Balance Performance with a Cognitive Task: A Dual-Task Testing Paradigm

Abstract

Purpose: 

Athletic performance demands simultaneous use of cognitive and postural control capabilities. Decrements to both systems have been observed following concussion. This study evaluated a dual-task methodology to establish the tenability of using this testing model in concussed athletes.

Methods: 

Nonconcussed subjects were assessed over 2 d. Subjects were introduced to the task-switching cognitive assessment test and a NeuroCom Smart Balance Master postural control assessment protocol on day 1. In the following session, subjects were evaluated on postural control and cognitive function tests independently (single task), and then concurrently (dual task).

Results: 

Significant improvements were seen in three of the four balance conditions and in three of the four reaction times when the cognitive and balance task were performed simultaneously (P < 0.05). No significant balance by cognitive task interaction was revealed (P > 0.05); however, significant differences existed in reaction time based on stimulus position and increasing balance demands (P < 0.05).

Conclusion: 

Combining the cognitive and balance assessments resulted in healthy subjects showing improved performance when compared with individual task performance. The dual-task methodology brought about systematic changes to reaction time in relation to increasing balance demands. The ability of this protocol to detect changes in postural control or cognitive function following concussive injury requires further study.

Recent large-scale investigations have suggested that 5–6% of football players will incur a concussion during a season (^8,24). The current concussion evaluation paradigm advocates a comprehensive approach to the injury evaluation to assist in making safe return to play decisions (^10,22). The majority of research, however, has focused on individual measures of self-reported symptomatology (^4,23), neuropsychological testing (^4,6,15,18), and posturography (^9,22). Although these approaches have provided valuable information regarding the recovery curve following concussion, participation in sport requires integration among the cognitive, sensory, and motor systems to efficiently and effectively perform sport movements.

Recently, an updated definition of concussion was proposed, one tenet of which stated that concussion largely reflected a functional disturbance rather than structural damage to the brain (³). The metabolic cascade following concussion is widely thought to impair normal brain function (⁷), which could lead to cognitive slowing demonstrated by neuropsychological testing (^4,6,15). Further, it has also been frequently reported that athletes self-report symptoms of “feeling in a fog,” “feeling slowed down,” and “fatigue” as primary complaints following concussion (^9,22,23). Only recently has research begun to explore the relationship between symptoms and performance on neuropsychological testing following concussion (^4,6).

It is widely accepted that a concussed athlete should be symptom free at rest and during exertional maneuvers before returning to sport participation. A variety of exertional tests are used, including cycling, running, and sport participation without the threat of contact, as part of the return to play decision. However, these tests lack objective criteria and require the athletes to report their symptoms honestly to the health care practitioner.

Dual-task methodology is a testing model that requires a person to simultaneously perform a cognitive and motor task. Currently, this is the closest paradigm to replicate sport performance to evaluate multiple systems concurrently. Typical dual-task methodology models have used a mathematical task while performing a static motor task. The majority of research on healthy subjects has found a slight reduction in one or both measures during dual-task testing (^19,28,29). Only one study has evaluated concussed subjects using a dual-task testing model. Results indicated that athletes performed significantly worse on cognitive and motor tasks immediately following a concussive episode (²¹). However, this study was limited by not using instrumented tests for either the balance or cognitive tasks. Instead, field tests of walking on a balance beam combined with having the athlete count backward from a randomly selected number between 95 and 105 by 7s (serial-7s test) was implemented.

We sought to extend the previous research by replacing the tasks with more sophisticated measures. The Neurocom Smart Balance Master has shown to be a reliable and valid assessment instrument in the evaluation of balance following concussion (^9,10,22). The Neurocom sensory organization test uses six conditions with three trials per condition to stress the somatosensory, vestibular, and visual systems (³²), and has been described previously (^9,10). The task-switching cognitive performance test is a measure of cognitive flexibility, attention, and information-processing speed. Compared with younger adults, older adults demonstrated less attentional flexibility and greater switch-task costs when stimulus conditions were altered (³⁰). The purpose of this project was to evaluate a dual-task methodology in healthy young adults to determine the plausibility of using this testing model in concussed athletes.

METHODS

Participants

Twenty male (N = 10) and female (N = 10) subjects (age: 22.30 ± 2.06 yr, height: 1.75 ± 0.10 m, weight: 72.20 ± 17.29 kg) were recruited to participate in this study. Sample size was estimated using a large effect size (d = 0.8) from previous research in the areas of postural control and task switching (^2,9). Subjects were considered ineligible to participate if they had a previously diagnosed concussion, if English was not a primary language, or if they were currently receiving treatment for a lower-extremity injury.

Procedure

Each subject read and signed an institutional review board–approved informed consent, and then completed a brief questionnaire of self-reported demographics. Testing was completed over two sessions separated by 24–72 h. Session 1 was used to orient the participant to the balance and cognitive tasks, and to practice each task.

The balance test was conducted with a NeuroCom Smart Balance Master equipped with the Data Acquisition Toolkit (DATa) version 2.0 software (NeuroCom International, Inc., Clackamas, OR). Four test conditions presented in Figure 1 were developed for balance assessment: fixed surface and fixed visual reference (fixed-fixed), fixed surface and sway-referenced vision (fixed-sway), sway-referenced surface and fixed visual reference (sway-fixed), and sway-referenced surface and vision (sway-sway). Conditions that required the subjects to close their eyes were not included because of the visual requirement of the cognitive task. Sway gain was set at 1.0, exactly matching sway referencing to the subject’s sway as described in the NeuroCom System Operator’s Manual (²⁰). Before the practice session, each subject was instructed to stand on the NeuroCom platform as upright and motionless as possible during the testing session. The subject then completed each of the four conditions three times. In lieu of a counterbalanced design, the 12 trials were randomized. Each trial lasted 20 s, and was completed with the eyes open.

FIGURE 1— Modified Sensory Organization Test images used courtesy of NeuroCom International, Inc. a, fixed-fixed; b, fixed-sway; c, sway-fixed; d, sway-sway.

The task-switching test was a visual processing task in which a letter–digit pair was presented in a 2 × 2 matrix centered in the middle of a computer screen. The participant was required to perform an odd/even numerical judgment task for two trials, followed by performance of a vowel/consonant judgment task for two trials. When a letter–digit pair was shown in either of the two upper quadrants of the matrix, participants performed an odd/even judgment task, and when the letter–digit pair was shown in the two lower quadrants of the matrix, participants performed a consonant/vowel letter judgment task (Fig. 2). Each letter–digit pair was presented until the individual responded by depressing a computer mouse key. Participants’ response time and response choice were recorded. The letter–digit pairs were presented in the matrix in a continuous clockwise direction; thus, the location of the stimuli was predictable. Each participant performed two blocks of training during which 60 letter–digit pairs were presented. The task was modeled on one developed by Kramer et al. (¹³), and shown to be a sensitive measure of adults’ executive processing ability.

FIGURE 2— An example task-switch screen. Participants are asked to make an odd/even judgment in the top quadrants, and a vowel/consonant decision in the bottom quadrants, as the letter–digit combination rotates clockwise through the quadrants.

Session 2 consisted of two phases. In the first phase, participants completed separate administrations of the balance test and the task-switching test. In the single-task balance condition, the subject was provided with the same instructions as in the practice session. Data were collected for the duration of the 20-s trial. Each subject was randomly assigned into 1 of 10 balance assessment protocols, each with 12 randomly ordered trials. In the single-task cognitive test, the participant was seated in front of a computer monitor, provided a description of the task, and a brief 12-trial practice trial. The task-switching test consisted of 12 sets of 16 letter–number pairs of approximately 20 s. Participants’ response time and response choice for each letter and number pair were recorded from a two-button computer mouse, which was located on a pad in front of the computer.

In the second phase, participants performed a dual-task assessment that required the simultaneous completion of the two tests. A monitor was mounted to the visual sway-referencing partition of the NeuroCom Smart Balance Master, and hidden behind a sliding cover. The cover was opened during the dual-task portion of the test. The subject viewed the monitor that displayed the task-switching test, while simultaneously performing the balance test, as shown in Figure 3. Twelve sets of sixteen letter and number pairs were presented to each subject. The response time and response choice for each letter and number pair were recorded through a two-button computer mouse that was held by the participant at waist level. Participants were instructed to focus on the balance test, but to perform the task-switching test as quickly and accurately as possible. Each trial was started with the tester’s verbal signal. Balance data were collected over a 20-s period. Each of the 12 trials was separated by approximately 15 s.

FIGURE 3— The dual-task device. A NeuroCom Smart Balance Master equipped with a second computer to run the task-switch test.

Data Analysis

Balance assessment.

During the dual-task condition, the task-switching test was often completed before the balance assessment ended. As such, only the first 10 s of each trial were used, to ensure that task-switching coincided completely with the balance assessment. A condition balance score was therefore calculated from the first 10 s of each trial as described in the NeuroCom System Operator’s Manual (²⁰). Multiple repeated measures t-tests were used to assess differences between balance scores obtained under the single-task and dual-task test conditions for each of the four balance conditions.

Cognitive assessment.

Two analyses were conducted. The first employed repeated measures t-tests to compare participants’ average response time to letter–digit stimuli under the single-task and dual-task test conditions. Comparisons were made between response times to letter–digit stimuli at each of the four spatial locations for each of the four balance conditions.

Dual-task assessment.

The second analysis assessed the response times of 19 participants via a 4 (balance conditions) × 4 (letter–digit position) within-subject, repeated measures ANOVA. A computer malfunction resulted in the loss of one participant’s performance. All repeated measures analyses employed the Greehouse–Geisser correction for sphericity violations (W₍₄₄₎ = 0.006, χ² = 80.56, P = 0.001). Post hoc comparisons were performed using a Bonferroni correction for Type I error at the 0.05 level of significance. Both correct and incorrect response times were included in the analyses as previously reported (³⁰). Incorrect responses accounted for 3.5% and 2.0% of the total answers in the single- and dual-task conditions, respectively. An evaluation of the data based on sex was considered, but not conducted due to small sample size. Statistical analyses were conducted using SPSS (Chicago, IL) version 11.0. Participants’ choice-response errors were considered negligible, and were not evaluated.

RESULTS

Balance Assessment

Condition score means and SD for 10-s assessment of the balance during single-task and dual-task conditions for each of the four balance conditions are provided in Table 1. The repeated measures t-tests revealed significant differences in balance scores between three of four balance conditions. Condition balance scores were significantly higher under the dual-task condition during the fixed-fixed (t₍₁₉₎ = −3.60, P = 0.002), fixed-sway (t₍₁₉₎ = −3.11, P = 0.006), and sway-fixed (t₍₁₉₎ = −2.34, P = 0.031) conditions. Analysis of the sway-sway condition was nonsignificant (P > 0.05).

TABLE 1: Condition balance score means ± standard deviations for the 10-s assessments.

Cognitive Assessment

As shown in Table 2, a comparison of response times made under single-task and dual-task conditions revealed that participants’ performance was significantly faster during the dual-task condition when responding to stimuli in three of the four stimulus locations: top left (t₍₁₈₎ = 4.09, P = 0.001), top right (t₍₁₈₎ = 4.06, P = 0.001), and bottom right (t₍₁₈₎ = 3.57, P = 0.002). No significant difference was obtained between response times to stimuli presented in the bottom left location (P > 0.05).

TABLE 2: Means ± SDs of task-switch reaction times (ms).

Dual-Task Assessment

The 4 × 4 ANOVA revealed a main effect for the balance factor (F_(1.7,32.31) = 4.44, P = 0.025) and a main effect for the letter–digit factor (F_{(1.16, 21.96)} = 35.37, P < 0.001). There was no significant interaction between factors. Mean response times for the task-switch assessment for each balance condition are presented in Table 2. Pairwise post hoc analyses revealed that response times under the fixed-fixed condition and the sway-sway condition differed significantly. Test of within-subjects contrasts indicated that response times increased in a linear fashion across the four balance conditions (F_(1,19) = 10.22, P = 0.005). These results indicate that the time required to identify and respond to stimuli was influenced by balance demands. Pairwise post hoc analysis of response times at each stimulus location indicated that participants’ response times were significantly faster on successive numerical judgment trials and on successive letter judgment trials. Response times were significantly slower on trials during which participants switched from making a numerical judgment to a letter judgment, and when switching from a letter judgment to a numerical judgment. The lack of a statistically significant interaction between the four balance conditions and four stimulus locations indicates that the time required to change categorical decisions was not influenced by balance demands.

DISCUSSION

This project was designed to assess a dual-task methodology and to determine the plausibility of using this testing model in concussed athletes. We intended to evaluate adults’ postural control with and without the addition of a cognitive task, and to assess the impact of balance perturbations on information processing. The normal healthy young adults participating in this study showed improvements in postural control when the balance and cognitive tasks were performed concurrently. These results are similar to those obtained by Hunter and Hoffman (¹²), who found a reduced level of mediolateral center of pressure motion when their subjects completed an additive number cognitive task while performing a balance assessment. Andersson et al. (¹) also found a decrease in sway with the addition of a cognitive task.

Results from the present experiment differ from those obtained by Peterson (²¹). Using a similar cohort, each subject completed an 8-foot balance beam walk for time while performing a memory task. The authors observed a compromise in balance beam walking time with the addition of a working memory task in healthy subjects. This assessment, however, provided only a gross measure of postural control and cognitive functioning. Other investigators have also reported deficits in postural control following the addition of a cognitive task. Shumway-Cook and colleagues (^28,29) and Melzer et al. (¹⁹) reported deficits in postural control following the addition of a cognitive task. These studies, however, involved older sample populations, and may not generalize to our study. Older individuals may lose the ability to switch attention between two tasks as necessary to maintain optimal performance as reported in a younger population (¹⁶).

There are several tenable explanations for participants’ improved postural control following the addition of a cognitive task exist. It is possible that participants’ improved balance under dual-task conditions reflects refinement of motor control strategies acquired from practice. The NeuroCom Smart Balance Master is a novel task, because the support surface and/or visual reference move in relation to the subject’s center of pressure. Participants were provided a period of orientation to the balance task during session 1 to minimize practice effects. Additional exposure and training, however, may be necessary to stabilize balance measures. Practice effects were observed in previous research evaluating balance in concussed athletes and healthy control subjects. Composite balance scores in the healthy control subjects improved by approximately 4–7% when tested serially across 3–5 d (^9,22). The authors did not provide a rationale for these improvements, but they likely resulted from new motor strategies to the novel balance task as a result of short-interval serial testing.

Alternatively, the improvement in postural control under the dual-task test conditions may be explained in terms of type of cerebral processing required of the participants. The visual display employed during the dual task required participants to track the location of the letter–number stimuli. As a consequence, the low-frequency, voluntary eye movement may have improved postural control (⁵). Although this finding has been reported in the literature, a valid explanation for the phenomenon has not been developed (¹²). The use of the eyes in maintaining balance is one part of the balance mechanism. Normal balance is maintained through the integrative process of afferent sensory information provided by the visual, vestibular, and somatosensory systems. The three subcomponents of the balance mechanism collect peripheral information to maintain postural control. Afferent information collected by these systems is transferred to the CNS, where differing anatomical regions of the brain integrate the signals and produce a motor output. The basal ganglia receive the first inputs and begin a motor response based on the current position of the limbs (¹⁷). The signal is integrated with the planned actions of the motor cortex in the cerebellum, where motor impulses are coordinated (¹⁷). A final efferent signal is generated and transmitted through the brainstem to alpha motor neurons. These nerves innervate the skeletal muscles that maintain postural control (¹¹). Under normal conditions, input from the somatosensory and visual components are the dominant senses used in maintaining balance (²⁵). In instances of conflicting stimuli from these two systems, the vestibular system provides an overriding or controlling signal (¹⁰) based on linear and angular accelerations of the head (²⁵).

Although our instructions to focus on the balance task during the second session of testing were given, it is doubtful that they directly influenced postural control scores. Lower postural control scores seen on the first day of testing may have resulted from the participants’ focusing singularly on the postural control task. Hunter and Hoffman (¹²) suggest that focusing solely on a balance task may increase muscle tension, leading to increased joint stiffness and center of pressure motion. The addition of the task-switching test also may have influenced balance by moving the process of postural control from a conscious level to a more sensorimotor process, making the subject more efficient at maintaining balance (³¹). The low cognitive demand of well-learned tasks such as ours would place minimal demand on the brain to maintain postural control at a high level, permitting maximal attention on the task-switching test (¹⁴).

This study design altered only one aspect of the balance mechanism (somatosensory input) in each subject. Altering the postural control assessment by eliminating the visual system (eyes closed) may influence subject scores on balance as the redundancy built into the balance mechanism is systematically removed. The nature of our cognitive task, however, would not permit this. Future studies should consider modifying the cognitive task to one that allows for the eyes to be closed during the balance assessment. Tests that have negligible practice effects when administered in succession, and that do not rely on a visual component, also should be considered. A complete evaluation of the balance mechanism would be essential if dual-task tests were implemented in a concussion assessment battery for athletes. A compromised balance mechanism following concussion has been linked to impaired visual and vestibular systems (⁹). Excluding a portion of the assessment known to be related to concussion would likely result in inaccurate findings following injury.

Balance perturbations systematically affected participants’ cognitive performance. The impact was selective, however. The task-switching test yields measures of two independent components of information processing: speed of response selection, and response inhibition (¹³). Participants’ response times to each letter–digit pair provide a composite index of speed of visual scanning, discrimination between two stimuli, and response selection. The longer response times obtained in the sway-sway condition, as compared with the firm-firm condition, suggests that balance perturbations influence speed of response selection.

Response inhibition is viewed as an executive process that determines an individual’s ability to withhold making one response and to initiate a different response. The task-switching test is designed to measure the costs associated with having to switch response-choice classification, for example, having to switch from responding to letter stimuli to digit stimuli and vice versa. Differences in response times on trials in which successive stimuli of the same category are presented (e.g., two successive digits), and responses times on trials that change category (e.g., a digit stimulus followed by a letter stimulus) provide a measure of the cost associated with having to engage executive processes and to shift response categories. As expected, participants’ choice-response times were most rapid when repetitive decisions were made (e.g., two successive letter stimuli or two successive digit stimuli). Also, there was clear evidence of a cost in response time when participants were required to switch responses from one category to the other. Importantly, these inhibitory costs were similar under each of the four balance conditions. These results suggest that normal healthy adults are able to maintain executive processing and to alter categorical judgments despite perturbation of balance. Executive processes are considered to reflect higher mental processes that are essential for maintaining attention, problem solving, and adaptation to complex environments.

The construct of executive function has been supported by empirical evidence gathered in a variety of research domains. Evidence from animal studies, clinical research, and neuroimaging studies indicates that executive function is related to structures in the frontal and prefrontal cortical regions of the brain (²⁶). It remains to be determined whether insult to cortical areas involved in executive processing selectively affects components of information processing.

The dual-task methodology examined in this project showed systematic changes in balance and cognitive function. Our healthy subjects showed an increase in their ability to maintain postural control and improved reaction time when the tasks were combined. The current testing paradigm, however, may not be suitable for the evaluation of concussed athletes. Our results are limited by using a cognitive test that did not allow for the subjects to close their eyes during the balance assessment. Without removing the visual component of balance (eyes closed), we have obtained an incomplete view of the balance mechanism. Removing the visual component of balance may provide additional information under dual-task conditions. As such, an evaluation of the complete balance mechanism performed simultaneously with a cognitive task is warranted. This may be accomplished by combining the complete NeuroCom Sensory Organization Test (²⁰) with a verbal cognitive task that can be performed with the eyes closed. In addition, the cognitive task should last the duration of the balance trial (20 s), allowing for a full evaluation of both components of the dual-tasking methodology. Under these conditions, we speculate that both the cognitive function and the postural stability of concussed athletes would decrease following injury when administered independently, as previously reported (⁹).

Although dual-tasking methodologies ultimately seek to combine the physical and mental demands of athletics, our model is not currently advocated for concussion assessment. An assessment of a protocol evaluating the complete balance mechanism and comparing baseline measures to concussed individuals has not been performed. Until a thorough understanding of concussion is obtained, sports medicine clinicians are encouraged to collect baseline scores on athletes before their competitive season. Preseason baseline values will allow for individual comparison when making return to play decisions following injury. At the present time, a baseline evaluation should include a battery of tests that individually assess several aspects of cerebral functioning. At a minimum, examinations of self-reported symptomatology, neuropsychological functioning, and postural control are recommended.