Assessment of working memory abilities using an event-related brain potential (ERP)-compatible digit span backward task

Abstract

Objective

This study investigated the effectiveness of an ERP-compatible Digit Span Backward (ERP-DB) task to determine working memory abilities in healthy participants.

Methods

Participants were administered both the standard digit span backward and ERP-DB tasks. The ERP-DB task was divided into two sections, consisting of 2, 4, 6 and 8 (Group 1) and 3, 5, and 7 (Group 2) set sizes. A set of digits was aurally presented, followed by a second set that either corresponded to the reverse order of the first set (correct condition) or had one digit in the sequence replaced by an incorrect digit (incorrect condition).

Results

Two posterior positive components were found to distinguish the two conditions; an earlier positive component (P200/P300) was elicited in the correct condition, whereas a comparatively robust and prolonged positive slow wave (PSW) was elicited in the incorrect condition. Furthermore, the PSW and the difference in PSW amplitude between incorrect and correct conditions (dPSW) dissipated as working memory load increased and were related to working memory capacity.

Conclusions

The PSW, dPSW and P200/P300 components were found to be associated with working memory abilities and may have the potential to act as neurophysiological markers for the assessment of working memory capacity.

Significance

This research lends support for the utility of the ERP-DB task as a means of assessing working memory abilities, which may have implications for testing patients with expressive communication impairments.

Introduction

After receiving a brain injury, an accurate and valid assessment of a patient's level of cognitive functioning is essential in order to formulate treatment and rehabilitation strategies (Lezak, 1995, Sohlberg and Mateer, 2001). However, most standard neuropsychological tests of cognitive functioning require verbal or behavioral responses from the patient. Unfortunately, following neurological trauma, patients frequently have speech and/or motor disabilities (Morse and Montgomery, 1992, Pedersen et al., 1995, Wade et al., 1986) and thus the ability to assess cognitive functioning with standard neuropsychological tests is difficult (or impossible) in patients with communication and/or behavioral impairments.

To bypass some of these assessment challenges, researchers advocated decades ago for computer-automated testing procedures to overcome the need for verbal responses (Miller, 1968, Thompson and Wilson, 1982). However, computerized paradigms of this type have generally required fine motor control (i.e. typing skill) and have had limited clinical applicability to date. Another, more recent approach to overcoming communication barriers is the use of a brain–computer interface (see Kübler et al., 2001 for review), where patients learn to emit specific electrophysiological responses that subsequently drive a communication device (e.g. a spelling device; Birbaumer et al., 1999). This method is still in its rudimentary stages, demands numerous training sessions and imposes a high cognitive load. Therefore, before implementing such a program, it would be sensible to demonstrate that the patient is capable of such a high level of cognitive functioning. Overall, there continues to be a necessity for a method for assessment of cognitive functions independent of verbal and behavioral responses (Connolly et al., 2000).

Event-related brain potentials (ERP) have the potential to significantly contribute to clinical neuropsychology by providing a neurophysiological index of patients' on-line cognitive functioning (Connolly and D'Arcy, 2000, Reinvang, 1999). Also, cognitive ERP paradigms have been modified or developed for application to neurotrauma populations (e.g. Allen et al., 1992, Ellwanger et al., 1996, Kotchoubey et al., 2001, Lang and Kotchoubey, 2002). One way to extend the clinical utility of cognitive ERP paradigms is to adapt standardized neuropsychological tests (see Connolly and D'Arcy, 2000, Connolly et al., 2000 for reviews). This allows one to (1) target the same cognitive processes assessed by the standard test, thereby reducing the inferences necessary to interpret the results; (2) have access to a large normative database for comparison purposes and; (3) assess various patients with the same test materials regardless of their communication impairments.

A research program initiated by Connolly and colleagues in the late 1990's has focused on adapting several standard neuropsychological tests for computer presentation and simultaneous ERP recordings in order to assess language functioning. These tasks include the Peabody Picture Vocabulary Test – Revised (PPVT; Dunn and Dunn, 1981), the Vocabulary and Similarities subtests of the Wechsler Intelligence Scale for Children—Third Edition (WISC-III; Wechsler, 1991), the Wechsler Adult Intelligence Scale—Revised as a Neuropsychological Instrument (WAIS-R-NI; Kaplan et al., 1991), the Token Test (Boller and Vignolo, 1966, De Renzi and Vignolo, 1964), and the Psycholinguistic Assessments of Language Processing in Aphasia Test (PALPA; Kay et al., 1992).

In all cases, the ERP-adapted tests were designed so that they did not require verbal or behavioral responses (although button-press responses were employed in some studies in order to ensure that performance on the adapted tests was comparable to the original versions). This was achieved by aurally or visually presenting either correct or incorrect answers to the test items. Results demonstrated that different ERP patterns were elicited by correct and incorrect answers when the questions were within participants' ability range but not when the task demands exceeded their capabilities (as determined by their performance on the traditionally administered tests). In addition, performance on the ERP-adapted tests correlated strongly with the standard versions. Irrespective of the type of test, these patterns of results were found in healthy adults (Connolly et al., 1995, Connolly et al., 1999a, D'Arcy and Connolly, 1999, D'Arcy et al., 2000), children (Byrne et al., 1995a, Byrne et al., 1999) and stroke patients (D'Arcy et al., 2003) as well as in case studies of communication-impaired patients with profound dyslexia, traumatic brain injury (TBI) (Connolly et al., 1999b, Connolly et al., 2000) and cerebral palsy (Byrne et al., 1995b). Moreover, using a newly developed statistical method, stroke patients' performance on the standard PPVT test was found to have a high correlation (r=0.86) with their ERP patterns (Marchand et al., 2002).

The next step in this research program was to develop a battery of ERP-adapted neuropsychological tests to assess other aspects of cognitive functioning beyond language abilities. In the same manner that a battery of standardized neuropsychological tests is used to highlight a patient's pattern of strengths and weaknesses to guide the development of an individualized rehabilitation program, it would be ideal to have a battery of ERP-adapted neuropsychological tests that are independent of language abilities for the same purpose.

Currently, there are a handful of ERP-adapted standardized tests that have been developed to assess other aspects of cognitive functioning. These include the adaptation of the Wisconsin Card Sorting Task (WCST), the Delayed Recall section of Verbal Paired Associates subtest from the Wechsler Memory Scale-Revised (WMS-R) and the Continuous Visual Memory Test, to assess, respectively, executive functioning (Barceló et al., 1997), recognition memory (Holamon et al., 1995) and figural memory (Retzlaff and Morris, 1996).

Working memory is another key aspect of cognitive functioning, which involves the temporary storage and effortful manipulation of information (Baddeley and Logie, 1999). Working memory is a core cognitive ability shown to be essential to and a strong predictor of learning, and intellectual and fluid reasoning abilities (de Jong and Das-Smaal, 1995, Fry and Hale, 2000, Kyllonen, 1987, Kyllonen and Christal, 1987, Kyllonen and Christal, 1990, Sternberg, 1980, Woltz, 1988). Because working memory is vital for learning, a necessity for everyday functioning, and deficits are very prominent (Morse and Montgomery, 1992) and disruptive to patients following brain injury (Schwartz et al., 2003), an accurate assessment is critically important for developing individualized rehabilitation and treatment programs (Gioia and Isquith, 2004). In addition, the development of a method to evaluate working memory independent of communication abilities would be of benefit not only to non-communicative patients but also to patients with slow or delayed expressive language abilities because standardized neuropsychological tests of working memory require immediate verbal responses. Lastly, the use of an ERP-adapted measure may reveal differences in neurophysiological processing that may be complementary to, or of benefit over and above the use of standardized tests even in communicative patients.

Numerous studies have investigated the neurophysiological and theoretical aspects of working memory processing using ERP. An early description of the functional significance of the P300 proposed that it was associated with the updating of working memory. Subsequent work supported this proposal by demonstrating a correlation between P300 amplitude and subsequent recall or recognition of items when rote memory strategies are utilized (Fabiani et al., 1985, Fabiani et al., 1986, Johnson and Donchin, 1985). Further studies have found an increase of P300 amplitude and latency (and/or subsequent posterior positive slow waves (PSW) within 300–1000 ms post-stimulus) as memory load increased (García-Larrea and Cézanne-Bert, 1998, Kusak et al., 2000, Nittono et al., 1999); which has been interpreted as a reflection of the additional processing related to the number of items to be retrieved and manipulated. P300 amplitude was also found to correlate with working memory abilities as assessed by standardized neuropsychology tests (Howard and Polich, 1985, Nittono et al., 1999, Polich et al., 1983) and to increase as task demands increased as long as performance remained high (Johnson and Donchin, 1985, Nittono et al., 1999).

Additional links between the P300 (and/or aspects of the PSW) and various stages of working memory processing have been investigated using an ERP version of the ‘Sternberg task’ (Sternberg, 1966). For this task, participants are presented with a set of digits/letters to memorize, then following a brief delay, a ‘probe’ digit is presented and participants indicate by button press whether it was part of the set. There is a sizeable literature demonstrating that a large, sustained, and parietally distributed positivity is elicited by the probe with a linear increase in latency as memory set size increases (e.g. Blumhardt, 1996, Pelosi et al., 1992, Pelosi et al., 1995, Pelosi et al., 1998, Starr and Barrett, 1987, Verleger, 1997). Moreover, a P300 has been found to be evoked during the ‘study phase’ or encoding phase to digits subsequently retrieved, suggesting that elicitation of a P300 during encoding can be used as a predictor of successful retrieval (Chao and Knight, 1996, Golob and Starr, 2004, Kotchoubey et al., 1996, Patterson et al., 1991).

Past ERP studies have been helpful in elucidating the neurophysiological and theoretical aspects of working memory processing and in establishing links between specific ERP components and working memory functions. However, this research does not offer the clinical assessment-related benefits associated with the use of stimuli from standardized neuropsychological tests. Therefore, the purpose of the present study was to investigate the effectiveness and potential clinical utility of an ERP-compatible version of a subsection of the Digit Span task from the Wechsler Adult Intelligent Scale—Third Edition (WAIS-III; Wechsler, 1997a) in order to determine working memory capabilities in healthy participants. The use of healthy participants is an essential step in determining the validity and reliability of this technique before it can be applied to a patient population. The standard Digit Span task involves the recall of a series of digits, either in the order presented (Digit Span Forward [DF]) or reverse order (Digit Span Backward [DB]), and has been shown to have high construct validity and reliability (Wechsler, 1997b).

Similar to the ERP-adapted neuropsychological paradigms by Connolly and colleagues, the major goal of the ERP adapted Digit Span task was to target the same cognitive processes involved in successful completion of the standard task without requiring verbal responses from participants. To accomplish this, the ‘recall’ segment of the test was presented aurally rather than requiring participants to respond verbally as is done in the traditional version of the tests. For this study, only the DB task was selected for ERP adaptation (ERP-DB task) since it demands more effort from working memory resources than the DF task (Gardner, 1981, Mishra et al., 1985). In addition, due to the nature of the adaptation, the DB task was selected instead of the DF task in order to avoid possible facilitation of working memory performance by the use of a simple auditory pattern-matching strategy from hearing the digits replayed back in the exact order. Successful completion of a DB trial is believed to involve cognitive manipulation (Sattler, 1992, Wechsler, 1997), therefore replaying the digits in the reverse order is not expected to lead to the use of simple strategies that may inflate working memory performance.

For the ERP-DB task digits are played back either in the exact reverse order (correct condition) or with one error (incorrect condition). This is methodologically important because we predict that successful performance will be reflected by the elicitation of different ERP patterns across conditions within participants' ability range; thereby providing a neurophysiological marker capable of assessing a patient's ability level on the task. This prediction is based on the results of the ERP-adapted standardized tests by Connolly and colleagues (discussed above). In further support of our prediction, the ERP-DB task can be viewed as a sequence learning task and previous ERP sequence learning studies have demonstrated the elicitation of a robust parietal positive component to a violation of the expected pattern when the sequence is explicitly known by participants (Lang and Kotchoubey, 2002, Polich, 1985, Schlaghecken et al., 2000, Squires et al., 1976). Specifically, we hypothesized that: (1) healthy subjects' performance on the ERP-DB task will be comparable to the standard test; (2) different ERP patterns will be elicited in correct and incorrect conditions; and, (3) ERP patterns will change as a function of working memory load and will be related to working memory capacity.