of body mass), compared with less-accomplished performers during treadmill running at absolute
submaximal workloads (6). In regard to brain activity, fewer cerebral cortical resources are required for processing psychomotor task demands as expertise develops (20). The application of EEG to the study of expert motor behavior provides a high-resolution metric to capture the temporal dynamics of regional cerebral cortical activity associated with performance (5,16–18,21,22,24). Examination of EEG during the readiness period for various selfpaced visuomotor tasks like golf, archery, and target shooting has revealed that experts generally exhibit lower cerebral cortical activation when compared with novices (14). Specifically, Haufler et al. (18) observed higher alpha power and lower gamma power across the entire cerebral cortex (i.e., bilateral frontal, central, temporal, parietal, and occipital regions) in expert marksmen when compared with the results observed in novices. These findings are indicative of economical cortical activation in experts during the aiming period. Additionally, Kerick et al. (20) observed a progressive rise in cortical relaxation during 3 months of marksmanship training. The essential finding from a number of psychophysiological studies is that refinement and relaxation of cerebral cortical activity, particularly in nonessential regions, is highly related to skilled motor behavior (5,16–18,21,22,24). Furthermore, Deeny et al. (7) observed lower functional connectivity
Address for correspondence: Bradley D. Hatfield, Department of Kinesiology, Health and Human Performance Building, University of Maryland, College Park, MD 20742; E-mail:
[email protected]. Submitted for publication February 2007. Accepted for publication November 2007. 0195-9131/08/4004-0752/0 MEDICINE & SCIENCE IN SPORTS & EXERCISEÒ Copyright Ó 2008 by the American College of Sports Medicine DOI: 10.1249/MSS.0b013e318162c49d
752
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
EEG AND PSYCHOMOTOR PERFORMANCE
thinking when compared with convergent thinking. The former is a complex exercise involving the generation of many ideas on a given topic, a process that putatively requires a large number of synchronously oscillating cell assemblies (25). On the other hand, convergent thinking requires only one solution to a given problem, which logically would involve fewer cell assemblies (1,26,27). As such, these studies support that EEG dimension is a valid index of the complexity of cognitive operations. In light of the established notion that experts are characterized by automaticity (i.e., little reliance on explicit cognitive operations) (10), the nonlinear approach to the study of brain dynamics appears useful to elucidate the relationship between the level of functional complexity in the cerebral cortex and the quality of cognitive motor performance. The present study employed an expert–novice contrast (31), along with the application of nonlinear signalprocessing techniques (the correlation dimension metric (13)), to the EEG obtained during visuomotor performance. Specifically, we predicted that expert marksmen would show reduced complexity of cerebral cortical activity relative to novices and, thereby, maintain superior and more consistent psychomotor performance. In addition, the magnitude of difference between the groups was expected to be greater in the left relative to the right hemisphere, particularly in the temporal region, in light of previous reports of a positive relationship between EEG alpha synchrony (i.e., left temporal relaxation) and visuomotor expertise (14). Such a finding would support the notion of simplification or refinement of the involved neural processes to orchestrate skilled performance and provide a critical test for the psychomotor efficiency hypothesis (15). Finally, a negative relationship was expected between the dimensional complexity of brain activity and target shooting performance. That is, lower complexity would be associated with higher accuracy scores and reduced variability of performance. Furthermore, this relationship was expected to be more clearly established in the experts in light of the more established and stable taskrelated neuron–motor processes.
METHODS Participants. Thirteen male and two female marksmen ranging in age from 18 to 60 yr (mean age 26.5 T 11.1 yr), who held an average of 13.4 yr (SD = 10.6) of competitive experience at the national and international levels, were compared with 21 male novice volunteers ranging in age from 18 to 35 yr (mean age 23.1 T 5.5 yr). The groups were not different in age. The inclusion of more novice participants (N = 21) than experts (N = 15) was based on the expectation of greater variability in brain activity and behavioral performance in the novice group. Therefore, a larger group of novices was employed to achieve greater stability in the measures of interest for the purpose of group comparison. All expert group participants had at least 6 yr of national and international competitive shooting
Medicine & Science in Sports & Exercised
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
753
APPLIED SCIENCES
between the left temporal association (T3) and motor planning (Fz) regions of the cerebral cortex, as indexed by EEG coherence, in skilled marksmen with superior performance history when compared with those who performed less well under competition. This finding is important in that refined cortico–cortical communication between these regions would likely result in the reduction of nonessential input (neuromotor noise) to the motor region, resulting in less variability of neural activity, more stable efference to skeletal muscle, and greater consistency of performance. Hatfield and Hillman (15) refer to this notion as the psychomotor efficiency hypothesis. However, the EEG studies on brain activity and motor performance published to date have relied exclusively on linear signal-processing tools such as spectral analysis (14,18,20) and generation of event-related potentials (ERP) (8,28). Such linear analytical approaches extract information from the frequency and time domains of the EEG record, respectively, and provide metrics of regional activation (i.e., spectral analysis) as well as chronometric indices of cognitive- and movement-related processes (i.e., ERP). Unfortunately, the measures derived from linear analyses are limited to one dimension, while the working brain, in actuality, is characterized by multiple dimensions of complexity (9). According to Hebb (19), the cell assembly is the functional processing unit of the human brain, and, in this regard, higher mental processes (e.g., attention and motor planning) emerge from complex interactions of many cell assemblies or collections of neurons that cooperate as neural generators over time. Such processes would result in greater dimensional complexity. In accord with this view, the theory of dynamic systems posits that any time series recorded from the brain (e.g., EEG or magnetoencephalography (MEG)) reflects the collective activity of all currently active units or cell assemblies in the system (9). In this regard, Lutzenberg et al. (25) have demonstrated that the EEG is, in fact, derived from the oscillations of many cell assemblies. In this view, the resultant EEG record is the product of constituent components or ‘‘vectors’’ (ranging from one to many) that collectively determine the signal recorded at the scalp from moment to moment. To effectively determine the multiple components or ‘‘ingredients’’ of the signal, nonlinear measurement approaches such as dimensional analysis are critical. Therefore, the dimensional analysis of EEG can be conceptualized as an estimate of the number of active cell assemblies that produce the signal through their independent oscillations (26). Failure to estimate or extract these constituent influences precludes understanding of the underlying complexity of the EEG signal. Nonlinear analysis has been employed to address the association between EEG dimension and purely cognitive processes (27). For example, a study by Muller et al. (27) reveals that EEG dimension was positively related to the complexity of stimulus processing. In addition, Molle et al. (26) observed higher EEG dimension during divergent
TABLE 1. Mean and standard deviations of D2 and performance at 10 sites for experts and novices. Expert
APPLIED SCIENCES
Novice
F3
F4
C3
C4
T3
T4
P3
P4
O1
O2
Performance
5.11 (0.95) 5.68 (0.61)
4.81 (0.96) 5.27 (0.72)
4.90 (0.86) 5.57 (0.57)
4.82 (0.86) 5.10 (0.76)
5.63 (0.74) 6.10 (0.34)
5.09 (0.93) 5.60 (0.71)
4.88 (0.94) 5.40 (0.70)
4.88 (0.80) 5.12 (0.71)
5.05 (0.80) 5.59 (0.54)
5.04 (0.93) 5.48 (0.61)
8.60 (1.15) 2.27 (0.97)
experience with a 0.22-caliber small-bore rifle. They were qualified at the ‘‘sharpshooter’’ level or better in threeposition shooting (the progression as identified by the Education and Training division of the National Rifle Association proceeds from marksman, to sharpshooter, to expert, to master). None of the novices held any experience with position shooting (i.e., shooting under regulated conditions while in a standardized posture). All participants were screened with a health history questionnaire and were determined to satisfy the following inclusion criteria: (a) no history of neurological, cardiovascular, or other major disorders; (b) no history of psychiatric disorders; (c) no current use of medications; and (d) no hospitalizations or experience of general anesthesia within the last 12 months. All participants were right-hand dominant, as determined via procedures developed by Chapman and Chapman (3), and ipsilateral eye dominant. Prior to testing, all participants provided written consent on a form approved by the institutional review board for the protection of human subjects. Procedures. Participants were asked to refrain from consuming any alcohol or caffeine on the day of testing. After arriving at the laboratory, participants were informed about the requirements of the study, and they provided written informed consent. Novice shooters viewed a 20-min videotape on proper shooting technique. Participants were then fitted with the lycra electrode cap, and impedances were checked (see EEG recording). All of the participants shot right-handed and assumed a standardized posture while shooting that served to minimize body sway while aiming at the target. The details of the task requirements have been reported by Haufler et al. (18). Participants first had to calibrate an optical shooting simulator unit (Noptel ST2000, version 2.33) that emitted a light beam that was
FIGURE 1—Mean dimension (T SEM) for expert (N = 15) and novice (N = 21) groups at 10 electrodes.
754
Official Journal of the American College of Sports Medicine
subsequently reflected back to the device from a stationary target (positioned 4 m away) to indicate the alignment and the positioning of the aiming point of the rifle in relation to the center of the target (i.e., the bull`s eye). The proportional size of the target and the distance from the barrel of the gun simulated a standard small-bore International A-17 target at a distance of 50 feet. Participants then completed a minimum of 10 shots for practice. On initiation of the task, they were instructed to do their best and were informed that their shooting accuracy for 40 self-paced shots would be compared with that of the others in their group. In each trial, EEG and event-marker data sampling, via a microphone that detected the sound of the trigger pull, were initiated after the participant had adopted a steady aiming stance such that the real-time display of EEG recordings were free of motion artifact. A trial terminated with the trigger pull, which was recorded by the event-marker channel. EEG recording. EEG was recorded from the left and right frontal (F3, F4), central (C3, C4), temporal (T3, T4), parietal (P3, P4), and occipital (O1, O2) locations and referred to the right earlobe (A2) online. An additional recording of the left earlobe (A1) referenced to A2 was obtained to derive an averaged-ear reference offline. This was done to reduce lateral bias. The ground electrode was located at the frontal pole (Fpz). Scalp recordings were obtained with surface tin electrodes housed in a stretchable lycra cap (Electrode-Cap International, Inc.). Vertical and horizontal electrooculograms (VEOG and HEOG, respectively) were recorded with bipolar configurations of 10-mm Grass cup electrodes (model E5GH) located superior and inferior to the right eye and on the left and right orbital canthi. Impedance at each electrode site was
FIGURE 2—Mean dimension (T SEM) for the pooled expert and novice groups (N = 36) at 10 electrodes. Light bars designate left cerebral hemispheric regions; dark bars designate right cerebral hemispheric regions.
http://www.acsm-msse.org
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
FIGURE 3—Pairwise comparisons of the mean dimension (T SEM) for the pooled expert and novice groups (N = 36). The mean values are shown in descending order.
maintained at or below 5 k6. The differences between homologous pairs of scalp electrodes was within 500 6. EEG was amplified 50,000 times using Grass model 12A5 Neurodata Acquisition amplifiers with band-pass filter settings of 1–100 Hz. EOG was amplified 5000 times. A 60-Hz notch filter was also employed during the data collection. Amplifiers were calibrated prior to each testing session with a 10-Hz, 50-V sinusoidal input signal that was presented to all channels simultaneously. Data were acquired at a sampling rate of 256 Hz, using Neuroscan software installed on a Gateway 2000 Pentium computer. Data preparation and signal processing. Shooting performance was quantified as the total score of 40 shots (maximum possible was 400 points). Each shooting trial was scored from a computer display of the target with shot placement and was assigned the value corresponding to the innermost ring of 10 equally spaced rings broken by the shot. A shot breaking the innermost ring (a bull`s-eye) was scored as a 10; a shot placed outside of the outermost ring (a miss) was scored as a 0. The following steps were applied to EEG data to prepare for dimensional analysis. First, ocular artifact reduction was applied to the continuous records of the EEG activity. More specifically, the data were subjected to an algorithm (29) for the detection and, when appropriate, reduction of eye-movement artifact. Second, the longest aiming interval that was common to all participants during the shooting task was 5–6 s. Therefore, a 5-s window terminating with the trigger pull was selected for analysis, to achieve a common data structure across trials and participants. Third, the 5-s window was corrected for baseline, using the average of the interval. Fourth, epochs were examined to exclude those in which EEG amplitudes exceeded T 80 KV. Finally, epochs were visually inspected for the presence of artifact. Across all participants, 18 % of trials were rejected from analysis.
TABLE 2. Correlation between shooting performance and EEG dimensionality.
Performance experts (N = 15) Performance novices (N = 21)
F3
F4
C3
C4
T3
T4
P3
P4
O1
O2
j0.34 0.45*
j0.20 0.35
j0.13 0.22
j0.52* 0.37
j0.26 0.13
j0.24 0.33
j0.43 0.21
j0.45 0.12
j0.49 j0.08
j0.35 j0.11
* P G 0.05.
EEG AND PSYCHOMOTOR PERFORMANCE
Medicine & Science in Sports & Exercised
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
755
APPLIED SCIENCES
The correlation dimension (D2) was determined from the EEG recorded at the individual electrode sites during each trial, using Dataplore software (Datan GmbH, 1998). Evidence in support of stationarity of the EEG time series, one of the primary requisites for the determination of D2, has been provided in a previous report (18) of the spectral analysis of the data that revealed no change in the amplitudes of the major frequency bands over the time of the 5-s aiming period. In addition, the construction of the state–space trajectory of the system from the scalar EEG time series involves derivation of (a) the time delay and (b) the embedding dimension. An autocorrelation procedure was employed to determine the appropriate time delay (23). Generically, the decay of the autocorrelation function will not be strictly monotonous. As such, there will be local minima in the function, and also a first minimum. The geometric structure of the reconstructed attractor should be maintained as simple as possible, and it is required that the time delay not be too long. Specifically, the ‘‘first minimum in the autocorrelation function’’ method was used to estimate an appropriate time delay for the attractor reconstruction. In the present study, the sampling rate of 256 Hz applied to the EEG translates into a sampling interval of 3.9 ms. The first significant minimum, derived from the autocorrelation function, was determined to lie at a delay of five sampling intervals corresponding to 19.5 ms (i.e., five samples 3.9 ms). The time delay also corresponds to one quarter of the period of the dominant oscillation, which was then derived by multiplying 19.5 ms 4. (Note that the resultant period of 78.1 ms corresponded to 12.8 Hz, a frequency that lies in the high-alpha band of EEG oscillations and is typically associated with alert, but relaxed, cerebral cortical activity.) Accordingly, this identified time delay (five sampling intervals) was applied to the EEG obtained from the experts and novices at each electrode site during each 5-s trial, to determine the estimated dimensions. The embedding dimension (De) was derived by consideration of the total number of data points per trial (5 s 256 samples per second = 1280 data points), which is of the order of 103. Based on the minimum sampling requirement for each dimension (~10 data points for each dimension), the maximum for reliable estimation of the attractor dimension (Da) is 3. Accordingly, a uniformly filled cube with 10 data points for each dimension would contain 103 points. For De, we selected a minimum value of De = 2Da + 1 = 7. Therefore, the value 7 (i.e., 2 3 + 1) was applied to the EEG time series following the Whitney embedding theorem. Finally, Grassberger and Procaccia`s (13) procedure was used to estimate D2.
TABLE 3. Correlation between shooting standard deviation and EEG dimensionality.
Performance experts (N = 15) Performance novices (N = 21) Performance pooled (N = 36)
F3
F4
C3
C4
T3
T4
P3
P4
O1
O2
j0.18 0.48* 0.08
j0.14 0.35 0.06
j0.35 0.12 j0.05
j0.12 0.25 0.04
j0.01 0.25 0.17
0.12 0.28 0.23
j0.05 0.16 0.11
0.04 0.04 0.09
j0.07 j0.05 0.09
j0.07 j0.09 0.04
APPLIED SCIENCES
* P G 0.05.
Statistical analysis. Target shooting scores for each participant were subjected to a one-tailed t-test for independent means to determine whether the experts were more accurate than the novices on the shooting task. In addition, estimates of performance variability (i.e., standard deviation for shooting score) for each participant were subjected to a one-tailed t-test for independent means to determine whether the experts were more consistent with shot placement than the novices on the shooting task. Average D2 for each participant was subjected to a 2 (group: experts vs novices) 2 (cerebral hemisphere; left vs right) 5 (region: frontal, central, temporal, parietal, and occipital) ANOVA with repeated measures on the last two factors. In the event of violation of the sphericity assumption, the Greenhouse–Geisser correction was applied to the degrees of freedom. Conventional degrees of freedom are reported in the Results section, along with the Greenhouse–Geisser adjustment (i.e., epsilon) when necessary. The least square difference method was employed for post hoc comparison of means. Effect sizes (D) were also calculated for any significant contrasts. Furthermore, shooting performance scores (i.e., both accuracy and variability) and D2 estimates (at each recording site) were subjected to Pearson correlation analysis in each group, separately, to assess any differences in this relationship attributable to skill level. Pooled correlational analyses across all participants (i.e., experts and novices) were conducted to determine the relationship between brain activity and the quality of performance if the distributions between the variables of interest were overlapping for the two groups. Such an approach was deemed to achieve a more stable assessment of the relationship in light of the increased number of observations. The criterion alpha level was set to 0.05.
lower D2 than the novice shooters at all electrode sites. In addition, a hemisphere region interactive effect was revealed (F(4,31) = 3.00, P = 0.03, epsilon = 0.59, eta = 0.28). Post hoc comparisons for all possible electrode pairs involved in the interaction were conducted. As shown in Figure 2, the general pattern of this interaction, which was common to both groups, was characterized by significantly higher levels of D2 in the anterior regions of the left hemisphere (i.e., frontal–F3, central–C3, and temporal–T3 regions) relative to each of the respective homologous sites in the right hemisphere, while no such differences occurred in the posterior regions (i.e., the parietal and occipital pairings). Furthermore, the highest level of D2 occurred at T3, which was significantly higher than that recorded at all other sites in both hemispheres. The descending order of magnitude for the remaining sites was F3, T4, O1, O2, C3, P3, F4, P4, and C4. Figure 3 illustrates the pairwise comparisons. Correlation between D2 and shooting performance. As shown in Table 2, Pearson`s correlation analysis revealed a negative relationship between D2 and shooting scores at C4 (r = j0.52) in the experts, while a positive relationship was found between D 2 and shooting performance at F3 (r = 0.45) in the novices when the groups were examined separately. In addition, Table 3 contains the results of the Pearson`s correlation analysis between D2 and shooting score variability (i.e., SD) for the groups separately as well as pooled. Generally, there were no significant correlations between D2 and SD, except at site F3 in the novice group. A pooled correlational analysis between D2 and performance scores was contraindicated because of nonoverlapping distributions between these variables in the expert and novice groups, while a pooled analysis was indicated between D2 and shooting score variability in light of overlapping distributions between these variables for the two groups.
RESULTS Task performance. Experts achieved significantly better performance, both in terms of mean (339.8 T 44.7 vs 90.7 T 38.9, t(34) = 17.80, P = 0.001) and standard deviation (1.80 T 0.53 vs 2.84 T 0.10, t(34) = j2.25, P = 0.031) on the shooting task than the novices. EEG dimensionality. The D2 measures and performance scores (as well as standard deviations), which were obtained at each electrode site from the expert and novice participants, are provided in Table 1. The 2 2 5 (group hemisphere region) ANOVA revealed a group main effect (F(1,34) = 5.17, P = 0.03, eta = 0.36) such that experts exhibited lower D2 (5.02 T 0.16) than the novices (5.49 T 0.13). As shown in Figure 1, the expert shooters exhibited
756
Official Journal of the American College of Sports Medicine
DISCUSSION The purpose of this study was to determine whether expertise in the cognitive-motor domain is related to the complexity of cerebral cortical activity during performance. More specifically, it provided a test of the proposition that the quality of performance is inversely related to the complexity of brain processes, using a nonlinear dimensional analysis of the time series data recorded from the cerebral cortex of experts and novices during a visuomotor aiming task (i.e., rifle shooting). As predicted, the experts were more accurate during target shooting than the novices, and the experts negotiated the task with lower complexity
http://www.acsm-msse.org
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
EEG AND PSYCHOMOTOR PERFORMANCE
brain activity, resulting in more consistent performance (i.e., less opportunity for variability), while greater complexity is likely associated with increased degrees of freedom and less consistent motor behavior (i.e., more opportunity for variability across trials). Perhaps this view of brain activity during motor performance explains why the experts performed with more accuracy and less variability than the novices. From this perspective, it is logical that the novices were characterized by a less stable or ‘‘noisy’’ brain state, owing to the lack of refinement of brain processes, resulting in greater variability in performance. In addition to the differences noted in the complexity of the cortical activity between the groups, we also examined the relationship between EEG dimensionality and motor (i.e., shooting) performance. As predicted, the correlational analysis revealed that D2 and target shooting accuracy were inversely related in the experts, such that lower EEG dimension was associated with elevated shooting performance. The directionality of the relationship was reversed in the novice group. That is, a higher level of complexity (i.e., higher dimension) was associated with greater accuracy in the novices, although the relationship was confined to site F3. The prefrontal lobe is related to working memory, planning, and emotion (12), and it is also located close to the premotor cortex, which is critically involved with motor behavior. Furthermore, the premotor cortex plays a prominent role during the early stages of motor skill learning (11,30). Studies in motor skill learning generally have found greater involvement at structures such as premotor the cortex (30), along with the primary somatosensory and motor cortices, and the basal ganglia (11). In this manner, the engagement of the prefrontal region in novices is necessitated. In contrast, the contribution of the prefrontal region would be supplanted by subcortical processes as skill acquisition proceeds, such that the role of prefrontal involvement is lessened. With learning, the internal representation of motor skills has been found migrating to subcortical or other cortical areas (30). In support of this notion, we failed to observe any significant relationship between dimensional complexity at F3 and shooting accuracy in the experts. Therefore, it would seem that more cortical resources are needed for the novice shooter to process the unfamiliar cues and engage in feature extraction of task-related stimuli in the target shooting environment (i.e., both external and internal). It appears that the efficiency hypothesis, strictly speaking, is valid only for those who have attained a reasonable level of skill, perhaps the associative stage as defined by Fitts and Posner (10). As such, a moderating effect of skill level on the relationship between complexity of cortical processes and performance is implied by the results. That is, the novice may be aided by engagement of increasingly complex neural processes up to a certain level of skill acquisition, beyond which performance is likely compromised with any additional increase in complexity as skill progresses. In this regard, the present results strongly support a reduction of complexity with advancing expertise.
Medicine & Science in Sports & Exercised
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
757
APPLIED SCIENCES
of cortical activity compared with that exhibited by the novice shooters (Fig. 1 and Table 1). According to Cohen`s categorization of effect sizes, the magnitude of difference in dimensionality between the groups was large (4). The results suggest that practice and skill acquisition were associated with a significant change in the neural architecture and processes underlying the brain`s response to the challenge (aiming task). However, such a conclusion requires caution in light of the cross-sectional nature of the study design. One could argue that the examination of differences in the complexity of cerebral cortical activity between expert and novice target shooters also provided for a conservative test of the study prediction, as the task requirements were likely challenging to both groups. That is, experts likely maintain a high degree of attention or focus during such a precision-based aiming task, and attainment of an advanced level of skill would also likely involve a shift in strategic thinking and a refinement of cognitive processes to the point of automaticity, but not a state of mindless action (10). Despite this continuing cognitive demand, the experts apparently accomplished the task with fewer neural resources while achieving a higher level of behavioral output (i.e., accuracy score) than the novices. This finding implies that expert marksmen negotiate the task demands in an efficient manner. In contrast, novices require effortful feature extraction from the environment and activation of a larger number of neuronal assemblies to complete the task (10,19), thus negotiating the task in a relatively inefficient manner. Collectively, the findings from the current study and those of Haufler et al. (18) provide convergent evidence for the psychomotor efficiency hypothesis from both linear and nonlinear signal-processing strategies. A number of investigators have validated the expectation that dimension of EEG is, indeed, positively related to the complexity of mental processes. For example, Aftanas and Golocheikine (1) report that experienced meditators exhibited lower dimensional complexity during meditation, when compared with that observed during a resting state. They suggest that the lowered dimensional complexity during meditation involved ‘‘switching off’’ or pruning irrelevant networks for the maintenance of focused internalized attention. Bell and Fox (2) describe the adaptive nature of the pruning of irrelevant networks in the context of motor development (i.e., learning to walk). In a similar manner, the lower dimensional complexity, which accompanied the superior performance of the expert marksmen in the current study, suggests inhibition of neural assemblies that are nonessential. Accordingly, the expert group in the present study showed a reduction or refinement of neural activity to achieve high-quality motor performance, as evidenced by lower D2, just as Lutzenberger et al. (25) show that simplified or convergent thinking, as opposed to divergent thinking, was associated with lower D2. Theoretically, such a simplified or refined mental state is likely associated with a reduction in the degrees of freedom of
APPLIED SCIENCES
Although the experts exhibited greater consistency of performance than the novices, as indicated by the lower standard deviations for the shot placement in the former group, there was no significant correlation evidenced between variability of target shooting performance and D2. In this regard, and contrary to expectation, performance variability was unrelated to the complexity of cerebral cortical activity when the relationship was examined across both groups (Table 3). However, a positive correlation between these variables was found in the novice group only, again in the left prefrontal region, at site F3. One explanation for this finding lies in the strategic approach of the novice who likely employs explicit or controlled and effortful processing in their motor behavior. Such an approach would lead to increased complexity of cortical activity in light of the increased effort, and would also lead to greater performance variability in light of the unstable and changing strategies employed during this early stage of skill acquisition, resulting in the observed positive relationship. Regardless of the level of expertise, there was a consistent pattern of topographical variation in the complexity (D2), as derived from different brain regions, that was common to both groups. This statement is based on the absence of a statistical interaction between group status and recording site location. In both groups, the highest dimensional complexity was generally observed in the anterior regions of the left hemisphere relative to the homologous right regions as well as the perceptual and sensory regions (parietal and occipital, respectively) of the brain. Such a finding is indicative of greater engagement of the neural resources devoted to frontally-mediated executive processes, temporally mediated feature detection and memory storage, and contralateral motor control of the right trigger finger. In fact, the left temporal region was the region of highest complexity and was differentiated from all other recording sites. Accordingly, the medial temporal lobe, which contains memoryrelated structures such as the hippocampus, is critically involved in analytical feature detection (left hemisphere) and in encoding the visuospatial task demands of target shooting (right hemisphere) (16). Although the experts exhibited lower D2 at the left temporal location than the novices, it appears that both groups are relatively dependent on such mental processes compared with those mediated by other brain regions (i.e., visual and motor functions) and that it is largely a matter of degree, with experts engaging in less of this type of processing. It is also noteworthy that the complexity of cortical activity in the occipital region was relatively high and undifferentiated from the left frontal and central sites. This finding makes sense in light of the premium placed on visual processes for successful marksmanship; again, it seems to be a matter of the degree of dependency diminishing with expertise. In conclusion, the findings from this study are consistent with the psychomotor efficiency hypothesis (15). Skilled shooters exhibited lower complexity of cerebral cortical
758
Official Journal of the American College of Sports Medicine
activity during a precision aiming task than did their inexperienced counterparts. As predicted, EEG dimensionality was negatively related to shooting accuracy in the expert study participants, albeit at only one site of the 10 possible cortical regions. In this manner, higher complexity of cerebral cortical activity was associated with progressive degradation in motor performance. The results suggest that refinement and ‘‘simplification’’ of cerebral cortical activity facilitates visuomotor performance, and, conversely, that increased complexity of brain activity contributes to the presence of neuromotor ‘‘noise’’ in the brain, resulting in interference with intended action. However, such a principle tends to be constrained to the level of motor skill, because, in contrast, the complexity of cortical activity in the left frontal region was positively related to shooting accuracy in the novices, although joined by greater variability. One must be cautious in arriving at such a conclusion in light of the constrained topography for this relationship (i.e., only one recording site), but it seems reasonable that a novice would derive benefit by more effortful engagement of neural resources, particularly those that relate to attention and response inhibition (i.e., executive processes). However, the increased effort would likely incur a cost or a negative side effect of greater variability in psychomotor performance. In this regard, the expert is characterized by relative automaticity, and any undue activation of neural assemblies could be detrimental to psychomotor performance, through an unnecessary recruitment of neural resources or a ‘‘noisy’’ system. Finally, it is remarkable how conceptually convergent the findings from the present study, using a nonlinear metric of brain activity, are with those derived from linear (i.e., spectral) signal analysis (14,18). For example, a number of studies have revealed relative relaxation in the brains of experts during motor performance using both EEG and magnetic resonance technology, while the present study is supportive of refinement and simplification of the involved cortical processes. As such, the collective findings from this area of research provide convergent validation for greater economy in the central nervous system with skill acquisition. However, the linkage between D2 dimensional complexity and EEG spectral power is not a simple one, as evidenced by the finding of greatest complexity of cortical activity at site T3, while many studies have revealed the highest levels of alpha power at this very same site, which is indicative of relaxation. It is plausible that a larger ensemble of neurons could be engaged in the left temporal region (as suggested by the D2 data) while exhibiting relatively higher synchrony, thus resulting in cortical relaxation. Although speculative, such a possibility calls for the employment of complementary neuroimaging techniques, to more fully and accurately understand the neural basis of skilled motor performance. The work of Tsung-Min Hung was supported in part by the National Science Council (Taiwan) under grant NSC92–2413-H154–009 and NSC94–2413-H-154–002.
http://www.acsm-msse.org
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
REFERENCES
EEG AND PSYCHOMOTOR PERFORMANCE
17. Hatfield BD, Landers DM, Ray WJ. Cardiovascular-CNS interactions during a self-paced, intentional attentive state: elite marksmanship performance. Psychophysiology. 1987;24:542–9. 18. Haufler AJ, Spalding TW, Santa Maria DL, Hatfield BD. Neurocognitive activity during a self-paced visuospatial task: comparative EEG profiles in marksmen and novice shooters. Biol Psychol. 2000;53:131–60. 19. Hebb DO. The Organization of Behavior: A Neuropsychological Theory. New York (NY): Wiley; 1949. 20. Kerick SE, Douglass LW, Hatfield BD. Cerebral cortical adaptations associated with visuomotor practice. Med Sci Sports Exerc. 2004;36(1):118–29. 21. Landers DM, Han M, Salazar W, Petruzzello SJ. Effects of learning on electroencephalographic and electrocardiographic patterns in novice archers. Int J Sport Psychol. 1994;25: 313–30. 22. Lay BS, Sparrow WA, Hughes KM, O`Dwyer NJ. Practice effects on coordination and control, metabolic energy expenditure, and muscle activation. Hum Mov Sci. 2002;21:807–30. 23. Liebert W, Schuster HG. Proper choice of time delay for the analysis of chaotic time series. Phys Lett A. 1989;142:107–11. 24. Loze GM, Collins D, Holmes PS. Pre-shot EEG alpha-power reactivity during expert air-pistol shooting: a comparison of best and worst shots. J Sports Sci. 2001;19:727–33. 25. Lutzenberger W, Preissl H, Pulvermuller F. Fractal dimension of electroencephalographic time series and underlying brain processes. Biol Cybern. 1995;73:477–82. 26. Molle M, Marshall L, Wolf B, Fehm HL, Born J. EEG complexity and performance measures of creative thinking. Psychophysiology. 1999;36:95–104. 27. Muller V, Lutzenberger W, Preissl H, Pulvermuller F, Birbaumer N. Complexity of visual stimuli and non-linear EEG dynamics in humans. Brain Res Cogn Brain Res. 2003;16:104–10. 28. Radlo SJ, Janelle CM, Barba DA, Frehlich SG. Perceptual decision making for baseball pitch recognition: using P300 latency and amplitude to index attentional processing. Res Q Exerc Sport. 2001;72:22–31. 29. Semlitsch HV, Anderer P, Schuster P, Presslich O. A solution for reliable and valid reduction of ocular artifacts, applied to the P300 ERP. Psychophysiology. 1986;23:695–703. 30. Shadmehr R, Holcomb HH. Neural correlates of motor memory consolidation. Science. 1997;277:821–5. 31. Starkes JL, Helsen W, Jack R. Expert performance in sport and dance. In: Singer RN, Hausenblas HA, Janelle CM, editors. Handbook of Sport Psychology. New York (NY): John Wiley & Sons; 2001. p. 174–201.
Medicine & Science in Sports & Exercised
Copyright @ 2008 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
759
APPLIED SCIENCES
1. Aftanas LI, Golocheikine SA. Non-linear dynamic complexity of the human EEG during meditation. Neurosci Lett. 2002;330:143–6. 2. Bell MA, Fox NA. Crawling experience is related to changes in cortical organization during infancy: evidence from EEG coherence. Dev Psychobiol. 1996;2:551–61. 3. Chapman LJ, Chapman JP. The measurement of handedness. Brain Cogn. 1987;6:175–83. 4. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale (NJ): Lawrence Erlbaum Associates; 1988. 5. Crews DJ, Landers DM. Electroencephalographic measures of attentional patterns prior to the golf putt. Med Sci Sports Exerc. 1993;25(1):116–26. 6. Daniels JT. A physiologist`s view of running economy. Med Sci Sports Exerc. 1985;17(3):332–8. 7. Deeny SP, Hillman CH, Janelle CM, Hatfield BD. Cortico-cortical communication and superior performance in skilled marksmen: an EEG coherence analysis. J Sport Exerc Psychol. 2003;25:188–204. 8. Di Russo F, Pitzalis S, Aprile T, Spinelli D. Effect of practice on brain activity: an investigation in top-level rifle shooters. Med Sci Sports Exerc. 2005;37(9):1586–93. 9. Elbert T, Ray WJ, Kowalik ZJ, Skinner JE, Graf KE, Birbaumer N. Chaos and physiology: deterministic chaos in excitable cell assemblies. Physiol Rev. 1994;74:1–47. 10. Fitts PM, Posner MI. Human Performance. Belmont (CA): Brooks/Cole; 67. 11. Floyer-Lea A, Matthews PM. Distinguishable brain activation networks for short- and long-term motor skill learning. J Neurophysiol. 2005;94:512–8. 12. Fox NA, Schmidt LA, Henderson HA. Developmental psychophysiology: conceptual and methodological perspectives. In: Cacioppo JT, Tassinary LG, Berntson GG, editors. Handbook of Psychophysiology. Cambridge (UK): Cambridge University Press; 2000. p. 665–86. 13. Grassberger P, Procaccia I. Measuring the strangeness of strange attractors. Physica D. 1983;9:189–208. 14. Hatfield BD, Haufler AJ, Hung TM, Spalding TW. Electroencephalographic studies of skilled psychomotor performance. J Clin Neurophysiol. 2004;21:144–56. 15. Hatfield BD, Hillman CH. The psychophysiology of sport: a mechanistic understanding of the psychology of superior performance. In: Singer RN, Hausenblas HA, Janelle CM, editors. Handbook of Sport Psychology. New York (NY): John Wiley & Sons; 2001. p. 362–88. 16. Hatfield BD, Landers DM, Ray WJ. Cognitive processes during self-paced motor performance: an electroencephalographic profile of skilled marksmen. J Sport Psychol. 1984;6:42–59.
