Event-related potentials index cognitive style differences during a serial-order recall task

Abstract

Working memory and attentional inhibition processes (jointly symbolized here as WM/I) have been proposed to explain cognitive style differences in Field Dependence–Independence (FDI). FI relative to FD subjects have been found to use more effectively WM/I to operate on task-relevant information. The purpose of this study was to determine whether cognitive style differences are revealed as differences in ERP activity in a novel WM/I task. A serial-order recall task served to manipulate memory load by varying the amount and kind of information to be elaborated and retained in WM in order of temporal appearance (S1, S2); recall demand of the serial-order judgment (S3) was also concurrently varied. FI subjects engaged in deeper WM processing during the high memory load conditions relative to FD subjects; and this was measured as a higher amplitude slow negative wave (SNW), over the centro-parietal scalp extending to the frontal scalp, during the retention interval. In contrast, P300 amplitude was larger for FD subjects in the high memory load conditions following S1, which corresponded with a reduced amplitude SNW. We suggest that inhibitory processes indexed by P300, which FD subjects must mobilize to change their usually global-perceptual (i.e. shallow) attentional strategy for processing task information, may have resulted in less mental-attentional (WM/I) resources available to them during the task's retention phase (Rosen and Engle, 1997). Thus, ERP methods can be used to investigate differences in cognitive style.

Introduction

The ability to utilize working memory and attentional inhibition (WM/I) for storage and manipulation of task relevant information is needed for the emergence of high cognitive functions such as language, planning, and problem-solving (Pascual-Leone, 1969, Pascual-Leone, 1970, Baddeley, 1986, Shallice, 1988). Although Baddeley's model (Baddeley, 1995) remains descriptively important, our present work adopted a more classic conception of working memory (Pascual-Leone, 2000) which construes WM, in the manner of attention for William James or Hebb, as the currently hyperactivated (i.e. dominant) set of ‘software’ processes — knowledge units which we interpret as information-carrying schemas or schemes — in the subject's cortical repertoire of these units (Cowan, 1999, Engle et al., 1995, Pascual-Leone, 1970, Pascual-Leone, 1987, Pascual-Leone, in press, Rosen and Engle, 1997). We then posit a number of functional ‘hardware’ resources in the brain that can update (i.e. adapt to the situation) the functioning of the WM/I mechanism. The two most important resources for our purpose are a mental-attentional activation mechanism that we call mental capacity M, and an attentional inhibition (or I) mechanism that executive processes (schemes) can monitor and allocate to change functioning of WM/I (Cowan (in press), Engle et al., 1999, Pascual-Leone, 2000, Pascual-Leone, in press, Pascual-Leone and Baillargeon, 1994).

There is some agreement that these executive processes include: (a) focusing attention on relevant information and inhibition of irrelevant processes; (b) switching focused attention between tasks; (c) planning a sequence of subtasks to accomplish some goal; (d) updating and checking the contents of WM; and (e) coding representations in WM for time and place of appearance (Smith and Jonides, 1999). Working memory paradigms have been important in revealing individual differences, evidenced both in behavioral studies (Cantor and Engle, 1993, Daneman and Carpenter, 1980, Kyllonen and Christal, 1990) and event-related brain potential (ERP) studies (Nittono et al., 1999, Ruchkin et al., 1990a, Ruchkin et al., 1992).

Working memory and attentional inhibition processes have been proposed to be factors in cognitive style differences found in field dependence–independence (Case and Globerson, 1974, Cochran and Davis, 1987, Pascual-Leone, 1969, Pascual-Leone, 1989, Pascual-Leone and Goodman, 1979). The field-dependence–independence (FDI) cognitive style refers to the cognitive-style disposition, within misleading situations, to process information either in a global perceptual manner (by FD subjects who favor right-hemisphere processing — Pascual-Leone, 1989, Waber, 1989) vs. an actively analytical deeper manner (by FI subjects who favor left-hemisphere processing). We mean by misleading a situation that elicits from the subject habitual global-perceptual schemes that interfere with the task at hand. In contrast, facilitating situations are those where no misleading schemes are activated; in these other situations one may not find performance differences between FD and FI subjects (Pascual-Leone, 1989, Pennings, 1991, Witkin et al., 1954, Witkin and Goodenough, 1981). Analytical/deep processing entails, especially in misleading situations, active segmentation of information into relevant parts and interrelations (Robinson and Bennink, 1978). Global processing, in contrast, is holistic and more passive in character — passive in the sense that it accepts for processing the perceptually salient aspects of the situation, whether they are relevant or irrelevant (as often happens in misleading situations). These styles of processing information have been linked to functional hemispheric dominance (Pascual-Leone, 1989, Silverman, 1991, Waber, 1989) as well as to differences in the WM/I processes (Nahinsky et al., 1979, Pascual-Leone, 1969, Pascual-Leone, 1970, Pascual-Leone, 1989, Pascual-Leone et al., in press). Pascual-Leone and associates call M-space the ‘region’ of WM where actively analytical/deep processing takes place. If a large amount of the available M-space is allocated to cognitive operations bearing on the visual representation of the input, such processing of perceptual information may leave less space available for other jobs (e.g. elaboration) required by the task. The analytical strategy of FI subjects is more effective than that of FD subjects at optimizing the distribution and allocation of attentional resources; and so they experience less cognitive demands and less need to use attentional inhibition in misleading situations than FD subjects. Pascual-Leone, 1969, Pascual-Leone, 1989 reached the conclusion across a variety of tasks that executive schemes of FI subjects adequately control their WM/I resources; in contrast, FD subjects’ executive controls are deficient in this regard, although their style has advantages in other situations (Baillargeon et al., 1998, Johnson et al., 2000). Consistent with these views Davis (1991) found that FD and FI performance differences are more apparent under conditions of high information load.

This is demonstrated in the rod and frame task (RFT), a traditional measure of FDI. The RFT presents a misleading situation (Pascual-Leone, 1989, Pascual-Leone, 1995) in which a tilted adjustable rod must be moved to the true vertical position when its surrounding frame is tilted and other visual cues of verticality have been eliminated. Error scores and subject reports (Pascual-Leone, 1989, Witkin and Asch, 1948, Witkin et al., 1954) suggest that FD subjects adopt a global strategy, focused on processing only the salient perceptual features incorporated into a perceptual gestalt. Thus their attention is captured by the salient tilted frame and they feel compelled (a gestaltist effect) to align the rod with a side of this frame, instead of aligning it with the true (gravity's) vertical line as required. Notice that here the tilted frame is misleading because all subjects, FI and FD, have an overlearned habit of judging verticality visually, by aligning objects (e.g. rods) with ‘apparently-vertical’ lines often found in the immediate urban environment (Witkin and Goodenough, 1981, Messick, 1994, Pascual-Leone, 1989, Pascual-Leone, 1992). This automatized habit of visual evaluation of verticality becomes a potent misleading factor for FD, but not for FI subjects, in this task. FI subjects, unlike FD subjects, can effectively mobilize the WM/I mechanisms and switch mental attention to deeper (proprioceptive, labyrinthic feedback) processing. And so they construct an estimate of verticality that is unaffected by misleading salient cues of ‘verticality’ (e.g. sides of the tilted frame) offered by the visual display. FD subjects, however, fail to switch attention away from misleading cues, and do not engage WM/I processes to restructure their perception using deeper information. While traditional measures of FDI such as RFT can be seen superficially as perceptual tasks, which exhibit individual differences, they have been found to predict subjects’ performance in very diverse cognitive, problem solving and personality tasks (Goodenough, 1986, Messick, 1994, Pascual-Leone, 1989, Pascual-Leone and Goodman, 1979, Witkin and Goodenough, 1981); this suggests that FDI measures are indexing a cognitive style. See Messick (1994) and Johnson et al. (2000) for recent reviews.

Neuroimaging studies have employed a verbal 2-back task to study storage and executive processes of WM (Cohen et al., 1994, Konishi et al., 1999, Smith and Jonides, 1999). The task visually presents a series of letters, each letter followed by a delay interval. The subject responds indicating whether each letter is the same as the one that occurred two-back in the sequence. Processes engaged during the task include WM storage processes and executive processes for temporal tagging and updating the contents of WM (Braver et al., 1997, Konishi et al., 1999). These WM processes are associated with a phonological storage system in the parietal area, a frontal rehearsal system comprised of Broca's area, the supplemental and premotor areas, and prefrontal executive processes (Smith and Jonides, 1999). ERPs have also been employed to investigate WM processes in paradigms requiring the storage and manipulation of information over a delay period. Ruchkin (Ruchkin et al., 1990a, Ruchkin et al., 1992) manipulated the phonological load by varying the number of consonants in the visual display of a delayed match-to-sample paradigm. ERPs revealed a parietal positive slow wave and a left anterior negativity that increased in amplitude with memory load, as well as a centro-parietal negativity that did not vary with load. The positive slow wave, dominant over the parietal scalp, was associated with long duration encoding processes. The latter two negativities were present during the retention interval and were associated with a phonological rehearsal loop.

The primary purpose of the present study was to determine whether strategy/cognitive style group differences would be revealed as differences in ERP activity in a WM/I task. Functional differences in FD and FI subjects were studied using a delayed-response serial-order recall task that manipulated WM/I processes. The task is similar to the verbal 2-back WM task (Smith and Jonides, 1999) in that it presents multiple stimuli in which verbal information and temporal order must be retained in WM, to respond to a probe requiring a temporal judgment. The serial-order recall task differs from the 2-back WM task in placing a greater demand on coding information from the task stimuli, and in the requirement to remember temporal order stimuli. The serial-order recall task demanded the subject to encode features (single or multiple) from geometric shapes presented with the task stimuli (S1, S2). This information has to be elaborated by forming a memory association and converting the features into letter codes to be retained so as to maintain temporal order for making the serial-order recall response (single or multiple) that the response-cue stimulus (S3) requires. Elaboration instructions varying the memory load were given to the subject at the beginning of each block of trials, asking to memorize via letter codes, either one (low memory load condition) or three (high memory load condition) for each stimulus. This elaboration instruction contradicted, in the case of FD subjects, their habitual tendency to generate passive (salient-feature driven) perceptual processing. Thus, this habitual scheme of FD subjects for passive global processing should transform our serial-order recall task into a misleading situation for them, since the task explicitly required elaboration of letter codes to be memorized, and passive processing is not conducive to it.

The task varied demand on WM/I by manipulating not only the memory load (one vs. three features) to be letter-coded and remembered, but also the number of features to be recalled. Recall demand stipulated that either a single or multiple features be recalled for the serial-order response. It was hypothesized that FI, but not FD subjects, would be able to function equally well across treatment conditions. Furthermore, it was predicted that the global perceptual style of FD subjects would require them to interrupt (i.e. actively inhibit) their habitual scheme for passive and global visual representation of geometric shapes; in particular when multiple features had to be encoded and elaborated during the high memory load conditions. This effort might conflict with FD subjects’ WM/I allocation of attention during the task's retention step.